Nickel(II)-carbene intermediates in reactions of geminal dihaloalkanes with nickel(0) reagents and the corresponding carbene capture as the phosphonium ylide

Article abstract of DOI:10.1002/ejic.200601106

In a previous study of geminal bond cleavages of substrates of the type R₂CE₂ [E₂ = X₂, O, S, Li(SO ₂Ph)] by nickel(0) reagents [L_nNi, L_n = (Cod)₂, (Et₄P)₄], leading to R ₂C=CR₂ as products, the tentative hypothesis had been proposed that such reactions likely proceed via nickel(0)-carbene intermediates (J. J. Eisch, Y. Qian, M. Singh, J. Organomet. Chem. 1996, 512, 207). Because such proposed nickel(0)-carbenes do not satisfactorily account for reactions encountered with such α-eliminations, a detailed reexamination of the reaction of these nickel(0) reagents with geminal dihalides has been undertaken. For example, two reactions of such presumed nickel(0)-carbenes remained anomalous: (1) the failure of 5,5-dibromotetraphenylcyclopentadiene to form its expected dimer, octaphenylphenylfulvalene and instead the formation of triethylphosphonium tetraphenylcyclopentadienide in its reaction with (Et ₃P)₄Ni; and (2) the presumed capture of intermediate R₂C=Ni⁰ in presence of the trapping agent (benzaldehyde or benzophenone). As to the first anomaly, a detailed study has shown that no trace of octaphenylfulvalene was formed. As to the second anomaly, the R ₂C fragment could be trapped by the carbonyl reagents only in reactions involving (Et₃P)₄Ni, but not in reactions with (Cod)₂Ni. This finding compels one to conclude that the carbonyl reagent is capturing the Wittig reagent, R₂C=PEt₃, and not R₂C=Ni⁰. Based upon all present data, the mechanism of such C=C bond dimerizations is best explicable in terms of nickel(II)-carbenes. The triethylphosphonium tetraphenylcyclopentadienide formed here has by single-crystal X-ray structure determination, complemented by ¹³C NMR spectroscopic data, been found to have the zwitterionic structure Ph ₄Cp^--+PEt₃ as its paramount resonance contributor. Wiley-VCH Verlag GmbH & Co. KGaA, 2007.

Full text of DOI:10.1002/ejic.200601106

FULL PAPER
DOI: 10.1002/ejic.200601106
Nickel(II)-Carbene Intermediates in Reactions of Geminal Dihaloalkanes with
Nickel(0) Reagents and the Corresponding Carbene Capture as the
Phosphonium Ylide[‡]
John J. Eisch,*[a] Yun Qian,[a] and Arnold L. Rheingold[b]
Keywords: Nickel(0) reagents / Nickel(II)-carbenes / Phosphonium ylides / α-Elimination / Carbon-carbon bond formation
In a previous study of geminal bond cleavages of substrates
of the type R2CE2 [E2 = X2, O, S, Li(SO2Ph)] by nickel(0) rea-
gents [LnNi, Ln = (Cod)2, (Et4P)4], leading to R2C=CR2 as
products, the tentative hypothesis had been proposed that
such reactions likely proceed via nickel(0)-carbene interme-
diates (J. J. Eisch, Y. Qian, M. Singh, J. Organomet. Chem.
1996, 512, 207). Because such proposed nickel(0)-carbenes
do not satisfactorily account for reactions encountered with
such α-eliminations, a detailed reexamination of the reaction
of these nickel(0) reagents with geminal dihalides has been
undertaken. For example, two reactions of such presumed
nickel(0)-carbenes remained anomalous: (1) the failure of
5,5-dibromotetraphenylcyclopentadiene to form its expected
dimer, octaphenylphenylfulvalene and instead the formation
of triethylphosphonium tetraphenylcyclopentadienide in its
reaction with (Et3P)4Ni; and (2) the presumed capture of in-
termediate R2C=Ni0 in presence of the trapping agent (benz-
aldehyde or benzophenone). As to the first anomaly, a de-
tailed study has shown that no trace of octaphenylfulvalene
was formed. As to the second anomaly, the R2C fragment
could be trapped by the carbonyl reagents only in reactions
involving (Et3P)4Ni, but not in reactions with (Cod)2Ni. This
finding compels one to conclude that the carbonyl reagent
is capturing the Wittig reagent, R2C=PEt3, and not R2C=Ni0.
Based upon all present data, the mechanism of such C=C
bond dimerizations is best explicable in terms of nickel(II)-
carbenes. The triethylphosphonium tetraphenylcyclopen-
tadienide formed here has by single-crystal X-ray structure
determination, complemented by 13C NMR spectroscopic
data, been found to have the zwitterionic structure Ph4Cp––
+PEt3 as its paramount resonance contributor.
(© Wiley-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim,
Germany, 2007)
carbon π-bond coupling also appears to involve such oxi-
dative addition by nickel(0) to produce 5[7] (Scheme 1).
However, in this situation adduct 5 could lead to C=C bond
coupling via two distinctly different routes: (1) 5 could itself
undergo an α-elimination yielding a nickel(II)-carbene com-
plex (6); or (2) alternatively, 5 could undergo a second oxi-
dative addition with 1 to produce dinickelaalkane 7, which
upon elimination of NiAB (8) would yield nickel(0)-carbene
complex 9. Then via a dimerization-elimination of nickel-
(II)-carbene 6 (path 1) or by a coupling process through
nickel(0)-carbene 9 and 4 (path 2) the olefin dimer 10 would
result. This previous global investigation of α-eliminations
of geminally disubstituted hydrocarbons by nickel(0) rea-
Introduction
The oxidative addition of nickel(0) complexes (LnNi, 1)
into σ-covalent bonds between carbon and a wide variety
of heteroatoms or even another carbon [2; E = X, O (of
epoxide ring or allylic ether), N, C, S, P, Al][2,3] produces
organonickel intermediates (3) pivotal in olefin isomeriza-
tion,[3] olefin and diolefin oligomerization,[3] desulfurization
of aromatic sulfur heterocycles,[4] epoxide deoxygenation[5]
and the Kumada carbon–carbon cross coupling[6] [Equa-
tion (1)].
