The other half of the Michaelis-Arbuzov reaction: A new synthetic route to tertiary phosphorus-carbon bond formation

Article abstract of DOI:10.1016/j.inoche.2006.01.021

A first-row transition metal assisted modification of the Michaelis-Arbuzov rearrangement has been discovered that allows the reaction to take place under ambient conditions using a mild phosphine nucleophile. This reaction has been used as a novel synthetic strategy for the assembly of P-C bonds in typically difficult to construct 3-methylpyridyl substituted phosphine and phosphonium ligands. Metal complexes of these ligands were characterized crystallographically. One of the products also contains the previously unknown pyridylFeCl3- species.

Full text of DOI:10.1016/j.inoche.2006.01.021

Inorganic Chemistry Communications 9 (2006) 418–422
www.elsevier.com/locate/inoche
The other half of the Michaelis–Arbuzov reaction:
A new synthetic route to tertiary phosphorus–carbon bond formation
*
Rodney P. Feazell, Cody E. Carson, Kevin K. Klausmeyer
Department of Chemistry and Biochemistry, Baylor University, Waco, TX 76798-7348, United States
Received 14 November 2005; accepted 23 January 2006
Abstract
A first-row transition metal assisted modification of the Michaelis–Arbuzov rearrangement has been discovered that allows the reac-
tion to take place under ambient conditions using a mild phosphine nucleophile. This reaction has been used as a novel synthetic strategy
for the assembly of P–C bonds in typically difficult to construct 3-methylpyridyl substituted phosphine and phosphonium ligands. Metal
complexes of these ligands were characterized crystallographically. One of the products also contains the previously unknown
pyridylFeClÀ3 species.
Ó 2006 Elsevier B.V. All rights reserved.
Keywords: X-ray diffraction; N,P ligands; Ligand design; Nucleophilic substitution
Trivalent phosphorus esters have been known for over a
century to react with alkyl halides to produce substituted
phosphinates, phosphonates, and phosphine oxides in what
has come to be known as the Arbuzov (or Michaelis–Arbu-
zov) rearrangement [1]. This type of reaction is typically
achieved by heating a phosphite in the presence of an alkyl
bromide or iodide, which is believed to dissociate and
nucleophilicaly attack the C–O bond in the most agreed
upon mechanism shown in Scheme 1 [2]. Although the
reaction conditions almost always involve high tempera-
tures, self-rearangements utilizing only the phosphorus
ester have been known to occur at moderate temperatures
using a Lewis acid catalyst [3]. The desired product of this
reaction is generally a phosphorus oxide, with the sacrifi-
cial RX group being discarded as an unwanted by-product.
We report here a novel, transition metal assisted version
of the Arbuzov reaction in which the alkyl halides are
replaced as nucleophiles by tertiary phosphines allowing
us to isolate the conventionally discarded alkyl portion of
the phosphinites as Zwitterionic phosphonium–metal com-
plexes. In an unprecedented instance of a nucleophilic
attack on the O–C bond of a phosphinite by triphenyl- or
diphenylphosphine, charge-neutral phosphonium=MCl3À
complexes (M = Fe2+ or Co2+) are formed when the
phosphine is added to a solution of MCl2 Æ 4H2O with diph-
enylphosphino-3-pyridylcarbinol, PCP-31 [4]. The triphe-
nylphosphine product contains the previously unreported
pyFeClÀ3 species while the diphenylphosphine product con-
tains an acidic proton that is easily removed to form a new
tertiary phosphine.
The cationic phosphonium PPh3(3-CH2C5H4N)+ was
formed by the nucleophilic attack of triphenylphosphine
on diphenylphosphino-3-pyridylcarbinol in the presence
of iron(II) or cobalt(II) chloride tetrahydrate as shown in
Scheme 2. The mechanism of the reaction appears to pro-
ceed through the initial coordination of the pyridyl-nitrogen
to the metal, activating the benzylic position towards the
Arbuzov reaction. It is observed that when the pyridyl ring
is replaced by phenyl, no phosphonium is formed, implying
that phosphorus–metal binding is not a determining step in
this reaction [5]. This nitrogen-donation into the metal cen-
ter increases the electrophilicity of the benzylic carbon to
such an extent that it readily dissociates from oxygen at
*
Corresponding author. Fax: +1 254 710 4272.
