An EPR Study of 2,3-Bis(diphenylphosphino)maleic Anhydride (BMA) Complexes and the BMA Radical Anion

Article abstract of DOI:10.1021/ic980581h

EPR spectra are reported for four metal complexes of 2,3-bis(diphenylphosphino)maleic anhydride (BMA), [Co₂(PhCCR)(CO)₄(η-BMA)^-, R = Ph. H, [Co₂(PhCCPh)(CO)₄(μ-BMA)^-, and [PhCW(CO)₂(BMA)Cl]^-, as well as the radical anions. [BMA]^- and [BPCD]^-, BPCD = 4,5-bis(diphenylphosphino)cyclopentene-1,3-dione. At room temperature, all spectra are 1:2:1 triplets due to hyperfine coupling to two equivalent ³¹P nuclei with coupling to two equivalent ¹H nuclei for [BPCD]^- and unresolved coupling to one or two ⁵⁹Co nuclei for the Co complexes with chelating or bridging BMA, respectively. The ³¹P couplings are temperature dependent, ca. -3 and -13 mG K^-1 for the metal complexes and ligand radical anions, respectively. At low temperature, the spectrum of [BMA]^- shows the presence of symmetric and asymmetric PPh₂ rotational conformers, related by the thermodynamic parameters ΔH° = -0.8 ± 0.2 kJ mol^-1 and ΔS° = 4 ± 1 J mol^-1 K^-1 and interconverted with activation parameters ΔH^? = 18.2 ± 0.4 kJ mol^-1, ΔS^? = -30 ± 2 J mol^-1 K^-1. The temperature dependence of the ³¹P couplings is explained by a negative spin-polarization contribution to (a^P) and a positive contribution due to P 3s character; the latter increases with the asymmetry of the PPh₂ conformations. The range of conformations accessible to the metal complexes is less than for the ligand radical anions, and accordingly the temperature dependence is significantly smaller.

Full text of DOI:10.1021/ic980581h

Inorg. Chem. 1998, 37, 4849-4856
4849
An EPR Study of 2,3-Bis(diphenylphosphino)maleic Anhydride (BMA) Complexes and the
BMA Radical Anion
Noel W. Duffy,† Ross R. Nelson,‡ Michael G. Richmond,§ Anne L. Rieger,‡ Philip H. Rieger,*,‡
Brian H. Robinson,† David R. Tyler,| Jian Cheng Wang,§ and Kaiyuan Yang§
Departments of Chemistry, Brown University, Providence, Rhode Island 02912, University of
North Texas, Denton, Texas 76203, University of Oregon, Eugene, Oregon 97403, and
University of Otago, Dunedin, New Zealand
ReceiVed May 22, 1998
EPR spectra are reported for four metal complexes of 2,3-bis(diphenylphosphino)maleic anhydride (BMA), [Co2-
(PhCCR)(CO)4(η-BMA)]-, R ) Ph, H, [Co2(PhCCPh)(CO)4(µ-BMA)]-, and [PhCW(CO)2(BMA)Cl]-, as well
as the radical anions, [BMA]- and [BPCD]-, BPCD ) 4,5-bis(diphenylphosphino)cyclopentene-1,3-dione. At
room temperature, all spectra are 1:2:1 triplets due to hyperfine coupling to two equivalent 31P nuclei with coupling
to two equivalent 1H nuclei for [BPCD]- and unresolved coupling to one or two 59Co nuclei for the Co complexes
with chelating or bridging BMA, respectively. The 31P couplings are temperature dependent, ca. -3 and -13
mG K-1 for the metal complexes and ligand radical anions, respectively. At low temperature, the spectrum of
[BMA]- shows the presence of symmetric and asymmetric PPh2 rotational conformers, related by the
thermodynamic parameters ∆H° ) -0.8 ( 0.2 kJ mol-1 and ∆S° ) 4 ( 1 J mol-1 K-1 and interconverted with
activation parameters ∆Hq ) 18.2 ( 0.4 kJ mol-1, ∆Sq ) -30 ( 2 J mol-1 K-1. The temperature dependence
of the 31P couplings is explained by a negative spin-polarization contribution to aP and a positive contribution
due to P 3s character; the latter increases with the asymmetry of the PPh2 conformations. The range of
conformations accessible to the metal complexes is less than for the ligand radical anions, and accordingly the
temperature dependence is significantly smaller.
Introduction
(3a-), H (3b-), [Co2(PhCCPh)(CO)4(µ-BMA)]- (4-), and
[PhCW(CO)2(BMA)Cl]- (5-), in addition to those of the ligand
radical anion, [BMA]- (6-), and the radical anion of the related
compound 4,5-bis(diphenylphosphino)cyclopentene-1,3-dione
(BPCD) (7-).
Several years ago, Mao, et al.,1,2 reported EPR spectra of [Co-
(CO)2L(BMA)], L ) CO (1), PPh3 (2); BMA ) 2,3-bis-
(diphenylphosphino)maleic anhydride, best described as Co(I)
complexes with a BMA radical anion ligand. The spectra
exhibited 31P and 59Co hyperfine couplings which are unusually
temperature- and solvent-dependent. The 31P coupling increases
in magnitude with decreasing temperature while the 59Co
coupling decreases in magnitude; both couplings decrease with
increasing solvent polarity. Qualitatively similar solvent effects
were observed for the C-O stretching frequencies, the lowest
energy electronic absorption frequency, and the rate of substitu-
tion of CO by PPh3 in [Co(CO)3(BMA)]. These effects were
attributed to stabilization of the Co+(BMA)- zwitterion with
increasing solvent polarity.
Subsequent work on other metal complexes, reported in this
paper, suggests that the above interpretation, while substantially
correct with respect to solvent effects, is oversimplified with
respect to the temperature-dependent hyperfine couplings.
Accordingly, we were led to a study of the free ligand radical
anion, [BMA]-, which has greatly added to our understanding
of the system.