(1)
A recent study of nickel(0)-induced α-eliminations from
geminal disubstituted alkanes 4 leading directly to carbon–
[‡] Organic Chemistry of Subvalent Transition Metal Complexes,
40; Part 39: Ref.[1]
[a] State University of New York at Binghamton, Department of
Chemistry,
P. O. Box 6000, Binghamton, NY 13902-6000, USA
Fax: +1-607-777-4865
E-mail: jjeisch@binghamton.edu
[b] University of California, San Diego, Department of Chemistry
and Biochemistry,
9500 Gilman Drive, m/c 0358, LaJolla, CA 92093-0332, USA
Scheme 1.
1576
© 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Inorg. Chem. 2007, 1576–1584

Reactions of Geminal Dihaloalkanes with Nickel(0) Reagents
FULL PAPER
gents has embraced in its survey dihalo, sulfonyl lithio, ke- pentadiene (19) by tetrakis(triethylphosphane)nickel(0),
tonic, thioketonic and even cyclopropene reagents.[7]
(Et3P)4Ni (20). Instead of dimerizing to octaphenylfulva-
Since the ring-opening and alkyne-insertion reactions of lene (21), the corresponding phosphonium ylide or Wittig
1,2,3-triphenylcyclopropene (11) with nickel(0) complexes reagent 22c was formed in high yield[7] (Scheme 4). Any ac-
to produce intermediate 12 seemed to be explicable only in ceptable reaction mechanism, in our opinion, would have
terms of path 2 (Scheme 2; Diels–Alder capture of 12 by to explain how such Wittig reagents could readily arise in
diphenylacetylene and subsequent Ni0-elimination to yield the course of such couplings. We will show that only the
13), we made the preliminary hypothesis that all such gemi- intermediacy of nickel(II)-carbene complexes (Scheme 1;
nal disubstituted derivatives would similarly involve path 1: 4 Ǟ 5 Ǟ 6 Ǟ 10) can readily lead to Wittig reagents
nickel(0)-carbene intermediates.
in situ.
Scheme 2.
But in extending path 2 to the reactions of LnNi (1) with
geminal dichlorides, for example, one would have to invoke
the unlikely polarization 14 to explain the coupling with
R2CCl2 (15) to form olefin dimer 10 or the purported cap-
ture of 9 as 17 with an aldehyde or ketone R2C=O (16),
benzaldehyde or benzophenone for 9 in ref.[7] (Scheme 3).
On the contrary, with late transition metal or Fischer carb-
enes, the carbene center is known to exhibit electrophilic
(18), rather than nucleophilic (14) character. Accordingly,
resonance structure 14 would be only a minor contributor
to the actual π-electron distribution in 9.
Scheme 4.
Results and Discussion
Anomalous Reaction of 5,5-Dibromotetraphenylcyclo-
pentadiene (19) with (Et3P)4Ni (20)
The reaction mixture of a 1:3 molar ratio of 19 and 20
in THF at room temp. was freed of volatiles and the re-
sulting residue was extracted with warm cyclohexane. The
extract deposited 70% of triethylphosphonium tetraphen-
ylcyclopentadienide (22c).[8a] From the concentrated and
cooled cyclohexane filtrate were obtained 80% of the ex-
pected (Et3P)2NiBr2. Finally, treating the last filtrate with
deoxygenated 6  aqueous HCl led upon hydrolytic workup
to the isolation of 1,2,3,4-tetraphenylcyclopentadiene (23,
24%) as the only detectable organic product with no sign
of fulvalene 21. The hydrocarbon 23 can be considered as
the expected reductive cleavage product of phosphonium
ylide 22c by 20 and acid, shown in Scheme 5, a reaction
having ample precedent.[8b] From this detailed analysis of
Scheme 3.
Because of these incongruities between the experimental the reaction products, we therefore conclude that the forma-
observations made in the reactions of geminal dihalides tion of phosphonium ylide in Scheme 3 has proceeded es-
(e.g. 15) with nickel(0) reagents 1 and the postulated role of sentially to completion without the generation of any 21.
nickel(0)-carbenes (Scheme 1; path 2: 4 Ǟ 5 Ǟ 7 Ǟ 9 Ǟ The failure of any fulvalene like 21 to form in such a reac-
10), we have reinvestigated these reactions in a more search- tion cannot be ascribed to any inherent instability of these
ing manner. We have tried to detect and identify all possible substituted fulvalenes. Both 1,2,3,4-tetraphenylfulvalene
by-products and side products. Most of all, we have striven (m.p. 202 °C)[9a] and octaphenylfulvalene [21, m.p. 170 °C
to explain the anomalous course of the attempted carbon– (dec.)][9b] have been prepared as brilliant orange-red solids
carbon π-bond coupling of 5,5-dibromotetraphenylcyclo- and have been found to be indefinitely stable below 100 °C.
Scheme 5.
Eur. J. Inorg. Chem. 2007, 1576–1584
© 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjic.org
1577

J. J. Eisch, Y. Qian, A. L. Rheingold
FULL PAPER
The Stability and Structure of Triethylphosphonium
Tetraphenylcyclopentadienide (22c)
Table 1. Selected bond lengths and angles of triethylphosphonium
tetraphenylcyclopentadienide (22c).
Bond lengths [Å]
Although the phosphonium ylide 22c has all the struc-
tural features of a typical Wittig reagent, it is stable to air
and moisture when stored at room temperature and does
not react with either benzaldehyde or benzophenone in
P–C(5)
1.756(5)
1.807(5)
1.812(6)
1.825(5)
1.399(6)
1.425(6)
C(1)–C(16)
C(2)–C(3)
C(2)–C(26)
C(3)–C(4)
C(3)–C(36)
C(4)–C(5)
1.492(6)
1.420(7)
1.483(7)
1.400(6)
1.476(6)
1.428(6)
P–C(51)
P–C(53)
P–C(55)
warm cyclohexane to give a fulvene, the expected Wittig C(1)–C(2)
methylenation reaction product. But the corresponding tri-
C(1)–C(5)
phenylphosphonium cyclopentadienide has also been
shown by Ramirez and Levy not to react with cyclohexa-
none even at elevated temperatures.[10] In both instances,
the delocalization of the anionic charge and Hückel aroma-
ticity of such a cyclopentadienyl ring could be viewed as
the cause for the inertness of such Wittig reagents.
