pubs.acs.

org/joc

An Oxidation Induced by Potassium Metal. Studies on the Anionic

Cyclodehydrogenation of 1,10 -Binaphthyl to Perylene
Michel Rickhaus,†,‡ Anthony P. Belanger,† Hermann A. Wegner,‡ and
Lawrence T. Scott*,†
†
Department of Chemistry, Merkert Chemistry Center, Boston College, Chestnut Hill,
Massachusetts 02467-3860, United States, and ‡Department of Organic Chemistry,
University of Basel, St. Johanns-Ring 19, 4056 Basel, Switzerland

lawrence.scott@bc.edu
Received August 18, 2010

Oxidative cyclization of 1,10 -binaphthyl (1) to perylene (2) can be achieved in essentially quantitative
yield by the action of three or more equivalents of potassium metal in hot tetrahydrofuran. An overall
reaction mechanism is proposed that accounts for all of the experimental observations reported by
previous investigators and those from the present studies. The trans-6a,6b-dihydroperylene dianion
(62-) is believed to be the pivotal intermediate from which H2 is lost. A radical chain reaction involving
free hydrogen atoms (H•) in the two-step propagation cycle is proposed to explain the formation of H2
from 62-. Anionic cyclodehydrogenations of this sort are complementary to those performed under
strongly acidic/oxidizing conditions, photochemically, or thermally (flash vacuum pyrolysis), and a
better understanding of how they occur, together with the optimized synthetic protocol reported here,
should encourage their wider use in organic synthesis.

Introduction homocoupling of unfunctionalized arenes, on the other

hand, can often be achieved by the oxidative dimerization
Methods for the selective formation of unsymmetrical
of electron-rich arenes, thanks to the pioneering work of
aryl-aryl bonds between reaction partners that require no
Scholl, Kovacic, and others,2,3 and intramolecular adapta-
functionality other than C-H bonds at the carbon atoms to
tions of these oxidative methods have been pushed to
be joined constitute some of the most exciting new advances
spectacular heights by M€ullen et al. for the synthesis of very
in the field of transition metal catalyzed cross-coupling reac-
large graphene substructures.4 Other noncatalytic, intramo-
tions.1 The formation of symmetrical aryl-aryl bonds by
lecular methods include the oxidative photocyclization of
stilbenes to phenanthrenes (the Mallory reaction)5 and, in
(1) (a) Stuart, D. R.; Fagnou, K. Science 2007, 316, 1172. (b) Campeau, special circumstances, thermal cyclodehydrogenations by
L.-C.; Stuart, D. R.; Fagnou, K. Aldrichimica Acta 2007, 40, 35. (c) McGlacken,
G. P.; Bateman, L. M. Chem. Soc. Rev. 2009, 38, 2447.
(2) (a) Musgrave, O. C. Chem. Rev. 1969, 69, 499. (b) Kovacic, P.; Jones, (4) (a) Watson, M. D.; Fechtenkoetter, A.; M€ ullen, K. Chem. Rev. 2001,
M. B. Chem. Rev. 1987, 87, 357. (c) McKillop, A.; Turrell, A. G.; Taylor, 101, 1267. (b) Wu, J.; M€ullen, K. In Carbon-Rich Compounds; Haley, M. M.,
E. C. J. Org. Chem. 1977, 42, 764. (d) McKillop, A.; Turrell, A. G.; Young, Tykwinski, R. R., Eds.; Wiley-VCH: Weinheim, 2006; p 90. (c) Mueller, S.;
D. W.; Taylor, E. C. J. Am. Chem. Soc. 1980, 102, 6504. M€ullen, K. Philos. Trans. R. Soc., A 2007, 365, 1453. (d) Wu, J.; Pisula, W.;
(3) See also: (a) Rempala, P.; Kroulik, J.; King, B. T. J. Am. Chem. Soc. M€ullen, K. Chem. Rev. 2007, 107, 718. (e) Zhi, L.; M€ullen, K. J. Mater. Chem.
2004, 126, 15002. (b) Rempala, P.; Kroulik, J.; King, B. T. J. Org. Chem. 2008, 18, 1472. (f) Feng, X.; Pisula, W.; M€ullen, K. Pure Appl. Chem. 2009,
2006, 71, 5067. (c) King, B. T.; Kroulik, J.; Robertson, C. R.; Rempala, P.; 81, 2203.
Hilton, C. L.; Korinek, J. D.; Gortari, L. M. J. Org. Chem. 2007, 72, 2279. (5) (a) Mallory, F. B.; Mallory, C. W. Org. React. 1984, 30, 1. (b) Mallory,
(d) Ormsby, J. L.; Black, T. D.; Hilton, C. L.; Bharat; King, B. T. Tetra- F. B.; Butler, K. E.; Evans, A. C.; Brondyke, E. J.; Mallory, C. W.; Yang, C.;
hedron 2008, 64, 11370. (e) Zhai, L.; Shukla, R.; Rathore, R. Org. Lett. 2009, Ellenstein, A. J. Am. Chem. Soc. 1997, 119, 2119. (c) Mallory, F. B.; Butler,
11, 3474. (f) Zhai, L.; Shukla, R.; Wadumethrige, S. H.; Rathore, R. J. Org. K. E.; Berube, A.; Luzik, E. D.; Mallory, C. W.; Brondyke, E. J.; Hiremath,
Chem. 2010, 75, 4748. R.; Ngo, P.; Carroll, P. J. Tetrahedron 2001, 57, 3715.

