Separating the racemic and meso diastereomers of a binucleating tetraphospine ligand system through the use of nickel chloride

Article abstract of DOI:10.1021/ic010035f

The reaction of a 1:1 mixture of rac- and meso-et,ph-P4 (et,ph-P4 = Et₂PCH₂CH₂)(Ph)PCH₂P(Ph) (CH₂CH₂PEt₂) with 2 equiv of NiCl₂·6H₂O in EtOH produces soluble rac-Ni₂Cl₄(et,ph-P4) and precipitates meso-Ni₂Cl₄-(et,ph-P4), allowing facile isolation of each bimetallic complex. Subsequent reaction with more than 250 equiv of NaCN in H₂O/MeOH releases the et,ph-P4 ligand and [Ni(CN)₄]^2-. The rac,trans- and meso,trans-Ni(CN)₂-(η^2.5-et,ph-P4) form as intermediates in the cyanolysis of rac- and meso-Ni₂Cl₄(et,ph-P4). These have been characterized by X-ray crystallography. The unusual partial isomerization of the meso- to rac-et,ph-P4 ligand via the monometallic trans-Ni(CN)₂(η^2.5-et,ph-P4) intermediate complex is discussed.

Full text of DOI:10.1021/ic010035f

5036
Inorg. Chem. 2001, 40, 5036-5041
Separating the Racemic and Meso Diastereomers of a Binucleating Tetraphosphine Ligand
System through the Use of Nickel Chloride
David A. Aubry, Scott A. Laneman, Frank R. Fronczek, and George G. Stanley*
Department of Chemistry, Louisiana State University, Baton Rouge, Louisiana 70803-1804
ReceiVed January 9, 2001
The reaction of a 1:1 mixture of rac- and meso-et,ph-P4 (et,ph-P4 ) (Et2PCH2CH2)(Ph)PCH2P(Ph)CH2CH2PEt2)
with 2 equiv of NiCl2‚6H2O in EtOH produces soluble rac-Ni2Cl4(et,ph-P4) and precipitates meso-Ni2Cl4-
(et,ph-P4), allowing facile isolation of each bimetallic complex. Subsequent reaction with more than 250 equiv
of NaCN in H2O/MeOH releases the et,ph-P4 ligand and [Ni(CN)4]2-. The rac,trans- and meso,trans-Ni(CN)2-
(η2.5-et,ph-P4) form as intermediates in the cyanolysis of rac- and meso-Ni2Cl4(et,ph-P4). These have been
characterized by X-ray crystallography. The unusual partial isomerization of the meso- to rac-et,ph-P4 ligand via
the monometallic trans-Ni(CN)2(η2.5-et,ph-P4) intermediate complex is discussed.
Introduction
highly regioselective bimetallic hydroformylation catalyst (90
°C, 90 psig, 1:1 H2/CO, acetone solvent, 1 mM catalyst, 1.8 M
1-hexene substrate, initial turnover frequency ) 640 turnovers/
h, 28:1 linear/branched regioselectivity, 8% alkene isomeriza-
tion, 3% alkene hydrogenation).5 In situ spectroscopic studies
have determined that [rac-Rh2(nbd)2(et,ph-P4)]2+ reacts with
H2/CO to rapidly generate [rac-Rh2H2(CO)2(µ-CO)2(et,ph-
P4)]2+ (along with other complexes), which we believe is the
active hydroformylation catalyst.6 Compound 1 was originally
referred to as eLTTP;3 however, because of the current syntheses
of P4 ligand derivatives to produce better dirhodium hydro-
formylation catalysts, the nomenclature has been modified to
more clearly indicate what substituents are attached to the
phosphines as these have a major impact on the catalytic activity
of the bimetallic complexes formed.
The racemic bimetallic catalyst precursor [rac-Rh2(nbd)2-
(et,ph-P4)](BF4)2 is 12 times faster for the hydroformylation of
1-hexene than [meso-Rh2(nbd)2(et,ph-P4)]2+, gives higher prod-
uct regioselectivity (28:1 vs 14:1 l/b), and gives fewer side
reactions (12% vs 34%).5 The higher rate of the racemic
bimetallic complex relative to that of the meso is proposed to
arise in part from the ability of the racemic bimetallic catalyst
to more readily form an edge-sharing bioctahedral hydride/
carbonyl-bridged intermediate that allows facile hydride transfers
from one metal center to the other. This, in turn, allows an
intramolecular hydride transfer step that leads to aldehyde
elimination.5 In marked contrast to the initially proposed
mechanism involving neutral bimetallic complexes, we have
determined that the most active bimetallic catalyst species is
not neutral, but rather dicationic.6 The presence of a localized
positive charge on each Rh center compensates for the strongly
electron-donating et,ph-P4 phosphine ligand that would normally
deactivate a neutral Rh center for hydroformylation catalysis.
Indeed, when one starts with a neutral bimetallic precursor such
as rac-Rh2(allyl)2(et,ph-P4), one generates a very poor bimetallic
hydroformylation catalyst.7
The tetraphosphine ligands, meso- and rac-(Et2PCH2CH2)-
(Ph)PCH2P(Ph)(CH2CH2PEt2) (et,ph-P4, 1m and 1r), are electron-
rich, powerful binucleating ligands designed to chelate and
bridge two transition metal centers. The resulting bimetallic
complexes can have either closed-mode geometries with M-M
bonds1 or open-mode structures in which the metals are rotated
apart from one another with M‚‚‚M separations of 5-7 Å.2,3
As with open-mode bimetallic complexes based on the previ-
ously reported binucleating hexaphosphine ligand,4 there is
considerable conformational flexibility in the open-mode bi-
metallic et,ph-P4 complexes. This is quite unlike most ligand-
bridged dinuclear complexes that usually have two bridging
ligands that impose a considerably more rigid framework
geometry. We believe that this conformational flexibility may
be quite important in encouraging bimetallic cooperativity
between two metal centers.