Here we report temperature-dependent EPR spectra of four
BMA complexes, [Co2(PhCCR)(CO)4(η-BMA)]-, R ) Ph
† University of Otago.
‡ Brown University.
‡ University of North Texas.
| University of Oregon.
(1) Mao, F.; Tyler, D. R.; Rieger, A. L.; Rieger, P. H. J. Chem. Soc.,
Faraday Trans. 1991, 87, 3113.
(2) Mao, F.; Tyler, D. R.; Bruce, M. R. M.; Bruce, A. E.; Rieger, A. L.;
Rieger, P. H. J. Am. Chem. Soc. 1992, 114, 6418.
S0020-1669(98)00581-3 CCC: $15.00 © 1998 American Chemical Society
Published on Web 09/04/1998

4850 Inorganic Chemistry, Vol. 37, No. 19, 1998
Duffy et al.
Experimental Section
The ligands BMA (6) and BPCD (7) were prepared from dichloro-
maleic anhydride and 4,5-dichlorocyclopentene-1,3-dione, respec-
tively,3,4 while the Ph2P(TMS) used in the synthesis of the diphosphine
ligands was prepared from PPh3 and TMSCl.5 The compounds 3a,
3b, and 4 were prepared from [Co2(CO)6(alkyne)] and BMA according
to published procedures.6,7 The carbyne complex [PhCW(CO)4Cl] was
synthesized according to the procedure of Mayr.8 All compounds were
handled and stored under argon using Schlenk techniques.
[PhCW(CO)2(BMA)Cl] (5). To a freshly prepared solution of
[PhCW(CO)4Cl] (ca. 1.0 mmol) in THF at -78 °C was added 0.56 g
(1.2 mmol) of BMA, after which the solution was allowed to warm to
room temperature with stirring continued overnight. IR analysis of
the crude reaction solution revealed the presence of the desired BMA-
substituted complex. THF was removed under vacuum and the product
taken up in 100 mL of toluene, followed by filtration over a short pad
of Celite. The solution was concentrated to a volume of ca. 25 mL and
then layered with 20 mL of pentane, which ultimately afforded 0.23 g
of green-yellow crystals of 5. Yield: 28%. IR (CH2Cl2): ν(CO) 2021
(s), 1958 (s), 1777 (s, BMA). 31P(1H) NMR (CDCl3): δ 32.64 (JW-P
) 117 Hz). Anal. Calcd (found) for C37H25ClO5P2W: C, 53.49
(52.54); H, 3.03 (3.46).
Metal BMA complexes, in 1:1 CH2Cl2/C2H4Cl2, 1:1 C2H4Cl2/THF,
or THF with 0.1 M [NBu4][PF6] or [NBu4][ClO4] supporting electrolyte,
were reduced electrochemically in situ in the EPR cavity. 6- was
generated by electrochemical reduction in acetonitrile or by [Co(C5H5)2]
reduction in THF with qualitatively similar results; because of the
greater temperature range accessible with THF, the data presented here
are of 6- in THF. 7- was generated electrochemically in THF with
[NBu4][ClO4] supporting electrolyte.
EPR spectra were recorded using a Bruker ER-220D X-band
spectrometer equipped with a Bruker variable temperature accessory,
a Systron-Donner microwave frequency counter, and a Bruker gauss-
meter.
Figure 1. (a) X-band EPR spectra of 3b- (i) and 4- (ii) at 210 and
240 K, respectively. (b) Comparison of 3b- central line (i) with least-
squares-fitted simulations assuming one (ii) or two (iii) unresolved 59Co
coupling. (c) Comparison of 4- central line (i) with least-squares-fitted
simulations assuming two (ii) or one (iii) unresolved 59Co coupling.
(d, e) Comparison of 3b- and 4- central lines (heavy lines) with least-
squares-fitted simulations assuming Gaussian (solid lines) or Lorentzian
(dashed lines) line shapes.
Results and Discussion
BMA Complexes. EPR spectra of [Co2(PhCCR)(CO)4(η-
BMA)]-, R ) Ph (3a-), H (3b-), and [Co2(PhCCPh)(CO)4(µ-
BMA)]- (4-), obtained on electrochemical reduction of the
neutral parent compounds in THF, 1:2 CH2Cl2/THF, or 1:1 CH2-
Cl2/C2H4Cl2, are all apparently simple 1:2:1 triplets correspond-
ing to hyperfine coupling to two equivalent 31P nuclei (I ) 1/2).
Typical spectra of 3b- and 4- are shown in Figure 1a. Although
small extra lines are usually seen in the spectra (see below),
the initially formed radical anions appear to be quite stable,
consistent with the completely reversible reduction processes
observed in cyclic voltammetry.6
is clear: unresolved coupling to one Co results in sharply
defined extrema whereas coupling to two Co nuclei gives more
rounded extrema. Since the unpaired electron resides primarily
in the maleic anhydride π-system,1 we expect coupling to only
one Co nucleus when BMA is a chelating ligand (3-) and to
two Co nuclei when BMA is bridging (4-) and can conclude
that there is no isomerization of the BMA ligation on reduction.