7358 J. Org. Chem. 2010, 75, 7358–7364 Published on Web 10/06/2010 DOI: 10.1021/jo101635z
r 2010 American Chemical Society
Rickhaus et al.
JOC Article
making it available for a broader scope of substrates and
facilitating its application. Herein, we describe studies on the
archetypical system in Figure 1 that have provided deeper
mechanistic insights and an optimized synthetic protocol.

Background
The cyclization of 1,10 -binaphthyl (1) to perylene (2) by
FIGURE 1. Anionic cyclodehydrogenation of 1,10 -binaphthyl (1)
to perylene (2) by alkali metals.
alkali metals was discovered accidentally in 1967 by Solo-
dovnikov et al. during their attempts to record the ESR
flash vacuum pyrolysis.6 Less well-known than all of these spectrum of the 1,10 -binaphthyl radical anion in solution.7,8
methods, however, is the anionic cyclodehydrogenation of Because the work was published originally in Russian
aromatic hydrocarbons, for which the cyclization of 1,10 - and the later English translation is not cited by Chemical
binaphthyl (1) to perylene (2) by alkali metals serves as the Abstracts, the experimental data described therein, unfortu-
classic example (Figure 1).7,8 nately, have been mostly overlooked for the last 40 years; the
None of the other methods mentioned above convert 1,10 - paper has been cited only three times. To rectify this situa-
binaphthyl (1) to perylene (2) very efficiently, if at all, and this tion, we summarize here some of the key findings that we
makes the anionic cyclodehydrogenation method unique, have gleaned from our own translation of the early Russian
complementary, and worthy of attention.9 Most intriguing is paper; the summary in Chemical Abstracts is not entirely
the seemingly incongruous fact that the starting material (1 = accurate.
C20H14) is being oxidized (2 = C20H12), whereas the reagents Solodovnikov et al. reduced 1,10 -binaphthyl (1) in 1,2-
used to induce this oxidation are alkali metals, which arguably dimethoxyethane (0.1 M) with an excess of potassium metal
qualify as some of the strongest reducing agents known. The under vacuum at room temperature. The ESR signal for the
alkali metal is also being oxidized during the course of this 1,10 -binaphthyl radical anion (1•-) grew in rapidly during the
reaction, so what is being reduced? The inescapable conclusion first hour but then began to diminish slowly, disappearing
is that the two lost hydrogen atoms must end up reduced to a essentially completely over a period of 48 h. After 72 h, the
lower oxidation state, either H2 or hydride. contents of the ampule were exposed to oxygen, and perylene
Almost nothing is known about the mechanistic details of (2) was isolated in 39% yield. Some other hydrocarbon
this reaction.10,11 Even the stoichiometry is uncertain, so the products, detected by TLC and tentatively identified as
metal is generally just used in large excess. Because this partially hydrogenated 1,10 -binaphthyl, were also obtained,
reaction is so poorly understood and the yields rarely exceed but the amount of recovered 1,10 -binaphthyl, if any, was not
50%,11 the method remains underutilized in synthesis. The specified. When the experiment was repeated and the reac-
most prominent application is probably the syntheses of tion mixture was filtered to remove the potassium metal after
rylenes by M€ ullen et al.,12 in which potassium metal was 1 h, the point at which the ESR signal had reached its maxi-
used to “zip up” specifically designed oligomers of 1,4-linked mum intensity, immediate exposure of the filtrate to oxygen
naphthalenes. A knowledge of the mechanism operating, the produced only a 0.3% yield of perylene (2). On the other
intermediates involved, and the optimum conditions to use hand, allowing the filtrate to stand at room temperature for
