Kilogram-scale production of corannulene

Article abstract of DOI:10.1021/op200387s

An efficient entry process for the synthesis of corannulene has been demonstrated on kilogram scale. Compared to the discovery and gram-scale syntheses, the amounts of solvents and reagents per gram of product were greatly reduced. Priority was given to implement the least toxic agents possible. Improvements in the purification of products obviated the need for column chromatography, alleviating four chromatographic operations. A new reduction method for the final step of the synthesis decreased reaction time from 6 to 0.5 days, and avoided the use of 100 equiv of zinc metal. The process now comprises nine steps, each of which runs smoothly at 100-L scale with a charging of 3-12 kg of educt. A total of 1.3 kg corannulene was isolated. This kilogram-scale process reduces material costs by over 2 orders of magnitude compared to that for the published gram-scale syntheses. Key opportunities in the process are identified for further improvements that should make synthesis on 100-kg scale feasible with a target price for 1 that is suitable for commercial production and engineering application.

Full text of DOI:10.1021/op200387s

Article
pubs.acs.org/OPRD
Kilogram-Scale Production of Corannulene
Anna M. Butterfield, Bruno Gilomen, and Jay S. Siegel*
Organic Chemistry Institute and Labor fur Prozessforschung, University of Zurich, Winterthurerstrasse 190, 8057 Zurich, Switzerland
̈
̈
S
* Supporting Information
ABSTRACT: An efficient entry process for the synthesis of corannulene has been demonstrated on kilogram scale. Compared
to the discovery and gram-scale syntheses, the amounts of solvents and reagents per gram of product were greatly reduced.
Priority was given to implement the least toxic agents possible. Improvements in the purification of products obviated the need
for column chromatography, alleviating four chromatographic operations. A new reduction method for the final step of the
synthesis decreased reaction time from 6 to 0.5 days, and avoided the use of 100 equiv of zinc metal. The process now comprises
nine steps, each of which runs smoothly at 100-L scale with a charging of 3−12 kg of educt. A total of 1.3 kg corannulene was
isolated. This kilogram-scale process reduces material costs by over 2 orders of magnitude compared to that for the published
gram-scale syntheses. Key opportunities in the process are identified for further improvements that should make synthesis on
100-kg scale feasible with a target price for 1 that is suitable for commercial production and engineering application.
reach the transition state en route to the final closure, and
overestimated the aromatic stabilization benefit one might obtain
from creating the corannulene unit. Indeed, higher-energy reaction
conditions or higher-energy synthetic precursors would be needed
to bring a new synthesis to fruition.
INTRODUCTION
■
Comprising 20 carbon atoms, corannulene (1) is one-third the
size of C60 and the smallest subunit of the buckyball motif that
still maintains a curved surface.1 The curvature of 1 and C60
give both compounds unique electronic properties that are not
observed in planar polyaromatic hydrocarbon (PAH) analogues
such as pyrene or naphthalene.2 The literature around corannulene
abounds with a wide diversity of mono- and multifunctionalized
derivatives.3 These derivatives serve as precursors for numerous
classes of materials, such as graphitic tubes/caps,4 liquid crystals,5
dendrimers,6 polymers,7,8 cruciforms,9 cyclophanes,10 and molec-
ular clefts.11 Despite the ever-growing academic interest in 1, the
need for a substantial synthetic commitment to prepare 1 and
derivatives creates a barrier to broad materials and engineering
applications. Thus, a “sustainable” synthesis on scale would go a
long way toward allowing corannulene to grow beyond being an
esoteric molecule of academic interest to an industrially interesting
prospect, with direct application in materials chemistry. Herein,
the optimization and kilogram production of corannulene intends
to provoke a commercially viable production.
In the early 1990s two new synthetic strategies to 1 appeared.
Both chose a bilaterally symmetric retrosynthetic strategy over
the ring-by-ring methods; however, these retrosynthetic
approaches differed in their choice of which bond type of 1
to disconnect: (a) flanking or (b) rim (Figure 1). The former
leads to a relatively low energy and unstrained 7,10-disub-
stituted fluoranthene, in contrast to the latter, which foresees a
sterically crowded 1,6,7,10-tetrasubstituted fluoranthene. Dis-
connection (a) had been an unsuccessful strategy when coupled
to Friedel−Crafts chemistry in the forward direction.
Disconnection (b) resembles the ultimate stage of the classic
Lawton-and-Barth strategy, wherein acyloin chemistry enabled
a successful synthesis.
The failure of disconnection (a) when coupled to Friedel−
Crafts chemistry can be seen in the additional strain needed to
obtain the transition state to attack on the pi face. The
molecule must fold to a deeper-bowled and higher-strained
conformation before bond formation can occur. Also the
reaction conditions are prone to rearrangements that would allow
for back reaction to ring-opened products. Disconnection (b)
requires a higher-energy synthetic precursor; however, in
combination with a reductive, radical, or insertion reaction,
wherein the transition state to carbon−carbon bond formation can
be reached by a more or less linear least-motion path of the
proximal units, this approach could be much more tractable. The
experience of Lawton and Barth with the acyloin reaction
empirically supports this analysis.
HISTORICAL DEVELOPMENT
■
Approaching half a century ago, Lawton and Barth reported the
first synthesis of corannulene in 17 linear steps (<1% overall
yield) starting from acenaphthene (Scheme 1).1 Their strategy
builds the structure up ring by ring and introduces strain as late
as possible in the synthesis by trading the stabilization from
aromatization of the rings for the destabilization of the strain
due to out-of-plane distortion. Albeit a pioneering tour de force,
the lengthy synthesis and low overall yield of 1 could not
stimulate a wave of exploration in corannulene chemistry. That
would have to wait for a new, more tractable synthesis.
From 1970 to 1990, several groups attempted to find alternative
routes to corannulene. A basic Friedel−Crafts strategy12 looked
attractive but went fallow. Creation of naphthyl-phenyl cyclo-
phanes as precursors to corannulene also never bore fruit.13 In the
glory of hindsight and physical organic chemical analysis, both
these unsuccessful approaches underestimated the strain needed to
Larry Scott’s and our laboratories independently developed
the two pathways simultaneously. The first strategy, from Scott
and co-workers,14 applied high-temperature, gas-phase flash
Received: December 30, 2011
Published: February 10, 2012
© 2012 American Chemical Society
664
dx.doi.org/10.1021/op200387s | Org. Process Res. Dev. 2012, 16, 664−676

Organic Process Research & Development
Article
Scheme 1. First synthesis of corannulene 1
to no functional group tolerance, limited possibility for scale-up
and undesired thermal rearrangements/side products.
The second strategy, pioneered in our group, used milder
and more selective solution phase chemistry to achieve the
synthesis of corannulene. The initial solution phase synthesis
(Scheme 3) follows the precedent of Buu-Hoi set back in
1942.16 It starts by preparing 2,7-dimethylnaphthalene (4) by
the addition of the Grignard of (2) to 4,4-dimethoxy-
2-butanone followed by acidic workup leading to deprotection
and ring closure.17 Chloromethylation of 4 occurs regioselec-
tively at the 1 position, and homologation with KCN and
hydrolysis in sulfuric acid/water leads to the acetic acid
derivative. Formation of the acid chloride and Friedel−Crafts
ring closure produces 2,7-dimethylacenaphthone, which is
converted to the analogous acenaphthaquinone (5) by
selenium dioxide oxidation.18 The crossed-aldol condensation
of 5 with 3-pentanone and subsequent Diels−Alder addition
yielded 1,6,7,10-tetramethylfluoranthene (7), which by Wohl−
Ziegler radical bromination was converted to 1,6,7,10-tetrakis-
(bromomethyl)fluoranthene (8a). The bridging thioether was
prepared by nucleophilic displacement with sodium sulfide,
from which sulfur extrusion and aromatization led to 1. This
expensive and lengthy synthesis proved the principle of the
Figure 1. Different retrosynthetic approaches to 1. Disconnection
between the flanking (a) or rim (b) carbons of 1 results in two
different fluoranthene precursors.
vacuum pyrolysis (FVP) to achieve high reaction energies
for short residence times. The idea that these conditions
could result in effective ring closure came from the
pioneering work of R. F. C. Brown, in which he showed
that under FVP conditions acetylenes could isomerize
and react as vinylenes.15 The high energy of the reaction
conditions opened a reaction path not available by
conventional chemical methods.
The optimized version of the procedure sets out from
the chlorovinyl compound (Scheme 2), instead of the
acetylene originally foreseen from the Brown chemistry.
Corannulene was produced on gram-scale in a 20−25%
yield starting with commercially available acenaphthene-
quinone. Although this route succeeded in a small number
of steps, it suffered from the following: modest yields, poor
bilaterally symmetric approach8 but lacked the practicality of
a
anything close to a scalable synthesis. Toxic reagents like
Scheme 2. Flash vacuum pyrolysis method developed by Scott
665
dx.doi.org/10.1021/op200387s | Org. Process Res. Dev. 2012, 16, 664−676

