Recycling the waste: The development of a catalytic wittig reaction

Full text of DOI:10.1002/anie.200902525

Communications
DOI: 10.1002/anie.200902525
Olefination
Recycling the Waste: The Development of a Catalytic Wittig
Reaction**
Christopher J. OꢀBrien,* Jennifer L. Tellez, Zachary S. Nixon, Lauren J. Kang, Andra L. Carter,
Stephen R. Kunkel, Katherine C. Przeworski, and Gregory A. Chass*
Dedicated to Avner and Marie OꢀBrien (nꢁe Yang)
The formation of carbon–carbon double bonds is among a
select group of key transformations on which much synthetic
chemistry is based. This is not surprising, as the fabrication of
many natural products and drugs necessitates their assembly
via alkenes.[1] Accordingly, numerous processes for their
construction have been developed; besides direct elimina-
tion[2] there are four widely employed methodologies for the
routine and reliable formation of alkenes:[3] 1) the Wittig
reaction,[4] 2) the Peterson reaction,[5] 3) the Julia–Lythgoe[6]/
Julia–Kocienski[7] olefination reactions, and 4) metathesis.[8]
Of the three stoichiometric olefination processes discussed,
one that may offer the possibility to evolve to a catalytic
process, coupled with selective formation of E or Z alkenes, is
the Wittig reaction.[4] Discovered in 1953 by Georg Wittig, the
reaction involves treatment of an aldehyde or ketone with a
phosphonium ylide,[4] yielding an alkene with the concomitant
generation of phosphine oxide. Since its discovery the Wittig
reaction has been used extensively in organic chemistry.[9]
Nevertheless, the Wittig reaction suffers from several limi-
tations: The current process is stoichiometric, and complete
removal of the phosphine oxide byproduct is not always
straightforward. The lack of a catalytic protocol, due to cost/
benefit ratio, removes from serious consideration the possi-
bility to control the olefination by alteration of phosphine
structure This is unfortunate, as the structure of the phosphine
has been shown to have a substantial impact on the
stereochemical outcome of the reaction. Therefore a carefully
designed phosphine may result in a selective process.[10]
Yet, the barriers to the development of a catalytic Wittig
reaction are formidable, and the successful construction of a
catalytic process relies on the completion of four steps
(Scheme 1): 1) formation of the phosphonium ylide precur-
Scheme 1. Proposed catalytic Wittig reaction.
sor, typically a phosphonium salt;[9,10] 2) generation of the
phosphonium ylide, normally by deprotonation;[9,10] 3) olefi-
nation with concomitant generation of phosphine oxide;[9,10]
and 4) reduction of the phosphine oxide byproduct producing
phosphine to re-enter the catalytic cycle (Scheme 1). The
most daunting aspect of the above processes is the requisite
chemoselective reduction of the phosphine oxide byproduct
to a phosphine in the presence of either aldehyde or ketone
starting materials and alkene product. One could ameliorate
this problem of chemoselective reduction by the replacement
of phosphorus with arsenic,[11] tellurium,[12] or antimony,[13] as
their corresponding oxides, owing to bond strength, are
appreciably easier to reduce.[14] In fact, such an approach has
led to the successful development of catalytic Wittig-type
processes employing arsines and tellurides.[15] Unfortunately,
significant drawbacks to the broad adoption of the afore-
mentioned methodologies are the intrinsic high toxicity and
carcinogenicity of arsenic,[16] tellurium,[17] and antimony
compounds;[18] environmental anthropogenic contamination
particularly of groundwater would be one concern if these
reactions were performed on large-scale.[19] Importantly, the
catalytic use of phosphine would not suffer from these issues;
therefore a Wittig reaction catalytic in phosphine would find
wider employment. Furthermore, this would marry the power
of the Wittig olefination protocol to the synthetic benefits of a
catalytic reaction without the poisoning issues that can plague
transition-metal-catalyzed processes.[10f] Hence, the successful
[*] Dr. C. J. O’Brien, J. L. Tellez, Z. S. Nixon, L. J. Kang, A. L. Carter,
S. R. Kunkel, K. C. Przeworski
Department of Chemistry and Biochemistry
The University of Texas at Arlington
Box 19065, Arlington, TX 76019 (USA)
Fax: (+1)817-272-3808
E-mail: cobrien@uta.edu
Dr. G. A. Chass[+]
School of Chemistry, University of Wales
Bangor, Wales, LL57 2UW (UK)
[+] Corresponding author responsible for computational work.