A total of 28 spectra in the temperature range 200-300 K
were subjected to least-squares analysis to determine the solvent
and temperature dependence of the 31P and 59Co couplings as
well as the component line widths and line shapes for 3a- (in
1:1 CH2Cl2/C2H4Cl2, 1:2 CH2Cl2/THF, and THF), 3b- (in THF),
and 4- (in 1:2 CH2Cl2/THF). The results are shown in Table
1, which also gives comparable results for 1 and 2 in THF
solution.1
The distinction between Gaussian and Lorentzian components
is subtle and relies on comparison of the wings of the line:
Lorentzian component shapes result in broader wings than do
Gaussian components. In general, Gausian line shapes are
expected if the spectrum is subject to “inhomogeneous broaden-
ing”, e.g., an unresolved 1H multiplet or an instantaneous
Gaussian distribution of coupling constants; Lorentzian shapes
are expected for “exchange-narrowed lines”, e.g., when the
radical species tumbles sufficiently rapidly to average anisotro-
pies in the g and hyperfine matrices.9 It appears that spectra
of 3a- in CH2Cl2/THF or CH2Cl2/C2H4Cl2 and 3b- in THF have
Gaussian components (Figure 1d), whereas the components for
On close inspection, the lines of the triplet spectra are seen
to be grossly distorted from either Gaussian or Lorentzian line
shapes. Least-squares analysis of digitized spectra result in very
good fits to the line shapes for 3a- and 3b- if we assume
7
unresolved hyperfine coupling to one 59Co nucleus (I ) /2);
for spectra of 4-, the fit is best assuming hyperfine coupling to
two 59Co nuclei. As seen in Figure 1b,c, which compares the
central lines of spectra of 3b- and 4- with least-squares fits
assuming one and two unresolved Co couplings, the distinction
(3) Mao, F.; Philbin, C. E.; Weakley, T. J. R.; Tyler, D. R. Organometallics
1990, 9, 1510.
(4) (a) Fenske, D.; Becher, H. J. Chem. Ber. 1974, 107, 117. (b) Fenske,
D. Chem. Ber. 1979, 112, 363.
(5) Kuchen, W.; Buchwald, H. Chem. Ber. 1959, 92, 227.
(6) Yang, K.; Bott, S. G.; Richmond, M. G. Organometallics 1994, 13,
3788.
(7) Yang, K.; Bott, S. G.; Richmond, M. G. Organometallics 1994, 13,
3767.
(8) (a) Mayr, A.; McDermott, G. A.; Dorries, A. M. Organometallics 1986,
4, 808. (b) McDermott, G. L.; Dorries, A. M.; Mayr, A. Organome-
tallics 1987, 6, 925.
(9) Atherton, N. M. Electron Spin Resonance; Ellis Horwood: Chichester,
1973; pp 44-45.

EPR Study of BMA Complexes and the BMA Radical Anion
Inorganic Chemistry, Vol. 37, No. 19, 1998 4851
Table 1. EPR Parameters for Metal Complexes
|aP|/Gb
|aCo|/Gd
w0/Ge
w(1/Ge
solvent
THF
gb
d|aP|/dTc
d|aCo|/dTc
dw0/dTc
dw(1/dTc
1f
2.0042
11.047 ( 0.005
-4.3 ( 0.1
9.69 ( 0.01
-2.7 ( 0.2
7.939 ( 0.004
-2.0 ( 0.1
8.00 ( 0.01
-2.2 ( 0.2
8.066 ( 0.004
-2.7 ( 0.1
8.20 ( 0.02
-3.6 ( 0.3
9.24 ( 0.02
-3.6 ( 0.3
8.644 ( 0.006
-3.8 ( 0.2
1.299 ( 0.001
1.57 ( 0.01g
2f
THF
2.0038
2.0036
2.0034
2.0036
2.0039
2.0035
2.0044
e0.26
3a-
3a-
3a-
3b-
4-
DCE/DCM
DCM/THF
THF
0.031 ( 0.001
0.01 ( 0.01
0.030 ( 0.001
0.02 ( 0.01
0.58 ( 0.08
-0.6 ( 1.4
0.34 ( 0.07
-4 ( 1
0.47 ( 0.06
-1.5 ( 1.1
0.26 ( 0.06
-5 ( 1
0.057 ( 0.002
-0.21 ( 0.05
0.09 ( 0.01
0.33 ( 0.02i
-1.2 ( 0.3
0.7 ( 0.1
0.30 ( 0.01i
-1.2 ( 0.3
0.7 ( 0.1
THF
-0.8 ( 0.2
0 ( 1
2 ( 2
DCM/THF
DCE/DCM
0.141 ( 0.005j
-0.02 ( 0.08
0.23 ( 0.04i
-1.8 ( 0.6
0.39 ( 0.03
-1.4 ( 0.8
0.25 ( 0.05i
-1.2 ( 0.9
0.39 ( 0.03
-1.2 ( 0.8
5-
a Coupling to two equivalent 31P nuclei at 300 K. b Uncertainty, (0.0002. c In units of mG K-1
.
d Coupling to one 59Co nucleus at 300 K (unless
f
otherwise noted). e Peak-to-peak derivative width of components of m ) 0 or (1 lines; Gaussian lines unless otherwise noted. From ref 1. g +(0.014
( 0.001 mG K-2) (T - 300). h +(0.032 ( 0.004 mG K-2) (T - 300). i Lorentzian line shapes. j Two 59Co nuclei.
3a- in THF and 4- in CH2Cl2/THF are more nearly Lorentzian
(Figure 1e). Most spectra contain small satellite lines (see
below) which may confuse the distinction between Gaussian
and Lorentzian component shapes (note that the 3b- line shown
in Figure 1a,d shows such satellites). Furthermore, there is a
strong cross-correlation in the least-squares fits between the
unresolved 59Co coupling and the component line widths.