Organic Process Research & Development
Article
Scheme 3. First corannulene synthesis by the Siegel group
Scheme 4. Solution phase synthesis of corannulene (2006)
selenium dioxide further weakened the practical aspects of this
synthesis.
pyrolysis.10,19 Here the first gram scale solution phase synthesis
of corannulene was established and shown to produce not only
the basic core but also specific derivatives.
Investigations toward the formation of the tetracarboxaldehdyde
cognate of 7 or 8b led Sygula and Rabideau to attempt
hydrolysis of 8b.20 Serendipitiously, hydrolysis did not lead to
the expected precursor but instead deprotonated the remaining
benzylic hydrogen and initiated carbon−carbon bond for-
mation, leading to tetrabromocorannulene (9). This reaction
was strategically important and offered certain advantages for
gram-scale syntheses; however, the use of dioxane and the low
Further investigations lead to the implementation of a direct
formation of 5 from 4 by Friedel−Crafts acylation with oxalyl
chloride (Scheme 4). The method yielded two closely related
isomers of dimethylacenaphthaquinone, which required
stringent chromatographic separation; however, it obviated
many steps and the use of selenium dioxide. Benzylic
bromination of 7 could be pushed to give the octabromide
8b, which afforded a ring closure of the flanking bonds by using
low valent titanium or vanadium, eliminating the need for
666
dx.doi.org/10.1021/op200387s | Org. Process Res. Dev. 2012, 16, 664−676

Organic Process Research & Development
Article
Table 1. Comparison of reagents required before (2006) and
after (2011) for the synthesis of 3 to produce 1 kg of
corannulene22
Table 2. Comparison of reagents required before (2006) and
after (2011) for the synthesis of 4 to produce 1 kg of
corannulene22
2006
reagent
2011
2006
2011
kg
reagent
kg
reagent
kg
reagent
kg
reaction:
workup:
2
8.0
41
2
7.7
40
reaction:
workup:
3
9.2
39
44
94
3
9.3
45
diethyl ether
magnesium
diethyl ether
magnesium
AcOH
AcOH
1.7
1.6
HBr (48%)
Water
H2SO4 (97%)
water
5.6
78
dimethoxybutanone 7.0
dimethoxybutanone 6.3
water
59
55
5.7
water
77
MgSO4
3.1
70
NH4Cl solution
MgSO4
NH4Cl solution
Na2SO4
MTBE
27
toluene
3.8
110
Na2CO3
2.3
0.43
sat. NaCl
solution
purifica-
tion:
N/A
N/A
purification: N/A
yield23
87% (4)
vol. yield:
E-factor24
ethanol
19
yield23
:
72% (3)
9.2
85% (3)
9.2
:
5.2
71% (4)
4.3
vol. yield: reaction: 16%
E-factor24
total: 5.1% reaction: 16%
17
total: 3.0%
28
reaction: 7.8% total: 3.2% reaction: 9.4%
34
total: 1.9%
52
:
:
volume yield of the reaction made this operation very costly
and not very “green” if contemplated for synthesis on multi-
kilogram scale.
Reduction of 9 was possible using ∼100 equiv of zinc and
potassium iodide in ethanol. Alternatives using lithium
aluminum hydride had also been reported. The excessive
waste of the former and harsh reagents of the latter also
required attention for a successful scale-up.
Such was the state of the art in corannulene synthesis when
the present investigation in scale-up began; eight steps (overall
yield 7.4%), four chromatographic separations (compounds 5,
7, 8b, and 1), several undesirable solvents, harsh or excessive
reagent and no step-volume yield above 3%. To produce 120 g
of corannulene required almost a metric ton of solvent.
Four general challenges were the focus of this initial
optimization: (1) Reduce or replace as many costly or toxic
reagents as possible; (2) Increase the volume yields toward a
target of 10% from values in some cases below 1%; (3)
Enhance the robustness throughout the synthesis by mecha-
nistic insight and testing of critical parameters; (4) Remove all
chromatographic separations, including that needed to separate
the physically similar isomers of dimethylacenaphthenequinone.
acetic acid and 48% hydrobromic acid. During production, 95%
sulfuric acid replaced the hydrobromic acid (Table 2) and a
higher concentration of 0.81 M H2SO4 in acetic acid was
achieved. Crystallization resulted in a 71% yield of 4, lower than
the literature yield of 87%. Two additional side products were
isolated and characterized, 2,5-dimethylnaphthalene and 1,2-
bis-(3-methylphenyl)ethane, accounting for 7% and 6% yields
respectively.
Acylation of Dimethylnaphthalene (4) to Acenaph-
thenequinone (5). The Friedel−Crafts acylation of 4 with
oxalyl chloride presented several problems. The initial synthesis25
used large excess of reagents like aluminum bromide and oxalyl
chloride in dichloromethane to yield a mixture of primarily two
isomers, 5 and 5b in a 55−60% combined yield. The ratio of
isomers ranged from 1:1 to 3:1 in favor of 5. Separation of these
isomers was possible by column chromatography, but required
large amounts of ethyl acetate/hexane and a silica gel-to-product
ratios of 150:1 because both isomers have similar elution times.
Furthermore, the reaction was not robust and yields could vary by
as much as 50%.
The systematic optimization of 5 began by screening a series
of Lewis acids in a variety of solvents. The initial screen of
Lewis acids included AlBr3, AlCl3, SnCl4, and BF3(C2H5)2O in
dichloromethane. Aluminum bromide showed the best
reactivity in dichloromethane. A wider screening of solvents
was done for each of the four Lewis acids. The solvents tested
included toluene, chlorobenzene, THF, hexane and cyclo-
hexane. The reactions were started at temperatures between
−30 and −10 °C then, if no consumption of starting material
could be detected by GC−MS using an internal standard of
octadecane, were slowly heated to 20−25 °C. The result of the
initial screening showed that of the Lewis acids tested, only
AlBr3 and AlCl3 showed any reactivity towards the acylation,
regardless of temperature, solvent or time.
Next, the AlCl3 was screened against a wider variety of
solvents including those previously mentioned as well as
nitrobenzene, 1,2-dichlorobenzene and CS2. Again, the initial
temperature ranged between −30 and −20 °C then was slowly
raised to −10 to 0 °C. The reactions showed consumption of
the starting material but a poor yield of <25% of 5. Upon
interpretation of the side products we speculated that the Lewis
acid was reactive enough to allow for the first acylation at low
RESULTS AND DISCUSSION
■
Synthesis of 2,7-Dimethylnaphthalene (4). Conversion
of α-chloro-m-xylene (2) to 2,7-dimethylnaphthalene (4) was
accomplished through a Grignard addition of dimethoxybuta-
none followed by acidic ring closure. The production of 3
closely followed the reported procedure17 and required
minimal optimization (Table 1). Important in this step were
the safety considerations associated with a Grignard reaction.21
Diethyl ether is not a preferred solvent on scale due to its
extremely volatile and flammable nature. Attempts to use
tetrahydrofuran in this step led to substantial amounts of 1,2-
bis-(3-methylphenyl)ethane, the product of homocoupling of
the Grignard reagent. Methyltetrahydrofuran and methyl tert-
butylether were not investigated. Suppression of the coupling
reaction in other solvents would not allow the replacement of
diethylether in this step. Fortunately, it was found that with
incremental scale-up of this step and close monitoring of the
reaction, production was possible without problems.
For the ring closure of 3 to yield 2,7-dimethylnaphthalene
the original paper reported using a 0.51 M mixture of glacial
667
dx.doi.org/10.1021/op200387s | Org. Process Res. Dev. 2012, 16, 664−676