would allow chemists to tame this rather obstreperous reaction, 72 h, after removal of the potassium at the 1 h point, pro-
duced a 10% yield of perylene (2). These results indicate that
(6) (a) Clar, E. Polycyclic Hydrocarbons; Academic Press: New York, the cyclization and rearomatization process is slow at room
1964. (b) Scott, L. T.; Bratcher, M. S.; Hagen, S. J. Am. Chem. Soc. 1996, 118, temperature and that it does not require the continued pres-
8743. (c) Hagen, S.; Bratcher, M. S.; Erickson, M. S.; Zimmermann, G.;
Scott, L. T. Angew. Chem., Int. Ed. 1997, 36, 406. (d) Scott, L. T.; Bronstein,
ence of potassium metal, once the solution contains a
H. E.; Preda, D. V.; Ansems, R. B. M.; Bratcher, M. S.; Hagen, S. Pure Appl. significant concentration of organic radical anions. The yield
Chem. 1999, 71, 209. (e) Boorum, M. M.; Vasil’ev, Y. V.; Drewello, T.; Scott, of perylene after 72 h was lower, however, if the solution was
L. T. Science 2001, 294, 828. (f) Scott, L. T.; Boorum, M. M.; McMahon,
B. J.; Hagen, S.; Mack, J.; Blank, J.; Wegner, H.; de Meijere, A. Science 2002, not left in contact with excess potassium after the first hour
295, 1500. (g) Scott, L. T. Angew. Chem., Int. Ed. 2004, 43, 4994. (10% vs 39%).
(h) Tsefrikas, V. M.; Scott, L. T. Chem. Rev. 2006, 106, 4868. (i) Xue, X.; Solodovnikov et al. also report the formation of hydrogen
Scott, L. T. Org. Lett. 2007, 9, 3937.
(7) (a) Solodovnikov, S. P.; Zaks, Y. B.; Ioffe, S. T.; Kabachnik, M. I. gas in this reaction. They monitored the production of H2 by
Radiospektrosk. Kvantovokhim. Metody Strukt. Issled. 1967, 106; Chem. GC analysis and the formation of perylene by UV-vis
Abstr. accession no. 1969:28235; CAN 70:28235. (b) Solodovnikov, S. P.;
Ioffe, S. T.; Zaks, Y. B.; Kabachnik, M. I. Izv. Akad. Nauk SSSR, Ser. Khim.
spectroscopy. Figure 2 reproduces the graph they published
1968, 442; Chem. Abstr. accession no. 1968:476168; CAN 69:76168; English in 1968 that shows the percent yields of H2 gas and perylene
translation Bull. Acad. Sci. USSR, Div. Chem. Sci. 1968, 442. (2), as well as the growth and decline of the ESR signal, as a
(8) See also: Gilman, H.; Brannen, C. G. J. Am. Chem. Soc. 1949, 71, 657.
(9) For related anionic cyclodehydrogenations, see: (a) Tamarkin, D.; function of time. It is noteworthy that (i) the amount of H2
Benny, D.; Rabinovitz, M. Angew. Chem. 1984, 96, 594. (b) Tamarkin, D.; detected was always significantly lower than the amount of
Cohen, Y.; Rabinovitz, M. Synthesis 1987, 196. (c) Rabinovitz, M.; Tamarkin, perylene and (ii) the formation of both products leveled off
D. Synth. Met. 1988, 23, 487. (d) Deselets, D.; Kazmaier, P. M.; Burt, R. A.;
Hamer, G. K. Can. J. Chem. 1995, 73, 325. (e) Yao, J. H.; Chi, C.; Wu, J.; Loh, with time; after 72 h, the yield of H2 reached 24.6%, whereas
K.-P. Chem.-Eur. J. 2009, 15, 9299. the yield of perylene (2) reached 39%.
(10) (a) Hnoosh, M. H.; Zingaro, R. A. J. Am. Chem. Soc. 1970, 92, 4388.
(b) Eisenstein, O.; Mazaleyrat, J. P.; Tordeux, M.; Welvart, Z. J. Am. Chem.
As the reaction proceeded, color changes were observed
Soc. 1977, 99, 2230. that gave a crude indication of the dominant species present.
(11) (a) Michel, P.; Moradpour, A. Synthesis 1988, 894. (b) Benshafrut, In the beginning, a green color developed that corresponded
R.; Hoffman, R. E.; Rabinovitz, M.; M€ ullen, K. J. Org. Chem. 1999, 64, 644.
(12) (a) Bohnen, A.; Koch, K. H.; L€ uttke, W.; M€ ullen, K. Angew. Chem., to the radical anion of 1,10 -binaphthyl (1•-); the radical
Int. Ed. 1990, 29, 525. (b) Scherf, U.; Muellen, K. Synthesis 1992, 23. anion of naphthalene itself is also green. The solution then
J. Org. Chem. Vol. 75, No. 21, 2010 7359
JOC Article Rickhaus et al.