Bond angles [°]
C(5)–P–C(51)
C(5)–P–C(53)
C(5)–P–C(55)
C(2)–C(1)–C(5)
C(1)–C(2)–C(3)
112.3(3)
112.0(3)
113.6(2)
107.4(4)
108.6(4)
C(4)–C(3)–C(2) 108.5(4)
C(3)–C(4)–C(5) 107.4(5)
C(1)–C(5)–C(4) 108.1(4)
C(1)–C(5)–P
C(4)–C(5)–P
126.6(4)
125.0(4)
The single-crystal X-ray structure determination of phos-
phonium ylide 22c (Figure 1 and Tables 1 and 2) sheds light
on the extent of delocalization of the negative charge in the
cyclopentadienyl moiety of 22c, namely the relative impor-
tance of resonance structures 22a, 22b and 22c, as well as
the relative importance of the neutral phosphorane reso-
nance structure 22d, often invoked with Wittig reagents. It
should be noted that this XRD structure of 22c is not the
first trisubstituted phosphonium cyclopentadienide whose
structure has been determined in the crystalline state. The
crystal structure of triphenylphosphonium cyclopentadien-
ide (24) has been determined by X-ray crystallography and
the question of the degree of ylide-ylene character in the
cyclopentadienyl carbon–phosphorus bond carefully ad-
dressed.[11a] This prior work has been most valuable in our
present attempt to evaluate the relative contributions of res-
onance structures 22a–22d.
Table 2. Crystal data and structure refinement for triethylphospho-
nium tetraphenylcyclopentadienide (22c).
Empirical formula
Formula weight
Diffractometer
Temperature
C35H35
486.60
P
Siemens P4 (scintillation counter)
296(2) K
Wavelength
0.71073 Å
Crystal system
Space group
monoclinic
P2(1)
Unit cell dimensions
a = 10.757(2) Å; α = 90°
b = 11.076(2) Å; β = 94.57(2)°
c = 11.988(3) Å; γ = 90°
Volume
1423.8(5) Å3
Z
2
Density (calculated)
Absorption coefficient
F(000)
1.135 mg/m3
0.117 mm–1
520
Crystal size
0.32ϫ0.28ϫ0.24 mm
θ range for data collection
Index ranges
2.45–22.50°
–11Յ hՅ 11, 0Յ kՅ 11, 0Յ lՅ 12
Reflections collected
Independent reflections
Completeness to θ = 22.50°
Absorption correction
2076
1970 [R(int) = 0.0205]
99.6%
none
Max. and min. transmission 0.9724 and 0.9635
Refinement method
full-matrix least squares on F2
Data/restraints/parameters
Goodness-of-fit on F2
Final R indices [IϾ2σ(I)]
R indices (all data)
1970/1/325
0.795
R1 = 0.0371, wR2 = 0.0645
R1 = 0.0715, wR2 = 0.0722
Absolute structure parameter 0.04(17)
Largest diff. peak and hole
0.118 and –0.168 eÅ–3
that of the similar phosphonium ylide 24 and that of naph-
thalene derivative 25.[13a]
In the first place, the clearly alternating C–C bond
lengths of the cyclopentadienyl ring (1.40Ϯ0.02 Å) of 22a–
d indicate that resonance structures 22a and 22b make rela-
Figure 1. ORTEP projection of triethylphosphonium tetraphen-
ylcyclopentadienide (22c) with ellipsoids at the 30% probability
level.
In Table 1 are listed the experimentally determined C(5)– tively lesser contributions to the π-electron distribution. In
P and C–C bond lengths of greatest relevance in evaluating the cyclopentadienyl anion of KCp(18-crown-6), however,
the relative contributions of resonance structures 22a–22d. the ring C–C bond lengths range from 1.37 to 1.41 Å with
The observed C–C bond lengths within the cyclopen- a mean of 1.40 Å. The distances in 22a–d might be consid-
tadienyl ring, as shown in 22d, can be profitably compared ered the same within experimental error, if their bond
with those of the cyclopentadienyl anion.[12] Of analogous length alternation did not so nicely parallel the C–C bond-
use is a comparison of C(5)–P bond length in 22a–d with ing in resonance structures 22c or 22d.
1578
www.eurjic.org
© 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Inorg. Chem. 2007, 1576–1584

Reactions of Geminal Dihaloalkanes with Nickel(0) Reagents
FULL PAPER
presence of electron-donating ethyl groups on P, instead of
phenyl groups, stabilizing the cationic P-center; and (2) the
presence of π-electron-withdrawing phenyl groups on the
cyclopentadienyl ring, instead of hydrogens, stabilizing the
anionic C-center. Assigning a reliable percentage of ylene-
ylide character to the C–P bond, however, seems to be beset
with great uncertainty.[11b]
The 13C NMR spectrum of 22c provides confirmation of
the higher charge density of carbon C(5) over that of cen-
ters C(1) or C(4).[13b] The doublet centered at δ =
124.85 ppm displays the largest 13C-31P coupling (60 Hz)
whereas the doublets at δ = 127.10 ppm and 127.80 ppm
are split by 25 Hz and 18 Hz, respectively. As expected, the
doublet with the largest splitting must be assigned to C(5),
which is closest to the phosphorus. Since C(5) is the closest
to phosphorus and its signal is the most shielded, it would
thus have a higher electron density than those of C(1) and
C(4).