Organic Process Research & Development
Article
Scheme 5. Decomposition of the monoacylated intermediate 5c
temperatures but required increased temperatures to overcome
the energy barrier for the second acylation. Reported
literature26 supported this theory and showed that increasing
the temperature to even 0 °C will cause decomposition in such
a molecule, which was evident in our reaction by the formation
of 5f as shown in Scheme 5.
Other proposed side products of this reaction were
determined by mass analysis and NMR spectroscopy
(5b−f).
of this step was the replacement of 50% of the AlBr3 with the
cheaper aluminum chloride. The disadvantages to this step
were that hexane is an undesirable solvent due to its
electrostatic discharge and its neurological toxicity.28 Also, the
aluminum chloride does not completely dissolve and the
mixture of oxalyl chloride and 4 in hexane must be added at
temperatures in excess of 40 °C to keep 4 from crystallizing.
Replacing hexane with heptane addressed the safety consid-
erations but by doing so also resulted in a lower yield for 5. The
downfalls of this reaction just described were problematic enough
that these conditions were thought to be a last resort and
aluminum bromide in dichloromethane was considered the best
option for moving further. Continued optimization of these
conditions and efficient separation of the isomers 5 and 5b were
studied in parallel.
A reaction profile was performed on 10-g scale with an in-
process control via HPLC every 30 min. The first reaction with
both Lewis acids present in an equal molar ratio (Figure 2)
Although aluminum chloride alone did not furnish
comparable yields, we anticipated that a mixture of the two
Lewis acids could be successful. Per mole, aluminum chloride
costs about 1/20th the price of aluminum bromide at present.27
The ratio of the Lewis acids was varied from 100:0 to 0:100
AlBr3:AlCl3 (Table 3).
Table 3. Friedel−Crafts acylation using mixture of AlBr3 and
AlCl3
a
lewis acid (%)
major products (%)
AlBr3
AlCl3
4
5
5b
Figure 2. In process control of 4, 5 and 5b−5e composition (HPLC)
100
75
50
25
0
0
25
5
5
2
2
4
38
31
30
19
15
20
16
13
6
along time/temp profile.
50
75
resulted in a 25% yield of 5, as much of it decomposed to 5e
(48%).
100
4
a
All reactions were performed on 500 mg scale using 1.2 equiv oxalyl
We hypothesized that as previous research showed aluminum
chloride to be suitable for the first acylation, aluminum bromide
could be added later in the reaction to promote the second
acylation. This delayed addition of AlBr3 would reduce the
amount of aluminum bromide needed while maintaining the
yield of 5. Surprisingly in the presence of only AlCl3, after two
hours more than 50% of the detected intermediate was 5c, the
precursor to 5. However, upon addition of AlBr3, most of 5c
was either converted to 5b or decomposed to 5e. Although it is
still not fully understood why the addition of aluminum
bromide caused the formation of side products and not the
desired diketone 5, the results were reproducible and in end
favored the formation of 5b over 5 in a 4:1 ratio. A stability test
of 5 was performed and no decomposition was detected in the
chloride and 2.6 equiv Lewis acid in dichloromethane. A temperature
ramp of −40 to −15 °C over 4 h was applied. Percentages of products
were determined by GC−MS using an internal standard of octadecane.
The screening results of the AlBr3/AlCl3 mixtures in
dichloromethane suggested that replacement of aluminum
bromide with aluminum chloride lowers the reaction yield
more than an acceptable amount. A 1:1 mixture of AlBr3/AlCl3
using hexane at 40 °C showed a 29% yield of 5 but almost no
5b (98:2). These conditions were important because, if an
efficient separation method for 5 and 5b could not be achieved,
then selectively synthesizing only 5 might have been the best
strategy for obtaining pure product on scale. Another highlight
668
dx.doi.org/10.1021/op200387s | Org. Process Res. Dev. 2012, 16, 664−676

Organic Process Research & Development
Article
presence of aluminum bromide and aluminum chloride over a
temperature range of −40 to 25 °C.
substantially different in a broad spectrum of solvents.30 The
key chemical difference is the steric environment in the plane of
the quinone function. Normal kinetic attack on carbonyls
occurs via Burgi−Dunitz31 approaches out of the carbonyl
plane, therefore such strategies seemed unlikely to help, and
screening revealed no exception to this expectation. Imine
formation would occur under thermodynamic control and
would place a new substituent in the plane, where the largest
steric difference between the isomers could be exploited. Such a
process showed greater mechanistic promise and indeed imine
formation proceeds for 5b better than 5. More importantly,
amino acid hydrazides had precedence in the steroid literature
as reagents to make water-soluble derivatives that could be
separated by simple aqueous extractions. These reagents
are sold under the commercial name of Girard’s reagents.32
In particular, limiting amounts of Girard’s reagent T (GRT)
formed a hydrazone selectively with 5b in AcOH over two
hours at 40 °C. The hydrazone of 5b was washed away from 5
with water, in a highly selective manner that eliminated the
need for column chromatography (Scheme 7). The isolation
yield for this step was 90% based on the initial amount of 5 in
the mixture. The aqueous extract could be hydrolyzed under
acidic condition to recover 5b containing a small amount of 5.
Total recovery of the isomers from the reaction and hydrolysis
was 96%. It was also discovered that when this separation
procedure was performed at reflux in acetonitrile, with only a
small amount of acetic acid for 4 h, one could obtain 5 in 98%
yield following the same workup conditions; however, due to
the price of acetonitrile at the time of production, it was not
advantageous to use these conditions.
Finally, it was decided to use only AlBr3. The concentration
and stoichiometry of reagents were varied to determine that the
optimal molarity of 0.23 M of 4 in dichloromethane (Table 4)
resulted in a reproducible combined yield of 73% (2:1) of 5 and
5b (Figure 3) when a temperature gradient of −38 to −15 °C
over 6 h was applied, as shown in Scheme 6.
Having accepted a 2:1 mixture of 5 and 5b as the best
synthetic outcome, the focus shifted to developing an efficient
method for isolation and purification of the product.
Chromatography was not seen as an option. Both isomers
have very similar physical properties; their solubilities were not
Table 4. Comparison of reagents required before (2006) and
after (2011) the optimization of 5 and 5b to produce 1 kg of
corannulene22
2006
reagent
2011
reagent
kg
kg
reaction:
workup:
4
5.2
6.6
20
4
4.3
4.2
16
oxalyl chloride
oxalyl chloride
aluminum
bromide
aluminum
bromide
dichloromethane
water
540
910
40
dichloromethane
water
150
92
MgSO4
Na2SO4
15
sat. NaCl sol.
240
Celite
7.8
2.9
67
Na2CO3
toluene
yield:
N/A
73% (5 + 5b 2:1) 4.2
Synthesis of 1,6,7,10-Tetramethylfluoranthene (7).
The next step of the synthesis was conversion of 5 to
1,6,7,10-tetramethylfluoranthene (7) through a crossed-aldol
condensation followed by a Diels−Alder, retro Diels−Alder
reaction cascade using norbornadiene (Scheme 8). Improve-
ments to this step significantly improved volumetric
productivity, which reduced the amount of solvents required.
The amount of methanol required for the Knoevenagel
condensation could be reduced by 50% compared to that
previously reported33 and still obtain the same yield (Table 5);
however, decreasing the equivalents of 3-pentanone34 and
KOH from the reported value led to a decrease in yield.
Another important factor in this step was the isolation of the
hydroxy intermediate 6. Careful neutralization of the reaction
must be done to avoid protonation and elimination, leading to
dimerization of the newly formed cyclopentadienone. Previous
attempts to crack the dimer in the subsequent step were
unsuccessful. Formation of the dimer dramatically decreases the
yield.
purification: silica gel
hexane
780
AcOH
29
1300
750
GRT
1.2
47
dichloromethane
36% (5)
reaction: 0.62%
water
yield:
2.5
42% (5)
reaction: 2.2%
2.5
total:
0.70%
vol. yield:
total:
0.07%
E-factor:
1800
170
Important to the optimization of the workup for the Diels−
Alder/retro-Diels−Alder step was the realization that, with slow
quenching of the final reaction mixture, one could precipitate
and filter 7 at the end of the reaction. This eliminated the need
for the extraction and column chromatography reported in the
literature procedure. The final product was obtained in 99.8%
purity after recrystallization from an i-PrOH/acetone mixture.
Radical Bromination of 6 To Yield 1,6,7,10-Tetrakis-
(dibromomethyl)fluoranthene (7b). The literature bromi-
nation of 6 proceeded through a radical bromination using NBS
and radical initiaton by benzoyl peroxide (BPO) in benzene (or
carbon tetrachloride) with a 375 W tungsten lamp as the light
source.35 This step was challenging to produce on scale due
to the safety issues associated with the solvents and with
Figure 3. In process control of 4, 5 and 5b−5e composition (HPLC)
along time/temp profile (Optimized).
Scheme 6. Preparation of isomers 5 and 5b
669
dx.doi.org/10.1021/op200387s | Org. Process Res. Dev. 2012, 16, 664−676

