Multivariate optimization of the Kharasch-Sosnovsky allylic oxidation of olefins

Article abstract of DOI:10.1016/j.tet.2011.11.083

The multivariate optimization method known as simplex is applied to the Kharasch-Sosnovsky allylic oxidation of double bonds. By applying this method, the amounts of three variables (copper source, oxidant, and additive) are optimized at the same time. Un

Full text of DOI:10.1016/j.tet.2011.11.083

Tetrahedron 68 (2012) 1105e1108
Contents lists available at SciVerse ScienceDirect
Tetrahedron
journal homepage: www.elsevier.com/locate/tet
Multivariate optimization of the KharascheSosnovsky allylic oxidation of olefins
*
ꢀ
Ruben Marín-Barrios, Francisco M. Guerra , Ana Leticia García-Cabeza, F. Javier Moreno-Dorado,
*
Guillermo M. Massanet
ꢀ
ꢀ
ꢀ
Departamento de Química Organica, Facultad de Ciencias, Universidad de Cadiz, Pol. Río San Pedro s/n, 11510 Puerto Real, Cadiz, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 12 October 2011
Received in revised form 18 November 2011
Accepted 25 November 2011
Available online 7 December 2011
The multivariate optimization method known as simplex is applied to the KharascheSosnovsky allylic
oxidation of double bonds. By applying this method, the amounts of three variables (copper source,
oxidant, and additive) are optimized at the same time. Under the conditions thus obtained the reaction
takes place in a considerable shorter time, being the alkene the limiting reagent. These conditions are
applied to some monoterpenes and sesquiterpenes leading regioselectively to the corresponding ben-
zoate esters, opening a route to the employment of this reaction in the synthesis of more complex
molecules.
Keywords:
Allylic oxidation
Olefins
Ó 2011 Elsevier Ltd. All rights reserved.
KharascheSosnovsky reaction
Simplex
Multivariate optimization
1. Introduction
Nevertheless, some items remain unsolved: (i) most of the reported
conditions requires the use of an excess of the alkene to be oxidized
A great interest in the development of new procedures aimed at
the allylic oxidation of alkenes has arose in the synthetic community
as a mean to increase the molecular complexity.1,2 This interest re-
lies on the possibility to oxidize a saturated carbon located next to
a double bond, enabling the access to remote positions located apart
from the functional groups of the starting material. Nevertheless,
despite the many procedures described in the literature, we still lack
a general and reliable method to perform this transformation.
The KharascheSosnovsky reaction3 is an interesting option to
perform this type of transformation. This reaction involves the oxi-
dation of the allylic position of an alkene by a perester in the pres-
ence of a copper or cobalt source, providing an allylic ester (Fig. 1).
(typically 10 equiv), a fact unacceptable when the alkene is valuable,
(ii) very long reaction times (from hours to weeks) are required in
order to get good conversions and yields, (iii) the reaction has been
rarely employed beyond methodological purposes and it has seldom
been considered as a tool in complex syntheses.4e6 Its use has been
restricted to the oxidation of poorly functionalized alkenes. On the
positive side, the KharascheSosnovsky reaction can be carried out in
an enantioselective manner, and excellent enantiomeric excesses
have been reported.7e13
Taking into account the potential of the KharascheSosnovsky
reaction and the benefits that the solving of the aforementioned
problems would provide, we have undertaken the task of developing
new reaction conditions that permits this reaction to be included in
the selected pool of reactions commonly employed in the synthesis
of complex natural products. Three main goals are sought in this
work: (i) the discovery of conditions in which the alkene is the
limiting reagent, (ii) the use of inexpensive conditions with regard to
the copper source and the oxidant and the development of the
simplest technical conditions, and (iii) the widening of the current
scope in terms of the complexity of the substrate.
Fig. 1. Original KharascheSosnovsky reaction.
This reaction has been thoroughly studied by different authors,
andaplethoraof conditions havebeendescribedinwhichthereaction
proceeds mostly in moderate to good yields and enantioselectivities.
2. Results and discussion
As it is usual in this type of study we first approached the
problem by keeping all variables but one constant and studying the
effect of the remaining variable, following a classical univariate
approach. We soon discovered that contradictory results were
* Corresponding authors. Tel.: þ34 956 016405; fax: þ34 956 016393; e-mail
addresses: francisco.guerra@uca.es (F.M. Guerra), g.martinez@uca.es (G.M.