E-mail address: Kevin_Klausmeyer@baylor.edu (K.K. Klausmeyer).
1387-7003/$ - see front matter Ó 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.inoche.2006.01.021

R.P. Feazell et al. / Inorganic Chemistry Communications 9 (2006) 418–422
419
R1
R2
its consumption of phosphinite and phosphine as seen by
their complete exclusion from the product’s 31P NMR
spectra.
O
R1
R2
R3X
X
R4
R4
P
P
O
R3
The molecular structure of 1 is shown in Fig. 1. Fe(1) is
in a distorted tetrahedral geometry. N–Fe–Cl(1) and N–
Fe–Cl(2) angles are constricted to 102.98 and 97.17°,
respectively, while the third N–Fe–Cl(3) angle, that
involves the chlorine directed away from the center of the
molecule, is slightly obtuse at 110.03°. The Fe–Cl(2) bond,
which is involved in the largest angular distortion, is seen
Scheme 1. Mechanism of the Michaelis–Arbuzov Reaction.
room temperature when beset by a relatively mild nucleo-
phile. This type of a-carbon [6] activation has been demon-
strated recently using oxophilic Lewis acids such as
BF3 Æ OEt2 to withdraw electron density from the carbon
closest to the ester oxygen via that oxygen to rearrange
phosphinites to phosphonates [3b]. Even so, elevated tem-
peratures and long reaction times are necessary for these
rearrangements to take place. In the present case, self-rear-
rangement is not observed due to the relatively intermediate
acidity of the metal chlorides as well as the low reaction
temperatures employed. The benzylic carbon atom, how-
ever, remains in a substantially activated state allowing tri-
phenylphosphine to easily displace the Ph2P(O)À group to
give the cationic Ph3PCH2C5H4N(MCl2)+ species as an
intermediate en route to the final Zwitterionic product.
Reaction of triphenylphosphine with PCP-31 was not
observed to proceed to any detectable extent in the absence
of a transition metal chloride [7], yet the substitution is seen
to be complete within seconds with metal activation.
˚
to be shorter than the other Fe–Cl distances by ꢀ.03 A at
˚
˚
2.2654(5) A. An N–Fe distance of 2.125(1) A is slightly
short for a pyridyl–Fe2+ complex [12] and can likely be
attributed to the cationic attraction to the extra electron
density associated with the anionic FeClÀ3 moiety. The
cobalt derived product is seen to be nearly identical in
structure and conformation [8].
The chelating 2-(diphenylphosphinomethyl)pyridyl
ligand is readily made by the action of the lithiated 2-pico-
line on chlorodiphenylphosphine at low temperatures [13].
Because of unfavored resonance, our attempts to use the
same synthesis for making 3-(diphenylphosphinom-
ethyl)pyridyl have resulted in poor yields and a large
amount of by-products. In the current case, addition of
diphenylphosphine as the nucleophile in the previously
described reaction was used as the initial step in a
synthetic pathway that resulted in the formation of the
phosphorus–carbon bond of 3-(diphenylphosphinomethyl)-
pyridyl under ambient conditions. Each step of the
synthesis shown in Scheme 3 was followed by 31P NMR
spectroscopy with the final product being characterized
by crystallography.
The initial attack on the phosphinite is the same using
diphenylphosphine as with triphenylphosphine. However,
the acidic P-bound hydrogen present on the former is eas-
ily removed by a mild base allowing for further function-
ality of the phosphonium product. Addition of an excess
of triethylamine readily removes the proton from the
phosphorus while concomitantly stripping a chloride from
the iron(II) center to leave the neutral metal complex. The
triethylammoniumchloride precipitates from acetonitrile
solution and is easily removed via filtration. The struc-
tural features of the iron complex are uncertain at this
stage with the tetrahedral coordination sphere of the
metal center potentially being filled by either solvent or
the newly formed phosphine. Attempts at crystallization
at this point have been thus far unsuccesssful, however,
the broadened 31P resonance which is shifted over
5 ppm upfield of the free ligand implies at least some
degree of interaction between the phosphine and the metal
center; possibly a dimeric head-to-tail coordination mode.