[**] We thank The University of Texas at Arlington (UTA) for funding this
work and ThalesNano (Hungary) for their collaborative in-kind
donation (in part) of an H-Cube Midi used to synthesize 1. C.J.O.B.
thanks Prof. Daniel W. Armstrong and Edra Dodbiba for separation
of 1. G.A.C. thanks GIOCOMMS for computational resources and
CAFMaD (Wales (UK)) for personal support.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200902525.
6836
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 6836 –6839

Angewandte
Chemie
Table 1: Effect of the silane on the catalytic Wittig reaction.[a]
development of a stereoselective Wittig reaction that is
catalytic in phosphine would represent a significant advance.
The first step toward this aim would be the construction of
a phosphine-catalyzed protocol. Initial studies focused on the
systematic evaluation of each step in the proposed catalytic
Wittig cycle (Scheme 1) and utilized phosphine oxides rather
than phosphines to remove ambiguity in the interpretation of
results. The generation of product would demonstrate the
completion of the steps necessary to achieve a catalytic cycle.
This required the first step to be the reduction of phosphine
oxide to phosphine. There are few reliable methods to reduce
a phosphine oxide;[20–22] most either use harsh reducing agents,
such as lithium aluminum hydride[21] or trichlorosilane, with
or without an amine base.[20] However, we felt that neither of
the aforementioned reduction protocols would be compatible
with the overall catalytic process.
Entry
Silane
Yield [%][b]
E/Z[c]
1
2
3
4
Ph3SiH
Ph2SiH2
PhSiH3
trace
75
46
n.d.
>95:5
>95:5
70:30
(MeO)3SiH
61
[a] Benzaldehyde (1.0 mmol), methyl bromoacetate (1.3 mmol), silane
(1.1 mmol), 3.0m in toluene. [b] Yields were determined by GC/MS/MS
against a calibrated internal standard (undecane) and were performed in
duplicate. [c] E/Z ratio was determined by GC/MS/MS. n.d.=not
determined.
We were also not willing to risk inversion of the
phosphorus center, as reduction with loss of stereointegrity
would significantly detract from the realization of a catalyzed
Wittig protocol with stereocontrol. From the literature we
highlighted two possible reducing agents, diphenylsilane and
phenylsilane, both known to reduce phosphine oxides with
retention of configuration.[23] It was anticipated that the use of
silanes would allow chemoselective reduction of the phos-
phine oxide in the presence of an aldehyde or ketone; it was
assumed that hydrosilylation would not occur, as this
normally requires a transition-metal catalyst.[24] In order to
ensure a reasonable rate of phosphine oxide reduction, we
decided to employ 3-methyl-1-phenylphospholane-1-oxide (1,
Scheme 2) rather than triphenylphosphine oxide, as the
former is more easily reduced.[25]
Evaluation of solvent and temperature (Table 2) revealed
that altering the solvent and temperature produced little
variation in yield. However, reducing the temperature to
808C (cf. Table 2, entries 3 and 4) led to significant amounts of
Table 2: Effects of solvent and temperature on the catalytic Wittig
reaction.[a]
Entry
Solvent
T [8C]
Yield [%][b]
E/Z[c]
1
2
3
4
5
6
DME
100
100
100
80
70
60
50
52
60
62
49
33
>95:5
>95:5
>95:5[d]
67:33[d]
67:33
CH3CN
toluene
toluene
toluene
toluene
67:33
[a] Benzaldehyde (1.0 mmol), methyl bromoacetate (1.1 mmol), diphe-
nylsilane (1.1 mmol), 3.0m in requisite solvent. [b] Yields were deter-
mined by GC/MS/MS against a calibrated internal standard (undecane)
and were performed in duplicate. [c] E/Z ratio was determined by GC/
MS/MS. [d] Identical results were obtained with diastereomerically pure
1 (major diastereomer), separated by HPLC; courtesy of Edra Dodbiba
and Prof. Daniel W. Armstrong, UTA. No isomerization was observed
during the reaction, indicating that reduction proceeded with retention.
