Reduction of monophosphaallenes: An EPR study and ab initio investigations of (HPCCH2)-· and (HPCHCH2)· radicals

Article abstract of DOI:10.1021/jp982519b

Cyclic voltammetry shows that monophosphaallene ArP=C=C(C₆H₅)₂ (where Ar = C₆H₂^tBu₃-2,4,6), 1a, undergoes irreversible reduction at 2266 mV in THF. The EPR spectra of the reduction products are obtained in liquid and frozen solutions after specific ¹³C enrichment of the allenic carbon atoms. The resulting hyperfine tensors are compared with those obtained from ab initio MP2, MCSCF, CI, and DFT calculations for the radical anion (HP=C=CH₂)^-· and for the monophosphaallylic radical (HP·-CH=CH₂) ? (HP=CH-·CH₂). The most elaborate treatments of the hyperfine structure (CI and DFT) indicate that the species observed by EPR is the monophosphaallylic radical.

Full text of DOI:10.1021/jp982519b

J. Phys. Chem. A 1998, 102, 10469-10475
10469
Reduction of Monophosphaallenes: An EPR Study and ab Initio Investigations of
(HPCCH2)-• and (HPCHCH2)• Radicals
Mostafa Chentit, Helena Sidorenkova, and Michel Geoffroy*
Department of Physical Chemistry, UniVersity of GeneVa, 30 quai Ernest Ansermet, 1211 GeneVa, Switzerland
Yves Ellinger*
Laboratoire d’Etude The´orique des Milieux Extreˆmes, Ecole Normale Supe´rieure, 24 rue Lhomond,
75005, Paris, France
ReceiVed: June 8, 1998; In Final Form: October 7, 1998
t
Cyclic voltammetry shows that monophosphaallene ArPdCdC(C6H5)2 (where Ar ) C6H2 Bu3-2,4,6), 1a,
undergoes irreversible reduction at 2266 mV in THF. The EPR spectra of the reduction products are obtained
in liquid and frozen solutions after specific 13C enrichment of the allenic carbon atoms. The resulting hyperfine
tensors are compared with those obtained from ab initio MP2, MCSCF, CI, and DFT calculations for the
radical anion (HPdCdCH2)-• and for the monophosphaallylic radical (HP•-CHdCH2) T (HPdCH-•CH2).
The most elaborate treatments of the hyperfine structure (CI and DFT) indicate that the species observed by
EPR is the monophosphaallylic radical.
Introduction
result from indirect mechanisms (spin polarization) are probably
not very suitable for providing information about the spin
The past decade has been marked by considerable develop-
ments in the chemistry of low-coordinated phosphorus com-
pounds.1 A large variety of new organic molecules containing
a trivalent dicoordinated phosphorus atom have been synthe-
sized. Most of these molecules present interesting properties in
coordination chemistry,2 and recent investigations have shown
that they are often good electron acceptors.3,4 In the present
paper, our objective is to explore the possibilities of adding an
electron to monophosphaallene5 1a and to determine the
electronic structure of the corresponding reduction compound.
delocalization. On the other hand, ab initio predictions of the
isotropic couplings using a single-determinant representation of
the wave function, even amended by Moller-Plesset perturba-
tion theory, are expected to be rather inaccurate for three-
electrons, three-centers π systems and spin-clean calculations
are necessary.
To facilitate the analysis and the interpretation of our results,
the following strategy has been adopted: (i) synthesis of
specifically 13C-enriched compounds ArPd13CdC(C6H5)2 (1a1)
and ArPdCd13C(C6H5)2 (1a2) followed by electrochemical
reduction have been carried out. (ii) The EPR spectra have been
recorded in both liquid and frozen solutions in such a way as
to measure the Fermi-contact interaction as well as the dipolar
couplings. (iii) Finally, ab initio and DFT calculations of the
isotropic and anisotropic coupling constants have been per-
formed for model compounds where the aryl groups have been
replaced by hydrogen atoms. Owing to the well-known but still
unpredictable artifacts that may spoil the calculations of EPR
parameters according to the method used, particular attention
has to be paid to this theoretical part.
Finally, it is shown that when all these conditions are fulfilled,
a joint interpretation of the EPR measurements and ab initio
calculations leads to an unambiguous identification of the
reduction product of 1a.
Two approaches are generally used to obtain structural
information about a paramagnetic species: (1) A spectroscopic
approach which characterizes the radical by its hyperfine
couplings and which interprets these parameters in terms of spin
densities. (2) Quantum mechanical investigations which lead
to the wave function of the open-shell system in its equilibrium
structure.
Several difficulties are expected to appear when applying
these methods to the products resulting from the reduction of
1a. On one hand, since this system is likely to be delocalized
on three atoms, a reliable analysis of the experiments requires
that the spin densities must be obtained on two carbons which,
a priori, do not bear any R-proton; moreover, an eventual π
structure implies that the isotropic hyperfine couplings which
Experimental Section
Compounds. Molecule 1a has been synthesized by reacting,
after lithiation with nBuLi, ArPdCCl2 with ClSiMe3 and
6
7
reacting the resulting phosphaalkene ArPdC(C1)SiMe3 with
tBuLi; finally, benzophenone was added to the reaction mixture.