Possible Mechanistic Routes to Triethylphosphonium
Tetraphenylcyclopentadienide (22c)
The question of ionic or double character of the car-
bene–carbon bonding with phosphorus can reasonably be
judged on the basis of the carbene C–P bond in 25, one of
Two quite divergent pathways leading from 5,5-dibro-
the few open-chain phosphoranes with alkyl substituents, motetraphenylcyclopentadiene (19) and (Et3P)4Ni (20) to
whose XRD structures have been measured. At 1.69 Å the 22c can be envisioned, the one an integral part of the
C–P bond in 25 should have more C=P bond character nickel(II)-carbene pathway proposed here for the coupling
than that in 22a–d, 0.08 Å longer. Another indication of of geminal dihalides to yield dimeric olefins via intermedi-
lessened C=P bonding in 22a–d results from comparing the ates analogous to 26 and 27 (cf. infra) (path a; Scheme 6)
bond lengths of the C–P bonds of the three ethyl groups and the other involving an initially formed quarternary
attached to phosphorus, 1.81 Å, with the carbene C(5)–P phosphonium salt (28), followed by oxidative addition of
bond, 1.77 Å. A change in the C(sp3) hybridization of the Ni0 from 20 to produce 29 and concluded by elimination of
ethyl in 22 to the C(sp2) hybridization of the phenyl in NiBr2 (path b).[14] Clearly, the latter route would involve no
Ph4P+ (1.81 Ǟ 1.79 Å) suggests that the sp2-hybridized car- nickel-carbene complex as such.
bene-carbon–phosphorus bond length would be 1.79 Å for
As will be discussed below, there is ample precedent for
an ordinary single C–P bond. This length is so close to the the coupling of geminal dihalides by Ni0 reagents, either
observed 1.77 Å as to indicate that there is markedly less (Et3P)4Ni (20) or (Cod)2Ni (34), via intermediates like 26
phosphorane character in 22, as suggested by 22c than there and 27, when steric factors do not prevent their coupling.
is in 25 (C–P bond of 1.69 Å) or even in 24 (C–P bond of Thus, the feasibility of path a is evident. On the other hand,
1.72 Å). Therefore, the carbon–phosphorus linkage is best path b requires as its first step the slow quaternization of
represented by the single resonance structure 22c. The Et3P (20a) by 19 to form 28, a process known to require a
greater zwitterionic character in the carbene-carbon–phos- combination of higher temperatures, peroxides and/or ul-
phorus bond of 22c, compared with that of 24, can be at- traviolet radiation to be initiated and widely considered to
tributed to the following electronic stabilization: (1) the involve free-radical chain reactions.[15a,15b] By contrast, it
Scheme 6.
Eur. J. Inorg. Chem. 2007, 1576–1584
© 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjic.org
1579

J. J. Eisch, Y. Qian, A. L. Rheingold
FULL PAPER
should be noted that the reaction of 19 with 20 proceeds trapping product was detected, we suggest an alternative
readily at room temperature in oxygen- and peroxide-free and mechanically more reasonable source of 33. We now
THF with no special illumination. Furthermore, the reac- propose, instead, that a nickel(II)-carbene (36) is generated
tion of putative intermediate 28 with 20 to yield the neces- in these coupling reactions and that coupling occurs
sary 29 is obviously fraught with steric hindrance. Finally through the reaction between 36 and oxidative-addition
and decisively, the attempted interaction of 19 with 20a in product 35. The matched polarities of carbon–nickel bond
THF, with the conditions under which 19 and 20 react com- of 35 and the Fischer-carbene-carbon–nickel(II) bond of 36
pletely (18 h at 25 °C), failed altogether. Since this proposed should foster rapid C–C bond formation to yield 31. Only
first step of path b has been attempted in this study without in the C–C couplings employing reagent 20 would the tri-
success, path b can be dismissed as a viable alternative to ethylphosphane be present to undergo a competitive reac-
path a.
tion with the electrophilic carbon of nickel(II)-carbene 36,
leading to the displacement of nickel(II) chloride as its Et3P
complex and the generation of phosphane ylide 37
(Scheme 7). The reaction of benzophenone (32) with this
Wittig reagent, rather than the dubious reaction with the
previously suggested nickel(0)-carbene, is therefore con-
cluded to be the actual source of the triphenylethylene (33)
[Equation (4)].
Reactions of α,α-Dichlorotoluene (30) with Nickel(0)
Reagents
In our previous report we observed that a 1:2 mixture of
α,α-dichlorotoluene (30) and (Et3P)4Ni (20) reacted in THF
at 25 °C to give 98% of trans-stilbene (31) [Equation (2)].[16]
(2)
(4)
In a parallel reaction conducted at 0 °C, the 1:2 mixture
of 30 and 20 was allowed to react in the presence of 1 equiv.
of benzophenone (32). Hydrolytic workup led to the isola-
tion of 40% of 31 and 47% of triphenylethylene (33). The
formation of 33 was at first thought to have originated from
a Wittig-like reaction of the nickel(0)-carbene, PhCH=Ni,
with 32 through the elimination of NiO [Equation (3)].
The observation of phosphonium ylide 22c as the ulti-
mate Wittig-like reagent from dibromo compound 19 and
nickel reagent 20, as well as the aforementioned detection
of 37 in the reaction of geminal dichloro compound 30 with
20, show that phosphonium ylides can occur as intermedi-
ates in such C–C bond-coupling reactions. That they are
necessary intermediates, however, seems most unlikely.[17]
(3)
But the present findings, employing (Cod)2Ni (34) with
30, tend to rule out this interpretation. Certainly, a 1:2 mix-
ture of 30 and 34 reacted equally smoothly in THF at 25 °C
to produce a high yield of trans-stilbene (96%). But the cor-
responding attempted trapping experiment involving a 1:2:1
Generalized Mechanism for the Nickel(0)-Promoted
Coupling of Geminal Disubstituted Hydrocarbons
The foregoing coupling mechanism, having nickel(II)-
mixture of 30, 34 and 32 again led only to a high conversion carbenes as key intermediates, can be rationally applied to
to trans-stilbene (94%) with no detectable amount of the the other geminal dihalohydrocarbons already studied,
trapping product 33. If the nickel(0)-carbene, PhCH=Ni, namely dibromomethane, dichlorodiphenylmethane and
were the source of 33, it should also have been present in 5,5-dibromo-1,2,3,4-tetraphenylcyclopentadiene
(19)
this mixture and therefore have been trapped. Since no such (Scheme 8).
Scheme 7.
1580
www.eurjic.org
© 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Inorg. Chem. 2007, 1576–1584

Reactions of Geminal Dihaloalkanes with Nickel(0) Reagents
FULL PAPER
oxidative addition process depicted in Scheme 9, which
would lead to a nickel(II)-carbene (40) and thus to C–C
bond coupling analogous to that in Scheme 7.