The bound ligand is then liberated from the iron center by
the addition of a large excess of cyanide. Favorable for-
mation of iron cyanide complexes frees the 3-(diphenyl-
phosphinomethyl)pyridyl ligand, which is then separated
by extraction into hexanes. The 31P spectra of the free
ligand at this point displays a proton coupled pentet in
The yellow complex FeCl3PPh3(3-CH2C5H4N) 1,1 as
well as the structurally identical violet complex CoCl3PPh3
(3-CH2C5H4N) [8] are obtained as neutral coordination
complexes that are formed as the MCl2 in solution dissoci-
ates to neutralize the charge of both species being formed
from the Arbuzov reaction. The pyridine donor makes a
convenient bridge linking the charge balancing MClÀ3 ion
to the phosphonium, creating a static charge separation of
considerable distance [9]. While the MClÀ3 species is well
known, especially in the case where M = Co(II), this is
the first known instance where FeClÀ3 is found bound by a
pyridyl ligand [10]. MCl+ fragments absorb the anionic
product (the phosphine oxide that is ordinarily the desired
product of these reactions) by forming the sparingly soluble
MClPPh2(O) complexes. These are then seen to be further
hydrolyzed by water liberated from the starting metal
hydrates to diphenylphosphinic acid and an as yet uniden-
tified insoluble metal containing precipitate. The fairly
rapid reaction progress can be confirmed by 31P NMR by
observing the disappearance of the phosphinite and triphe-
nylphosphine resonances with a concurrent emergence of
the three other signals [11]. This reaction is quantitative in
1
Crystal data for 1: C26H24Cl3FeN2P, M = 557.64, Triclinic, space
ꢀ
˚
group P1, a = 8.2649(6), b = 10.7127(7), c = 15.741(1) A, a = 94.275(À41),
3
˚
b = 102.809, c = 96.550(4)°, V = 1343.0(2) A , Z = 2, l = 0.936 mm
,
37648 data collected, 8066 data unique (Rint = 0.0387), 6362 data with
I > 2r(I), R1 = 0.0339, wR2 = 0.0859, CCDC deposit number: 275,131.
For 2: C40H34Ag2F12N2O12P2S4, M = 1368.61, Monoclinic, space group
˚
C2/c, a = 19.224(1), b = 13.0339(7), c = 20.434(1) A, b = 95.176(2)°,
3
V = 5099.5(5) A , Z = 4, l = 1.098 mmÀ1, 30,273 data collected, 6366
˚
data unique (Rint = 0.0352), 5359 data with I > 2r(I), R1 = 0.0275,
wR2 = 0.0661, CCDC deposit number: 275,133.

420
R.P. Feazell et al. / Inorganic Chemistry Communications 9 (2006) 418–422
PPh3
1
O
CH2
MCl2(H2O)4
Ph2P
Ph3P
N
MCl3
N
MCl2
+
Ph2P(O)MCl
H2O
O
Ph2POH
MXY
+
X,Y = Cl, OH,or [XY = O]
Scheme 2. Reaction of PCP–31 with PPh3.
accounting for the extra triflate present in the crystal struc-
ture. The discrete structure of 2 contains an anion bridged
ring connecting two equivalent metal positions. Here the
metal environment is a trigonal pyrimidal arrangement
˚
with the metal center being only 0.246(3) A removed from
the plane made by the atoms P1, O4 and O3A. The bond
lengths to these three atoms are all similar in in length with
˚
Ag-donor lengths of 2.3626(5), 2.321(7), and 2.353(1) A,
respectively. O1 from the g2, l2-triflate forms the apex of
the pyramid which has a notably longer bond length at
˚
2.517(1) A. The angles around the trigonal base sum to
356° but are distorted by the constriction of the bridging
anions as well as H-bonds to the anions involving the pro-
tonated pyridyl rings giving them values that range from
78.1(1) to 143.7(1)°. These H-bonds are of average distance
˚
with lengths of 3.012(2) and 2.898(7) A for N(1)–
HÁ Á ÁO(2)A and N(1)–HÁ Á ÁO(4)A, respectively. It is likely
that these H-bonds are imperative to the crystallization
of this compound, as we have been so far unable to isolate
a structure in the complete absence of any residual water.