The product 1a was purified by recrystallization from hexane.
These reaction sequences were also followed for the synthesis
of 1a1 and 1a2, using, respectively, H13CC13 and Od13C(C6H5)2
for the preparation of these two isotopically enriched com-
pounds.
10.1021/jp982519b CCC: $15.00 © 1998 American Chemical Society
Published on Web 12/01/1998

10470 J. Phys. Chem. A, Vol. 102, No. 51, 1998
Chentit et al.
Figure 1. EPR spectra obtained at 285 K with an electrochemically reduced solution of (a) ArPdCdC(C6H5)2, (b) ArPd13CdC(C6H5)2, (c) ArPd
Cd13C(C6H5)2.
Instrumentation. Cyclic voltammetry measurements were
carried out on a BAS station (model CV-50W) with a platinum
electrode, an SCE reference electrode, and tetrabutylammonium
hexafluorophosphate (10-1 M) as the electrolyte. EPR spectra
were recorded on a Bruker 200D spectrometer (X-band, 100
kHz field modulation) equipped with a variable-temperature
attachment. NMR spectra of all compounds were obtained prior
to EPR studies. The analysis of all EPR spectra was controlled
by numerical simulation.8
Computations. The ab initio calculations were performed
9
with GAUSSIAN94 for UHF, MP2, and density functional
levels of theory. MCSCF calculations used HONDO8.5,10 and
CI treatments were carried out with the MELDF11 package.
Results and Discussion
EPR Spectra. Cyclic voltammetry of 1a shows that this
compound undergoes irreversible reduction at -2266 mV in
THF and at -1965 mV in DMF. The in situ electrolysis, inside
the EPR cavity, of a solution of 1a in THF leads, at 285 K, to
an EPR spectrum characterized by a 31P isotropic coupling
constant of 262 MHz (Figure 1a). As shown in Figure 1b, when
this experiment is carried out with the 13C-enriched compound
1a1, an additional splitting of 33 MHz is observed, whereas
isotopic enrichment of the terminal carbon (1a2) gives rise to
an additional coupling of 34 MHz (Figure 1c).
Figure 2. EPR spectra obtained at 110 K with an electrochemically
The EPR spectra are drastically modified when the temper-
reduced solution of (a) ArPdCdC(C6H5)2, (b) ArPd13CdC(C6H5)2,
ature of the electrolyzed solution is lowered from 285 to 150
K. A broadening of the signals occurs between 285 and 160 K,
which principally affects the external lines. Below 150 K, any
motion of the paramagnetic species is frozen and the resulting
“powder-type” spectra are characterized by an axial 31P hyper-
fine tensor. The trace of this tensor is totally consistent with
the value of Aiso measured in the liquid phase. The 13C coupling
is easily measured on the external lines (“parallel” components
of the phosphorus hyperfine structure). For 1a1, this splitting is
equal to 45 MHz; this value together with the isotropic constant
of 33 MHz imply, in the hypothesis of an axial 13C coupling
tensor, a “perpendicular” coupling of 27 MHz. Similarly, the
13C splitting of 79 MHz observed on the external lines of 1a2
together with the isotropic coupling of 34 MHz imply a
“perpendicular” coupling of 11.5 MHz for the terminal 13C.
It is worth mentioning that the simulation of the frozen
solution spectra8 with the above parameters, on the hypothesis
of aligned axial g,31P, and 13C coupling tensors, leads to a fitting
which is not perfect for the central parts of the spectra of 1a1:
when the line width is adjusted to fit the external lines, the
simulation predicts that, in contrast to the experimental spectrum,
some 13C splitting would be detected in the central pattern. This
implies that an additional small interaction is probably present
which affects the perpendicular region more than the parallel
one. As shown in Figure 2, satisfactory simulations were
obtained by using the tensors given in Table 1 together with an
(c) ArPdCd13C(C6H5)2. a′, b′, and c′ are the corresponding simulated
spectra (see text).
additional proton coupling tensor: Tx ) 8 MHz, Ty ) 20 MHz,
Tz ) 4 MHz (with the z axis aligned along the 31P-T| direction).
The decomposition of the 31P and 13C hyperfine tensors into
isotropic and anisotropic coupling constants (Table 1) clearly
shows that, in the reduced species, the unpaired electron is
mainly localized on the phosphorus atom. Comparison of the
experimental isotropic and anisotropic interactions with the
corresponding atomic coupling constants12 (31P, A*iso ) 13 306
MHz, 2Bo ) 733 MHz; 13C, A*iso ) 3777 MHz, 2Bo ) 214
MHz) leads to the spin densities given in Table 2.