Massanet).
0040-4020/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tet.2011.11.083

1106
R. Marín-Barrios et al. / Tetrahedron 68 (2012) 1105e1108
obtained, which led to us assuming that a multivariate analysis
optimizing several variables at once would lead to better results as
we will explain in detail below.
Not only DBU but also other bases, such as pyrrolidine or proline
provided the corresponding esters in excellent yields. The results
suggest that better s
-donors increase the reaction rate.23 Although
these reactions produced high yields, they required 48 h. The yield
at 24 h using DBU as an additive was only 64%. It was clear that
further optimization was needed if we wanted to shorten the re-
action time.
2.1. Univariate approach: variable to variable optimization
For the sake of simplicity, cyclohexene was selected as an olefin
model and tert-butyl peroxybenzoate (TBPB) as an oxidant agent.
The reactions were run for 24 or 48 h at room temperature.
An essential requisite was to employ only 1 equiv of the olefin,
aiming at the application to worthy substrates. The first variable
under study was the copper source. A survey of the literature dis-
plays that a wide range of copper sources can be employed14,15 and
that those more suitable for the enantioselective version of the
reaction are air sensitive and expensive. We decided to start reas-
sessing the performance of the most popular copper salts (Table 1),
using DBU since Muzart and Le Bras demonstrated that this addi-
tive increases significantly the reaction rate.16 The initial results
were rather disappointing, leading to low yields. Nevertheless,
since CuCl resulted to provide the highest yields, we decided to
continue our investigation using this salt.
Furthermore, the question of whether we were using the opti-
mal amounts of reagents or further adjustments were still needed
remained unsolved. The influence of a substantial lowering either
of the amounts of copper, DBU or both had to be investigated.
Consequently, we decided to perform an optimization of the
stoichiometry of the reaction. Several concentrations of CuCl and
DBU were examined. The results are shown in Table 3. The in-
spection of this table shows that a minimum amount of copper and
DBU is needed although a drop of the yield is observed for amounts
higher than 20 mol % employing only 1 equiv of TBPB. A preferential
formation of different multinuclear copper species24 under these
conditions could explain this outcome.
Table 3
Optimization of the catalytic charge
Table 1
Optimization of the copper source
Catalytic charge (mol %)
CuCl
Yielda (%)
DBU
Copper source
Yielda (%)
10
20
20
30
40
70
100
10
20
40
30
20
22
64
49
45
37
31
17
Cu(CH3CN)4(OTf)
Cu(CH3CN)4(PF6)
Cu(CH3CN)4(BF4)
Cu(OTf)1/2PhH
Cu(OTf)2
5
2
16
10
9
70
100
CuCl
CuCl2
23
19b
a
Determined by GC analysis.
a
Determined by GC analysis.
4 days of reaction.
b
2.2. Multivariate approach: the simplex method
It is also described in the literature that the presence of amines
other than DBU, used as additives, can accelerate notably the re-
action rate.17e22 According to Singh’s group previous reports,17
phenylhydrazine was also tested, however this additive showed
lower yields than DBU. We checked the effect of the addition of
different nitrogen additives to the reaction. The results are dis-
played in Table 2.
At this point, given the uncertainty that we were using reagents
and reactants in the correct proportions, we decided to use a dif-
ferent approach. This is not so simple as many issues are involved in
the outcome of this reaction. It is noteworthy to point out that the
mechanism of the KharascheSosnovsky oxidation is not fully
known in detail. It is a complex mechanism involving free radical
generation, coordination to metallic species in different oxidation
states, and a concerted rearrangement. Recently Mayoral and co-
workers25 have proposed some changes to the mechanism pre-
viously reported by Zavitsas.26 An additional issue is the need to
adjust the addition of the catalyst and the additive with the ho-
molysis of the oxidant agent for an optimal performance of the
catalytic cycle.
Table 2
Additive optimization
Taking into account these considerations, we carried out the
optimization of three variables (amount of copper, oxidant, and
additive) employing the mathematical method for multivariate
optimization known as simplex, previously used by Muzart et al. in
an asymmetric version.27,28
The use of the simplex method allows the optimization of the
yield of a reaction by the simultaneous modification of all variables.