Conclusions
Scheme 8.
The bond cleavages of geminal dihalides of the type
R2CX2 by nickel(0) reagents is best depicted as leading to
dimers R2C=CR2 via nickel(II)-carbenes by the following
sequence of reactions: (1) oxidative addition of Ni0 to pro-
duce XR2C–Ni–X;to provide the nickel(II)-carbene, R2C+–
–NiX2; (3) C=C bond formation between XR2Cδ––δ+NiX2
and R2C+––NiX2 because of opposite polarities. If step 3 is
too slow in the reactions involving (Et3P)4Ni, then any free
Et3P can displace the NiX2 from R2C+––NiX2 and capture
the R2C ligand as the Wittig reagent, R2C=PEt3. Such a
Wittig reagent, rather than the previously assumed
nickel(0)-carbene, R2C=Ni0, is responsible for the forma-
tion of R2C=CRЈRЈЈ when RЈRЈЈC=O is added to such re-
action mixtures.
Considering 38 as an appropriate activated complex,[18]
it is clear that when R and RЈ are space-filling groups as in
the reaction of 5,5-dibromotetraphenylcyclopentadiene (19)
(R and RЈ combined as a 1,4-tetraphenylbutadienylene
group), then C–C bond formation will be relatively slow
and the Et3P attack will form the Wittig reagent selectively.
With R and RЈ as phenyl groups, C–C bonding is still pos-
sible but backside attack at the pseudo-tertiary carbene-car-
bon by Et3P should be slow. This view is in agreement with
our previous finding that the reaction of (Et3P)4Ni with
dichlorodiphenylmethane in the presence of benzaldehyde
leads to 95% of tetraphenylethylene with no sign of triphen-
ylethylene, the expected Wittig reaction product expected to
arise from benzaldehyde and Ph2C=PEt3.[19]
Triethylphosphonium
tetraphenylcyclopentadienide
(22c), the Wittig reagent obtained in high yield from the
unsuccessful attempt to couple 5,5-dibromotetraphenylcy-
clopentadiene to produce octaphenylfulvalene (cf. supra:
slow step 3), was subjected to X-ray crystallographic struc-
ture determination. Such structural data, together with its
13C NMR chemical shifts and 13C-31P coupling constants,
are clearly consistent with the electronic structure 22c,
where the ring C(5)–P bond is essentially a zwitterionic sin-
gle C–P bond and the negative charge of the pseudo-cyclo-
pentadiene anion greatest on C(5).
Finally, with R = Ph and RЈ = H, as was presented in
Scheme 5, C–C bond formation would occur along the in-
ternuclear line in 38 with the phenyl groups trans to each
other for steric reasons. In this case, the carbene-carbon is
pseudo-secondary and hence more accessible to backside
attack by Et3P and thus the formation of the corresponding
Wittig reagent. With R and RЈ as H in dihalomethanes, the
frontside C–C coupling in 38 and the backside attack by
Et3P should be of comparable facility and therefore C–C
coupling and Wittig-reagent formation both are possible
and in fact observed.
The coupling mechanism operative with other geminal
disubstituted hydrocarbon derivatives, such as thioketones
and α-lithioorganic sulfones, is also most likely one involv-
ing analogous nickel(II)-carbene intermediates.[20] Even the
behavior of benzylic sulfones toward nickel(0) reagents indi-
cates the possible intermediacy of nickel(II)-carbene com-
plexes. Benzyl phenyl sulfone (39), for example, reacted with
(Cod)2Ni in toluene at 25 °C over an extended period to
give the deposition of nickel metal. Hydrolytic workup gave
a 40% yield of trans-stilbene (31) with no trace of bibenzyl
(Scheme 9). The absence of bibenzyl is consistent with the
absence of reductive cleavage of the C–S bond. The forma-
tion of trans-stilbene can be readily explained through the
Experimental Section
General Procedures and Techniques: All reactions were carried out
under a positive pressure of anhydrous, oxygen-free argon. All sol-
vents employed with organometallic compounds were dried and
distilled from a sodium metal/benzophenone ketyl mixture prior
to use. The preparation and purification of oxygen- and moisture-
sensitive reagents were carried out under argon with the use of
Schlenk techniques.[21] The IR spectra were recorded with a Perkin–
Elmer instrument, model 457 and samples were measured either as
mineral oil mulls or as KBr films. The NMR spectra (1H and 13C)
were recorded with a Bruker spectrometer, model AM-360 and tet-
ramethylsilane (TMS) was used as the internal standard. The chem-
ical shifts reported are expressed on the δ-scale and in parts per
million (ppm) from the reference TMS signal. The gas-chromato-
graphic analyses were carried out with a Hewlett–Packard instru-
ment, model 5880, provided with a 6 ft OV-101 packed column or
with a Hewlett–Packard instrument, model 5890, having a 30 m
SE-30 capillary column, respectively. Melting points were deter-
mined with a Thomas-Hoover Unimelt capillary melting point ap-
paratus and are uncorrected. The (Cod)2Ni was prepared according
to a procedure adapted from those of Semmelhack[22] and of
Schunn.[23] The yield of product has been improved to 95% by
purifying the starting materials with extreme care.
The (Et3P)4Ni was prepared according to the method of Cundy[24]
and the (Cod)(Bpy)Ni by the procedure of Dinjus et al.[25] The α,α-
dichlorotoluene, dichlorodiphenylmethane and dibromomethane
Scheme 9.
Eur. J. Inorg. Chem. 2007, 1576–1584
© 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjic.org
1581

J. J. Eisch, Y. Qian, A. L. Rheingold
FULL PAPER
were purchased at 98+% of purity and their purity further checked
by gas chromatography.
tracts were filtered, concentrated to 2/3 of volume and chilled to
deposit 1.24 g of amber-colored 22c (70%). Recrystallization from
cyclohexane yielded pale orange 22c, m.p. 200–202 °C, which was
submitted to X-ray single-crystal determination (Table 2). 1H
NMR (C6D6): δ = 0.41–0.50 (m, 9 H), 1.02–1.21 (m, 6 H), 6.82–
7.59 (m, 304) ppm. 13C NMR (C6D6): δ = 15.98 (d), 20.17 (3),
124.85 (d), 127.10 (d), 127.80 (m, high intensity), 132.45 (d) ppm.