The ligand rac-et,ph-P4, 1r, reacts in high yield with 2 equiv
of [Rh(nbd)2]BF4 (nbd ) norbornadiene) to produce [rac-Rh2-
(nbd)2(et,ph-P4)](BF4)2, which is a precursor for an active and
(1) Laneman, S. A.; Fronczek, F. R.; Stanley, G. G. Inorg. Chem. 1989,
28, 1206.
(2) Laneman, S. A.; Fronczek, F. R.; Stanley, G. G. J. Am. Chem. Soc.
1988, 110, 5585.
(3) Laneman, S. A.; Fronczek, F. R.; Stanley, G. G. Inorg. Chem. 1989,
28, 1872-1878.
(4) (a) Askham, F. R.; Stanley, G. G.; Marques, E. C. J. Am. Chem. Soc.
1985, 107, 7423. (b) Saum, S. E.; Askham, F. R.; Fronczek, F. R.;
Stanley, G. G. Organometallics 1988, 7, 1409. (c) Saum, S. E.;
Fronczek, F. R.; Laneman, S. A.; Stanley, G. G. Inorg. Chem. 1989,
28, 1878. (d) Saum, S. E.; Askham, F. R.; Laneman, S. A.; Stanley,
G. G. Polyhedron 1990, 9, 1317. (e) D’Avignon, A.; Askham, F. R.;
Stanley, G. G. Inorg. Chem. 1990, 29, 3363. (f) Saum, S. E.; Laneman,
S. A.; Stanley, G. G. Inorg. Chem. 1990, 29, 5067.
(5) Broussard, M. E.; Juma, B.; Train, S. G.; Peng, W.-J.; Laneman, S.
A.; Stanley, G. G. Science 1993, 260, 1784.
(6) Matthews, R. C.; Howell, D. H.; Peng, W.-J.; Train, S. G.; Treleaven,
W. D.; Stanley, G. G. Angew. Chem., Int. Ed. Engl. 1996, 35, 2253-
2256.
10.1021/ic010035f CCC: $20.00 © 2001 American Chemical Society
Published on Web 08/15/2001

Separating Diastereomers Using Nickel Chloride
Inorganic Chemistry, Vol. 40, No. 19, 2001 5037
water was degassed by purging with nitrogen gas for 20 min. NaCN
and NiCl2‚6H2O were purchased from Aldrich Chemicals and used
without further purification.
It is clear from these results that one wants to work with the
pure rac-et,ph-P4 ligand and the associated bimetallic rhodium
catalyst for hydroformylation. Our reported synthesis of et,ph-
P4, however, generates a 1:1 mixture of rac- and meso-et,ph-
P4 ligands.3 The separation of these two ligand diastereomers,
therefore, is a matter of considerable importance. Our previous
work on Ni2Cl4(et,ph-P4) complexes had shown that the racemic
and meso bimetallic complexes crystallized from THF with
different crystal morphologies; the meso-Ni2Cl4(et,ph-P4), 2m,
complex readily loses some THF solvent of crystallization to
turn opaque, while the rac-Ni2Cl4(et,ph-P4), 2r, complex crystals
remain clear.3 This property of the crystals allowed us to separate
small quantities of each bimetallic diastereomer for crystal
structure determinations and to do a Very small scale ligand
isolation by cyanolysis of the bimetallic Ni2Cl4(et,ph-P4)
complexes to release the et,ph-P4 ligand but only with about
45% yield.3
One separation strategy to obtain pure meso- and rac-et,ph-
P4 ligands would involve production of a metal complex in
which one diastereomer would be soluble in a suitable solvent,
while the other would cleanly precipitate out allowing simple
isolation of each in high yields and in larger quantities. The
pure meso- and rac-et,ph-P4 ligands could then be released from
the metal complex by use of a suitable reagent to give the pure
ligand. Because nickel complexes had been successfully used
to separate other diastereomeric mixtures in the past, we decided
to concentrate our separation chemistry efforts on nickel
et,ph-P4 complexes.8
We spent considerable time investigating the use of Ni(NCS)2
as a separating agent, but this chemistry has proven to be
surprisingly complex.9 Although separations of the rac- and
meso-et,ph-P4 ligands have been achieved using both Ni(NCS)2
and NiCl2, the previous procedures offered inconsistent and often
low yields.3,9 The original NiCl2-based separation chemistry was,
therefore, reexamined in order to optimize the yields.3 We report
here the straightforward NiCl2-based route for the bulk separa-
tion and isolation of rac- and meso-et,ph-P4 ligands in pure
diastereomeric form, along with the isolation and structural
identification of the monometallic trans-Ni(CN)2(η2.5-et,ph-P4)
intermediates from the cyanolysis step. An unusual meso- to
rac-et,ph-P4 partial isomerization mediated by these mono-
metallic Ni-cyanide complexes has also been observed.