The very small s character of the magnetic orbital together
with the large participation of phosphorus and terminal carbon
p atomic orbitals are indicative of a π type SOMO, the isotropic
coupling constants being mainly due to spin polarization. As
shown in Table 3, these features are not in conflict with the
properties expected for a (RPdCdCR′2)-• radical anion.
Quantum Mechanical Calculations. Phosphaallenic deriva-
tives, such as 1a, in which three aromatic groups are attached
to the allenic PdCdC linkage are still far beyond the attainable
targets for today’s quantitative ab initio methods. Simpler model
systems have to be considered. For phosphorus species contain-
ing allenic systems, we have shown that replacement of the
cumbersome aryl groups by hydrogens provides model com-

Reduction of Monophosphaallenes
J. Phys. Chem. A, Vol. 102, No. 51, 1998 10471
TABLE 1: EPR Parameters from the Observed Spectra
isotropic couplinga
species
tensor
eigenvaluesa
liquid
frozen
anisotropic couplingb
1a
g
g// 2.0027, g 2.0094
T// 725, T 30
31P (MHz)
262
262
τ// 463, τ - 232
1a1
31P (MHz)
T// 725, T 30
T// 45, T 23
262
33
262
30
τ// 463, τ - 232
τ// 15, τ -7.5
13
C
(MHz)
(1)
1a2
31P (MHz)
T// 725, T 30
T// 79, T 12.5
262
34
262
35
τ// 463, τ -232
τ// 44, τ - 22
13
C
(MHz)
(2)
a Only the absolute values of the coupling constants are known. The T values are obtained from a simulation of the frozen solution spectra (see
text). b Only the relative signs of the anisotropic coupling constants are known.
TABLE 2: Experimental Spin Densities
atom
s spin density
p spin density
phosphorus
C1 (central carbon)
C2 (terminal carbon)
0.019
0.008
0.009
0.616
0.059
0.202
TABLE 3: Hyperfine Interaction for Diphosphaallene and
Monophosphaalkene Radical Anion
31P coupling
(MHz)
13C coupling
(MHz)
Figure 3. Optimized structure for monophosphaallene HPdCdCH2.
radical anion
orbitals:
(ArPd13CdPAr)-• 4
Aiso ) 215
Aiso ) 152
τ// ) 303
Aiso ) 27
Aiso ) 16
τ// ) 31
(ArPd13C(H)C6H5)-• 5
[core]14 [σPH σCH σCH σPC
σ
CC 1pP π π π* π* ×
σPH*σCH* σCH* σPC* σCC*]16
pounds which retain the main characteristics of the larger
systems for neutral molecules as well as for the positive13 and
negative ions.4 Applying the same philosophy to the present
class of allenic species points to 2 as the proper model system.
1. The Diamagnetic Molecule 2. Before we verify to what
extent the above interpretation is consistent with the EPR
experiments we first consider the parent molecule 2. The
In this treatment, the core consists of the inner shells of the
two carbons and phosphorus atoms. All antibonding orbitals
are used in the definition of the active space, and internal
excitations are limited to singles and doubles with respect to
the reference determinant. This level ensures a correct treatment
of electronic correlation for structural studies.14
The Cs geometry found for HPdCdCH2 is typical of allenic
compounds with the PH bond in a plane perpendicular to the
HCH plane (Figure 3).
The difference from allene is that the replacement of one of
the CH2 groups by PH induces a slight distortion of the Pd
CdC skeleton from linearity (∼5°). This distortion is inferior
to that observed for PdCdP in diphosphaallene13 at the same
level of theory (∼10°). The CdC bond is close to that of allene,
and PdC is the same as in HPdCdPH. Comparison of the
calculated geometry with the structural parameters deduced from
X-ray diffraction on the bulky compound 1a5 shows the validity
of the model. The main difference is the opening of the RPC
angle with respect to HPC as a consequence of steric repulsion.
It should also be mentioned that these results obtained with
extended basis sets and correlated wave functions do not change
the previous results15 significantly.
optimized geometry obtained at several levels of theory (MP2/
6-311G**, MCSCF/6-311G**, and B3LYP/6-311G**) is re-
ported in Table 4.
The results of the RHF/3-21G* calculations are presented only
to show the level of quality which can be expected for large
systems when using this cost-effective atomic representation.
Anticipating the breakdown of the allenic system with reduction,
we also performed MCSCF/6-311G** calculations whose active
space includes all valence orbitals. In terms of a localized orbital
description of the σ backbone, the electronic configurations are
generated by distributing the 16 valence electrons into 15
2. The NegatiVe Ion 2-. The same series of calculations was
performed on the negative ion, including correlation effects at
TABLE 4: Optimized Parametersa and Energies (au) for HPdCdCH2
RHF/3-21G*
MP2/6-311G**
MC/6-311G**
B3LYP/6-311G**
expb
E
-416.07288
1.292
-418.71830
1.316
-418.33372
1.319
-419.34557
1.304
C(1)-C(2)
P(3)-C(1)
C(2)-H(4)
P(3)-H(6)
P(3)C(1)C(2)
H(4)C(2)C(1)
H(4)C(2)H(5)
H(6)P(3)C(1)
H(6)P(3)C(2)H(4)
1.310
1.628
1.624
1.075
1.409
176.47
121.44
117.12
97.05
1.646
1.087
1.417
173.36
121.01
118.98
94.92
1.655
1.095
1.435
174.96
121.18
117.64
96.93
1.646
1.087
1.432
173.84
121.42
117.14
95.95
167.7
122.4
117.2
103.9
-91.5
91.12
91.52
91.19
91.60
a Bond lengths in Angstro¨ms; bond and torsion angles in degrees. b The experimental data are obtained from the crystal structure of 1a.

