Comparison of the thermal and reductive decarbonylation of a rhodium trifluoroacetyl diphosphine complex

Article abstract of DOI:10.1021/om1007055

Acyl complexes of rhodium(III) with chelating diphosphine ligands are well known for their stability toward decarbonylation. However, we have found that the corresponding perfluoro complexes do decarbonylate under mild conditions. In this report we address specifically Rh(dppp)(COCF₃)I₂ (dppp = 1,3-bis(diphenylphosphino)propane). Another difference between this compound and rhodium acyl complexes containing monodentate phosphines is that it does not undergo reductive elimination of alkyl halide after decarbonylation. Instead, slow CO dissociation from Rh(dppp)(CO)(CF₃)I₂ occurs to yield Rh(dppp)(CF₃)I₂ as the final product. Reduction of the trifluoroacetyl complex is a net two-electron process that also involves decarbonylation, but through a quite different mechanism. The products consist of Rh(dppp)(CO)I, CF₃^- anion, and iodide. X-ray crystal structures of both Rh(dppp)(COCF₃)I₂ and Rh(dppp)(CF₃)I₂ have been obtained and show the expected square-pyramidal geometry with the acyl/alkyl group in the apical position.

Full text of DOI:10.1021/om1007055

5890 Organometallics 2010, 29, 5890–5896
DOI: 10.1021/om1007055
Comparison of the Thermal and Reductive Decarbonylation of a Rhodium
Trifluoroacetyl Diphosphine Complex
Basu D. Panthi, Stephen L. Gipson,* and Andreas Franken
Chemistry Department, Baylor University, One Bear Place #97348, Waco, Texas 76798, United States
Received July 19, 2010
Acyl complexes of rhodium(III) with chelating diphosphine ligands are well known for their
stability toward decarbonylation. However, we have found that the corresponding perfluoro
complexes do decarbonylate under mild conditions. In this report we address specifically Rh(dppp)-
(COCF3)I2 (dppp = 1,3-bis(diphenylphosphino)propane). Another difference between this com-
pound and rhodium acyl complexes containing monodentate phosphines is that it does not undergo
reductive elimination of alkyl halide after decarbonylation. Instead, slow CO dissociation from
Rh(dppp)(CO)(CF3)I2 occurs to yield Rh(dppp)(CF3)I2 as the final product. Reduction of the
trifluoroacetyl complex is a net two-electron process that also involves decarbonylation, but through
a quite different mechanism. The products consist of Rh(dppp)(CO)I, CF3- anion, and iodide. X-ray
crystal structures of both Rh(dppp)(COCF3)I2 and Rh(dppp)(CF3)I2 have been obtained and show
the expected square-pyramidal geometry with the acyl/alkyl group in the apical position.
Introduction
transfer chain catalyzed (ETC6-9) decarbonylation. The reac-
tion might be expected to proceed as
Rhodium acyl complexes function as homogeneous catalysts
or intermediates in many important organic transformations
on both the laboratory and industrial scales.1 These reactions
typically involve migratory insertion and/or decarbonylation
steps. It has been known for quite some time that five-coordi-
nate rhodium(III) acyl dihalide complexes containing chelating
diphosphine ligands, particularly 1,3-bis(diphenylphosphino)-
propane (dppp), are unusually stable with respect to decarbo-
nylation.2-4 The reason for this stability has been proposed to
be the unlikely occurrence of a complex containing an alkyl
group trans to a phosphorus, which would necessarily result
from decarbonylation of the dppp complex.4 However, while
one may infer that this claim is based in some way upon the
strong trans effect of both phosphine and alkyl ligands,5 it was
presented without any supporting evidence and we consider
this explanation to be far from an established fact.
-
RhðdpppÞðCORÞX2 þ e- f ½RhðdpppÞðCORÞX2ꢀ ð1Þ
-
-
½RhðdpppÞðCORÞX2ꢀ f ½RhðdpppÞðCOÞðRÞX2ꢀ ð2Þ
-
½RhðdpppÞðCOÞðRÞX2ꢀ þ RhðdpppÞðCORÞX2
-
f RhðdpppÞðCOÞðRÞX2 þ ½RhðdpppÞðCORÞX2ꢀ ð3Þ
The Rh(III) acyl complexes can be expected to be reducible,10
and once they are reduced to odd-electron species their reactivity
would likely be enhanced. Such rate enhancements have been
observed for migratory insertion reactions,11-15 but not so far
as we are aware for decarbonylation, other than a reductively
initiated radical chain decarbonylation of a rhenium formyl
complex.16 Initiation by reduction in eq 1 was expected to be
followed by a catalytic cycle consisting of eqs 2 and 3. It is gene-
rally held that such a process can be catalytic only if the reduc-
tion potential of the product is more negative than that of the
We have a longstanding interest in organometallic reactions
initiated or catalyzed by electron transfer, and it occurred to us
that the decarbonylation of rhodium acyl complexes containing
dppp might be amenable to study. In particular, we were curi-
ous to see whether these complexes could undergo electron
(8) Battaglini, F.; Calvo, E. J.; Doctorovich, F. J. Organomet. Chem.
1997, 547, 1–21.
(9) Astruc, D. Angew. Chem., Int. Ed. Engl. 1988, 27, 643–660.
(10) Barriere, F.; Geiger, W. E. Organometallics 2001, 20, 2133–2135.
(11) Magnuson, R. H.; Meirowitz, R.; Zulu, S. J.; Giering, W. P.
Organometallics 1983, 2, 460–462.