Synthesis of Ni2Cl4(et,ph-P4), 2m and 2r. A 125 mL EtOH solution
of mixed rac,meso-et,ph-P4 ligand3 (10.01 g, 0.0215 mol) was added
dropwise to a rapidly stirred clear green solution of NiCl2‚6H2O (10.24
g, 0.0431 mol) in 135 mL of EtOH. The solution turned dark red as
the ligand solution was added. An orange precipitate began to form
after the addition was complete. After the mixture was stirred for 24
h, the orange precipitate was collected by filtration and washed with
three ca. 30 mL portions of EtOH to give 7.32 g (94% yield) of meso-
Ni2Cl4(et,ph-P4), 2m. Yields are typically 90-96%. The filtrate was
concentrated down to a dark red amorphous solid of mainly rac-Ni2-
Cl4(et,ph-P4), 2r. Spectroscopic properties have already been reported.3
Isolation of rac-et,ph-P4, 1r. A Schlenk flask containing rac-Ni2-
Cl4(et,ph-P4), 2r (2.0 g, 2.77 mmol), was charged with a solution of
NaCN (18.1 g, 0.369 mol, 133 equiv) in 125 mL of H2O and 50 mL
of MeOH. The resulting orange solution was stirred slowly for 3 h,
during which it became increasingly red. The flask was then charged
with more NaCN (20.4 g, 0.416 mol, 150 equiv), and the solution was
allowed to slowly stir until all the NaCN dissolved (ca. 15 min). This
two-step cyanide addition allows any meso-et,ph-P4 ligand impurities
present to be partially isomerized to rac-et,ph-P4 ligand giving slightly
higher yields. The free rac-et,ph-P4 ligand was then extracted into three
100 mL portions of benzene. The slightly yellow extracted solution
was then passed through a small neutral alumina column to remove
the yellow color, which is the monometallic Ni complex trans,rac-Ni-
(CN)2(η2.5-et,ph-P4), 3r. The colorless solution was then concentrated
to yield a clear viscous oil (1.16 g, 2.50 mmol, 85%) of free rac-et,-
ph-P4 ligand. Yields are typically 75-87%, and the purity level, based
on 31P NMR, is typically greater than 95%. 31P{1H} NMR (C6D6):
-25.5 (Pint, JP-P ) 10.2, 12.1 Hz), -18.4 (Pext, JP-P ) 10.2, 12.1 Hz).
Computer-simulated coupling constants based on an AXX′A′ spin
system: JP
included in Supporting Information.
) 22.5 Hz, JP
) 109.5 Hz.3 31P{1H} NMR is
-P
int
-P
ext
int
int
Isolation of meso-et,ph-P4, 1m. A Schlenk flask was charged with
40 mL of H2O, 20 mL of MeOH and NaCN (2.0 g, 0.408 mol, 393
equiv). To the white cyanide suspension, 30 mL of a brown 50/50
solution of H2O and MeOH containing meso-Ni2Cl4(et,ph-P4) (0.75 g,
1.04 mmol) was added dropwise with rapid stirring. Upon addition,
the cyanide suspension turned light orange. After slow overnight stirring,
the free meso-et,ph-P4 ligand was then extracted into three 75 mL
portions of hexane. The slightly yellow extracted solution was then
passed through a small neutral alumina column to remove the light
yellow color. The colorless solution was then concentrated to yield a
cloudy, highly viscous oil (0.30 g, 0.647 mmol, 62% yield) that is meso-
et,ph-P4 ligand. Yields are typically 45-65%, and the purity level,
based on 31P NMR, is typically only about 75-85% because of partial
isomerization of meso-et,ph-P4 ligand to rac-et,ph-P4. Shorter reaction
times give less isomerization to racemic ligand, but there is also less
ligand released from nickel giving lower yields of free ligand.
Recrystallization of the extracted crude ligand (0.30 g, 0.647 mmol)
from hexane (ca. 25 mL) provided 99% pure meso ligand as a white
powder, based on 31P NMR, (0.25 g, 0.539 mmol, 51% yield). Yields
after recrystallization are typically 30-60%, and purity is greater than
98%. 31P{1H} NMR (C6D6): -26.2 (Pint, JP-P ) 10.4, 12.2 Hz), -18.3
(Pext, JP-P ) 10.4, 12.2 Hz).3
Experimental Section
General Procedures. All manipulations were performed under inert
atmosphere (argon or nitrogen), unless otherwise noted, with standard
Schlenk or glovebox techniques. None of the nickel complexes reported
are particularly oxygen-sensitive. NMR spectra were recorded on Bruker
1
Avance-250 or ARX-300 instruments (δ ppm, H TMS reference, 31P
85% H3PO4 reference). IR spectra were run on a Perkin-Elmer 1760X
FT-IR spectrometer using GRAMS/32 software (Galactic Industries).
Oneida Research Services, Inc., Whitesboro, NY, performed the
elemental analyses.
Isolation of Impure rac-et,ph-P4 from meso-Ni2Cl4(et,ph-P4). A
Schlenk flask was charged with meso-Ni2Cl4(et,ph-P4) (1.0 g, 1.385
mmol) and 20 mL of H2O. After the mixture was stirred for 1 h to
dissolve the meso-Ni2Cl4(et,ph-P4), a 10 mL H2O solution of NaCN
(0.35 g, 7.14 mmol, 5.2 equiv) was added dropwise with stirring. During
the addition, an orange precipitate formed, which was rac,trans-Ni-
(CN)2(η2-et,ph-P4). MeOH (20 mL) was added to dissolve the
precipitate. More NaCN was added (3.50 g, 0.0714 mol, 52 equiv) to
free the ligand. The ligand was extracted with three 50 mL portions of
benzene. The slightly yellow extracted solution was then passed through
a neutral alumina column to remove the yellow tint. The colorless
solution was concentrated to yield a clear viscous liquid (0.41 g, 0.875
mmol, 70% yield). Yields are typically 60-75% (mixture of 1:1 to
1.4:1 rac- to meso-et,ph-P4 ligand).