FIGURE 3. Anionic cyclodehydrogenation of [5]helicene (3) by

alkali metals.14

FIGURE 2. Change in ESR signal intensity (relative units) and step, in which two hydrogen atoms are lost, could be acceler-
yields of perylene (2) and H2 over time in the cyclodehydrogenation
ated by warming the reaction mixture to 25-30 °C.
of 1,10 -binaphthyl (1) by potassium metal at room temperature.7b
Two important additional conclusions can be drawn from
turned a wine-red color, which Solodovnikov et al. attributed this report: (i) the two C-H bonds remain intact until after
to the dianion of 1,10 -binaphthyl (12-); however, the ESR the C-C bond has been formed and two electrons have been
spectrum of the 1,10 -binaphthyl radical anion (1•-) could also added, and (ii) rearomatization by rupture of the two C-H
still be recorded at this stage. After 1 d, the solution became bonds is slower than the initial cyclization (at least in this
purple, which the authors attributed to the dianion of perylene loaded case) but occurs under an inert atmosphere, before
(22-); by this time, the ESR signal had dropped to 10% of its the solution is exposed to oxygen. Against this background,
maximum amplitude. A UV-vis absorption spectrum of the we now add our observations.
purple solution matched that reported in the literature for the
dianion of perylene (22-).13 Heating the wine-red solution Results and Discussion
from 20 to 100 °C caused it to turn purple and caused the ESR
From the outset, we reasoned that this reaction should be
spectrum to change from that of the 1,10 -binaphthyl radical
conducted at elevated temperatures so that it could be pushed
anion (1•-) to that of the perylene radical anion (2•-). Expos-
to completion in less than a day. We began by screening a
ing purple solutions of the perylene dianion (22-) to oxygen
few solvents, using both sodium metal and potassium metal.
caused the color to change first to blue (perylene radical anion,
Owing to the high reactivity of alkali metals, only ethers,
2•-) and then to yellow (neutral perylene, 2). It seems safe to
hydrocarbons, and tertiary amine solvents were examined.
conclude that electron transfer reactions among the species
Thus, we heated 1,10 -binaphthyl (1) overnight in various
present in the solution at various stages allow interconversions
solvents with excess metal and compared the yields of perylene
between the radical anion and dianion of 1,10 -binaphthyl
(2) formed under the different conditions. Potassium metal in
(1•-þ e- / 12-) and likewise between the radical anion and
THF at 80 °C in a pressure vessel proved far superior to all of
dianion of perylene (2•- þ e- / 22-).
the other combinations examined (73% yield),15 so we settled
In 1992, a careful NMR study of the closely related
on potassium and THF as the metal and solvent for further
cyclization shown in Figure 3 was reported by Ayalon and
studies. An operating temperature above the melting point of
Rabinovitz.14 In this case, [5]helicene (3) was reduced in-
potassium metal (63 °C) in a pressure vessel with constant
dependently with sodium metal and with lithium metal in
stirring ensures a large area of clean metal surface throughout
THF-d8 at -33 °C over periods of several days to give the
the course of the reaction, and this undoubtedly contributes to
dianion of benzo[ghi]perylene (52-). Exposure of the final
the success of these conditions. On a small scale, the anionic
reaction mixture to oxygen gave the neutral hydrocarbon,
products were oxidized to the neutral hydrocarbons by ex-
benzo[ghi]perylene (5). By continuous NMR monitoring, the
posure to oxygen in the workup, but elemental iodine was
growth and disappearance of even-electron species could be
used as the oxidizing agent for larger scale reactions.
observed; no ESR or UV-vis absorption spectra were re-
With this basic protocol in hand, the optimization was
ported. As the first paramagnetic species formed (3•-), the
then turned toward determining the minimum number of
NMR signals for 3 disappeared. Then, over time, the NMR
equivalents of potassium metal required to effect complete
signals for 42- grew in; no signals were ever seen for the dianion
conversion. An excess of potassium metal (5 equiv)16 was
of the open [5]helicene (32-). After prolonged exposure of the
solution to the alkali metal, the NMR spectrum of the dihydro-
benzo[ghi]perylene dianion (42-) was slowly replaced by that (15) Other conditions examined (metal, solvent, temperature) all gave
yields of 10% or less: (a) potassium, 1,2-dimethoxyethane, 80 °C;
of the benzo[ghi]perylene dianion (52-). This rearomatization (b) potassium, diglyme, 80 °C; (c) potassium, toluene, 80 °C; (d) sodium,
THF, 66 °C; (e) sodium, diglyme, 162 °C; (f) sodium, tetramethylethylene-
(13) Balk, P.; Hoijtink, G. J.; Schreurs, J. W. H. Recl. Trav. Chim. Pays- diamine (TMEDA), 120 °C; (g) sodium, toluene, 110 °C.
Bas Belg. 1957, 76, 813. (16) The maximum amount of potassium that could be consumed in this
(14) Ayalon, A.; Rabinovitz, M. Tetrahedron Lett. 1992, 33, 2395. reaction should be 4 equiv, even if all of the H2 were reduced to KH.17