A simplex is an n-dimensional figure with nþ1 vertices, where n is
the number of variables to optimize. In the case of two variables it
is a triangle; a tetrahedron in the case of three variables. Each
vertex represents different reaction conditions. In the case of two
variables, the simplex starts with three initial reaction conditions
Additive (20 mol %)
Yielda (%)
Additive (20 mol %)
Yielda (%)
DBU
PhNHNH2
Pyrrolidine
100
33b
100
80
67
7
25
10
9
DBN
DBO
47
1
24
1
17
10
5
lmidazol
Hexamethylenetetramine
DMAP
2-(Methylamino)pyridine
Colidine
Diphenylguanidine
L-proline
Et3N
DIPEA
iPr2NH
D
-Valine
16
L-Tyrosine
a
Determined by GC analysis.
Yield (44%) in acetone.
b

R. Marín-Barrios et al. / Tetrahedron 68 (2012) 1105e1108
1107
A, B, and C, forming a triangle (Fig. 2). Once the three reactions
corresponding to such conditions have been carried out, the yields
are evaluated. The experiment that leads to the lowest yield is the
discarded vertex and new conditions are calculated by a reflection
operation using the formula Vnew¼2MꢀVdiscarded, where V stands
for vertex and M is the average of the vertices that are kept. This
process is repeated iteratively ruling out those conditions that
provide the worst yields. Thus we shall achieve those reaction
conditions that lead to the best results. The simplex method also
provides the ability to go faster in the optimization by other op-
erations, such as expansion or contraction of the figure when the
worst experiment fulfills certain conditions.29 In this case it is
called modified simplex method. The simplex method finishes
when the reaction yield does not improve further.
Fig. 3. Simplex optimization of cyclohexene.
reactions resulted to be regioselective and, in those cases in which
the substrate is chiral, stereoselective, only a single diastereomer
being isolated.
The results for the employed substrates are summarized in
Table 5.30
Table 5
Optimized conditions for the oxidation of other substrates
Substrate
Product
Yielda (TBPB/CuCl/DBU equiv)b
100 (3.54:0.49:0.29)
Fig. 2. Simplex optimization of a 2-variable system. The figures in the circles corre-
spond to yields.
45 (4.17:0.47:0.59)
46 (3.51:0.22:0.23)
The optimization of the three above-mentioned variables (TBPB,
DBU, and CuCl equivalents) implied that the simplex was a tetra-
hedron. The conditions for the first four trials were set as shown in
Table 4. After 13 experiments, a 100% yield for the allylic oxidation
of cyclohexene is achieved using 1 equiv of cyclohexene, 3.54 equiv
TBPB, 0.49 equiv of CuCl, and 0.29 equiv of DBU, in only 24 h at
room temperature. Fig. 3 displays the simplex optimization process.
Table 4
Conditions and results for the simplex optimization in the case of cyclohexene
56 (3.92:0.14:0.36)
55c (1.81:0.27:0.38)
Vertex
TBPBa
CuCla
DBUa
Yieldb (%)
1
2
3
4
5
6
7
8
1.50
2.00
1.75
1.75
2.17
2.50
2.42
2.44
2.67
1.75
3.86
3.45
3.54
0.20
0.20
0.29
0.23
0.28
0.32
0.31
0.41
0.51
0.29
0.45
0.48
0.49
0.20
0.20
0.20
0.28
0.25
0.28
0.17
0.24
0.25
0.20
0.29
0.37
0.29
21
26
27
23
44
53
42
70
64
86
92
72
100
a
Determined by GC analysis after 1 day of reaction. All values are the media of
three runs.
b
c
9
Equivalents of each reagent, 1 equiv of starting material.
75% based on recovered starting material.
10
11
12
13
Although the yields are moderated, it is noteworthy to mention
that the reaction time has been reduced compared to what is usually
reported in the literature and that the yields are based on the alkene
and not in the oxidant. Furthermore, our conditions employ in-
expensive copper chloride at room temperature. No exclusion of
moisture or drying of solvents is needed. The complexity level of the
substrates employed expands the scope of the KharascheSosnovsky
reaction beyond the typical cycloalkenes described in the literature.