However, in almost all cases, the fitting routine converged to
give similar results at all temperatures for spectra of 3a-, 3b-,
and 4- in a given solvent.
from decomposition of an oxidation product. Since the anode
and cathode of the in situ electrolysis cell are relatively close
together, diffusion of 1 (or 1+) from the anode into the cavity
is expected with time. Cyclic voltammograms of 3 and 4 show
a small reduction peak on the cathodic scan after traversing the
partially reversible oxidation.6 This peak increases with de-
creasing scan rate, and, when the scan is reversed at -0.3 V,
the corresponding oxidation peak is observed, E1/2 ) -0.04 V
vs Ag wire. The present EPR evidence strongly suggests that
this secondary oxidation product is 1+, consistent with the
observed reversible oxidation of 1 at 0.07 V vs SCE.10
Apparently 3+ or 4+ undergoes rapid Co-Co bond cleavage to
form an unknown mononuclear species X and 1+:
All spectra showed a relatively large dependence of the 31P
coupling on temperature, -2 to -4 mG K-1. The 59Co coupling
also increases with decreasing temperature for 3b- and 3a- in
THF but is essentially temperature-independent for 3a- in CH2-
Cl2/C2H4Cl2 or CH2Cl2/THF and 4-. The fitted component
widths show more scatter, but in general, the fitted widths for
3a- and 3b- are significantly greater than for 4-; the widths
for 3a- and 4- increase with decreasing temperature, whereas
those of 3b- are nearly temperature-independent. Furthermore,
the widths for 4- are nearly the same for m ) 0 and (1 whereas
the width of the m ) 0 line for 3a- and 3b- is greater than that
of the (1 lines. Although line widths were not accurately
measured for 1 and 2, in general they increased with decreasing
temperature. We will discuss these line width effects below.
Most of the spectra of 3- and 4- showed small satellite lines,
usually symmetrically arranged. Most of these are reproducible
and are probably 13C satellites, but in a few cases they appear
to be due to other BMA complexes, either present in the original
sample or resulting from radical decomposition. In a single
series of spectra of 3b-, a doublet of doublets was observed,
comparable in intensity with the expected spectrum, with a1 )
12.94 ( 0.01 G, a2 ) 13.91 ( 0.01 G (at 300 K), and somewhat
different temperature coefficients, -6.2 ( 0.3 and -2.2 ( 0.2
mG K-1, respectively. The lines were relatively sharp with no
evidence of unresolved 59Co coupling. The identity of the
species remains a mystery.
Co-Co bond cleavage is consistent with the expectation that
the LUMO in 3 or 4 is the major Co-Co bonding orbital.11,12
The process is thus reminiscent of the fragmentation of
[µ-(PhCCPh)Co2(CO)6]- to [(PhCCPh)Co(CO)3] and [Co-
(CO)4]-.13
EPR spectra of 5-, obtained by electrochemical reduction of
5 in 1:1 CH2Cl2/C2H4Cl2, are triplets with relatively narrow
Gaussian lines and a temperature-dependent 31P coupling. The
spectra were subjected to least-squares fitting as above; param-
eters are given in Table 1. Three pairs of satellites surround
each major line. Although they may be 13C satellites, they are
1
much too weak to result from 183W coupling (I ) /2, 14.3%
abundant). Since the satellite widths increase with decreasing
temperature more rapidly than do the major lines, it is likely
that they result from other BMA complexes present as impurities
in the original sample or resulting from decomposition of 5-.
The BMA Radical Anion. The room-temperature EPR
spectrum of [BMA]- (6-), prepared by [CoCp2] reduction in
THF, consists of the expected 1:2:1 triplet, corresponding to
In almost all cases, on continued electrolysis and/or on
standing of electrolyzed solutions of 3- and 4-, the characteristic
triplet-of-octets spectrum of 1 was observed. The measured
coupling constants correspond within experimental error to those
reported1 for 1 in the same solvent and at the same temperature.
The spectrum of 1 could be made much more prominent by
reversing the electrolysis leads briefly, suggesting that 1 results
(10) Mao, F.; Tyler, D. R.; Keszler, D. J. Am. Chem. Soc. 1989, 111, 130.
(11) Peake, B. M.; Rieger, P. H.; Robinson, B. H.; Simpson, J. J. Am.
Chem. Soc. 1980, 102, 156.
(12) Thorn, D. L.; Hoffmann, R. Inorg. Chem. 1978, 17, 126.
(13) Arewgoda, M.; Rieger, P. H.; Robinson, J.; Visco, S. J. J. Am. Chem.
Soc. 1982, 104, 5633.

4852 Inorganic Chemistry, Vol. 37, No. 19, 1998
Duffy et al.
Table 2. 31P Coupling Constants and Line Widths for 6-
widtha
m ) (1
T/K
a/G
m ) 0
Kcb
ac/G
280
270
260
240
3.076(3)
3.210(2)
3.318(2)
3.516(4)
2.86
2.22
2.34
1.87
3.05
2.39
2.45
2.08
0.051(1)
0.040(2)
6.5(2)
5.7(2)
T/K
aa/G
ab/G
av widtha
Kc
ac/G
230 3.117c
220 3.265c
210 3.413c
200 3.555(3) 6.21(1), 5.08(1)
190 3.751(1) 6.133(4), 5.252(4)
180 3.878(1) 6.352(4), 5.328(4)
170 4.007(1) 6.565(3), 5.366(3)
4.009(1) 6.582(4), 5.355(4)
5.82, 4.80c
5.95, 4.89c
6.08, 4.98c
1.36
1.32
0.94
0.88
0.75
0.70
0.63
0.61
0.62
0.034(6) 7.0(3)
0.030(6) 7.2(3)
0.036(5) 7.2(3)
0.018(4) 7.5(3)
0.004(2) 7.7(2)
0.008(2) 8.2(1)
160 4.142(2) 6.75(1), 5.45(1)
a Peak-to-peak derivative width in gauss. b Kc ) [6c]/([6a] + [6b]).
c Coupling constants extrapolated from lower temperatures used in
determination of Kba, kab, and Kc.