Organic Process Research & Development
Article
Scheme 7. Separation of 5 and 5b using Girard’s Reagent in AcOH
Scheme 8. Optimization of 1,6,7,10-tetramethylfluoranthene (7)
Table 5. Comparison of reagents required before (2006) and
after (2011) for the optimization of 7 to produce 1 kg of
corannulene22
Table 6. Comparison of reagents required before (2006) and
after (2011) for the optimization of 8b to produce 1 kg of
corannulene22
2006
2011
2006
2011
reagent
reagent
kg
reagent
kg
reagent
kg
kg
reaction:
workup:
5
2.5
88
5
2.5
30
16
8.0
87
30
reaction:
workup:
7
1.9
14
7
2.0
16
MeOH
KOH
MeOH
KOH
NBS
BPO
NBS
16
0.11
130
38
AIBN
0.06
3-pentanone
water
8.3
110
3-pentanone
water
benzene
water
chlorobenzene 46
ethyl acetate
15
sat. NaCl solution 34
HCl (32%)
dichloromethane
MgSO4
63
HCl (10%)
11
19
dichloromethane
MgSO4
120
9.3
30
sat. NaCl solution 46
Na2S2O3
87
reaction:
workup:
acetic anhydride
acetic
32
purification: silica gel
hexane
200
1400
4.8
N/A
anhydride
NBD
8.4
74
52
50
NBD
water
8.6
7.4
yield:
74% (8b)
73% (8b)
5.1
water
vol. yield.
reaction: 3.3%
total:
0.19%
reaction: 12%
total:
8.8%
NaOH (10%)
DCM
NaOH (30%) 6.2
MeOH
56
E-factor:
410
12
sat. NaCl solution 54
N/A
a
yield:
66% (7)
silica gel
2.1
2.1
28
use of light and a radical initiator, albeit clear that substantial
thermal activation seemed to occur. The better stabilized
radical initiator AIBN, available from the Vazo line of initiators
at Dupont, showed good results at these higher temperatures
on scale, and was selected (Table 6).
There were also hesitations that the light source would not
be sufficient on such scale due to the heating mantle and depth
of the 100-L reactor. Accordingly, many attempts to find bromi-
nating conditions without an extra light source were attempted
but unsuccessful, even in the presence of AIBN. Fortunately,
radical generation at the outer layer of the reactor sufficed to
give a steady and robust process. On 100-L scale, three lamps
were adequate to perform the bromination when shone directly
into the solution.
purification: silica gel
hexane
250
2000
cyclohexane
isopropanol
acetone
17
0.52
2.0
b
yield:
60% (7)
1.9
65% (7)
vol. yield:
reaction: 1.4%
total:
0.05%
reaction: 3.0% total:
0.64%
E-factor:
1500
160
a
b
Assay: 94%; uncorrected yield: 70%. Assay: 99.8%; purification
yield: 99%.
performing a photochemical radical reaction.36 Specifically, the
radical initiator used, BPO, is less effective at stable radical
production than other initiators and is not preferred on scale.37
To overcome these safety issues a series of reactions were
screened to try and avoid the need for radical initiators. The
solvent of choice turned out to be chlorobenzene at 90−95 °C,
but, unfortunately, the higher temperature did not obviate the
An unforeseen problem arose when scaling the bromination
from 10 to 50 g. The dark-red solution unexpectedly turned
black and the optimized workup conditions failed to yield 8b
directly. Column chromatography of the mixture resulted in a
25−30% isolated yield of 8b while the remainder of the crude
670
dx.doi.org/10.1021/op200387s | Org. Process Res. Dev. 2012, 16, 664−676

Organic Process Research & Development
Article
mixture stayed on the baseline. This was most likely due to the
presence of HBr in the headspace that was responsible for the
formation of side products in significant proportions. A slight
vacuum of 400 mbar was applied and the reaction proceeded as
expected (Scheme 9). The product was recrystallized using
ethyl acetate, eliminating the column chromatography that was
previously required. Given the sensitivity of the octabromo
compound, a large-scale chromatography would not only have
added cost and material waste from the silica, but likely would
have led to some product decomposition.
Table 7. Comparison of reagents required before (2006) and
after (2011) for the optimization of 9 to produce 1 kg of
corannulene22
2006
2011
reagent
kg
reagent
kg
reaction:
workup:
8b
4.8
2.2
190
73
8b
5.1
NaOH
dioxane
water
NaOH (30%) 6.8
isopropanol
40
water
380
16
water
19
11
HCl (10%)
isopropanol
Scheme 9. Benzylic bromination of fluoranthene 7
dichloromethane 250
acetone
150
2.5
yield:
83% (9)
79% (9)
2.5
vol. yield:
E-factor:
reaction: 0.95%
total: 0.24% reaction: 4.9% total: 3.0%
420 30
compared to other metals, the large zinc dust excess (100 equiv)
and amount of ethanol (8.8 mM) required to facilitate stirring
proved to be a challenging environmental waste problem. Another
difficulty with this reaction was the reported long reaction time of
six days. Even after the use of excessive reagents and long reaction
times, one would still isolate a mixture of side products that
required column chromatography to purify.
Synthesis of 1,2,7,8-Tetrabromocorannulene (9). A
strategic improvement for the ring closure of 8b to form the
first derivative containing the corannulene core, tetrabromocor-
annulene 9, was reported by Sygula et al. using a dioxane/water
mixture in the presence of base.20 Nonetheless, there were still
two main areas of optimization for this step: the low volume
yield due to reaction concentrations of 20 mM and the cost and
toxicity of dioxane as solvent.
In an attempt to overcome the issue of low concentration,
the reaction was performed in a ratio of 1 g 8b in 10 mL of a
solvent mixture (H2O:MeOH, H2O:Toluene, H2O:Dioxane,
Toluene:MeOH) with varying bases (NaOH, NaOMe, KOH).
These reactions resulted in polymerization of the starting
material. The next consideration was portionwise addition of
8b to the solvent mixture. Upon formation, the product 9
precipitates and allows the starting material to dissolve in the
solvent. This would decrease the amount of solvent needed
while having minimal effect on the yield. Unfortunately, the
strategy proved wrong and instead polymerization of the
starting material was the main product isolated upon filtration.
A wider screening of solvents showed that isopropanol was a
suitable replacement of the dioxane/water mixture, increasing
the allowable reaction concentration from 20 mM to 112 mM
(Scheme 10). Addition of 8b to an isopropanol solution with
sodium hydroxide at reflux led to the clean formation of 9 in a
79% yield (Table 7). The product was filtered and carried
through to the next step without further purification.
A recent paper by Thiemann39 used zinc dust in the presence
of ammonium formate and base to dehalogenated tetrabromo-
bisphenol A (TBBPA), a hazardous pollutant, via transfer
hydrogenation. These results led us to believe that a stoichiometric
quantity of ammonium formate and zinc could have the same
effect on corannulene, greatly reducing the amount of reagents
and solvents required and shortening reaction time to just a few
hours. Initial results appeared encouraging with 1 detected after
only 2 h. Unfortunately the reaction did not proceed to com-
pletion and even after 2 days and up to 20 equiv of zinc powder,
GC−MS and TLC identified that there was dibromocorannulene
and monobromocorannulene still present in the reaction. Analysis
of the reactor headspace indicated the presence of acid. It was
hypothesized that HBr formation during the reaction was pre-
venting further reduction of 9 and addition of zinc oxide might
solve this problem and push the reaction to completion. We
expected the ZnO and HBr to react and form zinc bromide and
water. Unfortunately, the addition of ZnO did not have an effect
on the reaction and an alternative reduction method was sought.
A paper by Spatola40 showed that ammonium formate and Pd/C
were successful at reducing haloaromatics through catalytic transfer
hydrogenation. Successful reduction of 9 to 1 was achieved by
replacing zinc with 5% palladium on carbon. A setback occurred
upon scaleup from 1 to 10 g since formation of ammonium
bromide was detected in the condenser. Due to our specific reactor
configuration, salt accumulation would not be easy to remove safely
during production, although these reaction conditions were
otherwise efficient and scalable. To solve our specific infrastructure
issue, we replaced ammonium formate with a triethylamine/formic
acid mixture and further solvent screening showed pyridine was the
best to solubilze the formed salts. A less costly alternative to pyridine
is 3-picoline, which produced similar results and was used (Table 8)
in the reduction of 9 (Scheme 11).41
Scheme 10. Ring closure of 8b to yield 1,2,7,8-
tetrabromocorannulene 9
Reduction of Tetrabromocorannulene (9) to Coran-
nulene (1). The final step of the synthesis, reduction of 9 to 1
had a number of hurdles to overcome.38 One principal problem
for the reaction was the excessive amounts of reagents, solvent,
and reaction time required. Although zinc is relatively inexpensive
CONCLUSION
■
The optimization and production of 1 has been demonstrated on
scale. The key successes of the synthesis were to eliminate the
671
dx.doi.org/10.1021/op200387s | Org. Process Res. Dev. 2012, 16, 664−676