7360 J. Org. Chem. Vol. 75, No. 21, 2010

Rickhaus et al.
JOC Article
TABLE 1. Screening of Different Equivalents of Potassium Used TABLE 2. Screening of Different Temperatures for Cyclization
to Cyclize 1,10 -Binaphthyl (1) to Perylene (2) at 85 °C in Tetrahydrofuran of 1,10 -Binaphthyl (1) Using 3 equiv of Potassium in Tetrahydrofuran
in a Pressure Vessel time (d) temp (°C) conversion (%)a
equiv of potassium time (d) conversion (%)a
1 95b quant
5.0 2 quant 1 85b 90-quant
3.0 2 quant 1 70b 45
3.0 1 90-quantb,d 1 55b 34
2.5 7 quantb 3 25 15
2.5 1 0-78d 1 25 8
2.2 7 quant a
Conversion based on NMR integration. bReactions run in a pressure
2.2 4 64-70c,d vessel.
2.0 3 58
1.8 5 58
1.8 4 48
1.0 6 28-40d
0.5 12 5
0.5 7 0
a
Conversion based on NMR integration. bQuantitative yield isolated.
c
70% yield isolated. dRanges resulting from multiple runs.

chosen as the starting point, and the stoichiometry was

progressively adjusted. With 5 equiv of potassium metal in a
pressure vessel at 85 °C in THF for 2 d, the cyclodehydro-
genation of 1,10 -binaphthyl (1) to perylene (2) is remarkably
clean and essentially quantitative; no starting material re-
mains, and no other significant products are detected by
NMR analysis of the crude material obtained after work-
up. Table 1 summarizes the results from lower amounts of
potassium at various reaction times, all in THF at 85 °C in
a pressure vessel.
We find that the conversion of 1 to 2 is still quantitative
with 3 equiv of potassium after 2 d in THF at 85 °C.
Shortening the reaction time to 1 d with 3 equiv of potassium
still gives quantitative conversion most of the time, but small
amounts of unconverted starting material were detected
in some experiments run under these conditions. With lesser
amounts of potassium (2.5 and 2.2 equiv), the conversion
of 1 to 2 can still be pushed to completion after 7 d in THF
at 85 °C in a pressure vessel, but the reaction is definitely
slower than with 3 equiv of potassium under the same
conditions.
Although fewer than 3 equiv of potassium may be im-
practical for this reaction, it is noteworthy that even 0.5 equiv
of potassium metal is enough to produce small amounts
(5%) of 2 from 1 at 85 °C in THF over a period of 12 d.
Finally, the temperature dependence of the reaction was
investigated, all with 3 equiv of potassium metal in THF in a
pressure vessel (Table 2). A temperature of 95 °C was found
to be the most reliable for ensuring 100% conversion of 1 to 2
in 1 d, but the NMR spectrum of the crude product was not
FIGURE 4. First steps in the mechanism of the anionic cyclodehy-
as clean as that from the 85 °C reactions. Like Solodovnikov drogenation of 1,10 -binaphthyl (1) to perylene (2) by alkali metals.
et al., we see the cyclodehydrogenation occurring only rela-
tively slowly at room temperature. The conditions we re- One can safely assume that the first step is a single-electron
commend as optimum for converting 1 to 2 are 3.0-3.5 equiv reduction of 1 to produce the radical anion 1•-. This odd-
of potassium metal in THF in a pressure vessel at 85 °C for electron intermediate might then close spontaneously to the
12 h (see Experimental Section). dihydroperylene anion radical (6•-), or it might live long
The mechanism by which 1,10 -binaphthyl (1) is converted enough to accept a second electron and get reduced to the
into the dianion of perylene (22-) by potassium metal dianion of 1,10 -binaphthyl (12-). Cyclization of 12- would
requires two single-electron reductions, one C-C bond- give the dianion of dihydroperylene (62-), whereas the same
forming step, two C-H bond-breaking steps, and a H-H intermediate could be formed alternatively by single-electron
bond-forming step. In principle, the two C-H bond-break- reduction of 6•- (Figure 4).
ing steps and the H-H bond-forming step could all be Whether the C-C bond forming step occurs at the anion
concerted, but they may not be; the order of the steps has radical stage (path A) or at the dianion stage (path B) is
never been determined. uncertain. Attempts to answer this question by molecular
J. Org. Chem. Vol. 75, No. 21, 2010 7361
JOC Article Rickhaus et al.