Scheme 2. Synthesis of phosphine oxide precatalyst.
After experimentation we discovered that 1 could be
reduced at a reasonable rate (reduction in < 30 min) to
phosphine by diphenylsilane in toluene at 1008C. Following
optimization of reduction conditions we proceeded to con-
struct the catalytic cycle. As shown in Table 1, the treatment
of benzaldehyde with methyl bromoacetate with a 10 mol%
loading of 1 in the presence of diphenylsilane and Na2CO3 at
1008C yielded the desired methyl cinnamate in high yield
(Table 1, entry 2).[26] Importantly, control experiments in the
absence of phosphine oxide yielded no product, also triphe-
nylphosphine oxide was found to be ineffective as a preca-
talyst, and both resulted in the recovery of aldehyde.[27] The
above result represents to the best of our knowledge the first
example of a Wittig reaction catalytic in phosphine. Further
investigation of silane structure revealed that trimethoxysi-
lane also functioned as an effective reducing agent; however,
phenylsilane was less efficient and triphenylsilane was
essentially ineffective (Table 1).
the Z product. This hints at a post-olefination isomerization
event, possibly phosphine-mediated. To provide evidence for
this hypothesis Z-methyl cinnamate was treated with the
phosphine derived from 1 in [D6]benzene at 1008C (sealed
tube). Complete isomerization to the E isomer was observed.
The lack of a selective process when trimethoxysilane was
employed (Table 1, entry 4) may result from insufficient rate
of phosphine formation to effect the complete isomerization
of the product after aldehyde consumption.
In order to understand the differing reactivity of 1 and
triphenylphosphine and to set design rules for protocol
improvement we conducted preliminary theoretical charac-
Angew. Chem. Int. Ed. 2009, 48, 6836 –6839
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
6837

Communications
terizations[28] by a standardized methodology.[29] These calcu-
the utilization of a secondary alkyl bromide that yielded a
trisubstituted alkene in high yield with moderate E/Z control
(product 8 in Scheme 3). The selectivity of the reaction
diminished when benzyl bromides and bromoacetonitrile
were utilized (compare products 3 and 18 to 2 in Scheme 3).
In these cases the phosphine isomerization pathway could be
significantly reduced or shut down; selectivity would then be
governed by the preceding Wittig reaction.[10]
In conclusion, the first Wittig reaction catalytic in
phosphine has been developed. A variety of heteroaryl, aryl,
and alkyl aldehydes could be efficiently converted to the
corresponding alkenes in moderate to high yield with the
utilization of the phosphine oxide precatalyst 1. This protocol
also facilitated the synthesis of practical quantities of alkene,
as 3.39 g of 5 was produced (67% yield) with 4 mol% loading
of 1 (product 5 in Scheme 3, see footnote [c]). These results
permit the development of catalytic phosphine-controlled
stereoselective Wittig reactions with their ensuing synthetic
benefits; such protocols and related processes are being
pursued.