Table 3. Equilibrium and Rate Constants for the 6a-/6b-
Equilibrium
T/K
(10-6)kab/s-1
Kba
230
220
210
200
190
180
170
9.4(2)
5.6(1)
3.51(5)
1.94(3)
2.55(2)
2.59(3)
2.52(2)
2.42(2)
2.68(2)
2.68(2)
2.85(2)
2.92(2)
3.05(2)
Figure 2. Experimental X-band EPR spectra of [BMA]- (6-), 1 mM
in THF, at 280-160 K.
hyperfine coupling of two equivalent 31P nuclei. As the
temperature is decreased, the spectrum changes, as shown in
Figure 2. At first, the triplet resolution improves and the
coupling constant increases in magnitude (d|a|/dT ) -10.8 (
0.3 mG K-1). However, at about 220 K, the outer lines of the
triplet develop pronounced shoulders which resolve below 210
K, where the spectrum appears to be a superposition of two
1:2:1 triplets, 6a- and 6b-, with the intensity of the spectrum
of 6a- 2-3 times the intensity of the spectrum of 6b-. Between
180 and 260 K, another pair of lines is partially resolved,
corresponding to another apparent triplet spectrum (6c-) with
an amplitude less than 10% of 6a-. These features are not
detectable below 180 K and merge into the broad major features
at higher temperatures. The coupling constants of the two major
spectra continue to increase with decreasing temperature. The
relative intensities of the two major spectra remain nearly
constant between 210 and 160 K. All spectra correspond to
the same g value, 2.0035, independent of temperature.
At 160 and 170 K, shoulders are visible on the central line.
Subtraction of a simulated spectrum of the 6a- triplet shows
that the spectrum of 6b- is a doublet of doublets. Accordingly,
all spectra in the 160-190 K range were analyzed by a nonlinear
least-squares procedure which fitted the digitized experimental
spectra to a simulation optimized with respect to B0 (the same
for 6a- and 6b-), aa, ab1, ab2, Kba ) [6a-]/[6b-], the line widths,
and, for the spectra at 180 and 190 K, ac and Kc ) [6c-]/([6a-]
+ [6b-]). The same procedure was used to analyze the spectra
in the 240-280 K range, but only the average triplet coupling
was determined. In all cases, the fitting errors were consistent
with the experimental signal-to-noise ratio; to the eye, simulated
spectra are indistinguishable from experimental spectra.14 The
resulting parameters are given in Tables 2 and 3. The
temperature dependence of aP is significantly different for the
two spectra: -13 mG K-1 for 6a- and -6 and -21 mG K-1
for 6b-. Satellite lines assignable to 13C coupling are observed
in the spectra at 160 K, an outer set with ac ) 4.6 G assignable
160
to 6b- and an inner set, visible only after subtraction of the
simulated spectra from the experimental spectra, with ac ) 4.2
G assignable to 6a-.
Lines due to 6a- and 6b- coalesce in the temperature range
200-230 K. Detailed analysis of these spectra with the nonlinear
least-squares procedure, modified to incorporate the modified
Bloch equations,15 yielded the equilibrium and rate constants
for the 6a-/6b- equilibrium shown in Table 3. The coupling
constants were fitted for the 200 K spectrum, but were held
constant at values extrapolated from lower temperatures in fitting
the 210, 220, and 230 K spectra. Again, the fitting errors were
consistent with signal-to-noise ratios. From these four temper-
q
atures, we estimate the activation parameters ∆Hab ) 18.2 (
0.4 kJ mol-1 and ∆Sab ) -30 ( 2 J mol-1 K-1. Including
q
the equilibrium constants at 160-190 K, we have ∆Hba° ) -0.8
( 0.2 kJ mol-1 and ∆Sba° ) 4 ( 1 J mol-1 K-1, i.e., 6a-
corresponds to slightly lower energy and higher entropy than
6b-.
Spectra of 6- were recorded at 10° intervals, half going down
in temperature, the other half coming back up. Since the
collection of spectra form a smoothly varying set, the temper-
ature effects are completely reversible, and there is no evidence
of radical decomposition. This suggests that all three of the
observed spectra correspond to conformations of the radical
anion. The two major conformations, 6a- and 6b-, correspond
to very similar energies, but 6c- apparently corresponds to a
higher energy conformation since Kc increases with increasing
temperature (a van’t Hoff analysis is precluded by the low
precision of the Kc values).
(15) The analysis included the spectrum of 6c-, assumed to be a triplet,
but only for the purposes of overlap corrections; i.e., 6c- was assumed
to be only slowly exchanging with 6a- and 6b-.
(14) The line shapes in spectra of 6- are clearly Lorentzian.

EPR Study of BMA Complexes and the BMA Radical Anion
Inorganic Chemistry, Vol. 37, No. 19, 1998 4853
Figure 4. Potential energy contour map for BMA, determined by
molecular mechanics calculations. The angles φ1 and φ2 represent the
average CdC-P-C dihedral angles for Ph2P1 and Ph2P2, respectively.
The energy of the most stable conformation (φ1 ) (50°, φ2 ) -50°)
is set equal to 0. The corners of the map, bounded by the 20 kJ mol-1
contour line, correspond to much higher energies.
Figure 3. Experimental X-band EPR spectra of [BPCD]- (7-), 1 mM
in THF, at 260-180 K.
The BPCD Radical Anion. The EPR spectrum of 7- at 260
K, prepared by electrochemical reduction in THF, consists of
the expected triplet of triplets, corresponding to hyperfine
1
coupling by two equivalent 31P and two equivalent H nuclei.
As the temperature is decreased, the spectrum changes, as shown
in Figure 3. This radical anion is considerably less stable than
6-, and it was necessary to electrolyze continuously to obtain
spectra with reasonable signal-to-noise ratios. Spectra of
solutions reduced at or near room temperature showed features
due to unidentified decomposition products. The methylene
proton coupling of 2.97 ( 0.05 G showed little temperature
dependence, but the 31P coupling increases with decreasing
temperature. Spectra in the 250-280 K range gave |aP| (300
K) ) 3.66 ( 0.08 G with d|a|/dT ) -14 ( 3 mG K-1. At
lower temperatures, the spectrum changes more rapidly with
the growth of small shoulders and the partial splitting of the
central triplet lines, suggesting a minor conformer with a poorly
characterized spectrum and a major conformer with nonequiva-
lent 31P nuclei. The lines are generally broader at low
temperatures than those of the 6- spectra, and this, together
with the complication of the additional 1H coupling, precludes
a detailed analysis. Nonetheless, the qualitative behavior of 7-
and 6- is similar, with evidence for two or more conformers at
low temperature which rapidly interconvert as the temperature
approaches 300 K.