FIGURE 5. A plausible mechanism for the trans-to-cis isomeriza-

tion of 62- by two symmetry-allowed suprafacial [1,4]-shifts of
hydrogen.

orbital calculations would likely be futile, because the partially

solvated counterions in such species are bound to play a signi-
ficant role that is difficult to model. Fortunately, from a broad
perspective, the order of the steps for the second electron addi-
tion and the cyclization is relatively inconsequential. The fact
that this reaction can be driven to 100% conversion does indi-
cate, however, that the dianion of 1,10 -binaphthyl (12-, pre- FIGURE 6. Highest occupied molecular orbital (HOMO, B3LYP/
sumed by Solodovnikov to be the red-wine colored species) 6-31G*) of perylene dianion (22-).
cannot be a mechanistic dead end; if 12- is unable to cyclize by
path B, then it must be able to return to the radical anion 1•-
by donating one of its electrons to another species in solution
and cyclize by path A. All of the electron transfer steps and the
cyclization steps in Figure 4 should be reversible, and the
intermediates in this manifold are removed irreversibly only
after C-H bonds begin to be broken as 62- goes to 22-.
The mechanism by which the cyclized dianion 62- loses the
two hydrogen atoms is more obscure. How and when is the
new H-H bond formed? The direct observation of H2 by
Solodovnikov et al. proves that the two hydrogen atoms are
lost as H2 and not as 2 KH, but there is no evidence that the
two hydrogen atoms in the observed H2 both originated from
the same molecule. Ayalon and Rabinovitz established that
the hydrogen atoms on the newly joined carbon atoms are
oriented trans with respect to one another;14 thus, a concerted
loss of H2 from dianion 62- appears geometrically impossible.
Figure 5 outlines a plausible pathway for the isomerization
of trans-62- to cis-62- by two symmetry-allowed suprafacial
[1,4]-shifts of hydrogen over the allylic anion moiety. A σ2s þ
2-
σ2s pericyclic loss of H2 from cis-6 is symmetry-forbidden
in the ground state, however, because the HOMO of the
perylene dianion is symmetric with respect to the plane FIGURE 7. Two-step propagation cycle in the proposed radical
between the two carbon atoms from which the hydrogens chain mechanism for conversion of the dihydroperylene dianion
are lost (Figure 6). (62-) to the perylene dianion (22-) and H2.
We propose a radical chain mechanism involving free hydro-
gen atoms (H•) to account for conversion of the cyclized dia- THF. On further consideration, however, we realized that
nion 62- to H2 and the perylene dianion (22-). Figure 7 depicts even the great abundance of THF is probably not enough
the two steps of our proposed propagation cycle. Thus, ab- to overcome the far greater hydrogen donor ability expected
straction of a hydrogen atom from 62- by a free hydrogen atom for 62-. Furthermore, the THF R-radical, if formed, could
(H•) would result in the formation of a very strong H-H bond likewise abstract a hydrogen atom from 62- and thereby carry
(104 kcal/mol) and the rearomatization of one of the benzene on the chain reaction. As with all radical chain reactions,
rings at the expense of a very weak C-H bond (tertiary and chain termination would result only from infrequent radical-
conjugated with a very electron-rich π-system). This step should radical encounters.
be strongly exothermic. Subsequent rearomatization of the Initiation of this chain reaction requires a source of either
other benzene ring by homolysis of the second (now even free hydrogen atoms (H•) or the other chain-carrying inter-
weaker) C-H bond would deliver the final product (22-) and mediate (7•2-). One source of these initiators could be the
release a new hydrogen atom (H•) to carry on the next cycle. thermal homolysis of one of the very weak C-H bonds in
At first, we did not entertain any mechanisms involving 62-, a process that would actually give both 7•2- and a free
free hydrogen atoms (H•), because we assumed that such a hydrogen atom (H•), thereby initiating two independent
reactive species would simply abstract an R-hydrogen from chain reactions simultaneously.
7362 J. Org. Chem. Vol. 75, No. 21, 2010
Rickhaus et al.
JOC Article
after 3 d at room temperature drops from 39% to only 10% if
the potassium is removed from the reaction mixture by
filtration after the first hour? According to Solodovnikov’s
ESR spectrum, the concentration of the initial radical anion