All the reactions were run for 24 h. The bold means that the best result (100% yield).
a
Equivalents of each reagent.
Determined by GC analysis.
b
2.3. Application of the simplex method to other substrates
Other more complex olefins were tested under the conditions
found for cyclohexene. The yields resulted to be lower than those
found in cyclohexene. Therefore, the reactions were subjected to
individual optimizations following the procedure developed for
cyclohexene. In all cases, the reaction time was limited to 24 h since
no improvement was observed by increasing the time. The
3. Conclusions
We have demonstrated that it is possible to carry out the effi-
cient allylic oxidation of double bonds present not only in simple

1108
R. Marín-Barrios et al. / Tetrahedron 68 (2012) 1105e1108
cycloalkenes, but also in monoterpenes and sesquiterpenes, under
mild conditions, employing inexpensive CuCl, and TBPB as an oxi-
dant in the presence of DBU. The conditions have been optimized
by employing the multivariate optimization method called simplex.
The limiting reagent is now the alkene, a fact that guarantees its
applicability in worthy substrates. In addition, the reaction time has
been considerably reduced, taking only 24 h to provide useful
yields.
(m, 2H), 5.66 (dq, J¼4.4, 1.4 Hz, 1H), 4.70 (q, J¼1.6 Hz, 2H), 2.43 (dt,
J¼8.7, 5.6 Hz, 1H), 2.36e2.32 (m, 1H), 2.30 (ddd, J¼4.4, 3.0, 1.6 Hz,
1H), 2.22 (td, J¼5.6, 1.5 Hz, 1H), 2.13 (dtd, J¼8.7, 2.9, 1.4 Hz, 1H), 1.30
(s, 3H), 1.24 (d, J¼8.7 Hz, 1H), 0.88 (s, 3H). 13C NMR (101 MHz,
CDCl3):
d 166.39, 142.96, 132.75, 130.41, 129.50, 128.27, 121.59,
67.45, 43.63, 40.71, 38.07, 31.46, 31.25, 26.11, 21.11. IR (film) nmax
(cmꢀ1): 3034, 2832, 1719, 1451, 1366, 1270, 1176, 1111, 1070, 1026,
800, 711. ½a 2D0
ꢂ
ꢀ28.7 (c 0.072, CHCl3). HR-MS (CIþ): calcd for
Although further work is required, the application to substrates
other than simple cycloalkenes opens the way to the popularization
of this reaction in the field of organic synthesis.
C17H20O2 256.1463, found 256.1461.
4.2.4. Benzoyloxyvalencene (10). Yellow oil. 1H NMR (400 MHz,
CDCl3):
d
8.06 (dd, J¼8.4, 1.4 Hz, 2H), 7.56e7.51 (m, 1H), 7.43 (t,
4. Experimental section
J¼7.5 Hz, 2H), 5.59 (d, J¼5.1 Hz, 1H), 5.38 (ddd, J¼7.0, 4.0, 2.0 Hz,
1H), 4.72e4.71 (m, 2H), 2.37 (tdt, J¼14.1, 4.5, 2.1 Hz, 1H), 2.28 (tt,
J¼12.6, 3.1 Hz, 1H), 2.20 (ddd, J¼14.3, 4.2, 2.6 Hz, 1H), 1.94 (dt,
J¼12.9, 2.8 Hz, 1H), 1.86e1.80 (m, 2H), 1.74 (t, J¼1.0 Hz, 3H),
1.38e1.03 (m, 4H), 0.96 (s, 3H), 0.93 (d, J¼6.6 Hz, 3H). 13C NMR
4.1. General considerations
Reactions were monitored through TLC on commercial silica gel
plates. Visualization of the developed plate was performed by
fluorescence quenching and/or aqueous ceric ammonium molyb-
date/anisaldehyde stains. GC was performed on a PerkineElmer
Clarus GC400 using a PerkineElmer Elite5 column. The commer-
cially available reagents and solvents were used without further
purification. NMR spectra were recorded on a Varian Inova 400. IR
spectra were recorded in a PerkineElmer Spectrum BX2, using NaCl
plates, data are reported in cmꢀ1. Mass spectra were obtained in
a VG Autospec-Q.