The JC,P values of the doublets at δ = 124.85, 127.10 and
127.80 ppm are 60, 25 and 18 Hz, respectively.
The 5,5-dibromotetraphenylcyclopentadiende (19) was prepared for
the first time by the following procedure developed in this labora-
tory. A solution of 3.76 g (10.0 mmol) of 1,2,3,4-tetraphenylcyclo-
pentadiene (23) in 80 mL of CCl4 was treated with 3.65 g
(20.5 mmol) of N-bromosuccinimide (99%) and the suspension
heated at reflux until all suspended solid had risen to the surface
(90 min). During the heating, a small illuminated UV lamp was
focused on the reaction flask. The hot solution was then filtered
from the succinimide and the filtrate freed of solvent in a rotary
evaporator. The pale yellow solid residue of 5.1 g was essentially
pure 19 because the peak at δ = 4.0 ppm in the 1H NMR spectrum
of the starting diene was absent. Recrystallization from 95% etha-
nol yielded 4.4 g (83%) of pure 19, m.p. 154–155 °C. C29H20Br2
(528.30): C 65.31, H 3.78; found C 65.69, H 3.82.
The original cyclohexane filtrate from which 22c had crystallized
was concentrated to dryness under reduced pressure and the resi-
1
due dissolved in dry, deoxygenated C6H6. The H NMR spectrum
of this solution displayed only very broad absorptions, characteris-
tic of the presence of paramagnetic components. Hence, the entire
C6D6 solution was treated with deoxygenated 6  aqueous HCl.
Hydrolytic workup yielded 107 mg of a reddish solid, whose 1H
NMR spectrum showed it to be largely 23 (m.p. 180–181 °C, 24 %,
1H NMR peak at δ = 4.0 ppm).
The sample of octaphenylfulvalene (21) was prepared by the follow-
ing route. A suspension of 1.50 g (2.80 mmol) of 19, prepared in
the foregoing method, and 5.0 g (76 mmol) of zinc dust in 50 mL
of benzene was treated with a crystal of iodine as an activator and
was then stirred under reflux for 6 h. The resulting deep red suspen-
sion was filtered hot. The filtrate was freed of solvent in a rotary
evaporator to leave a deep red gummy residue. Stirring this residue
with hot portions of a 1:1 (v/v) mixture of benzene/95% ethanol
gave a deep red extract. Concentration and cooling of the extract
gave 710 mg (69%) of brilliant orange-red solid (21), which had
multiplets of proton signals between δ = 7.0 ppm and 7.4 ppm in
the 1H NMR spectrum and whose mass spectrum at 70eV had a
parent peak at m/z = 736, characteristic of fulvalene 21. No peaks
were present indicative of any 19 or any other bromine-containing
component (with isotopic 79Br and 81Br components). This sample
melted at 170 °C with decomposition, despite several recrystalli-
zations from 95% ethanol/toluene. C58H40 (736.77): calcd. C 94.54,
H 5.46; found C 94.10, H 5.16.
The deep red reaction residue remaining after the cyclohexane ex-
traction of the reaction mixture was redissolved in THF and cooled
to –78 °C. Long dark red needles of (Et3P)2NiBr2 were formed
(440 mg, 80%).[26] The 1H NMR spectrum showed no aromatic
proton absorptions.
Attempted Reaction of 5,5-Dibromotetraphenylcyclopentadiene (19)
with Triethylphosphane (20a). Test of the First Proposed Step, 19 +
20 Ǟ 28, of Path b in Scheme 6: A solution of 3.88 g (7.32 mmol) of
19 and 0.87 g (7.4 mmol) of 20a dissolved in 120 mL of anhydrous,
deoxygenated THF was allowed to stand under an argon atmo-
sphere for 18 h at 25Ϯ5 °C in a Schlenk flask sheathed in alumi-
num foil, so as to exclude light. During this time, the initially pale
yellow solution did not change color nor did any solid precipitate.
All volatiles were then removed under reduced pressure at 25 °C
and the resulting residue was shown to be essentially pure 19 by
1H and 13C NMR spectroscopic analysis.
X-ray Crystallographic Study of Triethylphosphonium Tetraphen-
ylcyclopentadienide (22). Crystal Mounting and Data Collection:
Data were collected at room temperature with a Siemens P4 dif-
fractometer equipped with a scintillation counter. An orange crys-
tal was mounted on a glass fiber with epoxy cement and a unit cell
determined from the angular settings of 25 centered reflections.
Systematic absences in the diffraction data indicated either the mo-
noclinic space groups P21 or P21/m. E statistics initially indicated
the noncentrosymmetric alternative, and the structure that was de-
veloped in this setting proved to be incompatible with mirror-plane
symmetry. All non-hydrogen atoms were refined with anisotropic
thermal parameters and hydrogen atoms were treated as idealized
contributions. The correctness of the reported hand was affirmed
by the value of the refined Flack parameter: 0.04(17). All software
used was contained in the SHELXTL and data collection libraries
of programs provided by the Siemens Corporation (Madison, WI).
The pertinent crystal data and structure refinement are presented
in Table 2. CCDC-628202 contains the supplementary crystallo-
graphic data for this paper. These data can be obtained free of
charge from The Cambridge Crystallographic Data Centre via
www.ccdc.cam.ac.uk/data_request/cif.
Reaction of (Cod)2Ni (34) with α,α-Dichlorotoluene (30): In our pre-
vious report on the reactions of geminal dihaloalkanes with
nickel(0) reagents[7] the experimental details for the individual reac-
tions of α,α-dichlorotoluene, dichlorodiphenylmethane and dibro-
momethane with (Et3P)4Ni or for the reaction of dichlorodiphenyl-
methane with (Cod)2Ni have already been published. The following
experiment is described here for the first time.
A solution of (Cod)2Ni (34, 1.35 g, 4.90 mmol) and α,α-dichlo-
rotoluene (30, 0.40 g, 2.40 mmol) in 20 mL of THF was stirred for
2 h at 25 °C. Quenching with deoxygenated 6  HCl under argon.