(1•-, although possibly already closed, 6•-) had reached its
maximum when the potassium was removed. Before any
C-H bonds could be broken, however, the radical anion
would need to have been converted to 62- by acquiring a
second electron. In the absence of potassium metal, genera-
tion of dianions 12- and 62- could be accomplished by dis-
proportionation of either 1•- or 6•-, and the loss of H2 would
then produce perylene dianion (22-). Conversely, if the
solution were still in contact with potassium metal, all of
the starting material would automatically be reduced to the
closed dianion (62-) by the potassium, thereby accelerating
perylene production, which would then not be dependent on
disproportionation reactions of radical anions to generate
the pivotal intermediate 62-.
Similar arguments can account for our observation that
1,10 -binaphthyl (1) can be cyclized to perylene (2) even with as
little as 0.5 equiv of potassium metal, albeit only in low
FIGURE 8. Published mechanism for the reduction of H2 to NaH conversion at elevated temperatures over many days (Table 1).
by sodium naphthalenide.17 One final comment is in order concerning the hydrogen-
abstracting ability of the anionic species involved in this
Once the first H2 is formed, more free hydrogen atoms (H•) reaction. When the reactions are run in dry THF under a
will be generated in this reaction mixture from a background nitrogen atmosphere and are then quenched with either dry
reaction that must also be mentioned. For many decades, it oxygen or iodine, we obtain essentially no Birch reduction
has been known that the radical anions of aromatic hydro- products. On the other hand, in reaction mixtures that come
carbons can slowly reduce H2 (e.g., sodium naphthaleninde into contact with an adventitious proton source prior to
þ 1/2H2 f naphthalene þ NaH).17 The mechanism for this completion of the oxidation, we see varying amounts of 3,4-
reaction has been studied, and the intermediacy of anion 9- dihydro-1,10 -binaphthyl (11),18 unless the starting material
has been verified experimentally (Figure 8).17 Presumably, has been converted entirely to perylene dianion (22-). Mass
many of the anionic species present in the cyclization of 1 to spectral analysis of the 11 obtained from a reduction of 1,10 -
22- are also capable of reducing H2, and that background binaphthyl (1) run in THF-d8 showed no evidence for any
reaction could provide another source of free hydrogen deuterium in the 11, so none of the organic intermediates
atoms (H•) to initiate the radical chain reaction we have involved are capable of abstracting hydrogens, either as
proposed. protons or as hydrogen atoms, from THF.
The operation of this background reaction can also explain
why the yield of H2 reported by Solodovnikov et al. never
equaled the yield of perylene; as perylene and H2 were being
formed in their experiments, the amount of H2 was concur-
rently being depleted by reduction to KH throughout the
course of the reaction. The fact that the H2 level rises
continuously during this reaction (Figure 2), however, indi-
cates that the reduction of H2 is slow compared to the rate of
cyclization and rearomatization. It is intriguing to note that
H2 is serving as an oxidizing agent in this transformation.
Conclusions
If the reduction of H2 were fast compared to the rate of
cyclization and rearomatization, then H2 would not accu- Oxidative cyclization of 1,10 -binaphthyl (1) to perylene (2)
mulate, and the overall stoichiometry for the cyclodehydro- can be achieved in essentially quantitative yield by heating 1
genation would be with three or more equivalents of potassium metal in THF
4K þ 1 f 22 - =2Kþ þ 2KH at 85 °C in a pressure vessel overnight. The reaction can be
driven to completion at the same temperature with as little as
We have found that 2.2 equiv of potassium metal is 2.2 equiv of potassium, but it then requires much longer
sufficient to drive the formation of perylene to completion times (4-7 days). The use of sodium or potassium in various
(Table 1), and this would not be possible if a 4:1 stoichiom- other solvents proved inferior. An overall reaction mechan-
etry were required. ism is proposed that accounts for all of the experimental
How does our mechanistic proposal account for the ob- observations reported by Solodovnikov et al., by Ayalon and
servation by Solodovnikov et al. that the yield of perylene Rabinovitz, and from this study. The trans-dihydroperylene