(101 MHz, CDCl3):
d 166.25, 150.94, 150.04, 132.59, 131.00, 129.56,
128.18, 117.67, 108.61, 68.45, 44.42, 40.53, 38.15, 35.80, 33.04, 32.44,
32.40, 20.85, 16.85, 15.08. IR (film) nmax (cmꢀ1): 2926, 1718, 1450,
1313, 1269, 1175, 1109, 888, 712. ½a D20
þ148 (c 0.043, CHCl3). HR-MS
ꢂ
(CIþ): calcd for C22H28O2 324.2089, found 324.2085.
Acknowledgements
We are grateful to Junta de Andalucía for the financial support.
ꢀ
R.M.B. acknowledges the Universidad de Cadiz for a fellowship. We
4.2. General procedure for allylic oxidation
ꢀ
thank Dr. Jose Ma Palacios for his guidance with multivariate
optimization, Dr. Juan Antonio Poce for his simplex software, and
Ms. Lindsey Ray for her helpful assistance.
A solution of the copper source in 4 mL of acetonitrile is stirred
and the amino additive was added. Then, the substrate (1 mmol)
and the oxidant agent, TBPB, are added. The yield is determined by
GC analysis using octadecane as internal standard. After 24 h the
reaction is worked up by adding 15 mL of AcOEt and extracted with
10 mL of saturated NH4Cl (ꢁ2). The aqueous layer is then extracted
with 10 mL of dichloromethane (ꢁ2). The organic layers are com-
bined and dried over anhydrous Na2SO4. The solvent is removed
under reduced pressure. The yellow oily crude is purified by column
chromatography on silica gel and semipreparative HPLC.
References and notes
1. Andrus, M. B. In Science of Synthesis, Stereoselective Synthesis Allylic and Benzylic
Oxidation; Georg Thieme Verlag: Stuttgart, Germany, 2011; Vol 3, pp 469e482.
2. Harschneck, T.; Kirsch, S. F. Nachr. Chem. 2010, 58, 36e39.
3. Kharasch, M. S.; Fono, A. J. Org. Chem. 1958, 23, 324e325.
4. Andrus, M. B.; Lashley, J. C. Tetrahedron 2002, 58, 845e866.
5. Eames, J.; Watkinson, W. Angew. Chem., Int. Ed. 2001, 40, 3567e3571.
6. Andrus, M. B.; Zhou, Z. J. Am. Chem. Soc. 2002, 124, 8806e8807.
7. Gokhale, A. S.; Minidis, A. B. E.; Pfaltz, A. Tetrahedron Lett. 1995, 36, 1831e1834.
8. Andrus, M. B.; Argade, A. B.; Chen, X.; Pamment, M. G. Tetrahedron Lett. 1995, 36,
2945e2948.
9. Kawasaki, K.; Tsumura, S.; Katsuki, T. Synlett 1995, 1245e1246.
10. DattaGupta, A.; Singh, V. K. Tetrahedron Lett. 1996, 37, 2633e2636.
11. Ginotra, S. K.; Singh, V. K. Org. Biomol. Chem. 2006, 4, 4370e4374.
12. Andrus, M. B.; Asgari, D. Tetrahedron 2000, 56, 5775e5780.
13. Zhou, J.; Tang, Y. Chem. Soc. Rev. 2005, 34, 664e676.
14. Andrus, M. B.; Chen, X. Tetrahedron 1997, 53, 16229e16240.
15. Tan, Q.; Hayashi, M. Adv. Synth. Catal. 2008, 350, 2639e2644.
16. Le Bras, J.; Muzart, J. J. Mol. Catal. A: Chem. 2002, 185, 113e117.
17. Sekar, G.; DattaGupta, A.; Singh, V. K. J. Org. Chem. 1998, 63, 2961e2967.
18. Malkov, A. V.; Bella, M.; Langer, V.; Kocovsky, P. Org. Lett. 2000, 2, 3047e3049.
19. Lee, W. S.; Kwong, H. L.; Chan, H. L.; Choi, W. W.; Ng, L. Y. Tetrahedron: Asym-
metry 2001, 12, 1007e1013.