Organic Process Research & Development
Article
Table 8. Comparison of reagents required before (2006) and
after (2011) for the optimization to produce 1 kg of
corannulene22
EXPERIMENTAL SECTION
■
General. Starting materials were obtained from commercial
suppliers and were used without further purification. HPLC
analyses were performed on an Agilent Series 1100 liquid
chromatograph equipped with a UV detector. NMR data
obtained were identical to those reported in the literature. Yield
calculation based on HPLC purity, or assay if indicated. Mole
amount reported has been corrected based on purity.
2006
2011
reagent
kg
reagent
kg
reaction:
9
2.5
26
9
2.5
1.7
3.6
4% HCl
Zn
formic acid
30
triethylamine
5% Pd/C
3-picoline
water
3-Hydroxy-3-methyl-4-(3-methylphenyl)butanal Di-
methyl Acetal (3). Under a N2 atmosphere, a 100-L reaction
vessel, magnesium turnings (2.15 kg, 88.2 mol) and anhydrous
diethyl ether (11.3 kg; test for peroxides was negative) were
added at a temperature of 15−25 °C. To this suspension
α-chloro-m-xylene (311 g, 2.12 mol) was added at once. The
reaction started immediately as observed by temperature rise to
reflux and turbidity. When the reaction started, a solution of
α-chloro-m-xylene (9.72 kg, 66.4 mol) in diethylether (21.9 kg)
was added at an internal temperature (IT) of 33−36 °C over
2 h. The suspension was stirred for 30−60 min (IT = 33−36 °C);
then a solution of acetylacetaldehyde-dimethylacetal (95%, 8.25
kg, 59.3 mol) in diethylether (18.7 kg) was added (IT = 33−
36 °C) over 2 h. The suspension was cooled to an internal
temperature of 15−25 °C and stirred at this temperature for
50−70 min. The suspension was then added to a precooled
(0−5 °C) solution of ammonium chloride (35.6 kg, 666 mol)
in water (100 kg) while stirring vigorously (IT = 0−5 °C;
caution: H2 gas evolution!). The mixture was stirred until gas
evolution ceased and two clear layers were obtained. The
aqueous phase was extracted two times with MTBE, and the
organic layers were combined and dried over sodium sulfate.
The sodium sulfate was filtered, and the organic layer was
evaporated to dryness at a bath temperature of 30−40 °C
to yield 13.5 kg (HPLC purity 89% 3-hydroxy-3-methyl-
4-(3-methylphenyl)butanal dimethyl acetal (3) [12.0 kg, 85%
HPLC corrected yield]).
2,7-Dimethylnaphthalene (4). In a 100-L reaction vessel,
3 (15.1 kg, purity 86% [13.0 kg, 54.8 mol]) was dissolved in
acetic acid (74.2 kg) at 20−25 °C. Sulfuric acid (97%, 9.10 kg)
was added to the solution of 3 slowly over 10 min. The reaction
mixture was refluxed (IT = 118 °C) for 30 min then cooled
to an internal temperature of 38−43 °C and the acetic acid
(64.3 kg) was distilled (IT = 107 °C) under reduced pressure
(50−70 mbar). The remaining mixture was poured into ice
water (130 kg) and extracted twice with toluene (110 kg total).
The organic phases were combined and washed with a sodium
bicarbonate solution (3.7 kg) then sodium chloride solution
(700 g). The organic phase was then dried using magnesium
sulfate (5.0 kg) and filtered, and then the toluene was com-
pletely removed under reduced pressure. The resulting solid (10.1 kg)
was recrystallized using ethanol (31 kg) to yield 6.37 kg of a light-
brown solid (HPLC purity 99% 2,7-dimethylnaphthalene (4) [6.32
KI
11
0.13
27
EtOH
water
390
250
workup:
14
dichloromethane 200
HCl (32%)
1.7
MgSO4
silica gel
hexane
15
purification:
250
toluene
27
1900
activated
carbon
0.04
Celite
0.16
1.0
yield:
90% (1)
1.0
88% (1)
vol. yield:
reaction: 0.19%
total:
0.03%
reaction: 3.5% total:
1.3%
E-factor:
3100
75
Scheme 11. Reduction of 9 to corannulene using Pd/C
column chromatographies previously required for four of the eight
steps as well as to find safer and less costly solvents and reagents.
Careful evaluation of required solvent volume in the synthesis of 1
starting from [alpha]-chloro-m-xylene 2 improved the average
concentration of each step from 0.04 to 0.24 M. A total of 1.3 kg of
1 were isolated in an 8.7% yield over nine steps with the average
yield per step being 75% (Scheme 12).42
In retrospect, the present process has come a long way since
the discovery synthesis of Lawton and Barth. It offers a scalable
batch process amenable to a simple pilot-scale facility and has
led to the production of corannulene on kilogram scale. None-
theless, there is still much room for improvement and
optimization, which bodes well for the future. In particular, a
more selective diacylation of 4 to 5 could reap great benefits in
cost and scalability. In addition at present the conversion of 7 to 1
goes through a multistep process, but a dehydrogenative benzylic
coupling and aromatization in one transformation, perhaps by a
flow process may be feasible. Such a direct oxidative benzylic
coupling by CH activation of the unfunctionalized methyl groups
would consititute a new carbon−carbon bond-forming reaction
with great synthetic utility. If these two hurdles were to be
addressed by proper catalysts or reaction conditions, another
improvement by 2 orders of magnitude in material costs could be
gained. This conversion of the production of corannulene from
milligram scale in 1966 to kilogram scale in 2011 also marks the
transition of corannulene from an esoteric molecular object of
academic interest into a chemical entity with a potential
engineering/commercial future.
1
kg, 74% HPLC corrected yield]). Mp 95−96 °C, H NMR (250
MHz, CDCl3) δ: 2.49 (s, 6 H), 7.22 (d, 2 H, J = 8.0), 7.50 (s, 2 H),
7.65 (d, 2 H, J = 8.0); 13C NMR (62.9 MHz, CDCl3) δ: 21.7, 126.3,
127.2, 127.4, 130.0, 134.0, 135.4.43
3,8-Dimethylacenaphthenequinone (5) and 4,7-Dime-
thylacenaphthenequinone (5b). Dichloromethane (66.5 kg)
was charged to a 100-L reaction vessel and then cooled to an
internal temperature of −40 to −30 °C. Aluminum bromide
(10.3 kg, 38.2 mol) was added and stirred at this temperature
for at least 30 min. Meanwhile, a solution of 2,7-
dimethylnaphthalene (2.78 kg, purity 100% [2.78 kg, 17.8 mol])
672
dx.doi.org/10.1021/op200387s | Org. Process Res. Dev. 2012, 16, 664−676