=
À
lations showed near-identical P O/P Ph bond lengths (1.509/
=
1.830 ꢀ, 1.509/1.830 ꢀ, 1.507/1.830 ꢀ, respectively) and O
À
P Ph bond angles (113.5, 112.0, 112.3, respectively); AIM
analyses[28] reiterated this, showing little difference in the
electronic structure of the phosphorus–oxygen bond; bond
critical points were 1b = 0.2127, 1b = 0.2127, 1b = 0.2138,
respectively. This supports the suggestion that 1 is easier to
reduce than triphenylphosphine oxide owing to the relief of
ring strain, optimization of which may lead to more efficient
catalyst recycling.[25]
Following protocol optimization, a substrate evaluation
was undertaken (Scheme 3). Notable results were that
heterocyclic aldehydes could be employed, with 5 and 7
produced in high yield with diastereocontrol (E/Z); also the
synthesis of 10 and 19 from citronellal, and stilbenes 17 and 18
from the corresponding benzyl bromides proceeded in high
yields. As anticipated, the use of methyl chloroacetate did not
affect the yield (68% versus 74%). Of particular interest is
Received: May 12, 2009
Published online: August 17, 2009
Keywords: alkenes · homogeneous catalysis · olefination ·
.
Wittig reactions
[1] a) K. K.-C. Liu, J. Li, S. Sakya, Mini-Rev. Med. Chem. 2004, 4,
1105 – 1125; b) A. Saklani, S. K. Kutty, Drug Discovery Today
2008, 13, 161 – 171; c) K. C. Nicolaou, E. J. Sorensen, Classics in
Total Synthesis, VCH, Weinheim, 1996; d) K. C. Nicolaou, S. A.
Snyder, Classics in Total Synthesis II. More Targets, Strategies,
Methods, Wiley-VCH, Weinheim, 2003.
[2] J. Clayden, N. Greeves, S. Warren, P. Wothers, Organic
Chemistry, Oxford University Press, New York, 2001.
[3] L. Kꢁrti, B. Czakꢂ, Strategic Applications of Named Reactions in
Organic Synthesis, Elsevier Academic Press, San Diego, CA,
2005.
[4] a) G. Wittig, G. Geissler, Liebigs Ann. Chem. 1953, 580, 44 – 57;
b) G. Wittig, U. Schollkopf, Chem. Ber. 1954, 87, 1318 – 1330.
[5] D. J. Peterson, J. Org. Chem. 1968, 33, 780 – 784.
[6] a) M. Julia, J.-M. Paris, Tetrahedron Lett. 1973, 14, 4833 – 4836;
b) P. J. Kocienski, B. Lythgoe, S. Ruston, J. Chem. Soc. Perkin
Trans. 1 1978, 829 – 834.
[7] a) P. R. Blakemore, W. J. Cole, P. J. Kocienski, A. Morley, Synlett
1998, 26 – 28; b) J. B. Baudin, G. Hareau, S. A. Julia, Tetrahedron
Lett. 1991, 32, 1175 – 1178.
[8] a) N. Calderon, H. Y. Chen, K. W. Scott, Tetrahedron Lett. 1967,
8, 3327 – 3329; b) P. Schwab, R. H. Grubbs, J. W. Ziller, J. Am.
Chem. Soc. 1996, 118, 100 – 110; c) M. Scholl, S. Ding, C. W. Lee,
R. H. Grubbs, Org. Lett. 1999, 1, 953 – 956; d) S. B. Garber, S. R.
Kingsbury, B. L. Gray, A. H. Hoveyda, J. Am. Chem. Soc. 2000,
122, 8168 – 8179; e) J. A. Love, J. P. Morgan, T. M. Trnka, R. H.
Grubbs, Angew. Chem. 2002, 114, 4207 – 4209; Angew. Chem.
Int. Ed. 2002, 41, 4035 – 4037; f) J. S. Murdzek, R. R. Schrock,
Organometallics 1987, 6, 1373 – 1374; g) R. R. Schrock, Tetrahe-
dron 1999, 55, 8141 – 8153; h) For a review of metathesis in
synthesis see: K. C. Nicolaou, P. G. Bulger, D. Sarlah, Angew.