Mechanics Calculations for BMA. In an attempt to
understand the possible conformations available to BMA, we
performed molecular mechanics calculations, using the CAChe
Mechanics program,16 starting with a somewhat idealized
structure with C2V symmetry based on the x-ray crystal structure
of BMA.17 In the calculations, the maleic anhydride and
phosphorus atoms were locked in position, all C-P-C bond
angles and the bond lengths and bond angles of the phenyl rings
were locked, and the torsional rotations of the phenyl rings were
optimized as a function of the C-P-CdC dihedral angles. The
results of these calculations are shown in the potential energy
Figure 5. Schematic representations of the idealized “cis, cis” (φ1 )
φ2 ) 0°) and “cis, trans” (0°, 180°) conformations of BMA and the
conformations found at the energy minima of Figure 4.
contour map of Figure 4. The angles φ1 and φ2 correspond to
the average of the two C-P-CdC dihedral angles for P1 and
P2, respectively, with φ1 ) 0° corresponding to the phenyl rings
on P1 oriented away from P2 (phenyl rings cis to the nearest
carbonyl group); similarly φ2 ) 0° corresponds to the phenyl
rings on P2 oriented away from P1. To a first approximation,
the two most stable conformations correspond to φ1 ) φ2 ) 0°
(the cis, cis conformation) and to φ1 ) 0, φ2 ) 180° or φ1 )
180, φ2 ) 0° (the cis, trans conformation), Figure 5. The
conformations near φ1 ) φ2 ) 180° (trans, trans) have very
much higher energies. Relatively small distortions from these
idealized minima optimize the van der Waals interactions
between phenyl rings; thus Mechanics finds minima at φ1 ≈
(50, φ2 ≈ -50° and at φ1 ≈ 160, φ2 ≈ -30° or vice versa.
These minima differ by less than 1 kJ mol-1 and are separated
by a barrier on the order of 12-15 kJ mol-1, in surprisingly
good agreement with the analysis of the experimental spectra
described above. Two other minima at only slightly higher
energies are found near the cis, trans idealized conformation.
Recognizing that the mechanics calculations correspond to
the gas phase neutral molecule and that solvent interactions are
doubtless important in stabilizing conformations of the radical
anion and contributing to the activation barrier for interconver-
sion of conformations, the results are nonetheless consistent with
the EPR results, at least for the two major conformations. The
cis, cis conformation has equivalent P atoms, and the energy
minimum is broad. The cis, trans conformation has nonequiva-
(16) CAChe Scientific, Inc., Beaverton, OR; the computations described
here utilized MM2 parameters and the block-diagonal Newton Raphson
method to optimize structures, iterating until the energy change was
less than 0.01 kJ mol-1.
(17) Avey, A.; Schut, D. M.; Weakley, T. J. R.; Tyler, D. R. Inorg. Chem.
1993, 32, 233.

4854 Inorganic Chemistry, Vol. 37, No. 19, 1998
Duffy et al.
lent P atoms, and the energy minimum is somewhat sharper,
corresponding to slightly lower entropy. Thus it seems reason-
able to assign 6a- to the cis, cis conformation and 6b- to the
cis, trans conformation. The third minor conformation remains
unassigned.
of the 6- and 7- couplings, ca. -13 mG K-1, is considerably
greater than for the metal complexes, ca. -3 mG K-1. Although
reduced in range, many lower symmetry conformations for
which P 3s character is allowed are nonetheless accessible for
the metal complexes. For example, EHMO calculations predict
a P 3s contribution of 2.4 G for the +10°/-10° conformation.
We should also note that all EHMO calculations (as well as
the mechanics calculations) constrained the P atoms to the plane
of the OdC-CdC-CdO system; in the X-ray structure of 2,10
the BMA P atoms are displaced ca. 0.4 Å above and below the
mean plane of the OdC-CdC-CdO system. In any case,
thermal excursions from the minimum-energy conformation are
expected to broaden the distribution of aP and make the average
less negative.
Origins of the Phosphorus Hyperfine Coupling. Since the
unpaired electron resides primarily in a OdC-CdC-CdO π*
orbital, the 31P coupling most likely arises from (i) polarization
of P s orbitals by pπ spin density on the adjacent C atom and
(ii) the direct contribution of P 3s character in the singly
occupied MO, as shown by eq 1. The polarization parameter
P
QC is probably negative, but it has not been estimated either
P
P
3s
aP ) QC FCπ + QP FP
(1)
Line Width and Line Shape Effects. As the range of low-
symmetry conformations increases with increasing temperature,
we expect a decrease in |aP| for 6-, 7-, and the metal complexes.
In addition, the instantaneous distribution of aP is expected to
increase in width with increasing temperature and very likely
becomes quite asymmetric. The various line width and line
shape effects observed in the EPR spectra thus depend on the
rate of averaging of this (probably skewed) distribution.
Examination of the line widths in spectra of 6- as functions
of temperature is revealing. In general, the widths increase with
increasing temperature. At 160 K, the peak-to-peak line widths
are about 0.6 G, nearly independent of mI, suggesting negligible
g and hyperfine anisotropy. As the temperature increases,
however, the widths of the outer lines of the triplet and, to a
lesser extent, the width of the central line increase significantly
until, at 280 K, the triplet spectrum is only barely resolved.
The spin-rotational interaction, the most common line width
contribution which increases with increasing temperature, may
be ruled out in this case since this effect is proportional to the
g-matrix anisotropy21 and in any case would not affect the
hyperfine components unequally. At 280 K, line width con-
tributions from the 6a-/6b- exchange process are negligible,
but of course many other conformations are being explored at
this temperature. The rate of exchange among these conforma-
tions must increase with increasing temperature, and indeed the
lines remain Lorentzian, so that there is substantial motional
narrowing. However, the width of the aP distribution apparently
increases faster than the rate of averaging.
theoretically or experimentally. The corresponding parameter
describing polarization of P ns orbitals by dπ spin density on
Cr has been estimated at -40 G;18 the shorter C-P bond
suggests a significantly larger value for QCP, say -60 G.