*To whom correspondence should be addressed. E-mail: stephen_gipson@
baylor.edu.
(1) Simpson, M. C.; Cole-Hamilton, D. J. Coord. Chem. Rev. 1996,
155, 163–207.
(2) Moloy, K. G.; Wegman, R. W. Organometallics 1989, 8, 2883–
2892.
(3) McGuiggan, M. F.; Doughty, D. H. J. Organomet. Chem. 1980,
185, 241–249.
(4) Slack, D. A.; Egglestone, D. L.; Baird, M. C. J. Organomet. Chem.
(12) Morneau, A.; Donovan-Merkert, B. T.; Geiger, W. E. Inorg.
Chim. Acta 2000, 300-302, 96–101.
(13) Amatore, C.; Bayachou, M.; Verpeaux, J.-N.; Pospisil, L.;
Fiedler, J. J. Electroanal. Chem. 1995, 387, 101–108.
(14) Bly, R. S.; Silverman, G. S.; Hossaln, M. M.; Sly, R. K.
Organometallics 1984, 3, 642–644.
1978, 146, 71–76.
ꢀ
(5) Coe, B. J.; Glenwright, S. J. Coord. Chem. Rev. 2000, 203, 5–80.
(6) Magdesieva, T. V.; Butin, K. P. Russ. Chem. Rev. 2002, 71, 223–238.
(7) Delville, M.-H. Inorg. Chim. Acta 1999, 291, 1–19.
(15) Miholova, D.;Vlcek, A. A.J. Organomet. Chem. 1982, 240, 413–419.
(16) Narayanan, B. A.; Amatore, C.; Casey, C. P.; Kochi, J. K. J. Am.
Chem. Soc. 1983, 105, 6351–6352.
r
pubs.acs.org/Organometallics
Published on Web 10/25/2010
2010 American Chemical Society

Article
Organometallics, Vol. 29, No. 22, 2010 5891
reactant, making reaction 3 thermodynamically favorable.17
However, we have shown that this is not strictly true if there
are other factors favoring the forward reaction.18 In any event, it
is not obvious which species, the five-coordinate acyl complex or
the six-coordinate alkyl carbonyl complex, would have the more
negative reduction potential. We therefore decided to investigate
this reaction using both electrochemical and chemical reduction.
Unfortunately, preliminary studies of the reduction of the
well-known acetyl complex Rh(dppp)(COCH3)I2 failed to
show any evidence for the proposed ETC decarbonylation
reaction and also proved difficult to follow by NMR. We
therefore used our experience with perfluoroalkyl complexes
to prepare the previously unreported Rh(dppp)(COCF3)I2 (1),
which would allow both 31P and 19F NMR to be conveniently
used to follow the course of the reaction and the fate of the acyl
group. Such an approach proved quite successful in our studies
of the similar Co(CO)3(PPh3)(COCF3) system.19 Much to our
surprise, while studying its electrochemistry we discovered that
complex 1 underwent thermal decarbonylation in solution
under ambient conditions, though at a rate slow enough that
it did not interfere with the reductive studies. Therefore, we
have also studied this unexpected thermal decarbonylation and
found that it proceeds via a path quite different from that of the
reduction reaction. The results of both studies are presented
below.
Figure 1. ORTEP diagram of Rh(dppp)(COCF3)I2 (1). Phenyl
groups and solvent molecules are omitted for clarity. Thermal ellip-
˚
soids are drawn at the 40% probability level. Selected distances (A)
and angles (deg): Rh1-I1=2.6926(6), Rh1-I2=2.6798(7), Rh1-
C6=1.946(3), Rh1-P2 = 2.3102(9), Rh1-P1 =2.3140(9), O1-
C6=1.207(3), C6-C2=1.579(4), P1-C11=1.820(3), P1-C21=
1.825(3), P1-C3 = 1.832(3), P2-C41 = 1.823(3), P2-C31 =
1.824(3), P2-C5 = 1.831(3), C3-C4 = 1.528(4), C4-C5 =
1.538(4), F1-C2 = 1.319(4), F2-C2 = 1.333(3), F3-C2 =
1.331(4), O1-C6-C2=112.2(3); O1-C6-Rh1=127.8(2), C6-
Rh1-P2 = 92.08(9), C6-Rh1-P1 = 91.31(9), P2-Rh1-P1 =
89.86(3), C6-Rh1-I2=104.06(8), P2-Rh1-I2=163.75(2), P1-
Rh1-I2 = 87.90(3), C6-Rh1-I1 = 100.17(9), P2-Rh1-I1 =
89.65(2), P1-Rh1-I1=168.52(2), I2-Rh1-I1=89.370(18), C11-
P1-C21 = 105.20(13), C41-P2-C31 = 107.06(13).
Results and Discussion
Thermal Decarbonylation of 1. Complex 1 was synthesized
using a variation of the procedure reported by Miller and
Nelson20 to initially produce the trifluoroacetate complex Rh-
(dppp)(COCF3)(CO2CF3)I, followed by in situ replacement of
the trifluoroacetate ligand with iodide. This complex was char-
acterized by IR, NMR, elemental analysis, and X-ray crystal-
lography (Figure 1, Table 1). As with previously reported
rhodium acyl complexes of this type,3,21-24 complex 1 adopts
an essentially square-pyramidal geometry with the trifluoroa-
cetyl group in the apical position. Other than the somewhat
enlarged C-Rh-I bond angles, the structure is an essentially
ideal square pyramid (vide infra).