Organic Process Research & Development
Article
Scheme 12. Optimized synthesis of corannulene (1)
in dichloromethane (33.2 kg) was prepared. Oxalyl chloride
(2.73 kg, 21.0 mol) was added to this solution, and the solution was
added dropwise to the aluminum bromide solution, maintaining
an internal temperature below −35 °C. When approximately 90%
of the oxalyl chloride solution was added, a temperature gradient
was started. The jacket temperature was programmed to increase
from −45 °C to −15 °C within 4 h. Afterwards, the internal
temperature was kept at −15 to −20 °C and HPLC in-process
controls were taken every hour until there was less than 7%
starting material compared to product, approximately 7−8 h.
Water (12.0 kg) was slowly added then the reaction was allowed
to warm to room temperature. The reaction mixture was filtered
through Celite (5.0 kg) then the Celite filter cake was washed with
water (30 kg) and dichloromethane (30 kg). The layers were
separated and the aqueous layer was extracted once more with
dichloromethane (13 kg). The two organic layers were washed
consecutively with water (20 kg) and the combined layers were
dried with sodium sulphate (5.0 kg). The suspension was filtered
and the filter cake was washed with dichloromethane (6.4 kg).
A "100-L reaction vessel was charged with the filtrate and heated
to an internal temperature of 36−42 °C and the solvent was
distilled. After distilling 118 kg, toluene (32 kg) was added.
The jacket temperature was increased from 55 to 90 °C within 2
h. When the internal temperature reached 85−90 °C, the jacket
temperature was cooled to 80 °C then a solution of sodium
carbonate (2.8 kg, 26.2 mol) in water (37 kg) was added. The
mixture was heated until the internal temperature reached 75−80
°C, then the layers were separated and the organic layer was
washed with water (10 kg). The aqueous layer was separated and
the organic layer was distilled at an internal temperature of 81−86
°C until the product began to crystallize. The suspension was then
cooled to an internal temperature of 0−5 °C within 4 h and stirred
overnight at that temperature. The suspension was filtered and
the white filter cake was washed via reactor with toluene (10 kg).
The wet product was dried for 14 h at 50 °C and <10 mbar to yield
2.33 kg yellow needles (HPLC purity 71.1% 3,8-dimethylacenaph-
thenequinone (5) 28.9% 4,7-dimethylacenaphthene-
quinone (5b) [2.33 kg, HPLC corrected yield of the isomer
mixture: 62%]). The mother liquour was concentrated to 3 L,
cooled to 0 °C, and kept at this temperature overnight. The
suspension was filtered and washed with toluene (1 kg) then dried
at 50 °C and <10 mbar to yield 180 g yellow-brown powder
(HPLC purity 64.7% 3,8-dimethylacenaphthenequinone (5) 33.8%
4,7-dimethylacenaphthenequinone (5b) [0.177 kg, HPLC corrected
yield of the isomer mixture: 4.7%]).
Separation of 3,8-Dimethylacenaphthenequinone (5)
and 4,7-Dimethylacenaphthenequinone (5b). Acetic acid
(31.5 kg) and a mixture of isomers 5 and 5b (6.00 kg, purity
69% (5) [4.1 kg (5), 19.7 mol (5)]) was charged to a 100-L
reaction vessel and heated to 40 °C for at least 30 min. During
this time, a solution of Girard’s Reagent T (1.73 kg, 10.3 mol,
1.2 equiv of 5b) in 9.55 kg acetic acid was prepared. This
solution was added dropwise to the isomer mixture solution
over 2 h. The reaction continued to stir for an additional 2 h,
then water (60.0 kg) was slowly added to quench the solution
over a period of 30 min. The suspension was filtered, washed
with 8 kg water, and then dried at <2 mbar overnight to result
in 3.72 kg 5 (HPLC assay 97% 3,8-dimethylacenaphthenequi-
none (5) [3.67 kg, 90% HPLC corrected yield]). Mp 204−
205 °C, 1H NMR (500 MHz, CDCl3): δ 2.87 (s, 6H), 7.49 (d,
2H, J = 8.28 Hz), 8.03 (d, 2H, J = 8.28 Hz); 13C NMR (100
MHz, CDCl3): δ 18.12, 76.58, 77.00, 77.42, 124.29, 127.47,
130.63, 131.73, 137.57, 147.10, 188.80.
To the mother liquor, hydrochloric acid (55.7 kg) in water
(24.0 kg) was added and stirred at room temperature for 3 h.
The precipitate was filtered then dried at <2 mbar overnight to
673
dx.doi.org/10.1021/op200387s | Org. Process Res. Dev. 2012, 16, 664−676