(17) Bank, S.; Lois, T. A.; Prislopski, M. C. J. Am. Chem. Soc. 1969, 91, (18) Dal Zotto, C.; Wehbe, J.; Virieux, D.; Campagne, J.-M. Synlett 2008,
5407 and references therein. 2033.

J. Org. Chem. Vol. 75, No. 21, 2010 7363

JOC Article Rickhaus et al.

dianion (62-) is believed to be the pivotal intermediate from to 85 °C; the mixture turned dark purple within 2 h. After 12 h at
which H2 is lost. A radical chain reaction involving free 85 °C, the reaction mixture was allowed to cool to room
hydrogen atoms (H•) in the two-step propagation cycle is temperature, and the solution was removed from the remaining
proposed to account for the formation of H2 from 62-. potassium by vacuum cannulation into a flame-dried flask
Anionic cyclodehydrogenations of the sort investigated here under a nitrogen atmosphere. The pressure vessel was washed
with some dry THF, and the solution again was removed by
are complementary to those performed under strongly acidic/
vacuum cannulation. The dark purple to blue solution was
oxidizing conditions, photochemically, or thermally (flash quenched by adding a solution of iodine (560 mg, 4.41 mmol)
vacuum pyrolysis). It is our hope that a better understanding in ∼6 mL of dry THF over a period of 15 min, which caused the
of anionic cyclodehydrogenations such as this one will solution to turn yellow and then dark red. To the solution was
encourage their wider use in organic synthesis. then added slowly ethanol (5 mL) followed by sodium thiosul-
fate (10% in water, ∼50 mL) until the color no longer changed
Experimental Section (typically a yellow solution). The reaction was then exposed to
air, and the organic solvents were removed on a rotary eva-
General. The tetrahydrofuran (THF) used in this reaction was porator. To the remaining aqueous solution containing preci-
dispensed from a Glass Contour solvent purification system. 1H pitates was added dichloromethane (DCM, 100 mL), and the
NMR spectra were recorded in CDCl3 on a 500 MHz instru- aqueous layer was extracted 3 times. The combined organic
ment. CAUTION: Glass encased magnetic stirbars should be layers were washed with brine and dried over MgSO4, and the
used for all reactions involving potassium metal; potassium solvent was removed on a rotary evaporator. The residue was
reacts with standard Teflon-coated stirbars to produce KF adsorbed onto silica, and the mixture was subjected to flash
and highly reactive carbon as a finely divided black powder column chromatography (SiO2, 1:15 DCM/hexanes) yielding
that, in one instance, upon exposure to oxygen, resulted in a 219 mg (quantitative) of 2 as a yellow solid. The 1H NMR
vigorous explosion. spectrum matched that obtained from authentic perylene, with
Perylene (2) from Binaphthyl (1) by the Action of Potassium no more than traces of impurities (<5%). To the residues in the
Metal. To a flame-dried and nitrogen-flushed 350-mL pressure pressure vessel was added THF (50 mL), and any remaining
vessel equipped with a glass-encased magnetic stirbar19 was potassium was slowly quenched by the slow addition of EtOH
added dry tetrahydrofuran (THF, 250 mL) and 1,10 -binaphthyl and then water.
(1) (221 mg, 0.87 mmol). The solvent was degassed for 15 min
while a freshly cut piece of potassium metal20 (108 mg, 2.77 Acknowledgment. We thank Maria Eliseeva for her trans-
mmol) was rinsed with hexanes and THF and then added to the lation of the Russian papers in ref 7. This work was sup-
reaction mixture. The vessel was sealed and heated in an oil bath ported by the University of Basel, the Werenfels Fond der
Freiwilligen Akademischen Gesellschaft Basel, the U.S.
(19) Note the CAUTION in the General Section.
(20) Not cleanly cutting all sides of the potassium piece generally leads to National Science Foundation, and the U.S. Department of
a significant drop in yield. Energy.

7364 J. Org. Chem. Vol. 75, No. 21, 2010