The thermal reactivity of 1 was followed by NMR, with the
starting complex showing a doublet at 16.8 ppm in its 31PNMR
and a singlet at -69.6 ppm in its 19FNMRspectruminCD2Cl2.
Over the course of 3 days in CD2Cl2 solution at room tem-
perature these peaks decreased in intensity and major new ones
appeared at -21.7 (pd) and 0.5 (ddq) ppm in the 31P NMR and
1.7 (dt) ppm in 19F NMR. Coupling constants of these peaks
are presented in Table 2. During this time IR showed that the
carbonyl stretching band of 1 at 1679 cm-1 decreased in inten-
sity and was replaced by a new band at 2079 cm-1. These NMR
and IR results are consistent with decarbonylation of 1 to give
Rh(dppp)(CO)(CF3)I2 (2). While this type of reaction has been
used to prepare trifluoromethyl complexes of a number of other
metals,25 so far as we know it is the first time it has been used for
rhodium. Other Rh(III) trifluoromethyl complexes have been
prepared via oxidative addition of CF3I,26 Me3SiCF3,27 or an
aryl-CF3 diphosphine to Rh(I).28
Presumably at least initially the CF3 of 1 migrates to a
position trans to one of the phosphorus atoms, displacing one
of the iodides. It has been reported that trans phosphorus-
fluorine three-bond coupling constants are larger than those
of cis isomers.27 We observed both large (3JP-F = 60 Hz) and
small (3JP-F = 12 Hz) phosphorus-fluorine three-bond
coupling in the NMR spectra of complex 2, indicating that the
CF3 must be trans to one of the phosphorus atoms and cis to
the other. This arrangement would require the terminal CO to
be trans to the displaced iodide. We were able to grow single
crystals of 2, but X-ray crystallography failed to distinguish
between the CF3 and iodide ligands trans to the two phos-
phorus atoms, and thus a suitable solution could not be
obtained.27 Nevertheless, the partial solution confirmed our
proposed decarbonylation step. The CO stretching frequency
(17) Pevear, K. A.; Banaszak Holl, M. M.; Carpenter, G. B.; Rieger,
A. L.; Rieger, P. H.; Sweigart, D. A. Organometallics 1995, 14, 512–523.
(18) Liu, Z.; Gipson, S. L. J. Organomet. Chem. 1998, 553, 269–275.
(19) Gunawardhana, N.; Gipson, S. L. J. Organomet. Chem. 2007,
692, 3231–3235.
(20) Miller, J. A.; Nelson, J. A. Organometallics 1991, 10, 2958–2961.
(21) Moloy, K. G.; Petersen, J. L. Organometallics 1995, 14, 2931–
2936.
(22) Shie, J.-Y.; Lin, Y.-C.; Wang, Y. J. Organomet. Chem. 1989, 371,
383–392.
(23) Cheng, C.-H.; Spivack, B. D.; Eisenberg, R. J. Am. Chem. Soc.
(26) Hyde, E. M.; Kennedy, J. D.; Shaw, B. L.; McFarlane, W.
J. Chem. Soc., Dalton Trans. 1977, 1571–1576. Clark, H. C.; Reimer,
K. J. Can. J. Chem. 1976, 54, 2077–2084. Dart, J. W.; Lloyd, M. K.; Mason,
R.; McCleverty, J. A. J. Chem. Soc., Dalton Trans. 1973, 2039–2045.
(27) Vicente, J.; Gil-Rubio, J.; Guerrero-Leal, J.; Bautista, D. Dalton
Trans. 2009, 3854–3866. Vicente, J.; Gil-Rubio, J.; Guerrero-Leal, J.;
Bautista, D. Organometallics 2004, 23, 4871–4881.
1977, 99, 3003–3011.
(24) Søtofte, I.; Hjortkjær, J. Acta Chem. Scand. 1994, 48, 872–877.
(25) Brothers, P. J.; Roper, W. R. Chem. Rev. 1988, 88, 1293–1326.
(28) van der Boom, M. E.; Ben-David, Y.; Milstein, D. J. Am. Chem.
Soc. 1999, 121, 6652–6656.

5892 Organometallics, Vol. 29, No. 22, 2010
Panthi et al.
characteristic of a terminal carbonyl, the inequivalence of the
phosphine NMR resonances, and the appearance of Rh-F
coupling in the 19F NMR spectrum are all consistent with the
proposed structure. Nevertheless, it should be noted that the
19F NMR spectrum, for which the signal-to-noise ratio was
quite good, displayed additional minor resonances very similar
to the major resonances but at different chemical shifts, perhaps
representing different isomers of the trifluoromethyl complex
produced as minor products.
Despite the fact that we were able to grow single crystals of
complex 2, it was not stable in solution at room temperature.
Over long periods of time (ca. 20 days or more) the carbonyl
stretching band of 2 disappeared from the IR spectrum and
NMR spectra showed the formation of a new compound.
The 31P NMR spectrum showed a single resonance at 14.8
(dq) ppm indicative of equivalent phosphorus atoms, and
19F NMR showed a new resonance at 14.5 (q) ppm with
identical Rh-F and P-F coupling constants. The logical
conclusion is that over time the carbonyl ligand dissociates,
leaving Rh(dppp)(CF3)I2 (3). In fact, we were able to grow
single crystals of 3, and the structure is shown in Figure 2. As
with compound 1, compound 3displays an almost ideal square-
pyramidal geometry (vide infra). Compound 3 was not pro-
duced in sufficient quantities to allow its characterization by
elemental analysis, but a high-resolution mass spectrum was
obtained using ESI on a solution of 3 in methanol. While the
compound is not capable of forming a molecular ion due to the
lack of an appropriate protonation/deprotonation site, the
spectrum exhibited peaks for [3 - CF3]þ and [3 - I]þ, confirm-
ing the compound’s identity.