Usual workup (diethyl ether extraction, neutralization with solid
NaHCO3, solvent removal and flash chromatography) permitted
the isolation of only trans-stilbene (31) (225 mg, 96%).
Reactions of (Cod)2Ni (34) with α,α-Dichlorotoluene (30) in the Pres-
ence of Benzophenone (28): A reaction run almost identical to the
foregoing was conducted, except that 1 equiv. of benzophenone
(32) was present. The analogous workup of the reaction mixture
and flash chromatography led only to trans-stilbene (94%), with
90% of the benzophenone recovered.
Reaction of (Cod)2Ni (34) with Benzyl Phenyl Sulfone (39): A solu-
tion of 34 (1.24 g, 4.50 mmol) and 39 (1.05 g, 4.50 mmol) in 60 mL
of dry, deoxygenated toluene became black upon mixing the com-
ponents. The black mixture was stirred at 25 °C for 4 h and then
at reflux for 3 h, during which latter period a nickel mirror ap-
peared on the flask wall. Hydrolysis of the mixture with deoxy-
genated, aqueous 6  HCl led to gas evolution. Separation and
drying of the toluene layer, removal of solvent and flash
Preparation of Triethylphosphonium Tetraphenylcyclopentadienide
(22c) from 5,5-Dibromotetraphenylcyclopentadiene (19) and (Et3P)4-
Ni (20): A solution of 1.94 g (3.66 mmol) of 19 in 60 mL of THF
was treated with 640 mg (1.20 mmol) of 20 at 25 °C. The dark red
solution was stirred for 18 h and then the THF was removed under
reduced pressure. The dark residue was extracted with warm cyclo-
hexane (6 10-mL portions of deoxygenated solvent). The hot ex-
1582
www.eurjic.org
© 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Inorg. Chem. 2007, 1576–1584

Reactions of Geminal Dihaloalkanes with Nickel(0) Reagents
FULL PAPER
[15] a) F. Ramirez, N. McKelvie, J. Am. Chem. Soc. 1957, 79, 5829–
5830. b) Although this report appeared 50 years ago, it has not
been repudiated since and in fact is treated as reliable in a clas-
sic monograph on organophosphorus reactions: R. F. Hudson,
Structure and Mechanism in Organo-Phosphorus Chemistry, Ac-
ademic Press, New York, 1965, pp. 297–299.
chromatography led to the isolation of trans-stilbene (0.81 g, 39%)
and recovered 39. No sign of bibenzyl was seen in the 1H NMR
spectrum of the crude reaction product.
Acknowledgments
[16] When the THF solution of the final reaction mixture was con-
centrated to half its volume and then cooled to 0 °C, dark red
crystals of bis(triethylphosphane)nickel(II) chloride were de-
posited, mp. 113–115 °C (68%); D. T. Doughty, G. Gordon,
R. P. Stewart Jr, J. Am. Chem. Soc. 1979, 101, 2645–2648.
[17] A referee has proposed that in such coupling reactions of gemi-
nal dihalides with (Et3P)4Ni (20) perhaps metal carbenes are
not involved at all, that maybe they are just phosphonium ylide
adducts, R3P+–CR2––NiCl2 and, as Milstein and co-workers
have shown, that ylides could be converted to metal-carbenes
by late transition metal compounds (M. Gandelman, K. M.
Naing, B. Rybtchinski, E. Poverenov, Y. Ben-David, N. Ash-
kenazi, R. M. Gauvin, D. Milstein, J. Am. Chem. Soc. 2005,
127, 15265–15272). Admittedly, our work has shown that phos-
phonium ylide 22c is readily formed from dibromo compound
19 and (Et3P)4Ni (20) and that 30 and 20 produce Wittig rea-
gent 37 in solution, which is capable of being captured in a
Wittig reaction with benzophenone. Despite the presence of 37
in such reactions, such phosphonium ylides or their complexes
with NiCl2 cannot be essential intermediates in these coupling
reactions, because 30 can be converted into olefin dimer 31 in
over 90% yield either by (Et3P)4Ni or by (Cod)2Ni. Obviously,
the latter nickel(0) complex could not involve any phospho-
nium ylide. In fact, any phosphonium ylide generated with 30
may be in ready equilibrium with the crucial carbene interme-
diate 36a (36 with coordinated Et3P):
This research has received the financial support of Akzo Corporate
Research America Inc. and the Boulder Scientific Company, Mead,
Colorado. Furthermore, through an Alexander von Humboldt Se-
nior Scientist Award, the corresponding author was able to spend
his sabbatical leave during the academic year of 2005–2006 in the
Institute of Inorganic Chemistry of the Munich Technical Univer-
sity, Germany. There, in the research group of Wolfgang A. Herr-
mann, he received much insight into the chemistry of nitrogenr
heterocyclic carbene complexes of transition metal salts. Finally,
the authors are grateful for valuable orienting experiments with
nickel(0) and zirconium(II) carbenoid reactions to former co-
workers, Drs. Adetenu Adeosun, Tomasz Dluzniewski, Somnath
Dutta and Xin Ma.
[1] J. J. Eisch, A. A. Adeosun, J. M. Birmingham, Eur. J. Inorg.
Chem. 2007, 39–43.
[2] a) P. W. Jolly, G. Wilke, The Organic Chemistry of Nickel, Aca-
demic Press, New York, 1974, vol. 1, pp. 110ff., 139ff., 329ff.;
b) P. W. Jolly, G. Wilke, The Organic Chemistry of Nickel, Aca-
demic Press, New York, 1975, vol. 2, p. 400.
[3] J. J. Eisch, S. R. Sexsmith, Res. Chem. Intermed. 1990, 13, 149–
152.
[4] J. J. Eisch, L. E. Hallenbeck, K. I. Han, J. Am. Chem. Soc.
1986, 108, 7763–7767.
[5] J. J. Eisch, K. R. Im, J. Organomet. Chem. 1977, 139, C45–C50.
[6] T. Hayashi, M. Kumada, Acc. Chem. Res. 1982, 15, 395–401.
[7] J. J. Eisch, Y. Qian, M. Singh, J. Organomet. Chem. 1996, 512,
207–217.