10472 J. Phys. Chem. A, Vol. 102, No. 51, 1998
Chentit et al.
B3LYP-
TABLE 5: Optimized Parametersa and Energies (au) for the Radical Anion (HPdCdCH2)-
cis-like
trans-like
UHF-
MP2-
MC-
B3LYP-
UHF-
MP2-
MC-
3-21+G*
6-311++G** 6-311++G** 6-311++G**
3-21+G*
6-311++G** 6-311++G** 6-311++G**
E
-416.07139 -418.70022
-418.28706
1.8
-419.35056 -416.07367 -418.70503
-418.29000
0.0
-419.35369
0.0
∆E (kcal)
1.4
1.334
1.761
1.084
1.081
1.429
155.78
122.08
122.11
115.81
96.99
2.18
3.0
1.323
1.688
1.100
1.091
1.442
150.44
122.02
122.82
115.15
99.94
17.83
50.25
2.0
1.331
1.726
1.103
1.093
1.456
149.88
122.70
122.85
114.46
98.96
8.63
0.0
1.332
1.767
1.085
1.081
1.417
144.85
122.37
122.00
115.63
96.38
10.47
127.32
0.0
1.324
1.706
1.100
1.092
1.421
138.57
122.75
155.18
115.03
98.14
23.92
122.24
C(1)-C(2)
1.343
1.769
1.110
1.105
1.345
1.769
1.110
1.103
1.335
1.740
1.102
1.093
C(1)-P(3)
H(4)-C(2)
H(5)-C(2)
H(6)-P(3)
1.456
1.422
1.438
P(3)C(1)C(2)
H(4)C(2)C(1)
H(5)C(2)C(1)
H(4)C(2)H(5)
H(6)P(3)C(1)
H(4)C(2)C(1)P(3)
H(6)P(3)C(2)H(4)
149.05
122.56
122.61
114.83
98.48
9.31
46.92
140.65
123.22
122.02
114.74
97.75
16.07
119.76
140.31
123.09
122.32
114.58
97.80
18.32
121.64
20.19
53.64
a Bond lengths in Angstro¨ms; bond and torsion angles in degrees.
Figure 4. Optimized structures for the two isomers of the monophosphaallenic radical anion (HPdCdCH2)-.
a
TABLE 6: Infrared Frequencies (cm-1) and Intensities (km/mol) for HPdCdCH2 and Its Negative Ionsb
H-PdCdCH2
H-PdCdCH2- cis-like
H-PdCdCH2- trans-like
assignment
PCC bend
PCC bend
HPCCH2 tors
PC stretch
CH2 wag
HPC bend
CH2 rock
HCH bend
CC stretch
irep
ν
I
ν
I
ν
I
A′
A′′
A′′
A′
A′
A′
A′′
A′
A′
A′
A′
A′′
305
315
736
759
859
888
943
1367
1764
2259
3028
3097
8
12
0
191
275
498
605
712
764
920
1299
1461
1942
2672
2784
6
247
312
534
630
728
815
931
1295
1454
2063
2689
2789
13
14
35
28
43
23
25
60
91
12
35
33
44
53
25
50
108
326
289
84
0
53
33
2
0
74
88
0
PH stretch
CH2 stretch (sym)
CH2 stretch (asym)
242
248
84
0
a B3LYP/6-311G**. b B3LYP/6-311++G**
the MP2/6-311++G** and MCSCF/6-311++G** levels of
theory. The structural parameters are reported in Table 5.
The UHF/3-21+G* results are also presented in view of
possible application to larger systems. The MCSCF treatment
is adapted from the neutral system by adding one more electron
to the valence space. Diffuse functions have been systematically
added to the basis set to account for the spatial extension of
the wave function.
As for diphosphaallene radical anion HPdCdPH-, two
minimum energy structures, cis-like and trans-like (Figure 4),
are found which differ from the neutral allenic parent by
opposite rotation of the P-H bond. The trans-like minimum is
more stable by about 1.8-3.0 kcal/mol. The two structures also
differ in the value of the PCC bending angles. The trans-like
minimum has a PCC angle of ∼140°, whereas it opens by 10°
in the cis-like minimum due to the repulsion between the PH
and CH bonds. In accord with the lower strength of the
phosphorus-carbon π bond compared to the carbon-carbon π
bond, the formation of the anion leads to a larger increase of
the PdC bond than of the CdC bond.