Extended Hu¨ckel MO calculations on BMA suggest a π spin
density on the order of 0.2 on the C atoms to which the P atoms
are bonded, and thus a predicted coupling on the order of -12
G, comparable to the low-temperature couplings for BMA metal
complexes. Since the singly occupied MO is essentially the
same in [BMA]- and in its metal complexes, the polarization
contribution to the coupling should remain nearly constant.
The contribution of P 3s spin density is positive and
potentially large, QP ) 4750 G.19 In the idealized cis, cis or
P
cis, trans conformations with Cs symmetry, P 3s character is
symmetry forbidden but becomes allowed as the symmetry is
reduced. Using the minimum-energy conformations found from
the mechanics calculations, EHMO calculations predict P 3s
spin densities of 0.0035, 0.0035 for the +50°/-50° conformation
(approximately cis, cis) and 0.0003, 0.0017 for the +160°/-
30° conformation (approximately cis, trans), corresponding to
predicted P 3s contributions of 17, 17 G and 1, 8 G, respectively.
Extended Hu¨ckel calculations suggest that the P 3s character is
even greater for many thermally accessible conformations of
3-. Although these predictions are extremely approximate, they
are qualitatively consistent with the small, equivalent 31P
couplings for conformation 3a- and the larger, nonequivalent
couplings for conformation 3b-, and thus they support the
assignments based on the Mechanics calculations.
The problem of motional averaging of a Gaussian distribution
has been addressed by Freed.22 He found that the absorption
line shape can be expressed by an expansion in Lorentzian
functions, as shown in eq 2, where ∆ is the Gaussian width
parameter, τ-1 is a measure of the rate of averaging, and ω0 is
the center frequency of the absorption line. In practice, eq 2
Temperature Dependence of the Phosphorus Hyperfine
Coupling. From the above discussion, it is clear that the
temperature dependence of aP for 6- and 7- arises from
excursions into conformations far from Cs symmetry and the
consequent increasing contribution of P 3s character with
increasing temperature, i.e., the average observed coupling
becomes less negative. The phenomenon is thus analogous to
the temperature dependence of the H coupling in aromatic
hydrocarbon radicals,20 which arises through the C-H out-of-
plane bending mode, wherein the proton experiences a time-
averaged positive contact interaction with π spin density which
partially cancels the negative spin polarization contribution;
n
+ ∆2τ
2
2
(-1)n
n!
e∆ τ
τ
∞
I(ω) )
(∆τ)2n
(2)
∑
2
π
n
n)0
+ ∆2τ + (ω - ω0)2
(
)
τ
[C6H6]-, for example, has d|aH|/dT ) -0.36 mG K-1
.
converges for τ∆ e 4 with 70-80 terms in the sum. Line
shapes, computed using eq 2, are shown in Figure 6 for τ∆ )
0.25, 1, and 4. Also shown in Figure 6 are the correlation
coefficients, r2, for the fits of purely Gaussian or Lorentzian
line shapes to the line shape computed from eq 2 as functions
of τ∆, with an approximate extrapolation for τ∆ > 4. Gaussian
Many of the asymmetric conformations accessible for 6- are
excluded for metal complexes which must stay close to the cis,
cis idealized conformation. Indeed, the temperature dependence
(18) Cummings, D. A.; McMaster, J.; Rieger, A. L.; Rieger, P. H.
Organometallics 1997, 16, 4362.
(19) Morton, J. R.; Preston, K. F. J. Magn. Reson. 1978, 30, 577.
(20) (a) Lawler, R. G.; Fraenkel, G. H. J. Chem. Phys. 1968, 49, 1126. (b)
Reddoch, A. H.; Dodson, C. L.; Paskovich, D. H. J. Chem. Phys. 1970,
52, 2318.
(21) Atkins, P. W.; Kivelson, D. J. Chem. Phys. 1966, 44, 169.
(22) Freed, J. H. In Electron Spin Relaxation in Liquids; Muss, L. T., Atkins,
P. W., Eds.; Plenum: New York, 1972; p 165.

EPR Study of BMA Complexes and the BMA Radical Anion
Inorganic Chemistry, Vol. 37, No. 19, 1998 4855
Figure 6. (top) Absorption line shapes, computed using eq 2, for τ∆
) 0.25, 1, and 4. (bottom) Plots of absorption line half widths at half
height and correlation coefficients (r2) for fits of purely Gaussian and
purely Lorentzian line shapes to absorption lines computed using eq
2, all as functions of τ∆.
Figure 7. (a) Skewed distribution of aP with aj ) -3 G and the
resulting distributions of (b) ((a1 - a2)/2 and (c) ((a1 + a2)/2.
In spectra of the metal complexes, line widths generally
decrease with increasing temperature, but the line shapes are
variable with Gaussian lines observed for 1, 2, 3a- in CH2Cl2/
C2H4Cl2 or CH2Cl2/THF, 3b-, and 5- and Lorentzian lines for
3a- in THF and 4-. If the aP distribution increases in width
with increasing temperature, a decrease in line width suggests
that the rate of averaging increases faster than the increase in
the width of the distribution. This model is consistent when
motional averaging is fast enough that the lines are Lorentzian,
but as we have shown above, for somewhat slower motional
averaging, the line width may decrease but the line remain
essentially Gaussian in shape.
line shapes are expected from eq 2 for τ∆ > 2 and Lorentzian
line shapes for τ∆ < 0.5; for τ∆ ≈ 1, the line shape is midway
between Gaussian and Lorentzian. Thus the observation of
Gaussian or Lorentzian line shapes in the experimental spectra
is an indication of the rate of averaging relative to the width of
the distribution. For relatively slow averaging, some motional
narrowing is possible while retaining approximately Gaussian
lines, while for relatively rapid averaging, Lorentzian lines may
be observed but with rather large widths.