Solvents were dried and distilled under inert atmosphere from the
appropriate drying agents as follows: tetrahydrofuran, diethyl ether,
and hexane (potassium/benzophenone); dichloromethane (calcium hy-
dride); ethanol and methanol (magnesium). Distilled (or deionized)
(7) Peng, W.-J.; Train, S. G.; Howell, D. K.; Fronczek, F. R.; Stanley, G.
G. Chem. Commun. 1996, 2607.
(8) (a) Schlenk, T. G.; Downes, J. M.; Milne, C. R. C.; Mackenzie, P. B.;
Boucher, H.; Whalen, J.; Bosnich, B. Inorg. Chem. 1985, 24, 2334.
(b) Allen, D. L.; Gibson, V. C.; Green, M. L. H.; Skinner, J. F.; Baskin,
J. Grebenik, P. D. J. Chem. Soc., Chem. Commun. 1983, 895.
(9) (a) Alburquerque, P. R. Ph.D. Dissertation, Louisiana State University,
Baton Rouge, LA, 1997. (b) Alburquerque, P. R.; Aubry, D. A.; Juma,
B.; Bridges, N. N.; Taploo, N.; Fronczek, F. R.; Stanley, G. G. To be
submitted for publication.

5038 Inorganic Chemistry, Vol. 40, No. 19, 2001
Aubry et al.
rac,trans-[Ni(CN)2(η2.5-et,ph-P4)], 3r. Method A (THF Solvate).
This complex was a byproduct of the et,ph-P4 extraction from rac-
Ni2Cl4(et,ph-P4) using a benzene/H2O/NaCN reflux system for 30 min.3
Yellow 3r remained in the benzene phase and was isolated on a neutral
alumina column and then removed by elution with THF. After slow
evaporation of the THF solvent, orange-yellow crystals of 3r‚THF were
obtained.
Table 1. Key Crystallographic Data for rac- and
meso-Ni(CN)2(η2.5-et,ph-P4), 3r and 3m
3r‚THF
3m‚H2O
formula
fw
T, K
C31H48N2NiOP4
647.35
296 K
C27H42N2NiOP4
593.22
120 K
diffractometer
λ, Å
CAD4
Kappa-CCD
0.710 73 (Mo)
P1h (No. 2)
8.4310(2)
18.5593(3)
19.7142(4)
95.2808(8)
94.3847(7)
99.5244(10)
3016.08(11)
4
3r. Method B (H2O Solvate). NaCN (0.11 g, 2.08 mmol, 3 equiv)
in 10 mL of H2O was added dropwise without stirring to a Schlenk
flask containing rac-Ni2Cl4(et,ph-P4) (0.50 g, 0.693 mmol) in 10 mL
of H2O. After the addition, the flask was slowly swirled for 30 s. The
orange solid that formed was filtered (0.23 g, 0.400 mmol, 58%), rinsed
with ca. 2 mL of H2O, and redissolved in H2O/MeOH. Stirring during
cyanide addition limits the amount of precipitate that forms. Slow
evaporation yielded orange clusters of crystals of 3r‚H2O. Anal. Calcd
(found) for C27H44N2O2P4Ni (water present by X-ray): C, 53.05 (53.80);
H, 7.26 (6.95); N, 4.58 (4.58). 31P{1H} NMR (C6D6): 43.0 (2Pext, s),
-14.7 (2Pint, s). IR (KBr): VCN ) 2096 cm-1 (sharp and strong).
meso,trans-[Ni(CN)2(η2.5-et,ph-P4)], 3m (H2O Solvate). NaCN
(0.11 g, 2.08 mmol, 3 equiv) in 10 mL of H2O was added abruptly to
a Schlenk flask containing meso-Ni2Cl4(et,ph-P4) (0.50 g, 0.693 mmol)
in 10 mL of H2O. The flask was slowly swirled for 5 s. The orange
solid that formed was filtered (0.22 g, 0.382 mmol, 55%), rinsed with
ca. 2 mL of H2O, and redissolved in a 3:1 H2O/MeOH mixture. The
quick addition of cyanide to 2m causes more precipitation of 3m and
minimizes the isomerization to 3r. Slow evaporation produced two
crystal morphologies, yellow needles and orange clusters. The vast
majority of them were yellow needles, which were determined through
X-ray crystallography to be 3m‚H2O. The orange clusters were 3r‚
H2O. Anal. Calcd (found) for C27H42N2OP4Ni (water present by
X-ray): C, 54.67 (54.18); H, 7.14 (6.95); N, 4.72 (4.58); P, 20.89
(22.54). 31P{1H} NMR (CD2Cl2): -16.2 (P3), 12.8 (P1), 47.5 (P4, d,
JP2-P4 ) 219 Hz), 68.9 (P2, d, JP2-P4 ) 219 Hz). 31P resonances were
1.541 84 (Cu)
P21/n (No. 14)
15.128(2)
11.085(2)
20.663(2)
90
98.78(2)
90
3402(1)
4
space group
a, Å
b, Å
c, Å
R, deg
â, deg
γ, deg
V, Å3
Z
F
calcd, g cm-1
1.256
2.80 (Cu)
0.063
1.306
0.878 (Mo)
0.049
µ, cm-1
R(Fo)a
2
Rw(Fo )b
0.084
0.099
a R(Fo) ) ∑||Fo| - |Fc||/∑|Fo|. b Rw(Fo2) ) (∑w(Fo2 - Fc2)/∑wFo2)1/2
.