Chem. 2005, 117, 4564 – 4601; Angew. Chem. Int. Ed. 2005, 44,
4490 – 4527.
Scheme 3. Substrate study. For each product the compound number,
yield, E/Z ratio, and halide are given. The reactions were performed in
duplicate; see the Supplementary Information for details. [a] When the
reaction was performed with 4 mol% of 1, the yield was 73%.
[b] When the reaction was performed on a 10.67 mmol scale with
10 mol% of 1, the yield was 63%. [c] When the reaction was
performed on a 30 mmol scale with 4 mol% of 1, 908C, 48 h, the yield
was 67%. [d] A single reaction employing 2 equiv (MeO)3SiH resulted
in 85% yield.
[9] a) D. Edmonds, A. Abell in Modern Carbonyl Olefinations (Ed.:
T. Takeda), Wiley-VCH, Weinheim, Germany, 2004, pp. 1 – 17;
b) A. Abell, D. M. K. Edmonds in Organophosphorus Reagents
6838
www.angewandte.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 6836 –6839

Angewandte
Chemie
(Ed.: P. J. Murphy), Oxford University Press, Oxford, UK, 2004,
pp. 99 – 127; c) O. I. Kolodiazhnyi, Phosphorus Ylides: Chemis-
try and Applications in Organic Chemistry, Wiley-VCH, New
York, 1999; d) K. C. Nicolaou, M. W. Hꢃrter, J. L. Gunzner, A.
Nadin, Liebigs Ann. 1997, 1283 – 1301; e) N. J. Lawrence in
Preparation of Alkenes: A Practical Approach (Ed.: J. M. J.
Willians), Oxford Univesity Press, Oxford, UK, 1995; f) B. E.
Maryanoff, A. B. Reitz, Chem. Rev. 1989, 89, 863 – 927.
[10] For leading discussions on mechanisms and selectivity see: a) R.
Robiette, J. Richardson, V. K. Aggarwal, J. N. Harvey, J. Am.
Chem. Soc. 2006, 128, 2394 – 2409; b) R. Robiette, J. Richardson,
V. K. Aggarwal, J. N. Harvey, J. Am. Chem. Soc. 2005, 127,
13468 – 13469; c) V. K. Aggarwal, J. R. Fulton, C. G. Sheldon, J.
de Vicente, J. Am. Chem. Soc. 2003, 125, 6034 – 6035; d) E.
Vedejs, M. J. Peterson in Advances in Carbanion Chemistry,
Vol. 2 (Ed.: V. Snieckus), 1996, pp. 1 – 85; e) E. Vedejs, M. J.
Peterson, Top. Stereochem. 1994, 21, 1 – 157; f) For a review of
asymmetric Wittig-type olefinations see: T. Rein, T. M. Peder-
sen, Synthesis 2002, 579 – 584; g) For stoichiometric recycling of
triphenylphosphine in an industrial setting see: H. Pommer,
Angew. Chem. 1977, 89, 437 – 443; Angew. Chem. Int. Ed. Engl.
1977, 16, 423 – 429.
[17] A. Taylor, Biol. Trace Elem. Res. 1996, 55, 231 – 239.
[18] K. A. Winship, Adverse Drug React. Acute Poisoning Rev. 1987,
6, 67 – 90.
[19] a) P. A. Dasgupta, Talanta 2002, 58, 1 – 2 and the remainder of
this issue dedicated to arsenic; b) H.-Y. Chiou, S.-T. Chiou, Y.-H.
Hsu, Y.-L. Chou, C.-H. Tseng, M.-L. Wei, C.-J. Chen, Am. J.