If the 31P coupling were distributed in a Gaussian pattern,
somewhat exchange narrowed to provide Lorentzian line shapes,
and the two 31P couplings were correlated in phase such that a1
- a2 ≈ 0 at all times, the central line of the triplet would be
sharp and the outer lines broadened; similarly an out-of-phase
correlation with a1 + a2 ≈ constant would result in a broad
central line and sharp outer lines.23 If a1 and a2 are not
correlated at all, the Gaussian distributions of ((a1 ( a2)/2
would be identical with widths 1/21/2 times than that of the aP
distribution, i.e., the widths of the inner and outer lines would
be the same. On the other hand, our discussion above suggests
that the distribution of aP may well be skewed, with a broad
tail into positive couplings as shown in Figure 7. In this case,
the distribution of ((a1 - a2)/2 would be symmetric and
somewhat narrower than the skewed distribution of ((a1 + a2)/
2, as shown in the figure. Of course, these distributions are
somewhat exchange-narrowed and the experimental spectra have
Lorentzian shapes, but the difference in width will persist for
moderate rates of averaging. The actual situation may well be
a combination of a skewed distribution in aP and partial in-
phase correlation of a1 and a2.
In the case of the metal complexes, g-matrix anisotropy is
expected to the extent that metal orbitals participate in the singly
occupied MO. Indeed, frozen solution spectra of 1 show such
anisotropy1 and isotropic spectra at low temperatures have quite
broad lines as this anisotropy is averaged slowly. Frozen
solution spectra of 3- and 4- are broad triplets with no hint of
g anisotropy; nonetheless, this contribution may have a small
but significant effect on the temperature dependence of the line
widths of isotropic spectra.
It is interesting to note the appearance of Lorentzian lines
for 4- in the same solvent where 3a- gives Gaussian lines.
Lorentzian line shapes, associated with faster motional averag-
ing, may be due to the expected greater flexibility of the six-
membered ring of the bridging complex 4- compared with the
five-membered ring of the chelate complex 3a-. We also note
that, for the bridging complex 4-, the m ) 0 and (1 lines have
the same width, whereas the (1 lines are broader for the
chelating complexes, 3-. The broader m ) 0 lines for the
chelate complexes suggest that the two 31P nuclei may be
slightly nonequivalent, even when averaged; this is expected
for 3- where one P atom is trans to Co and the other trans to
one of the acetylene carbon atoms.
(23) These effects are analogous to the “alternating line width effects”
observed in EPR spectra of the dinitrodurene anion radical24 and the
durosemiquinone cation radical,25 where steric effects lead to trapping
of the unpaired electron on one nitro group or slow the rate of
conformational averaging of the hydroxyl groups.
(24) Freed, J. H.; Fraenkel, G. K. J. Chem. Phys. 1963, 39, 326.
(25) Bolton, J. R.; Carrington, A. Mol. Phys. 1962, 5, 161.
Summary
(1) Reduction of [Co2(PhCCR)(CO)4(η-BMA)], R ) Ph (3a),
H (3b), [Co2(PhCCPh)(CO)4(µ-BMA)] (4), or [PhCW(CO)2-
(BMA)Cl] (5) gives stable radical anions, 3-, 4-, and 5-, which

4856 Inorganic Chemistry, Vol. 37, No. 19, 1998
Duffy et al.
exhibit EPR spectra with temperature-dependent hyperfine
coupling to two equivalent 31P nuclei and to one (3-) or two
(4-) 59Co nuclei. Oxidation of 3 or 4 leads to Co-Co bond
cleavage with one of the products identified as [Co(CO)3-
(BMA)]+ (1+).
π system. The aP temperature dependences observed for BMA
metal complexes, -2 to -4 mG K-1, and for 6- and 7-, -6 to
-21 mG K-1, thus result from the increase in P 3s character
with increasing temperature as the PPh2 groups explore con-
formations increasingly removed from ideality.
(2) Reduction of BMA (6) or BPCD (7) gives radical anions,
6- and 7-, the EPR spectra of which exhibit complex temper-
ature dependence. Detailed analysis of the 6- spectra, together
with a mechanics calculation for 6, leads to the identification
of two conformational isomers corresponding to the symmetric
and asymmetric arrangements of the PPh2 groups. Line shape
analysis results in thermodynamic and activation parameters for
the interconversion of these rotational conformers.
(4) With increasing temperature, the width of the distribution
of instantaneous 31P couplings increases more rapidly than the
rate of averaging, leading to an increase in line widths in spectra
of 6-. Temperature effects on line widths and component
shapes in spectra of the BMA metal complexes arise from
similar, but relatively smaller, changes with temperature in the
distribution of aP and the rate of averaging.
(3) The 31P couplings observed in EPR spectra of BMA metal
complexes, as well as 6- and 7-, results from a negative spin-
polarization contribution, proportional to the π-spin density on
the adjacent C atom, and a (potentially large) positive contribu-
tion from P 3s character in the singly occupied MO. The latter
contribution is symmetry forbidden in idealized C2V structures
with the P atoms in the plane of the OdC-CdC-CdO group
of BMA or BPCD, but becomes allowed at finite temperatures
where the PPh2 groups rotate or deviate from the plane of the
Acknowledgment. We thank R. G. Lawler for helpful
discussions and J. H. Freed for guiding us to his work on
exchange-narrowing of Gaussian lines. We thank the Robert
A. Welch Foundation (Grant No. B-1039) for grant support to
M.G.R., the National Science Foundation for grant support to
D.R.T., and the NSF/New Zealand Exchange Program for
supporting six weeks’ work by N.W.D. at Brown University.
IC980581H