[18] The likelihood of a bridging halide between the nickel centers
in the proposed activated complex 38 is supported by the
known dimeric structure of η4-tetraphenylcyclobutadiene–
nickel(II) bromide where the two units of the monomer are
held together by two Ni···Br···Ni bridges.
[19] The failure to detect triphenylethylene in this trapping experi-
ment with benzaldehyde can stem from either of two causes:
(1) the supposed ylide, Ph2C=PEt2, is too slow to react with
PhCHO, before it reacts thus: Ph2C=PEt3 + Ph2C(Cl)NiCl Ǟ
Ph2C=CPh2 + (Et3P)2NiCl2; or (2) in analogy with Scheme 7,
Et3P is unable, for steric reasons, to react with Ph2C–NiCl2 (36)
before coupling occurs via 38.
[8] a) Cf. infra for the XRD and 13C NMR evidence justifying the
ylide resonance structure 22c as the paramount contributor to
the π-electron distribution of this molecule: b) M. P. Cooke Jr,
R. M. Parlman, J. Org. Chem. 1975, 40, 531–532.
[9] a) E. C. Schreiber, E. I. Becker, J. Am. Chem. Soc. 1954, 76,
6125–6127; b) the preparative procedure for 21 has been devel-
oped in this laboratory and is published here for the first time.
[10] F. Ramirez, S. Levy, J. Org. Chem. 1956, 21, 488–489.
[11] a) H. L. Ammon, G. L. Wheeler, P. H. Waats Jr, J. Am. Chem.
Soc. 1973, 95, 6158–6163; b) in ref.[11a], such percentages for
24 ranged from 20% to 80%, depending upon the method of
calculation.
[12] S. Neander, F. E. Tio, R. Buschman, U. Behrens, F. Olbrich, J.
Organomet. Chem. 1999, 582, 58–65. For the potassium (18-
crown-6) salt the C–C bonds range from 1.371–1.411 Å.
[13] a) L. M. Engelhardt, R. I. Papasergio, C. L. Raston, G. Salem,
C. R. Whitaker, J. Chem. Soc. Dalton Trans. 1987, 1647–1653;
b) R. M. Silverstein, G. C. Bassler, T. C. Morill, Spectroscopic
Identification of Organic Compounds, 5th ed., John Wiley, New
York, 1991, p. 239: “Shifts (in the 13C NMR signals) of the
aromatic carbon atom directly attached to the substituent have
been correlated with substituent electronegativity after correct-
ing for magnetic anisotropy effects”. On the Allred-Rochow
scale phosphorus at 2.06 is electron-releasing relative to carbon
at 2.50 and thus magnetically shielding towards carbon.
[14] A referee has been helpful in suggesting the possibility of
path b as an alternative possibility for producing 22c from 19
and 20 without the intermediacy of a nickel(II)-carbene. In ad-
dition, the same referee suggested the key experiment for decid-
ing on the viability of path b as a competing mechanism,
namely whether or not 19 and 20a would react with each other
at 25 °C.
[20] The α-elimination of lithium benzenesulfinate from Ph–CHLi–
SO2Ph cannot be effected by heating this dry salt under argon
at 90–120 °C for 5 h. No trace of either cis- or trans-stilbene
could be detected upon hydrolytic workup and 90% of the
starting sulfone was recovered. By contrast, irradiation of Ph–
CHLi–SO2Ph in THF under argon in a quartz tube in a
Rayonet rotating reactor (model 100), equipped with lamp-
tubes of 254 nm wavelength for 24 h and subsequent hydrolysis
yielded upon liquid chromatography on silica gel with a hexane
eluent: (a) a 1:1.3 mixture of cis- and trans-stilbenes (33%); and
(b) cis- and trans-1,2,3-triphenylcyclopropanes (25%), besides
recovered Ph–CH2–SO2–Ph. The cyclopropanes were identified
by their C-H proton singlets at δ = 2.82 (cis) and 2.73 (trans)
ppm (CDCl3) and the mass spectrum of the mixture: MS
(70eV): m/z = 270 [M], 179, 178, 152, 89, 77 (Y. Qian, Masters
Thesis, SUNY-Binghamton, 1992, p. 57; Y. Qian, Research
Notebook No. 1, May 22, 1991, p. 36). More to the point of
this study, however, is the finding that 1–10 mol-% of Ni-
(acac)2 can cause the catalytic elimination of LiOSOPh from
Ph–CHLi–SO2Ph in THF from 0 °C to reflux, leading to vari-
able proportions of solely trans-stilbene and trans-1,2,3-tri-
1
phenylcyclopropane, as attested to by their IR, H NMR and
Eur. J. Inorg. Chem. 2007, 1576–1584
© 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjic.org
1583

J. J. Eisch, Y. Qian, A. L. Rheingold
[22] M. F. Semmelhack, Org. React. 1972, 19, 178–179.
[23] R. A. Schunn, Inorg. Synth. 1974, 15, 5–9.
[24] C. S. Cundy, J. Organomet. Chem. 1974, 69, 305–310.
[25] E. Dinjus, I. Gorski, E. Uhlig, H. Walther, Z. Anorg. Allg.
Chem. 1976, 75, 422–428.
FULL PAPER
mass spectra on pure samples separated by column chromatog-
raphy (silica gel/hexane). In two separate further experiments,
one with 1,1-diphenylethylene and the other with diphenylace-
tylene, the benzylidene fragment from Ph–CHLi–SO2Ph was
trapped as 1,1,2-triphenylcyclopropane and 1,2,3-triphenylcy-
clopropene, respectively (ref.[7]; Y. Qian, Doctoral Dissertation,
SUNY-Binghamton, 1996, pp. 117–118; Y. Qian, Research
Notebook No. 1, October 19, 1991, p. 66). The detailed mecha-
nisms of these carbenoid oligomerizations are under further
investigation.
[26] T. Yamamoto, T. Kohara, K. Osakada, A. Yamamota, Bull.
Chem. Soc. Jpn. 1983, 56, 2147–2153.
Received: November 22, 2006
Published Online: February 26, 2007
[21] J. J. Eisch, Organometallic Syntheses, Academic Press, New
York, 1981, vol. 2, pp. 1–84.
1584
www.eurjic.org
© 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Inorg. Chem. 2007, 1576–1584