The thermal reactivity of compound 1 can thus be summar-
ized by Scheme 1. Decarbonylation of the square-pyramidal
trifluoroacetyl complex yields the octahedral trifluorome-
thyl carbonyl complex 2, most likely with the trifluoromethyl
group trans to one of the phosphines. Given time, this complex
dissociates CO to yield the square-pyramidal complex 3. It is
interesting to note that when nonfluorinated rhodium acyl
complexes containing monodentate phosphine ligands decar-
bonylate to alkyl complexes, the alkyl group subsequently
reductively eliminates with one of the halides to leave trans-
RhP2(CO)X type complexes.29 It is unclear whether the differ-
ent reactivity of 2 stems from the enforced cis geometry of the
phosphines or the much stronger metal-carbon bond of the
CF3 group.30 Note also that while the geometry of the com-
plexes alternates between square pyramidal and octahedral, the
oxidation state of rhodium remains constant at þ3.
Electrochemistry of 1. Cyclic voltammetry of complex 1
was performed in dichloromethane and tetrahydrofuran.
Table 1. Summary of Crystallographic Data for Rh(dppp)-
(COCF3)I2 (1) and Rh(dppp)(CF3)I2 (3)
1 CH2Cl2
3
3 3CH2Cl2
3
empirical formula
formula wt
C30H28Cl2F3I2OP2Rh
951.07
C31H32Cl6F3I2P2Rh
1092.92
cryst size (mm)
cryst syst
space group
0.21 ꢁ 0.19 ꢁ 0.13
0.13 ꢁ 0.09 ꢁ 0.07
monoclinic
P21/m
8.7676(5)
22.1871(13)
10.3947(6)
90.0
110.587(2)
90.0
1892.93(19)
2
110(2)
triclinic
P1
9.500(2)
˚
a (A)
˚
b (A)
13.385(4)
14.187(4)
74.522(14)
79.145(12)
70.165(12)
1625.8(7)
2
110(2)
1.943
2.728
1.66-25.34
˚
c (A)
R (deg)
β (deg)
γ (deg)
3
˚
V (A )
Z
temp (K)
dcalcd (Mg m-3
)
1.917
2.628
1.84-25.35
μ (mm-1
)
θ range for data
collecn (deg)
Figure 2. ORTEPdiagramofRh(dppp)(CF3)I2 (3). Phenyl groups
and solvent molecules are omitted for clarity. Thermal ellipsoids are
˚
drawn at the 40% probability level. Selected distances (A) and angles
(deg): Rh1-I1 = 2.6884(3), Rh1-P1 = 2.3012(6), Rh1-C1 =
2.014(4), P1-C11 = 1.825(2), P1-C21 = 1.825(2), P1-C2 =
1.827(2), F1-C1 = 1.327(3), F2-C1 = 1.341(4), C2-C3 =
1.526(3); C1-Rh1-P1=93.36(7), P1-Rh1-P1A=91.84(3), P1-
Rh1-I1A = 166.619(19), C1-Rh1-I1 = 99.98(7), P1-Rh1-
I1=88.391(17), P1A-Rh1-I1=166.619(19), I1A-Rh1-I1=
88.319(11), C11-P1-C21 = 105.32(11).
no of data measd
no. of unique data
no. of data/restraints/
params
max/min transmissn
Rint
26 253
5890
5890/0/370
24 456
3561
3561/0/211
0.7113/0.5942
0.0440
0.0505, 0.0266
0.0480, 0.0223
1.040
0.8354/0.7230
0.0335
0.0478, 0.0239
0.0451, 0.0194
1.051
wR2, R1 (all data)
wR2, R1 (I > 2σ(I))
GOF on F2
-3
˚
max/min ΔF (e A
)
0.806/-0.691
0.779/-0.362
Table 2. Summary of 31P and 19F NMR Data
complex
31P (ppm); J (Hz)
19F (ppm); J (Hz)
Rh(dppp)(COCF3)I2 (1)a
Rh(dppp)(CO)(CF3)I2(2)a
16.8 (d); 117 (Rh-P)
-69.6 (s)
-21.7 (pd, PA), 0.5 (ddq, PB); 60 (Rh-PA, F-PA),
102 (Rh-PB), 11 (F-PB), 29 (PA-PB)
14.8 (dq); 114 (Rh-P), 13 (F-P)
1.7 (dt); 56 (PA-F), 12 (Rh-F, PB-F)
Rh(dppp)(CF3)I2 (3)a
14.5 (q); 12 (Rh-F, P-F)
-10.2 (dt); 19 (Rh-F, PA-F), 45 (PB-F)
[Rh(dppp)(CO)(CF3)I]- (E)b
9.2 (ddq, PA), 13.8 (ddq, PB); 141 (Rh-PA), 19 (F-PA),
95 (Rh-PB), 48 (F-PB), 53 (PA-PB)
3.5 (dd, PA), 22.0 (dd, PB); 115 (Rh-PA),
156 (Rh-PB), 56 (PA-PB)
Rh(dppp)(CO)Ib
a Solvent CD2Cl2. b Solvent THF-d8.