The volume of EtOH used and the concentration of the nickel
complexes are quite important because the meso-Ni2Cl4-
(et,ph-P4) complex precipitates as an orange powder and is only
slightly less soluble than rac-Ni2Cl4(et,ph-P4). The current work
has optimized these factors to give nearly quantitative separation
of 2m and 2r after rapid stirring for 24 h. The yield of isolated
meso-Ni2Cl4(et,ph-P4) solid obtained by vacuum filtration was
always greater than 90%. The filtrate, which contains rac-Ni2-
Cl4(et,ph-P4), was concentrated down to a black amorphous tarry
substance that still contains some EtOH, so the yield cannot be
accurately determined.
broad and no further coupling could be determined. IR (KBr): VCN
)
2119 cm-1 (sharp and strong).
After experimentation with numerous solvents and NiCl2 and
et,ph-P4 ligand concentrations, it was determined that the
optimal solvent for the separation of meso- and rac-Ni2Cl4-
(et,ph-P4) is EtOH (although MeOH works almost as well). A
total of 260 mL of EtOH solvent is required for each 10 g of
mixed et,ph-P4 ligand used. Interestingly, at higher Ni/P4 ligand
concentrations, no separation occurs. Instead, the concentrated
solution remains black and tarry with no precipitate formation.
The highly complex 31P NMR spectrum of this concentrated
solution contains coordinated and uncoordinated resonances and
is unlike any other mono- or bimetallic P4 complex that we
have characterized. This may be due to the formation of partially
coordinated, low-symmetry nickel-et,ph-P4 oligomeric chains.
Fortunately, addition of the appropriate amount of EtOH solvent
and overnight stirring do lead to the formation of well-defined
meso- and rac-Ni2Cl4(et,ph-P4) and produce the desired separa-
tion.
X-ray Crystallography. Suitable crystals (all air-stable) were
mounted on glass fibers using epoxy. Data collection was performed
on Enraf-Nonius CAD4 and Nonius Kappa CCD diffractometers using
Cu KR and Mo KR radiation and graphite crystal monochromators.
Structure solving was done using the Enraf-Nonius Molin and Bruker
ShelXTL program sets on DEC Alpha or PC computers. Both structures
were solved using MULTAN direct methods program set and refined
by using full-matrix least-squares refinement. The hydrogen atoms were
not refined but included in the calculations at fixed positions with B )
1.3Beq for their bonded atoms. Compound 3r had a disordered THF
that could only be partially modeled. Because the oxygen atom could
not be uniquely identified, all atoms were refined as carbons. There
are two independent molecules in the asymmetric unit for 3m and one
water solvent associated with each. Key crystal and data collection
information is presented in Table 1. Full crystal data, experimental
details, and positional and thermal parameters for 3r and 3m are given
in Supporting Information as CIF files.
Results and Discussion
Cyanolysis of rac-Ni2Cl4(et,ph-P4). Reaction of Ni2Cl4-
(et,ph-P4) with excess CN- is a convenient method for removing
the nickel centers from the ligand. The previous published
cyanolysis procedure for 2r only produced a 45% yield of rac-
et,ph-P4 ligand.3 This original separation chemistry employed
heating, lower concentrations of cyanide, and H2O/benzene as
a mixed solvent system. All these factors contributed to the low
yield and have now been corrected to provide an optimized
procedure for obtaining pure rac-et,ph-P4.
Heating during the cyanolysis procedure is not needed, as
the strongly σ-donating cyanide anions readily displace the
phosphine ligands, even a strongly coordinating, chelating,
mainly alkylated phosphine like et,ph-P4. Heating also has the
undesirable effect of promoting the decomposition of NaCN.10,11
rac- and meso-Ni2Cl4(et,ph-P4). Treating the mixed
et,ph-P4 ligand with 2 equiv of NiCl2‚6H2O produces the
bimetallic nickel complexes meso- and rac-Ni2Cl4(et,ph-P4), 2m
and 2r, in almost quantitative yield. Both diastereomeric
bimetallic complexes have symmetrical, chelated, bridged, open-
mode type structures (i.e., nonbonded Ni‚‚‚Ni separations of
6.272(1) and 5.417(1) Å).3 In the original work, the synthesis
was performed in EtOH solvent, but the concentrations of the
bimetallic complexes formed were not optimized to get the
cleanest possible separation. We had to resort to recrystalliza-
tions from THF that gave rac- and meso-Ni2Cl4(et,ph-P4)
crystals with different morphologies. Picking out single crystals
under a microscope allowed us to get pure 2r and 2m. This is
clearly not suitable for larger-scale separations.
(10) Wieland, G. H.; Tremelling, M. J. Org. Chem. 1972, 37, 914.

Separating Diastereomers Using Nickel Chloride
Inorganic Chemistry, Vol. 40, No. 19, 2001 5039
Not surprisingly, higher concentrations of CN- promote better
yields for the liberation of the et,ph-P4 ligand. The low yields
of isolated 1r in the previous work3 resulted from the formation
of a stable monometallic nickel complex, rac,trans-Ni(CN)2-
(η2.5-et,ph-P4), 3r, which will be discussed more fully below.