These two geometries are true minima as proved by the
vibrational analysis reported here at the B3LYP/6-311++G**
level together with that of the neutral parent (Table 6).
Frequencies have been scaled by 0.975 according to the
prescription by Bauschlicher and Langhoff.16 There are two
important consequences following the addition of one electron
in the π orbitals: first, a general shift of all vibrations to lower
frequencies, which shows the weakness introduced in the allenic
system, and second, a strong increase in the band strengths since
the total integrated intensity in the ion is 1065 (cis-like) or 906
km/mol (trans-like) compared to 270 km/mol in the neutral. The
breaking of the conjugated system also changes the coupling
between vibrations. The PC stretching which was strongly mixed
with the CC stretching in the neutral molecule is well-separated
in the ion, the CC stretching at 1441-1434 cm-1 now being

Reduction of Monophosphaallenes
J. Phys. Chem. A, Vol. 102, No. 51, 1998 10473
TABLE 7: Calculated Isotropic and Anisotropic Coupling Constants (MHz) for (HPdCdCH2)-
cis-like
trans-like
B3LYP-
UHF-
MP2-
B3LYP-
SCI-
SCI-
UHF-
MP2-
SCI-
SCI-
3-21+G* 6311++G** 6311++G** 631+G* aug-cc-pVDZ 3-21+G* 6311++G** 6311++G** 631+G* aug-cc-pVDZ
Aiso 13C(1)
Aiso 13C(2)
Aiso 31P(3)
165
-34
125
-118
217
344
119
22
291
190
2
113
184
-8
180
183
-22
-14
-104
228
118
139
28
-12
133
44
73
133
45
70
τ
τ
τ
13C(1)
13C(2)
31P(3)
113
-47
-66
-15
2
13
-11
20
122
4
91
-44
-47
-2
110
-58
-52
-2
112
-52
-60
-3
96
-58
-38
-15
2
13
84
-38
-46
120
-129
9
-55
-55
110
200
-76
-124
90
-48
-42
-5
88
-45
-43
-5
90
-46
-44
-5
-126
-54
-53
107
-57
90
-6
-7
-6
-3
-2
-2
8
9
9
-40
83
8
7
7
-78
149
-71
-62
122
-60
191
-92
-99
171
-81
-90
167
-80
-87
-9
-33
-43
coupled with the nearby HCC bending at 1281-1277 cm-1
.
of the frozen solution spectra. The most likely possibility that
would provide such an additional coupling is that a hydrogen
atom could be attached to the central carbon of the allenic
structure. The new model compound 3 can exist, a priori, under
the two mesomeric forms of an allylic system and is likely to
exhibit a preferential localization of the unpaired electron at
the phosphorus atom.
The EPR parameters calculated for the two minima are reported
in Table 7. In view of the large spin contamination of the UHF
wave function, CI calculations in a basis of spin-adapted
configurations were performed to ensure a spin-clean treatment
of the spin-dependent observable, i.e., a CI wave function that
is an eigenfunction of Sz as well as S.2 In this latter case, the
MCSCF/6-311++G** optimized geometry were used.
Two basis sets were employed: 6-31+G* and aug-cc-
pVDZ.17 The first one is a smaller but well-balanced basis set
which is well-suited to the CI treatment of negative ions;18 the
second one was selected after systematic investigations of
phosphorus-containing simple radicals. It was verified that the
choice of the method used for the geometry optimization does
not change the results by more than a few percent.
Geometry optimizations have been carried out at the same
levels of theory as for the negative ion. The MCSCF valence
space
From the values of Table 7 (and other calculations not
reported here), it is clear that none of the wave functions can
reproduce the experimental spectra. This is true not only for
UHF, UMP2, and DFT treatments but also for the spin-adapted
CI calculations.
[core]14 [σPH σCH σCH σCH σPC
σ
CC 1pP π π π* ×
σPH*σCH* σCH* σCH* σPC* σCC*]17
This result is not too surprising for isotropic coupling
constants. It is well-known, indeed, that the values of Aiso are
difficult to obtain with accuracy since they depend on the local
value of the spin density at the nuclei. UHF and MP2
calculations are irrelevant in this study because of spin
contamination. However, SCI calculations which contain all
single excitations from the reference determinant in order to
account for the dominant spin-polarization effects generally
provide better than qualitatively correct Aiso values. Here, they
are often off by 1 order of magnitude (experimental values
(MHz) Aiso(13C(1)) ) 33; Aiso(13C(2)) ) 34; Aiso(31P) ) 262;
cis-like SCI(MHz) Aiso(13C(1)) ) 190; Aiso(13C(2)) ) 2; Aiso-
(31P) ) 113; trans-like SCI(MHz) Aiso(13C(1)) ) 133, Aiso(13C-
(2)) ) 44; Aiso(31P) ) 73). DFT values do not show any better
agreement.