Article
Scheme 1. Thermal Decarbonylation of Complex 1
Organometallics, Vol. 29, No. 22, 2010 5893
The peak potential for the chemically irreversible reduction
observed in both solvents was -1.36 V vs Fcþ/Fc. When
cyclic voltammograms were recorded at a slow scan rate (200
mV/s), only one oxidation peak at ca. -0.1 V was observed
on the return scan, and it was confirmed to originate from
iodide by adding tetrabutylammonium iodide as an addi-
tional source of iodide.31 At faster scan rates (e.g., 2 V/s) an
additional oxidation peak for a short-lived intermediate
could be observed at -0.75 V on the return scan. When the
forward scan toward negative potentials was reversed at a
potential just before the reduction of 1, no oxidation peaks
were observed during the return scan. These results demon-
strate that a short-lived oxidizable intermediate and iodide
are produced by the reduction of compound 1.
Figure 3. 31P NMR spectrum after partial reduction of Rh-
(dppp)(COCF3)I2 (1) with K2BP in THF-d8.
The number of electrons transferred during the reduction
of complex 1 was investigated electrochemically by linear
scan voltammetry (LSV) in dichloromethane and acetoni-
trile and controlled-potential electrolysis in THF and acet-
onitrile. The LSV limiting current of 1 was compared with
those of cyclopentadienyliron dicarbonyl dimer (2e) and
tris(dibenzoylmethanato)iron (1e) using 0.1 and 0.01 mm
microelectrodes. While not an exact match, the LSV limiting
current of compound 1 corresponded most closely with that
of the two-electron reduction. Likewise, controlled-potential
electrolyses consumed approximately 2 mol of electrons/mol
of 1 reduced. We thus conclude from its electrochemistry that
the reduction of compound 1 is a two-electron process with
iodide as one of the major products.
Chemical Reduction of 1. Complex 1 was reduced in THF-
d8 with THF solutions of the mono- and dipotassium salts of
benzophenone (KBP and K2BP), and the progress of the reac-
tion was monitored by IR, 31P NMR, and 19F NMR. IR spectra
showed that during the reduction of 1 its carbonyl stretching
band at 1680 cm-1 (cf. 1679 cm-1 in CH2Cl2) disappeared and
was replaced by a peak at 2009 cm-1. This new IR band is in the
region expected for a terminal metal-carbonyl stretching fre-
quency and in fact matches that of Rh(dppp)(CO)I.2 No
intermediate species were detected by IR, perhaps because IR
lacked either the sensitivity or the selectivity to do so.
NMR spectroscopy proved to be a much more sensitive
way to follow the progress of the reduction reaction. After
partial reduction of 1 with K2BP, the 31P NMR spectrum
displayed resonances at 3.5 (dd), 9.2 (ddq), 13.8 (ddq), and
22.0 (dd) ppm, in addition to that of compound 1 at 16.8 (d)
ppm (Figure 3). Coupling constants of these peaks are presented
in Table 2. As more reductant was added, the starting material
disappeared, followed by loss of the peaks at 9.2 and 13.8 ppm,
leaving those at 3.5 and 22.0 ppm from the primary product after
complete reduction. A singlet at 32.3 ppm corresponding to 1,3-
bis(diphenylphosphino)propane dioxide was the sole remaining
31P NMR resonance after exposing the product mixture to air.
The 19F NMR spectrum after partial reduction with K2BP
Figure 4. 19F NMR spectrum after partial reduction of Rh-
(dppp)(COCF3)I2 (1) with K2BP in THF-d8.
(Figure 4) displayed peaks at -10.2 (dt), -74.9 (s), -78.1 (d),
and -78.8 (1:1:1 t) ppm, in addition to that of compound 1 at
-68.7 (s) ppm (cf. -69.6 ppm in CD2Cl2). As more reductant
was added, the starting material disappeared, followed by loss
of the peak at -10.2 ppm, leaving those at -74.9, -78.1, and
-78.8 ppm from the primary fluorine-containing products after
complete reduction. These 19F NMR peaks correspond to
Ph2C(OH)CF3, CF3H, and CF3D, respectively.19
When compound 1 was reduced with K2BP in the presence
of ethyl acetate, the 19F NMR peak for CF3H was much larger
than that in its absence while those for Ph2C(OH)CF3 and the
intermediate at -10.2 ppm were much smaller. This result
suggests a rapid reaction between the intermediate and ethyl
acetate, most likely consisting of proton transfer from the ethyl
acetate to CF3-. A similar result was obtained when compound
1 was reduced in acetone-d6, though in this case the yield of
CF3D was enhanced. Both esters with R-hydrogens and acetone
-
have been reported to react with CF3 to produce CF3H and
deprotonated solvent as major products.32
In order to quantify the number of electrons transferred
during the chemical reduction of 1, it was titrated with a
standardized solution of KBP. Experiments were performed in
which the consumption of 1was followed by either NMR or CV,
and in each case titrations consumed 2 equiv of KBP/mol of
compound 1. These results are in agreement with the electro-
chemical studies, which indicated that the reduction of 1 is a net
two-electron process.
Proposed Mechanism for the Reduction of 1. Scheme 2
shows the proposed mechanism for the reduction of compound
(29) Lau, K. S. Y.; Becker, Y.; Huang, F.; Baenziger, N.; Stille, J. K.
J. Am. Chem. Soc. 1977, 99, 5664–5672.
(30) Hughes, R. P. Adv. Organomet. Chem. 1990, 31, 183–267.
(31) Yaraliev, Ya. A. Russ. Chem. Rev. 1982, 51 (6), 566–581.