The stability of this intermediate complex requires higher
concentrations of CN- in order to fully liberate the rac-
et,ph-P4 ligand. Extensive experimentation with differing
concentrations of CN- has resulted in a procedure that employs
two separate CN- additions. The first addition of 133 equiv
serves to cleanly release one of the nickel centers from 2 to
produce trans-Ni(CN)2(η2.5-et,ph-P4) and orange, square-planar
[Ni(CN)4]2-. Addition of another 150 equiv of CN- frees the
remaining nickel center to give free rac-et,ph-P4 ligand. The
higher cyanide ion concentration also favors the formation of
the pentacyano-complex [Ni(CN)5]3- as evidenced by the
solution turning from orange to red.
For the intermediate monometallic complex 3r to efficiently
react with CN-, it must be completely dissolved. The rac,trans-
Ni(CN)2(η2.5-et,ph-P4) is insoluble in H2O but is soluble in
MeOH. Therefore, a 2.5:1 ratio of H2O/MeOH is employed,
which keeps rac,trans-Ni(CN)2(η2.5-et,ph-P4) in solution. The
removal of the free ligand from the H2O/MeOH solvent is
achieved via extractions with benzene. Hexane can also be
employed to extract the ligand; however, yields are 20-30%
lower. The use of a H2O/MeOH mixed solvent to keep 3r
dissolved, along with a second CN- addition, gives consistent
and good to excellent isolated yields of the rac-et,ph-P4 ligand.
Cyanolysis of meso-Ni2Cl4(et,ph-P4). The removal of the
nickel from 2m through cyanolysis is more difficult because of
an unexpected partial meso- to rac-et,ph-P4 ligand isomerization
side reaction that will be discussed below. Although 1m is not
particularly useful as a ligand for dirhodium-catalyzed hydro-
formylation, it is potentially useful in other applications. To
minimize the meso- to rac-et,ph-P4 isomerization, a 1:1 H2O/
MeOH solution of meso-Ni2Cl4(et,ph-P4), 2m, is added drop-
wise to a well-stirred, saturated NaCN solution in 2:1 H2O/
MeOH. When 2m is added, it quickly reacts with cyanide to
form a mixture of [Ni(CN)4]2- and [Ni(CN)5]3-, along with free
meso-et,ph-P4, 1m. The CN- saturation of the aqueous layer
aids the extraction process and increases the yield of the isolated
1m by 5-10%.
Benzene and hexane are both suitable solvents for the
extraction of 1m, but it is considerably more soluble in the
aqueous phase than 1r. The yield, therefore, of meso P4 ligand
from the aqueous extraction process is correspondingly lower:
45-65% for 1m, compared to 75-87% for 1r. The meso ligand
is evident in 31P NMR spectra of the aqueous layer both before
and after the extraction. In addition, the purity of the extracted
meso-et,ph-P4 is typically lower than that of 1r because of the
meso to rac ligand isomerization that occurs. Recrystallization
of the extracted meso-et,ph-P4 ligand from hexane, however,
does provide pure 1m, typically in isolated yields of 30-60%.
The 31P{1H} NMR spectra of mixed et,ph-P4, 1m, and 1r are
available in Supporting Information.
Figure 1. ORTEP plots of rac-Ni(CN)2(η2.5-et,ph-P4), 3r (top), and
the corresponding meso complex, 3m (bottom). Hydrogen atoms are
omitted for clarity. The carbon and nitrogen atoms in 3m were refined
isotropically.
determination revealed the presence of monometallic rac,trans-
Ni(CN)2(η2.5-et,ph-P4). More reproducible production of these
monometallic intermediates in the cyanolysis procedure is
obtained via the use of only 3 equiv of CN- in the cyanolysis
of rac- or meso-Ni2Cl4(et,ph-P4) allowing the isolation and
recrystallization of rac,trans- and meso,trans-Ni(CN)2(η2.5
-
et,ph-P4), 3r and 3m, in 55% yield, both of which have been
structurally characterized.
The structures (Figure 1) show that these monometallic
complexes have square-pyramidal coordination geometries with
two basal cyanide ligands trans to one another on the Ni(+2)
center. The other two basal coordination sites are occupied by
the external phosphorus atoms of the trans-spanning et,ph-P4
ligand, which forms a 10-membered ring system about the nickel
center. Trans-spanning phosphine ligand-metal complexes are
relatively well-known.12 Most have been characterized with
group 8-10 metals,13 but at least one Mo complex is known.14
Our system is unusual in that one of the internal phosphorus
atoms of the et,ph-P4 ligand forms a longer and weaker Ni-P
interaction (Ni-P1 ) 2.395(2) Å for 3r, 2.3526(9) Å for 3m)
meso,trans- and rac,trans-[Ni(CN)2(η2.5-et,ph-P4)], 3m and
3r. During our original studies into the cyanolysis of the rac-
Ni2Cl4(et,ph-P4) solution, a two-phase H2O/benzene solvent
system was used.3 A distinct yellow color almost always rapidly
appeared in the organic phase, and when this occurred, low
yields of the isolated et,ph-P4-ligand resulted. Concentration
of the yellow benzene solution yielded orange-yellow crystals,
which were recrystallized from THF. A single-crystal structural
(12) Compare: Bessel, C. A.; Aggarwal, P.; Marschilok, A. C.; Takeuchi,
K. J. Chem. ReV. 2001, 101, 1031-1066.
(13) Compare: (a) Kapoor, P. N.; Pregosin, P. S.; Venanzi, L. M. HelV.