Turning to anisotropic coupling constants does not show any
improvement either. These observables which are linked to the
spatial spin distribution are generally handled better at lower
levels of wave functions. UHF single-determinant calculations,
even with a small spin contamination, are known to provide a
good first approximation when a single determinant representa-
tion is sufficient (which is not true of the present system).
Obviously, the calculated tensors are not related to the measured
spectra. Since neither isotropic nor anisotropic couplings could
be even approached here, it was clear that the recorded spectra
were not the signature of the postulated negative ion.
where there are one more σCH and σCH* pair of orbitals but
only three π orbitals still contains 17 electrons.
Two minimum-energy structures are also found here, trans
and cis, both planar and corresponding to 2A′′ radicals (Figure
5), but contrary to the previous ions, they are of equal energy.
The geometrical parameters are reported in Table 8, and the
vibrational analysis is given in Table 9 for both structures.
The two radicals are of allylic type with three electrons in
the π system, the unpaired spin being mainly localized on
phosphorus as expected. The PdCdC backbone makes an angle
of 120-130° characteristic of allylic radicals instead of 140-
150° in the ions. The 10° difference between the cis and trans
neutral radicals accommodates the repulsion between the PH
bond and one of the CH bonds of the methylenic group. The
bond lengths are larger in the neutral compounds since addition
of a hydrogen atom to the central carbon has broken the in-
plane conjugation still partially effective in the phosphaallenic
systems. The IR frequencies reflect the structural modifications.
The PCC bending shifts to higher wavenumbers in the neutral
systems. The opposite shifts of CC (-100 cm-1) and PC (+50
cm-1) with respect to the negative ion is also an illustration of
the more balanced electronic distribution in the neutral radical.
The EPR parameters calculated for the two minima are
reported in Table 10. As for the negative ion, CI calculations
were performed to ensure a spin-clean treatment of the spin-
dependent observables using the MCSCF/6-311++G** opti-
mized geometry and the 6-31+G* and aug-cc-pVDZ basis sets.
3. The Neutral Allylic Radical 3. As mentioned in a preceding
section, the simulation of the spectra suggested an additional
but small hyperfine interaction that would explain the line width

10474 J. Phys. Chem. A, Vol. 102, No. 51, 1998
Chentit et al.
Figure 5. Optimized structures for the two isomers of the monophosphaallylic radical (HP-CHdCH2)•.
TABLE 8: Optimized Parametersa and Energies (au) for the Allylic Radical (HP-CHdCH2)•
cis-like
trans-like
UHF-
MP2-
MC-
B3LYP-
UHF-
MP2-
MC-
B3LYP-
3-21+G*
6-311++G** 6-311++G** 6-311++G**
3-21+G*
6-311++G** 6-311++G** 6-311++G**
E
-416.69654 -419.31854
-418.93014
0.05
-419.96154 -416.69658 -419.31845
-418.93022
0.00
-419.96158
0.00
∆E (kcal)
C(1)C(2)
C(1)P(3)
H(4)C(2)
H(5)C(2)
H(6)P(3)
H(7)C(1)
P(3)C(1)C(2)
H(4)C(2)C(1)
H(5)C(2)C(1)
H(6)P(3)C(1)
H(7)C(1)P(3)
0.03
1.367
1.772
1.074
1.076
1.405
1.077
122.48
121.43
121.66
97.10
119.48
-0.06
1.332
1.790
1.085
1.087
1.413
1.087
121.32
121.79
121.49
95.89
119.93
0.03
1.354
1.785
1.085
1.086
1.427
1.087
122.20
121.61
121.60
96.14
119.39
0.00
1.367
1.769
1.075
1.075
1.407
1.079
127.34
121.25
121.76
97.09
115.11
0.00
1.332
1.787
1.086
1.086
1.415
1.089
126.63
121.55
121.59
94.75
115.29
1.357
1.820
1.095
1.096
1.430
1.097
121.96
121.56
121.59
96.69
1.357
1.815
1.095
1.095
1.433
1.098
127.00
121.40
121.97
96.39
1.355
1.781
1.085
1.085
1.430
1.089
127.73
121.44
121.68
95.80
119.37
114.88
114.43
a Bond lengths in Angstroms. b Bond and torsion angles in degrees.
TABLE 9: Infrared Frequencies (cm-1) and Intensities (km/mol) for HPdCHdCH2
H-PdCHdCH2 cis-like
a
H-PdCHdCH2 trans-like
assignment
irep
ν
I
ν
I
PH rot
PCC bend
C-PH wag
PC stretch
HPC bend
CH2 wag
CH wag
CH2 rock
HCC bend
CC stretch
HCH bend
PH stretch
CH2 stretch (sym)
CH stretch
CH2 stretch (asym)
A′′
A′
A′′
A′
A′
A′′
A′′
A′
A′
A′
A′
A′
A′
A′
A′
297
328
510
682
876
912
968
1025
1264
1347
1512
2295
3044
3062
3131
45
0
11
6
16
38
30
9
1
11
2
96
7
5
360
353
516
666
868
907
979
1037
1264
1360
1506
2279
3041
3048
3130
0
2
7
6
17
38
24
15
0
14
1
93
8
7
7
10
a B3LYP/6-311++G**.