Organic Process Research & Development
Article
yield a mixture of isomers 5:5b (1.88 kg) of 1:4. Total recovery
for this step for 5 was 99% (5 + 5b 93%).
and NBS (20.0 kg, 112 mol) were added, and a vacuum
(400 mbar) was applied. The mixture was heated to an internal
temperature of 90 °C and after 15−30 min an exothermic
reaction was detected. The jacket temperature was set to <98 °C
until the internal temperature decreased since decomposition of
the product occurs at temperatures above 100 °C. The mixture
was stirred vigorously at 90−95 °C, and the reaction was moni-
tored by thin layer chromatography. Additional AIBN (25.0 g,
0.152 mol) was added after 3 and 10 h. After 23 h the
suspension was cooled to 40−45 °C, and the suspension was
filtered. The filtrate was evaporated at 70 °C (120 mbar to
8 mbar), and the distillate (20 kg) was used to wash the filter
cake. Ethyl acetate (8 kg) was added, and then the solvent was
evaporated at 70 °C (120 mbar to 8 mbar). Additional ethyl
acetate (2 kg) was added, and then the suspension was stirred for
1 h at 70 °C before cooling to 18−25 °C. The suspension was left
to stir at this temperature overnight, then filtered and washed with
ethyl acetate (3.8 kg). The wet product was dried at 40 °C and
<10 mbar for 12 h to yield 6.40 kg (HPLC purity 92% 1,6,7,10-
tetrakis(dibromomethyl)fluoranthene (8) [5.93 kg, 70% HPLC
corrected yield]). Mp > 300 °C, 1H NMR (500 MHz, CDCl3): δ
7.10 (s, 2H), 7.22 (s, 2H), 7.99 (d, 3J = 8.5 Hz, 2H), 8.22 (s, 2H),
8.28 (d, 3J = 8.5 Hz, 2H). 13C NMR (100 MHz, CDCl3): δ (ppm)
38.5, 39.0, 127.6, 130.1, 130.2, 131.5, 131.8, 132.3, 136.5, 137.9.
1,2,7,8-Tetrabromocorannulene (8). A 50-L reaction
vessel was charged with isopropanol (24.9 kg) and 1,6,7,10-
tetrakis(dibromomethyl)fluoranthene (3.19 kg, purity 93%
[2.97 kg, 3.35 mol]). The suspension was heated to an internal
temperature of 77−82 °C. A NaOH solution (30% aq, 4.25 kg,
31.9 mol) was added dropwise over 40 min; the reaction
mixture was allowed to stir at 77−82 °C for 55−65 min. The
suspension was slowly cooled to 20−25 °C over 90−120 min
and then filtered through a glass filter, and the filter cake was
washed with water (12.0 kg) then isopropanol (6.9 kg). The
product was dried at 50 °C and <10 mbar for 24 h to yield 1.90 kg
(HPLC purity 79% 1,2,7,8-tetrabromocorannulene (9) [1.49 kg,
1,6,7,10-Tetramethylfluoranthene (6). Methanol (20 kg)
was charged to a 100-L reaction vessel, then KOH (11.0 kg,
177 mol) was added in portions. The solution was heated to
60−80 °C until the KOH dissolved. The solution was then
cooled to 20 °C, and 3-pentanone (5.45 kg, 61.9 mol) was
added. 3,8-Dimethylacenaphthenequinone (1.70 kg, assay 98%
[1.67 kg, 7.86 mol]) was slowly added, and a brown solution
was obtained. The mixture was stirred at 18−25 °C for at least
2 h. The reaction mixture was cooled to 0 °C, and a previously
cooled (−3 °C) solution of water (44 kg) and HCl (32%,
20 kg) was added dropwise (so that the internal temperature
was less than 20 °C) until the color of the solution changed to a
yellow-green and pH was between 5 and 7. The suspension
was stirred for 30−60 min at 18−22 °C and then filtered, and
the filter cake was washed with water (15 kg). The product
was dried at 40 °C and <10 mbar for 24 h. The vessel was
charged with the dried intermediate (2.17 kg), acetic anhydride
(21.0 kg), and 2,5-norbornadiene (5.8 kg). The reaction
mixture was heated to reflux (135−140 °C) for 2−3 days and
then cooled to 20−25 °C. A solution of 30% NaOH (4.2 kg)
and water (2.5 kg) was added slowly dropwise. The reaction
mixture was allowed to stir overnight at 15−30 °C and then
distilled at 200 mbar and at an internal temperature of 70 °C
(distillate 16 kg). Water (2.5 kg) was added, and the internal
temperature was raised to 88 °C; then vacuum (300 mbar) was
slowly applied. The mixture was cooled to an internal tempera-
ture of 20−25 °C, and distillate (6.7 kg) was collected. Meth-
anol (2.5 kg) was added to precipitate 6, and the dark-brown
suspension was filtered and washed with additional MeOH
(2.5 kg). The reactor was charged with the filter cake and
methanol (25.5 kg) and then was heated to reflux for 1 h. The
suspension was cooled to 20−25 °C and stirred at this tem-
perature overnight and then filtered, and the filter cake was
washed with additional MeOH (2.1 kg). The product was
dried at 40 °C and <10 mbar for 4 h to yield 1.35 kg (HPLC
assay 94.5% 1,6,7,10-tetramethylfluoranthene (7) [1.28 kg, 63%
HPLC corrected yield]).
1
79% HPLC corrected yield]). Mp 338−340 °C dec, H NMR
3
(400 MHz, CDCl3): δ 7.98 (s, 2H), 7.95 (d, J = 8.9 Hz), 7.85
3
(d, J = 8.9 Hz, 2H). 13C NMR was not obtained due to poor
For further purification a 50-L reaction vessel was charged
with 7 (1.99 kg, assay 94% [1.87 kg, 7.22 mol]) and
cyclohexane (26.5 kg). A dark-brown suspension was obtained
and then filtered through a silica gel plug (2.0 kg) with 800
mbar. The filtrate was evaporated, and the distilled cyclohexane
was used to wash the silica gel until no 6 can be detected on
TLC. The residue was then heated in isopropanol (13.9 kg) at
79 °C. Acetone (0.50 kg) was added dropwise until a solution
was obtained; then the solution was cooled to 20−25 °C and
stirred at this temperature overnight. The suspension was
filtered, and the filtrate was washed with cooled isopropanol
(2.3 kg). The product was dried at 50 °C and <10 mbar over-
night to yield 1.84 kg (HPLC assay 99.8% 1,6,7,10-tetramethyl-
fluoranthene (7) [1.84 kg, 99% HPLC corrected yield]). Mp
140.5−143.5 °C, 1H NMR (500 MHz, CDCl3): δ 2.74 (s, 6H),
2.84 (s, 6H), 7.13 (s, 2H), 7.35 (d, 2H, J = 8.28), 7.67 (d, 2H,
J = 8.28); 13C NMR (100 MHz, CDCl3): δ 24.30, 25.13, 126.15,
126.60, 129.63, 130.67, 131.83, 132.00, 133.67, 134.88, 139.91
1,6,7,10-Tetrakis(dibromomethyl)fluoranthene (7). A
100-L reaction vessel was charged with chlorobenzene (55.5 kg)
and 1,6,7,10-tetramethylfluoranthene (2.47 kg, assay 99.8% [2.47
kg, 9.56 mol]). The solution was heated to 70−75 °C while
illuminating with three 350 W spot bulbs in a distance of 30 cm
for the duration of the reaction sequence. AIBN (25.0 g, 0.152 mol)
solubility.
Corannulene (9). A 50-L reaction vessel was charged with
1,2,7,8-tetrabromocorannulene (4.0 kg, purity 77%, [3.32 kg,
5.87 mol44]), 3-picoline (42.0 kg), formic acid (2.62 kg, 55.8
mol), triethylamine (5.76 kg, 56.6 mol), and 5% Pd/C (50%
water, 200 g, 0.41 mol). The suspension was heated to reflux
and stirred for 16 h, until the reaction was complete. The
mixture was cooled to an internal temperature of 20−25 °C,
and the solid was filtered through a glass filter and then washed
with 5.7 kg 3-picoline. The filtrate was returned to the reactor
and heated to an internal temperature of 73 °C while applying a
vacuum of (pstart = 100 mbar) to remove 45 kg of 3-picoline so
that crystallization occurred. Upon crystallization of the
product, toluene (23 kg) and a mixture of HCl (32%, 2.7 kg)
in water (14 kg) was added until the pH of the water phase was
3. The mixture was heated to an internal temperature of 83−
87 °C, and activated carbon (57 g) was added. This suspension
was stirred for 10 min, and then Celite (250 g) was added and
stirred for an additional 10 min. The mixture was filtered
through a Celite plug under pressure (700 mbar), and the
filtrand was washed with toluene (10 kg). The filtrate was
transferred back into the reactor and heated to an internal
temperature of 83−87 °C. The organic phase was washed twice
with water (5.7 kg total) and then filtered (IT = 86−90 °C)
674
dx.doi.org/10.1021/op200387s | Org. Process Res. Dev. 2012, 16, 664−676