Chim. Acta 1982, 65, 654-660. (b) Elding, L. I.; Kellenberger, B.;
Venanzi, L. M. HelV. Chim. Acta 1983, 66, 1676-1690. (c) Kapoor,
P. N. J. Organomet. Chem. 1988, 341, 363-366. (d) Tani, K.; Yabuta,
M.; Nakamura, S.; Yamagata, T. J. Chem. Soc., Dalton Trans. 1993,
2781-2789. (e) Gray, G. M.; Varshney, A.; Duffey, C. H. Organo-
metallics 1995, 14, 238-244.
(11) Kramer, V. W. U.S. Patent 2,773,752, 1957.
(14) Gray, G. M.; Duffey, C. H. Organometallics 1994, 13, 1542-1544.

5040 Inorganic Chemistry, Vol. 40, No. 19, 2001
Aubry et al.
Table 2. Selected Bond Distances (Å) and Angles (deg) for
rac,trans- and meso,trans-Ni(CN)2(η2-et,ph-P4), 3r and 3m
the internal and external phosphorus atoms. The -14.7 ppm
resonance corresponds to the internal phosphines and lies almost
exactly intermediate between the uncoordinated P3 resonance
(-23.5 ppm) and the weak apically coordinated P1 atom in 3m
(15.8 ppm). The exchange between the internal phosphorus
atoms can be slowed, but not stopped, by lowering the
temperature to -80 °C. New resonances are observed in the
31P{1H} NMR at -80 °C, but the coupling constants could
not be obtained because of line broadening, probably caused
by partial precipitation of the complex at this low tempera-
ture.
The room-temperature solution 31P NMR spectrum of the
meso monometallic complex 3m, on the other hand, shows four
somewhat broad resonances indicating an η2.5-meso-et,ph-P4-
bound static structure consistent with that seen in the ORTEP
plot (Figure 1). The uncoordinated lone pair on phosphorus atom
P3 in 3m is oriented almost completely opposite from the nickel
center and cannot coordinate easily to the metal. This is in
marked contrast to the P3 lone pair in 3r, which, on the basis
of the dynamic NMR behavior, can very readily coordinate and
dissociate. The somewhat broad nature of the 31P resonance for
3m probably indicates some fluxional behavior but on a rather
different time scale relative to 3r. This stereochemical difference
leads to a rather unusual reaction in the case of Ni(NCS)2 in
which we see a nickel-mediated selective sulfur transfer reaction
from thiocyanate to the racemic P3 phosphorus atom to produce
a PdS group. This thiocyanate sulfur transfer chemistry is not
seen for the meso mono- or bimetallic Ni-et,ph-P4 ligand
complexes.9 This is one reason we abandoned the Ni(NCS)2-
based et,ph-P4 separation procedures and concentrated on the
much cleaner NiCl2 chemistry.
Isomerization of meso- to rac-et,ph-P4. During the inves-
tigation of the cyanolysis of the meso-Ni2Cl4(et,ph-P4) to
produce pure meso-et,ph-P4, 1m, unexpectedly high amounts
of racemic ligand were occasionally obtained (50-70% racemic
ligand), although the starting Ni2Cl4(et,ph-P4) complex used was
pure meso based on 31P NMR spectra. There turns out to be an
isomerization pathway available that can partially convert 3m
to 3r during the cyanolysis. The monometallic intermediate 3r
appears to be thermodynamically favored over 3m, and that,
coupled with a kinetically accessible pathway, allows isomer-
ization from meso- to rac-et,ph-P4. The question naturally arose
whether it would be possible to isolate pure rac-et,ph-P4 ligand
(1r) from meso-Ni2Cl4(et,ph-P4), 2m. This could represent
a useful pathway for producing more of the valuable rac-
et,ph-P4 ligand.
Slow dropwise addition of NaCN to 2m in H2O provides the
optimal condition to form the maximum amount of the rac,-
trans-Ni(CN)2(η2.5-et,ph-P4), 3r, intermediate. During the ad-
dition, 3r forms as an orange precipitate. Further addition of
NaCN to this solution frees the et,ph-P4 ligand. However, the
highest conversion achieved was only 70:30 racemic to meso.
This ratio appears to be thermodynamically controlled for this
Ni-based system. We have observed complete meso to racemic
ligand isomerization in the case of reacting Rh(+1) complexes
(1 equiv of [Rh(nbd)2](BF4) (nbd ) norbornadiene) or 0.5 equiv
of Rh2(µ-Cl)2(CO)4) with 1 equiv of mixed or meso-et,ph-P4
ligand in CH2Cl2. The monometallic racemic Rh(+3) complex
[rac-RhCl2(η4-et,ph-P4)]+ is produced in nearly quantitative
yield.16 This selective metal-assisted stereochemical isomeriza-
tion is highly unusual and is the first reported example for a
phosphine ligand.