4.2.1. 3-Phenylcyclohex-2-enyl benzoate (4). Yellow oil. 1H NMR
(400 MHz, CDCl3):
d
8.00 (dd, J¼8.4, 1.3 Hz, 2H), 7.50e7.44 (m, 1H),
7.39e7.35 (m, 2H), 7.35e7.17 (m, 5H), 6.15 (dt, J¼3.6, 1.7 Hz, 1H),
5.69e5.60 (m, 1H), 2.52 (dddd, J¼6.0, 4.8, 3.5, 1.8 Hz, 1H), 2.36
(dtdd, J¼9.6, 7.5, 3.5, 1.8 Hz, 1H), 2.02e1.77 (m, 4H). 13C NMR
(101 MHz, CDCl3):
d 166.28, 142.27, 141.05, 132.76, 130.73, 129.60,
128.30, 128.25, 127.65, 125.43, 122.30, 69.47, 28.12, 27.41, 19.50. IR
(film) nmax (cmꢀ1): 2938, 1712, 1450, 1316, 1270, 1176, 1111, 1069,
1026, 915, 758, 712. HR-MS (CIþ), calcd for C19H18O2 278.1307,
found 273.1301.
4.2.2. Benzoyloxylimonene (6). Colorless oil. 1H NMR (400 MHz,
20. Bolm, C.; Frison, J. C.; Le Paih, J.; Moessner, C. Tetrahedron Lett. 2004, 45,
5019e5021.
CDCl3):
d
8.08 (dd, J¼8.4, 1.4 Hz, 2H), 7.58e7.53 (m, 1H), 7.47e7.42
21. Ramalingan, B.; Neuburger, M.; Pfaltz, A. Synthesis 2007, 572e582.
22. Hoang, V. D. M.; Reddy, P. A. N.; Kim, T. J. Organometallics 2008, 27, 1026e1027.
23. Sekar, G.; DattaGupta, A.; Singh, V. K. Tetrahedron Lett. 1996, 37, 8435e8436.
24. Le Bras, J.; Muzart, J. Tetrahedron: Asymmetry 2003, 14, 1911e1915.
25. Mayoral, J. A.; Rodríguez-Rodríguez, S.; Salvatella, L. Chem.dEur. J. 2008, 14,
9274e9285.
(m, 2H), 5.86e5.75 (m, 1H), 5.59e5.50 (m, 1H), 4.80e4.68 (m, 2H),
2.49e2.37 (m, 1H), 2.31e2.27 (m, 1H), 2.27e2.22 (m, 1H), 2.11 (ddd,
J¼14.2, 4.1, 2.4 Hz, 1H), 1.99e1.87 (m, 1H), 1.78e1.75 (m, 3H),
1.75e1.73 (m, 3H). 13C NMR (101 MHz, CDCl3):
d 166.30, 148.58,
26. Beckwith, A. L. J.; Zavitsas, A. A. J. Am. Chem. Soc. 1986, 108, 8230e8234.
27. Betteridge, D.; Wade, A. P.; Howard, A. G. Talanta 1985, 32, 709e722.
28. Levina, A.; Muzart, J. Tetrahedron: Asymmetry 1995, 6, 147e156.
29. For a more detailed explanation of the modified simplex method, applied to an
organic reaction, see: Chubb, F. L.; Edward, J. T.; Wong, S. C. J. Org. Chem. 1980,
45, 2315e2320.
132.75, 131.03, 130.66, 129.57, 128.26, 127.88, 109.20, 71.19, 35.99,
33.76, 30.92, 20.85, 20.67. IR (film) nmax (cmꢀ1): 2932, 1714, 1450,
1314, 1268, 1110, 1025, 711. ½a D20
ꢂ
ꢀ8.6 (c 0.025, CHCl3). HR-MS (CIþ):
calcd for C17H20O2 256.1463, found 256.1458.
30. Isolated yields for (R)-limonene (5) and (S)-b-pinene (7) were only around 30%
4.2.3. Benzoyloxypinene (8). Colorless oil. 1H NMR (400 MHz,
due to volatility issues. In the case of valencene (9), a 75% of isolated yield was
CDCl3):
d
8.04 (dd, J¼8.4, 1.4 Hz, 2H), 7.57e7.52 (m, 1H), 7.46e7.40
obtained after recovery of unreacted starting material.