(32) McDonald, R. N.; Chowdhury, A. K. J. Am. Chem. Soc. 1983,
105, 7267–7271.

5894 Organometallics, Vol. 29, No. 22, 2010
Panthi et al.
Scheme 2. Reduction Mechanism of Complex 1
produces CF3 radicals.19 However, in this case both chemical
and electrochemical evidence confirm the two-electron nature
-
of the reduction, requiring CF3 anion as the product. The
increased yield of CF3H in the presence of ethyl acetate and
CF3D in the presence of acetone-d6 also support the production
of CF3-. Thus, we feel confident in assigning the final step in the
-
mechanism of the reduction of 1 to CF3 dissociation from
complex E. The fact that this step involves loss of CF3- rather
than iodide may be the result of a trans effect of phosphine
weakening the Rh-CF3 bond.
It should be noted that Rh(dppp)(CO)I could also be
produced by loss of CF3- directly from intermediate D. This
pathway would produce the correct products, derived from
CF3-, with the correct electron stoichiometry and is analo-
-
gous to the loss of CF3 from the trifluoroacetyl ligand of
Cp*Ir(PMe3)(H)(COCF3).34 However, the observation of an
intermediate with NMR spectroscopy consistent with the
structure shown for E, but not D, makes this pathway an
unlikely alternative.
X-ray Crystal Structures of Complexes 1 and 3. Crystals of
1 and 3 suitable for X-ray crystallography were grown by
slow diffusion of hexanes into dichloromethane solutions at
low temperature. Compound 1 crystallized in the triclinic
space group P1 with two formula units per unit cell, while
compound 3 crystallized in the monoclinic space group P21/
m with two formula units per unit cell. The crystallographic
data are summarized in Table 1. Selected bond lengths and
bond angles are presented in the captions of Figures 1 and 2,
which show perspective views of Rh(dppp)(COCF3)I2 (1)
and Rh(dppp)(CF3)I2 (3), respectively. As can be seen in
Figures 1 and 2, both compounds have a square-pyramidal
geometry with the phosphorus and iodine atoms occupying
the basal sites and either a trifluoroacetyl or a trifluoromethyl
group occupying the apical position. This has been reported to
be the preferred orientation, over trigonal bipyramidal, for five-
coordinate d6 complexes.35
When the bond angles in complexes 1 and 3 are compared
with those of an ideal square-pyramidal structure, the major
deviation from ideality is in the larger than expected apical
carbon-rhodium-iodine angles. This distortion most likely
results from contact between the iodine atoms and C2 and F1
in 1 and F1 and F1A in 3. Contact between the acyl oxygen and
the phosphorus atoms in 1 and the single F2 atom and the
phosphorus atoms in 3 is less significant, and so those angles are
closer to the ideal 90° value. These slight distortions are quite
typical of this class of rhodium complexes and agree with several
related structures that we have solved,36 as well as literature
reports for similar dichloro complexes.3,22 Interestingly, Moloy
and Petersen’s reported structure for Rh(dppp)(COCH3)I2
shows significant distortion toward a trigonal bipyramid,21
but Søtofte and Hjortkjær have reported the structure as having
only minor deviations from square-pyramidal geometry.24 We
have independently solved the structure of this compound and
found it in agreement with that of 1.36
1. The initial electron transfer produces an unstable 17e anion
(A). It is clear from both IR and NMR that decarbonylation
occurs eventually, but it is unlikely to do so at this point, since the
resulting anionic trifluoromethyl complex would be a 19e
species. The anion A thus almost certainly undergoes loss of
iodide next to yield a neutral 15e intermediate (B). This com-
plex may then decarbonylate to yield a 17e trifluoromethyl
complex (C). However, there is no evidence to support the
intermediacy of C. In fact, it is also possible that complex B
instead undergoes a second electron transfer. This second
reduction would produce the intermediate labeled D, a 16e
anion, through a typical ECE mechanism.33 This complex
should be observable by NMR, but no appropriate resonances,
inequivalent 31P signals lacking P-F coupling, were observed.
Cyclic voltammetry does give evidence for a short-lived oxidize-
able intermediate. Such an intermediate would more likely be D
than C, given its ease of oxidation and observable lifetime.
However, we cannot conclusively state whether the reduction of
1 passes through either intermediate C or D.
Either decarbonylation of D or a second electron transfer
to C would yield an 18e anionic trifluoromethyl complex (E).
Good NMR evidence does exist to support the intermediacy of
complex E (see Table 2). The 31P resonances observed at 9.2 and
13.8 ppm after partial reduction of 1 as well as the 19F resonance
at -10.2 ppm are consistent with the proposed structure in terms
of both chemical shifts and coupling constants. The 3JP-F values
of 45 and 19 Hz are consistent with one phosphorus atom being
trans to the CF3 and one being cis. A terminal CO stretch which
could be associated with complex E was not observed, but this
may be due to its low intensity and/or overlap with the stretching
frequency of the final product, Rh(dppp)(CO)I.
While NMR spectra consistent with complex E were
observed, this compound was not stable. Over time its 31P NMR
resonances gave way to those of Rh(dppp)(CO)I, while 19F
NMR indicated the formation of compounds derived from the
CF3 group, including CF3H, CF3D, and the anion of trifluoro-
methyl benzhydrol. We have reported these same products from
the one-electron reduction of Co(CO)3(PPh3)COCF3, which
The bond lengths in complexes 1 and 3 are quite typical.