The ease with which the phosphine substitution takes
place on the metal coordinated PCP-31 is demonstrative
of the electron withdrawing effects of the metal on the ben-
zylic carbon. The use of this modified Arbuzov reaction for
the facile addition of the 3-methylpyridyl group to a diphe-
nylphosphine ligand gives access to an asymmetric bridging
ligand not easily obtained by other means. We are cur-
rently in the process of optimizing the reaction conditions
in order to use our synthesis for the production of multi-
gram quantities of the ligand. This synthetic route is also
being examined as a means of constructing other difficult
to achieve tertiary amines, arsines and stibines.
Fig. 1. Molecular structure of 1, with thermal ellipsoids shown at the 40%
˚
probability level. Selected bond lengths (A) and angles (deg): Fe–N,
2.125(2); Fe–Cl(1), 2.2935(5); Fe–Cl(2), 2.2654(5); Fe–Cl(3), 2.3096(5); N–
Fe–Cl(1), 102.98(4); N–Fe–Cl(2), 97.17(4); N–Fe–Cl(3), 110.03(4); Cl(1)–
Fe–Cl(2), 123.37(2); Cl1–Fe–Cl(3), 111.57(2); Cl(2)–Fe–Cl(3), 109.86(2).
Solvent acetonitrile is not shown.
the expected region for a diphenylbenzylphosphine. When
an equivalent of silver triflate dissolved in CH3CN is
added to the free ligand and then precipitated with ether
2 is obtained as a glassy solid.
Small colorless block shaped crystals of 21 grown from
CH2Cl2 solutions layered with ether produce the structure
shown in Fig. 2. Superficial water likely destroys one equiv-
alent of silver and protonates the pyridyl-nitrogen atom,
PPh2H
O
CH2
H
Ph2P
Ph2
Et3N
Ph2P
XS KCN
Ph2P(O)FeCl
AgOTf
Ph2P
Ph2P
N
FeCl2
H
P
N
FeCl2
N
N
FeCl3
N
Et3NHCl
(TfO)2Ag
-4
Fe(CN)6
+
FeCl2(H2O)4
2
Scheme 3. Synthetic route to compound 2.

R.P. Feazell et al. / Inorganic Chemistry Communications 9 (2006) 418–422
421
Fig. 2. Molecular structure of 2 with thermal ellipsoids shown at the 40% probability level. Symmetry equivalent positions are shown shaded. Selected
˚
˚
bond lengths (A), angles (deg) and interatomic distances (A): Ag1–O1, 2.517(2); Ag1–O4, 2.321(7); Ag1–P1, 2.3626(5); Ag1–O3A, 2.353(2); O3A–Ag1–O4,
78.1(1); O3A–Ag1–O1, 91.58(5); O3A–Ag1–P1, 134.12(5); O1–Ag1–O4, 93.7(2); O1–Ag1–P1, 100.18(4); O4–Ag1–P1, 143.7(1). N1–HÁ Á ÁO2A, 3.012(2);
N1–HÁ Á ÁO4A, 2.898(7). Symmetry operator used to generate equivalent positions: Àx + 1/2, Ày + 1/2, Àz + 1. Disorder of the terminal triflate has been
omitted for clarity.
[2] T.B. Brill, S.J. Landon, Chem. Rev. 84 (1984) 577–585.
[3] (a) A.K. Bhattacharya, G. Thyagarajan, Chem. Rev. 81 (1981) 415–
Acknowledgements
430;
This research was supported by funds provided by grant
(b) P.Y. Renard, P. Vayron, E. Leclerc, A. Valleix, C. Mioskowski,
from the Robert A. Welch Foundation (AA-1508). The
Angew. Chem. Int. Ed. 42 (2003) 2389–2392.
Bruker X8 APEX diffractometer was purchased with funds
received from the National Science Foundation Major
Research Instrumentation Program Grant CHE-0321214.