A cursory examination of the calculated hyperfine constants
(Table 10) shows that the two isomers have practically the same
couplings. The average values obtained from SCI (aug basis
set) calculations are in good accordance with the experimental
couplings: (1) the three dipolar tensors exhibit axial symmetry
and their eigenvalues are similar to the measured values 13C-
(2), τ|calc ) 44.5 MHz, τ|exp ) 43 MHz; 31P, τ|calc ) 464.5 MHz,
MHz, Aiso,exp ) -32.7 MHz. The calculated isotropic values
are satisfactory at the CI-level only, which was to be expected
in view of the strong spin contamination of allylic systems. DFT
couplings, though more remote from experiments, are still
qualitatively correct.
As expected for a π radical, the τ| eigenvectors of the 31P,
13C(1), and 13C(2) couplings are oriented perpendicular to the
molecular plane. Due to the specificity of the allylic structure,
the calculated values of the Fermi contact (Aiso) and of the
dipolar interaction (τ|) for the central carbon are negative.
Moreover, the calculated Aiso for H(7), attached to the central
atom, is positive while the anisotropic coupling of this proton,
along the 31P - τ|, is negative (ca. - 4 MHz). It leads to a
τ
|exp ) 465 MHz;13C(1) τ|calc ) -14 MHz, τ|.exp ) -15 MHz
in the hypothesis of three negative eigenvalues for Texp (13C-
(1)). It can also be seen that the B3LYP method provides dipolar
tensors which are close enough to experiment to justify its use
as a substitute to the SCI treatments. (2) The calculated Fermi
contact interactions agree reasonably with the experimental
results: 13C(2), Aiso,calc ) 24.5 MHz, Aiso,exp ) 34 MHz; 31P
Aiso,calc ) 216 MHz, Aiso,exp ) 262 MHz; 13C(1) Aiso,calc ) -30
1
total H coupling which is small in this direction, whereas the
maximum interaction lies in the molecular plane for this proton.

Reduction of Monophosphaallenes
J. Phys. Chem. A, Vol. 102, No. 51, 1998 10475
TABLE 10. Calculated Isotropic (AIso) and Anisotropic (τ) Hyperfine Coupling Constants (MHz) for the Monophosphaallylic
Radical (HP-CHdCH2)•
cis-like
trans-like
UHF-
MP2-
B3LYP-
SCI-
SCI-
UHF-
MP2-
B3LYP-
SCI-
SCI-
3-21+G* 6311++G** 6311++G** 631+G* aug-cc-pVDZ 3-21+G* 6311++G** 6311++G** 631+G* aug-cc-pVDZ
13
A
A
A
A
( C(1))
-49
54
93
20
7
89
-89
-59
-53
35
-32
27
109
9
9
12
-21
-29
-30
59
-273
-255
528
-2
5
-30
33
196
4
-30
24
221
4
-48
54
93
19
7
89
-89
-62
-51
-15
34
-32
28
106
8
8
13
-21
-29
-31
60
-254
-271
525
-3
6
-30
34
192
4
-30
25
212
3
iso
iso
iso
13
( C(2))
31
( P)
iso (1H(7))
τ (13C(1))
τ (13C(2))
τ (31P)
6
8
6
8
6
9
6
9
11
-15
-20
-3
11
-18
-25
-27
52
-209
-194
403
-3
7
-14
-21
-22
43
-249
-229
477
-1
4
-14
-21
-23
44
-242
-223
465
-1
5
-18
-25
-27
52
-195
-208
403
-3
7
-19
-14
-21
-23
44
-227
-248
475
-2
5
-14
-22
-24
45
-222
-242
464
-2
5
0
0
3
-3
3
-249
-210
459
-4
-210
-248
458
-5
τ (1H(7))
17
17
-4
-13
-3
-3
-3
-4
-12
-3
-3
-3
This implies a broadening of the central pattern which is
consistent with the absence of 13C hyperfine splitting in the
perpendicular region of the experimental spectrum and justifies
the additional proton coupling used for the simulation of the
frozen solution spectra.
Acknowledgment. We thank the Swiss National Science
Foundation for financial support. The support of French CNRS-
IDRIS is also greatly acknowledged.
References and Notes
These results unambiguously show that the neutral radical 4
obtained by capture of a proton by the primary negative ion is
a much better candidate for the EPR spectra.
(1) (a) Multiple Bonds and Low Coordination in Phosphorus Chemistry;
Regitz, M., Scherer, O., Eds.; Thieme: New York, 1990. (b) Canac, Y.;
Baceiredo, A.; Schoeller, W. W.; Gigmes, D.; Bertrand, G. J. Am. Chem.
Soc. 1997, 119, 7579.