When the two complexes are compared, the average Rh-I
bond length of complex 3 is very slightly longer than that of
1, while the average Rh-P bond length of 3 is slightly shorter
(34) Cordaro, J. G.; Bergman, R. G. J. Am. Chem. Soc. 2004, 126,
16912–16929.
(35) Hoffman, P. R.; Caulton, P. R. J. Am. Chem. Soc. 1975, 97,
4221–4228.
(36) Panthi, B. D.; Gipson, S. L.; Franken A. Manuscript in pre-
paration.
(33) Battaglini, F.; Calvo, E. J.; Doctorovich, F. J. Organomet. Chem.
1997, 547, 1–21.

Article
Organometallics, Vol. 29, No. 22, 2010 5895
than that of 1. The major difference, as expected, is in the
Rh-C bond length, with the acyl complex having the shorter
bond (1.946 vs 2.014 A). This trend is in agreement with two
Instrumentation. NMR spectra were acquired on a Varian
VNMRS 500 instrument. Proton chemical shifts were refer-
enced to residual protons in the solvent, 19F chemical shifts to
external CFCl3, and 31P chemical shifts to external 85% H3PO4.
IR spectra were obtained using a Mattson Instruments Genesis
II FTIR or Thermo Nicolet Avatar 370 FTIR instrument and
a cell having CaF2 windows separated by a 0.1 mm spacer.
Electrochemical measurements were performed with a CH
Instruments Model 1140 electrochemical analyzer using a Pt-
disk working electrode, a Pt-wire auxiliary electrode, and
a silver-wire quasi-reference electrode. Solutions contained
approximately 0.1 M [Bu4N]PF6 supporting electrolyte. All
potentials are expressed relative to the formal potential of the
ferrocenium-ferrocene couple (Fcþ/Fc). Controlled-potential
electrolyses were conducted in a standard H cell with compart-
ments separated by a medium-porosity glass frit using 25 ꢁ
25 mm platinum-foil working and auxiliary electrodes and a
silver-wire reference electrode. The working electrode potential
was set at a value ca. 0.2 V more negative than the CV peak
potential, and electrolysis was continued until CV indicated that
the starting material had been consumed. The total charge
consumed was corrected for a background current taken as
equal to the final current. High-resolution mass spectra were
obtained in the Baylor University Mass Spectrometry Core
Facility on a Thermo Scientific LTQ Orbitrap Discovery using
an ESI source in positive ion mode.
˚
previously reported pairs of perfluoroacyl and perfluoroalk-
yl complexes of rhodium and platinum37-39 but is in contrast
to our results with cobalt, where the Co-C bond length of
the trifluoromethyl complex was shorter than that of the
trifluoroacetyl complex.40 On the other hand, the Rh-C
bond length for the acetyl complex reported by Moloy
and Petersen (1.981 A)21 is, as expected, significantly longer
˚
than the trifluoroacetyl bond length of 1. However, the
˚
˚
carbon-oxygen (1.182 A) and carbon-carbon (1.513 A)
bond lengths of the acyl ligand are shorter in the acetyl complex
˚
than in the trifluoroacetyl complex (1.207 and 1.579 A,
respectively).
Conclusions
Unlike previously reported acyl complexes of rhodium
with diphosphine ligands,2-4 the trifluoroacetyl complex 1
undergoes spontaneous thermal decarbonylation at room
temperature. And unlike acyl complexes with monodentate
phosphine ligands which do decarbonylate, the resulting
alkyl carbonyl complex loses CO instead of undergoing
reductive elimination of alkyl halide.29 Thus, the reactivity
of this perfluoro complex is quite different from that of
nonfluorinated analogues, in line with results from a number
of other organometallic systems.30 The reduction of 1 also
results in decarbonylation, though we cannot say whether
this step occurs from a 15e Rh(II) or a 16e Rh(I) intermedi-
ate. In either case, a net two-electron reduction yields a Rh(I)
carbonyl complex, iodide, and trifluoromethyl anion. Good
evidence exists for the intermediacy of an unstable 18e Rh(I)
trifluoromethyl complex in this reaction.
X-ray Crystallography. Diffraction data were acquired at
110(2) K using a Bruker-Nonius X8 Apex CCD area-detector
diffractometer (graphite-monochromated Mo KR radiation,
˚
λ = 0.710 73 A). Several sets of data frames were collected at
different θ values for various initial values of φ and ω, each
frame covering a 0.5° increment in φ or ω. The data frames were
integrated using SAINT;42 the substantial redundancy in data
allowed empirical absorption corrections (SADABS42) to be
applied on the basis of multiple measurements of equivalent
reflections.
The structures were solved (SHELXS-97) via conventional
direct methods and were refined (SHELXL-97) by full-matrix
least squares on all F2 data using SHELXTL.43 All non-hydro-
gen atoms were assigned anisotropic displacement parameters.
All of the hydrogen atoms were set riding on their parent atoms
in calculated positions and were assigned fixed isotropic thermal
parameters calculated as Uiso(H) = 1.2[Uiso(parent)].