[4] Diphenylphosphino-3-pyridylcarbinol was prepared as previously
reported in K.K. Klausmeyer, R.P. Feazell, J.H. Reibenspies, Inorg.
Chem. 43 (2004) 1130–1136.
[5] Diphenylphosphinocarbinol was prepared in an analogous manner as
diphenylphosphino-3-pyridylcarbinol, substituting benzyl alcohol for
3-pyridyl carbinol. Details of synthesis are given in supplementary
material.
Appendix A. Supplementary material
[6] a-to a phosphinite oxygen.
Synthesis details, spectral and analytical data for all
compounds described herein as well as extra figures and
tables of crystallographic data in CIF format are available.
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/j.inoche.2006.
01.021.
[7] Using typical Arbuzov procedures: triphenyl phosphine was heated to
110 °C with diphenylphosphino-3-pyridylcarbinol for several hours
with no phosphonium formed.
[8] Synthesis, characterization, and X-ray data for the Co analog of 1 are
contained in supplementary information.
˚
˚
[9] Fe1–P1 = 6.2875(6) A; Co1–P1 = 6.211(5) A. Both distances involve
5-bond separations.
[10] Searches of Chemical Abstracts and Cambridge Crystallographic
Database, ConQuest version 1.6, Apr 2004, show no instances of an
iron (II) trichloride anion with any sort of pyridyl donor
coordinated.
[11] 31P NMR ds: The spectra of diphenylphosphino-3-pyridylcarbinol is
discussed in Ref. [4]. Spectra of Ph2P(O)M compounds can be found
in Organometallics 1996, 15, 3259–3241. Comparable spectra to all
other compounds can be found in: M. Crutchfield, C. Dungan, J.
Letcher, V. Mark, J. Van Wazer, Topics in Phosphorus Chemistry,
vol 5, 1967. 31P NMR spectra consist of broad singlets at d = 48.4 and
39.7 ppm for the Fe and Co complexes, respectively. 1H NMR spectra
consist of numerous indiscernible peaks in the aromatic region due to
the paramagnetic nature of the tetrahedral metals..
References
[1] (a) A. Arbuzov, J. Russ. Phys. Chem. Soc. 38 (1906) 687;
(b) A. Michaelis, R. Kaehne, Chem. Ber. 31 (1898) 1048;
(c) Recent examples of use include: T. Erker, N. Handler, Synthesis 5
(2004) 668–670;
(d) P. Even, E. Guenin, M. Benramdame, P. Quidu, D. El Manouni,
M. Lecouvey, Lett. Org. Chem. 1 (2004) 75–77;
(e) A. Poap, G. IIia, S. IIiescu, C.M. Davidescu, A. Pascariu, A. Bora,
Rev. Chim. 54 (2003) 834–836;
(f) M.D. Sorensen, L.K.A. Blaehr, M.K. Christensen, T. Hoyer, S.
Latini, P.J.V. Hjarnaa, F. Bjorkling, Bioorg. Med. Chem. 11 (2003)
5461–5481;
[12] (a) Typical unidentate pyridyl ligands bound to Fe(II) have an
˚
average N–Fe distance of ꢀ2.27 A. G.J. Long, P.J. Clarke, Inorg.
(g) A. Tellenbach, M. Jansen, Eur. J. Inorg. Chem. 20 (2003) 3759–
3766.
Chem. 17 (1978) 1394–1401;

422
R.P. Feazell et al. / Inorganic Chemistry Communications 9 (2006) 418–422
(b) I.I. Bulgak, V.E. Zubareva, K.I. Turte, Y.A. Simonov, D.G. [13] (a) B. Akermark, B. Krakenberger, S. Hansson, Organometallics 6
Batir, M.D. Mazus, T.I. Malinovski, Koord. Khim. 13 (1987)
960–962;
(c) P. Karsten, J. Strahle, Acta Cryst. Sect. C. 54 (1998) 1406–1408.
(1987) 620–628;
(b) P. Braunstein, B.T. Heaton, C. Jacob, L. Manzi, X. Morise, J.
Chem. Soc., Dalton Trans. (2003) 1396–1401.