Organic Process Research & Development
Article
(11) (a) Sygula, A.; Fronczek, F. R.; Sygula, R.; Rabideau, P. W.
J. Am. Chem. Soc. 2007, 129, 3842. (b) Kobryn, L.; Henry, W. P.;
Fronczek, F. R.; Sygula, R.; Sygula, A. Tetrahedron Lett. 2009, 50, 7124.
(12) Craig, J. T.; Robins, M. D. W. Aust. J. Chem. 1968, 21, 2237.
(13) Davy, J. R.; Iskander, M. N.; Reiss, J. A. Aust. J. Chem. 1979, 32,
1067.
(14) (a) Scott, L. T.; Hashemi, M. M.; Meyer, D. T.; Warren, H. B.
J. Am. Chem. Soc. 1991, 113, 7082. (b) Scott, L. T.; Cheng, P.-C.;
Hashemi, M. M.; Bratcher, M. S.; Warren, H. B. J. Am. Chem. Soc.
1997, 119, 10963. (c) Scott, L. T.; Hashemi, M. M.; Bratcher, M. S.
J. Am. Chem. Soc. 1992, 114, 1920.
(15) (a) Brown, R. F. C.; Harrington, K. J.; McMullen, G. L. J. Chem.
Soc., Chem. Commun. 1974, 123. (b) Brown, R. F. C.; Eastwood, F. W.;
Jackman, G. P. Aust. J. Chem. 1977, 30, 1757. (c) Brown, R. F. C.;
Eastwood, F. W.; Jackman, G. P. Aust. J. Chem. 1978, 31, 579.
(d) Brown, R. F. C.; Eastwood, F. W.; Harrington, K. J.; McMullen,
G. L. Aust. J. Chem. 1974, 27, 2393.
through a glass funnel to remove the precipitate. The filtrate
was transferred back to the reactor, and the organic phase was
washed one last time with 15 kg water (IT = 86−90 °C); then
the organic layer was distilled (24 kg) at an internal tem-
perature of 87−93 °C under reduced pressure (pstart = 800 mbar,
pend = 530 mbar) until the product started to crystallize. The
mixture was heated to reflux, and additional toluene (3.5 kg) was
added. A clear orange-brown solution was obtained. The solution
was cooled using a ramp (jacket temperature = 100 °C to jacket
temperature = 70 °C) during 3 h and then cooled to jacket
temperature = 15 °C and stirred for 15 h. The precipitate was
filtered, and the filtrand was washed with toluene (0.16 kg) and
then dried at 40 °C under vacuum for 21 h to yield 1.33 kg pale-
yellow solid (HPLC purity 98% corannulene (1) [1.30 kg, 88%
HPLC corrected yield]). Mp 268−269 °C, 1H NMR (500 MHz,
CDCl3): δ 7.82 (s, 10H). 13C NMR (100 MHz, CDCl3): δ 127.2,
130.8, 135.8.
(16) Buu-Hoi; Cagniant, P. Rev. Sci. 1942, 80, 130.
(17) Wolinska-Mocydlarz, J.; Canonne, P.; Leitch, L. C. Synthesis
1974, 556.
ASSOCIATED CONTENT
■
(18) Borchardt, A.; Hardcastle, K.; Gantzel, P.; Siegel, J. S.
Tetrahedron Lett. 1993, 34, 273.
(19) (a) Seiders, T. J.; Elliot, E. L.; Grube, G. H.; Siegel, J. S. J. Am.
Chem. Soc. 1999, 121, 7804. (b) Sygula, A.; Rabideau, P. W. J. Am.
Chem. Soc. 1998, 120, 12666.
S
* Supporting Information
HPLC methods. This material is available free of charge via the
Internet at http://pubs.acs.org.
(20) Sygula, A.; Rabideau, P. W. Tetrahedron 2001, 57, 3637.
(21) Tilstam, U.; Weinmann, H. Org. Process Res. Dev. 2002, 6, 906.
(22) Reaction and workup quantities from 2006 were calculated
using reported literature. Purification values, when listed, were
determined using the assumption that 50−120 g silica gel were
needed to purify 1 g of product, depending on the difficulty of
separation. Yield (2006) is uncorrected; yield (2011) is corrected
based on assay.
(23) Based on dimethoxybutanone (assay 95%) as the limiting
reagent.
(24) E-factor determined by kg waste/kg product. “Waste” included
water used and did not assume solvents were recycled.
(25) Elliot, E. L. Ph.D. Dissertation, UCSD: San Diego, CA, 2003.
(26) (a) Varvoglis, G.; Alexandrou, N. Xhmika Xponika 1961, 26,
AUTHOR INFORMATION
■
Corresponding Author
jss@oci.uzh.ch
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was supported by the Swiss National Foundation.
We thank Rolf Sigrist for scouting of conditions toward the
formation of 5. We thank Dr. Thomas Bader of the LPF for
stimulating discussions and critical reading of the manusript.
137. (b) Anthes, E.; Scholler, M. Ber. Dtsch. Chem. Ges. 1912, 1594.
̈
REFERENCES
■
(27) Price per mole was based on the current value from Sigma-
Aldrich at the time of production for comparable qualities and 100 g.
(28) Lolin, Y. Hum. Toxicol. 1989, 4, 293.
(29) HPLC analyses were performed by quenching the reaction
samples with ethanol to form the corresponding esters of 5c, 5d, and
5e.
(30) Extremely careful recrystallization from ethyl acetate allowed
separation, but the conditions were not robust enough for a reliable
scale-up.
(1) (a) Barth, W. E.; Lawton, R. G. J. Am. Chem. Soc. 1966, 88, 380.
(b) Lawton, R. G.; Barth, W. E. J. Am. Chem. Soc. 1971, 93, 1730.
(2) (a) Hirsch, A.; Brettreich, M. Fullerenes: Chemistry and Reactions;
Wiley-VCH: Weinheim, 2005. (b) Wu, Y.-T.; Siegel, J. S. Chem. Rev.
2006, 106, 4843. (c) Baldridge, K. K.; Hardcastle, K. I.; Seiders, T. J.;
Siegel, J. S. Org. Biomol. Chem. 2010, 8, 53.
(3) (a) Cheng, P.-C. Ph.D. Dissertation, Boston College: Boston,
MA, 1996. (b) Wu, Y.-T.; Maag, R.; Linden, A.; Baldridge, K. K.;
Siegel, J. S. J. Am. Chem. Soc. 2008, 130, 10729. (c) Seiders, T. J.;
Baldridge, K. K.; Grube, G. H.; Siegel, J. S. J. Am. Chem. Soc. 2001, 123,
517. (d) Reisch, H. A.; Bratcher, M. S.; Scott, L. T. Org. Lett. 2000, 2,
1427. (e) Scott, L. T. Pure Appl. Chem. 1996, 68, 291.
(4) (a) Scanlon, L. G.; Balbuena, P. B.; Zhang, Y.; Sandi, G.; Back, C.
K.; Feld, W. A.; Mack, J.; Rottmayer, M. A.; Riepenhoff, J. L. J. Phys.
Chem. B 2006, 110, 7688. (b) Hayama, T.; Wu, Y.-T.; Linden, A.;
Baldridge, K. K.; Siegel, J. S. J. Am. Chem. Soc. 2007, 129, 12612.
(5) Miyajima, D.; Tashiro, K.; Araoka, F.; Takezoe, H.; Kim., J.; Kato,
K.; Takata, M.; Aida, T. J. Am. Chem. Soc. 2009, 131, 44.
(6) Pappo, D.; Mejuch, T.; Reany, O.; Solel, E.; Gurram, M.; Keinan,
E. Org. Lett. 2009, 11, 1063.
(31) Burgi, H. B.; Dunitz, J. D.; Shefter, E. J. Am. Chem. Soc. 1973,
̈
95, 5065.
(32) (a) Girard, A.; Sandelescu, G. Helv. Chim. Acta 1936, 19, 1095.
(b) Henbest, H. B. Nature 1946, 158, 950.
(33) Borchardt, A.; Hardcastle, K.; Gantzel, P.; Siegel, J. S.
Tetrahedron Lett. 1993, 34, 273.
(34) The self-condensed product of 3-pentanone was never detected
but could be one reason that excess reagent was required.
(35) Sygula, A.; Rabideau, P. W. J. Am. Chem. Soc. 1999, 121, 7800.
(36) Bretherick, L. Hazards in the Chemical Laboratory; The Royal
Society of Chemistry: London, 1986.
(37) Denisov, E. T.; Denisova, T. G.; Pokidova, T. S. Handbook of
Free Radical Initiators; Wiley & Sons: NJ, 2003.
(7) Quimby, J. Ph.D. Dissertation, Boston College: Boston, MA,
2011.
(38) Sygula, A.; Rabideau, P. W. Tetrahedron 2001, 57, 3637.
(39) Liu, G.-B.; Thiemann, T. J. Hazard. Mater. 2009, 169, 1150.
(40) Anwer, M. K.; Sherman, D. B.; Roney, J. G.; Spatola, A. F. J. Org.
Chem. 1989, 54, 1284.
(41) The importance of fitting the process to the infrastructure at
hand is highly emphasized here. We are certain that in an appropriate
(8) Stuparu, M. C. Chimia 2011, 65, 799.
(9) Sygula, A.; Karlen, S. D.; Sygula, R.; Rabideau, P. W. Org. Lett.
2002, 4, 3135.
(10) (a) Seiders, T. J.; Baldridge, K. K.; Siegel, J. S. J. Am. Chem. Soc.
1996, 118, 2754. (b) Seiders, T. J.; Baldridge, K. K.; Siegel, J. S.
Tetrahedron 2001, 57, 3737.
675
dx.doi.org/10.1021/op200387s | Org. Process Res. Dev. 2012, 16, 664−676

Organic Process Research & Development
Article
facility ammonium formate and a cheaper solvent could be
implemented with great benefit to the overall process.
(42) For reference, the campaign retained an additional 400 g of 9,
2 kg of 7, 2 kg of 5, 2.5 kg of 5b, and 650 g of 4.
(43) Dallaire, C.; Kolber, I.; Gingras, M. Org. Synth. 2000, 78, 42−50.
(44) Moles of starting material also include 6% of di- and
tribromocorannulene that are present by HPLC analysis of the
previous step and should also reduce to 1.
676
dx.doi.org/10.1021/op200387s | Org. Process Res. Dev. 2012, 16, 664−676