Epidemiol. 2001, 153, 411 – 418; c) A. H. Welch, D. B. Westjohn,
D. R. Helsel, R. B. Wanty, Ground Water 2000, 38, 589 – 604.
[20] K. Naumann, G. Zon, K. Mislow, J. Am. Chem. Soc. 1969, 91,
7012 – 7023.
[21] T. Imamoto, S. Kikuchi, T. Miura, Y. Wada, Org. Lett. 2000, 3,
87 – 90.
[22] G. Keglevich, A.-C. Gaumont, J.-M. Denis, Heteroat. Chem.
2001, 12, 161 – 167.
[23] F. G. Mann, The Heterocyclic Derivatives of Phosphorus,
Arsenic, Antimony and Bismuth, Vol. 1, 2nd ed., Wiley-Inter-
science, London, 1970.
[24] S. Diez-Gonzalez, S. P. Nolan, Org. Prep. Proced. Int. 2007, 39,
523 – 559.
[25] G. Keglevich, M. Fekete, T. Chuluunbaatar, A. Dobꢂ, V.
˝
Harmat, L. Toke, J. Chem. Soc. Perkin Trans. 1 2000, 4451 – 4455.
[26] A study employing isolated phosphonium salt derived from
reduced 1 and methyl bromoacetate revealed that Na2CO3 was
the optimal base and the diastereoselectivity was 3:1 (E/Z).
Under identical conditions the catalytic protocol yields the same
selectivity, 3:1 (E/Z).
[11] a) P. Cao, C.-Y. Li, Y.-B. Kang, Z. Xie, X.-L. Sun, Y. Tang, J. Org.
Chem. 2007, 72, 6628 – 6630; b) S. Zhu, Y. Liao, S. Zhu, Org. Lett.
2004, 6, 377 – 380; c) W.-D. Dai, A. Wu, H. Wu, Tetrahedron:
Asymmetry 2002, 13, 2187; d) L. Shi, W. Wang, Y. Wang, Y.
Huang, J. Org. Chem. 1989, 54, 2027 – 2028 and references
therein.
[27] J. M. Smith, M. F. Greaney, Tetrahedron Lett. 2007, 48, 8687 –
8690.
[12] Z.-Z. Huang, Y. Tang, J. Org. Chem. 2002, 67, 5320 – 5326.
[13] Y. Liao, Y.-Z. Huang, Tetrahedron Lett. 1990, 31, 5897 – 5900.
[14] J. A. Kerr in CRC Handbook of Chemistry and Physics, 81st ed.
(Ed.: D. R. Lide), Boca Raton, US, 2000.
[15] For a review on catalytic aldehyde olefinations see: F. E. Kꢁhn,
A. M. Santos, Mini-Rev. Org. Chem. 2004, 1, 55 – 64.
[16] a) K. K. Kroening, M. J. Solivio, M. Garcꢄa-Lꢂpez, A. Puga, J. A.
Caruso, Metallomics 2009, 1, 59 – 66; b) Y. Hu, X. Jin, E. T. Snow,
Toxicol. Lett. 2002, 133, 33 – 45; c) J. Henriksson, A. Johannis-
son, P.-A. Bergqvist, L. Norrgren, Arch. Environ. Contam.
Toxicol. 1996, 30, 213 – 219.
[28] Density functional theory computations were carried out at the
[B3LYP/6-31G(d,p)(SCRF-PCM, solvent = toluene) (e = 2.379)]
level, using the Gaussian 03 (G03) program package, with
default convergence thresholds; all structures were frequency-
confirmed as residing at minima. Baderꢅs Atoms-In-Molecules
(AIM) analyses of the wavefunctions generated from the
geometry-optimized structures were performed.
[29] G. A. Chass, M. A. Sahai, J. M. S. Law, S. Lovas, O. Farkas, A.
Perczel, J.-L. Rivail, I. G. Csizmadia, Int. J. Quantum Chem.
2002, 90, 933 – 968.
Angew. Chem. Int. Ed. 2009, 48, 6836 –6839
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 6839