(2) (a) Nixon, J. F. Chem. ReV. 1988, 88, 1327. (b) Bourissou, D.;
Canac, Y.; Collado, M. I.; Baceiredo, A.; Bertrand, G. J. Am. Chem. Soc.
1997, 119, 9923.
(3) (a) Geoffroy, M.; Jouaiti, A.; Terron, G.; Cattani-Lorente, M.;
Ellinger, Y. J. Phys. Chem. 1992, 96, 8241. (b) Al Badri, A.; Chentit, M.;
Geoffroy, M.; Jouaiti, A. J. Chem. Soc., Faraday Trans. 1997, 93, 3631.
(4) Sidorenkova, H.; Chentit, M.; Jouaiti, A.; Terron, G.; Geoffroy,
M.; Ellinger, Y. J. Chem. Soc., Perkin 2 1998, 71.
(5) Appel, R.; Fo¨lling, P.; Josten, B.; Siray, M.; Winkhaus, V.; Knoch,
F. Angew. Chem., Int. Ed. Engl. 1984, 23, 619.
(6) Goede, S. J.; Bickelhaupt, F. Chem. Ber. 1991, 124, 2677.
(7) Yoshifuji, M.; Sasaki, S.; Inamoto, N. Tetrahedron. Lett. 1989, 30,
839.
Conclusion
Although the EPR parameters measured with an electro-
chemically reduced solution of 1a are not, a priori, inconsistent
with the formation of the radical anion, the quantum-mechanics
calculations clearly show that the hyperfine interactions for 3
are quite different from those calculated for 2- and that the
experimental couplings are considerably more in accordance
with the neutral allylic structure 4 than with that of the
negatively charged species 2-. In fact, this identification agrees
(8) WINEPR SimFonia (version 1.25); Bruker Analytische Messtecknik
Gmbh, 1996.
(9) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;
Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T.; Petersson, G.
A.; Montgomery, J. A.; Raghavachari, K.; Al-Laham, M. A.; Zakrzewski,
V. G.; Ortiz, J. V.; Foresman, J. B.; Ciolowski, J.; Stefanov, B. B.;
Nanayakkara, A.; Challacombe, M.; Peng, C. Y.; Ayala, P. Y.; Chen, W.;
Wong, M. W.; Andres, J. L.; Replogle, E. S.; Gomperts, R.; Martin, R. L.;
Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart, J. P.; Head-
Gordon, M.; Gonzalez, C.; Pople, J. Gaussian 94, Revision B.1.; Gaussian
Inc.: Pittsburgh, PA, 1995.
(10) Dupuis, M.; Johnston, F.; Marquez, A. HONDO 8.5 from CHEM-
Station; IBM Corp.: Kingston, NY, 1994.
(11) McMurchie, L.; Elbert, S.; Langhoff, S.; Davidson, E. R. (modified
by Feller, D.; Cave, R. J.; Rawlings, D.; Frey, R.; Daasch, R.; Nitzche, L.;
Phillips, P.; Iberle, K.; Jackels, C.; Davidson, E. R.) MELDF Suite of
Programs; 1990/91.
(12) Morton, J. R.; Preston, K. F. J. Magn. Reson. 1978, 30, 577.
(13) Chentit, M.; Sidorenkova, H.; Jouaiti, A.; Terron, G.; Geoffroy,
M.; Ellinger, Y. J. Chem. Soc., Perkin Trans. 2 1997, 921.
(14) Feller, D.; Huyser, E. S.; Borden, W. T.; Davidson, E. R. J. Am.
Chem. Soc. 1983, 105, 1459.
(15) Nguyen, M. T.; Hegarty, A. F. J. Chem. Soc., Perkin Trans. 2 1985,
1999.
(16) Bauschlicher, C. W.; Langhoff, S. R. Spectrochim. Acta A 1997,
53, 1225.
(17) (a) Dunning, T. J. Chem. Phys. 1989, 90, 1007. (b) Kendall, R.
A.; Dunning, T. H.; Harrison, R. J. J. Chem. Phys. 1992, 96, 6796.
(18) Parisel, O.; Giessner-Prettre, C.; Ellinger, Y. Chem. Phys. Lett. 1996,
250, 178.
with two experimental results: (1) the cyclic voltammogram
which reveals an irreversible reduction consistent with a rapid
transformation of the anion. (2) The simulations of the frozen
solution spectra which are consistent with an additional small
coupling with a proton bound to the central carbon atom.
It is worthwhile commenting on the theoretical predictions
for the two observables associated to the magnetic hyperfine
interaction. Whereas single-determinant calculations may lead
to acceptable values of the dipolar coupling tensors, we have
seen here that the Fermi contact interaction varies considerably
on passing from simple SCF, to Moller Plesset, and to CI
calculations. Anisotropic coupling constants, which are better
approximated by a single-determinant representation, are there-
fore considerably more pertinent than the Fermi contact for the
identification of paramagnetic reaction intermediates at lower
levels of theory. DFT calculations appear as an alternative to
the more expensive CI treatments, especially for dipolar
interactions.