Experimental Section
General Procedures. Chlorodicarbonylrhodium(I) dimer and
1,3-bis(diphenylphosphino)propane were obtained from Strem
Chemicals. Trifluoroacetic anhydride was obtained from Acros
Organics. Solutions of K2BP (potassium salt of benzophenone
dianion)41 were prepared by reaction of benzophenone in THF
with an excess of potassium metal and were standardized by
reaction with aqueous ethanol followed by titration with stan-
dardized HCl. Solutions of KBP (potassium salt of benzophe-
none monoanion) were prepared by 1:1 dilution of K2BP with a
THF solution of excess benzophenone and were also standar-
dized with HCl. The THF was first dried with CaH2 and then
distilled under nitrogen from sodium benzophenone anion
before use. Dichloromethane and acetonitrile were distilled
from CaH2 under nitrogen. The supporting electrolyte for
electrochemical experiments, [Bu4N]PF6, was obtained from
Alfa Aesar and was dried at 100 °C under vacuum before use.
All other reagents were obtained commercially and were used as
received. All chemical reactions were carried out in a nitrogen
atmosphere glovebox.
Rh(dppp)(COCF3)I2 (1). First, Rh(dppp)(CO)I was synthe-
sized from [Rh(CO)2Cl]2 and dppp by the published procedure.2
Rh(dppp)(COCF3)(CO2CF3)I was then synthesized by a pro-
cedure adapted from Miller and Nelson.20 Rh(dppp)(CO)I
(0.452 g, 0.21 mmol), trifluoroacetic anhydride (0.5 mL, 3.6
mmol), and 10 mL of toluene were added to a 100 mL round-
bottom flask, and the mixture was stirred for about 3 h inside the
glovebox. The trifluoroacetate complex was converted to 1 in
situ by adding an excess of solid NaI (0.2 g, 1.3 mmol). After the
mixture was stirred for about 4 h, 15 mL of hexanes was added
and the solution was taken out of the glovebox and cooled in a
freezer for about 1/2 h. The resulting precipitate was then filtered
and washed with hexanes. The collected precipitate was dis-
solved in a small amount of methylene chloride and the solution
filtered. The filtrate was concentrated under reduced pressure,
and an orange precipitate of Rh(dppp)(COCF3)I2 was obtained
by the addition of hexanes. This precipitate was filtered, washed
with hexanes, and dried overnight under vacuum at room
temperature. The yield was 0.455 g (0.52 mmol, 78% based on
Rh(dppp)(CO)I). Single crystals suitable for X-ray structural
determination were grown by the slow diffusion of hexanes into
a solution of 1 in CH2Cl2 at low temperature. Anal. Calcd for
C29H26F3I2OP2Rh (866.12): C, 40.21; H, 3.03; F, 6.58. Found:
(37) Bourgeois, C. J.; Hughes, R. P.; Husebo, T. L.; Smith, J. M.;
Guzei, I. M.; Liable-Sands, L. M.; Zakharov, L. N.; Rheingold, A. L.
Organometallics 2005, 24, 6431–6439.
(38) Bennett, M. A.; Chee, H.-K.; Robertson, G. B. Inorg. Chem.
1979, 18, 1061–1070.
(39) Bennett, M. A.; Ho, K.-C.; Jeffery, J. C.; McLaughlin, G. M.;
Robertson, G. B. Aust. J. Chem. 1982, 35, 1311–1321.
(40) Gunawardhana, N.; Gipson, S. L.; Franken, A. Inorg. Chim.
Acta 2009, 262, 113–116.
1
C, 39.78; H, 3.01; F, 6.86. IR (CH2Cl2): 1679 cm-1. H NMR
(41) Kamaura, M.; Inanaga, J. Tetrahedron Lett. 1999, 40, 7347–
7350.
(42) APEX 2, version 1.0; Bruker AXS, Madison, WI, 2003-2004.
(43) SHELXTL, version 6.12; Bruker AXS, Madison, WI, 2001.

5896 Organometallics, Vol. 29, No. 22, 2010
Panthi et al.
(CD2Cl2): 1.8 (m), 2.5 (m), 3.3 (m), 7.2-8.0 (m) ppm. 19F NMR
(CD2Cl2): -69.6 (s) ppm. 31P NMR (CD2Cl2): 16.8 (d) ppm
(JRh-P = 117 Hz).
768.8649, dev 0.75 ppm), [M - I]þ found 710.9547 (calcd 710.9556,
dev 1.2 ppm).
Rh(dppp)(CF3)I2 (3). Single crystals of Rh(dppp)(CF3)I2 were
grown in an NMR tube by layering hexanes over a CH2Cl2
solution of the solid residue remaining after evaporation of CD2Cl2
from a solution of 1 used for NMR studies of the thermal
decarbonylation reaction. The mass spectrum of this residue was
also collected in methanol solution. The compound was not other-
wise isolated but was studied by NMR during the approximately
20 days it took for decarbonylation to go to completion. 1H NMR
(CD2Cl2): 1.9 (m), 2.7 (m), 3.0 (m), 7.2-7.7 (m) ppm. 19F NMR
(CD2Cl2): 14.5 (q) ppm (JRh-F, JP-F = 12 Hz). 31P NMR (CD2-
Cl2): 14.8 (dq) ppm (JRh-P = 114 Hz, JF-P = 13 Hz). High-
resolution ESI/MS (MeOH): [M - CF3]þ found 768.8643 (calcd
Acknowledgment. We thank the Robert A. Welch Foun-
dation (Grant No. AA-1083) and Baylor University (URC
Grant No. 0301533DE) for support of this research. We also
thank the National Science Foundation (Award No. CHE-
0420802) for funding the purchase of our 500 MHz NMR.
Supporting Information Available: A CIF file giving all crystal
data and refinement parameters, atomic coordinates, bond
lengths, bond angles, and thermal displacement parameters
for complexes 1 and 3. This material is available free of charge
via the Internet at http://pubs.acs.org.