3r
3m
Ni-P1
2.395(2)
2.183(2)
2.186(2)
1.876(7)
1.859(7)
1.164(7)
1.156(7)
87.54(7)
105.67(7)
94.6(2)
107.3(2)
165.48(8)
85.6(2)
91.9(2)
87.3(2)
2.3526(9)
2.1900(9)
2.1829(9)
1.865(3)
1.880(3)
1.163(4)
Ni-P2
Ni-P4
Ni-C1
Ni-C2
N1-C1
N2-C2
1.149(4)
88.07(3)
P1-Ni-P2
P1-Ni-P4
P1-Ni-C1
P1-Ni-C2
P2-Ni-P4
P2-Ni-C1
P2-Ni-C2
P4-Ni-C1
P4-Ni-C2
C1-Ni-C2
Ni-C1-N1
Ni-C2-N2
P1-C′-P3
102.93(3)
106.69(10)
97.36(10)
167.58(3)
91.82(10)
86.85(9)
90.51(10)
86.04(9)
155.86(13)
177.4(3)
89.9(2)
157.9(3)
176.0(3)
178.7(3)
122.1(3)
178.0(3)
115.82(16)
with the apical coordination site. Weak apical interactions in
Ni(+2) complexes with phosphine ligands have been reported
before by several groups.15 This weaker apical phosphine
coordination is why we designate the coordination of the
et,ph-P4 ligand as η2.5. Selected bond distances and angles for
rac,trans- and meso,trans-[Ni(CN)2(η2.5-et,ph-P4)] are listed in
Table 2.
The compound trans-Ni(CN)2(η2.5-et,ph-P4) can be described
as a partially liberated monometallic complex in which the
cyanide anions have only displaced one nickel center, with the
ligand rearranging to an η2.5-coordination mode. The solubility
of this complex in organic solvents limits further attack of
aqueous cyanide ion and nicely explains the relatively low yield
of isolated et,ph-P4 ligand in the original H2O/benzene solvent
cyanolysis procedures. The IR spectrum has a strong absorption
at 2096 cm-1 for 3r and 2119 cm-1 for 3m, characteristic of
the CtN stretches.
In the solid state, 3r contains four inequivalent phosphorus
nuclei. This is not the case, however, in solution where the room-
temperature 31P{1H} NMR spectrum only shows two sharp
singlets. These singlets are produced from a temperature-
dependent solution fluxional process where the weak apical
Ni-P bond is easily broken. At room temperature, the ligand
on the molecule is rapidly wagging in solution so both internal
phosphorus atoms alternate in their weak bonding interactions
with the metal center. This generates an average symmetrical
structure, 3r*, that yields the simple singlets observed for both
(15) (a) Stalick, J. K.; Ibers, J. A. Inorg. Chem. 1969, 8, 1084. (b) Stalick,
J. K.; Ibers, J. A. Inorg. Chem. 1969, 8, 1090. (c) Raymond, K. N.;
Corfield, P. W. R.; Ibers, J. A. Inorg. Chem. 1968, 7, 1362. (d)
Hirshfeld, F. L.; Hope, H. Acta Crystallogr., Sect. B 1980, B36, 406.
(e) Allen, D. W.; Mann, F. G.; Millar, I. T.; Powell, H. M.; Watkins,
D. J. J. Chem. Soc. D 1969, 1004. (f) Hope, H.; Olmstead, M. M.;
Power, P. P.; Viggiano, M. Inorg. Chem. 1984, 23, 326,
(16) Hunt, C., Jr.; Fronczek, F. R.; Billodeaux, D. R.; Stanley, G. G. Inorg.
Chem., in press.

Separating Diastereomers Using Nickel Chloride
Inorganic Chemistry, Vol. 40, No. 19, 2001 5041
The stronger Rh-P bonding and increased Rh(+3) octahedral
coordination environment that allows full η4-rac-et,ph-P4
coordination to the metal center provide a considerably greater
thermodynamic impetus for isomerizing the meso to racemic
ligand. The meso ligand cannot readily coordinate in an η4-
fashion to a single metal center. Molecular modeling studies
showed that the η4-meso-et,ph-P4 coordination was 13 kcal/
mol higher in energy relative to that of the monometallic racemic
ligand complex.16 Unfortunately, the [rac-RhCl2(η4-et,ph-P4)]+
complex is extremely robust, and we have not yet been able to
release the rac-et,ph-P4 ligand or convert it into an active
bimetallic catalyst precursor complex.
Similarly, free meso-et,ph-P4 ligand is not isomerized simply
by cyanide in solution. The stereochemical inversion of the one
internal meso phosphorus atom is most likely tied into the trans-
2
spanning structure present in 3m. The presence of filled dz and
empty pz orbitals on the Ni(+2) center and their ability to
interact with the internal phosphorus atoms of the et,ph-P4
ligand in 3 should assist the inversion of the one internal
phosphorus center to form the somewhat better coordinating
rac-et,ph-P4 ligand. The 31P NMR spectrum of pure 3m shows
little evidence of conversion from 3m to 3r in solution at room
temperature, arguing for the active role of CN- anions in
conjunction with 3m for this isomerization process.
The current work strongly suggests that the formation of the
rac,trans- and meso,trans-Ni(CN)2(η2.5-et,ph-P4) intermediates
promotes the partial isomerization of the meso to the racemic
ligand diastereomer during the cyanolysis procedure. Increasing
the CN- concentration minimizes the amount and/or lifetime
of 3m limiting the meso to racemic ligand isomerization which
is important if one wants to isolate meso-et,ph-P4. The bimetallic
complex meso-Ni2Cl4(et,ph-P4), 2m, is stable in cyanide-free
solutions and does not show any tendency to isomerize.
Acknowledgment. This work was supported by the National
Science Foundation (Grants CHE-92-01051 and CHE-97-
00526).
Supporting Information Available: CIF files for 3r and 3m and
31P{1H} NMR spectra of mixed, rac-, and meso-et,ph-P4 ligands. This
material is available free of charge via the Internet at http://pubs.acs.org.
IC010035F