Cycloaddition and electron transfer: On a synthetically useful aspect of pnictogen(I) reactivity

Article abstract of DOI:10.1016/j.ica.2006.07.047

The behavior of several bi- or tri-dentate sp²-nitrogen-containing ligands with phosphorus(I) and arsenic(I) centers is investigated using computational and, in some cases, experimental methods. Whereas the chelation of putative Pn^I cations (Pn = P, As) by 1,4-diazabutadienes (DAB) generate stable, isolable phosphenium or arsenium cations that contain Pn^III centers, the corresponding reactions of Pn^I cations with β-diketiminate (NACNAC) ligands do not yield similar planar heterocycles. The behaviors of the ligands 2,6-diiminopyridine and 2,2′-bipyridine fall between the extremes delimited by the DAB and NACNAC ligands. The results of computational investigations of potential products provide a framework to rationalize the chemistry observed.

Full text of DOI:10.1016/j.ica.2006.07.047

Inorganica Chimica Acta 360 (2007) 329–344
www.elsevier.com/locate/ica
Cycloaddition and electron transfer: On a synthetically useful
aspect of pnictogen(I) reactivity
*
Bobby D. Ellis, Charles L.B. Macdonald
Department of Chemistry and Biochemistry, University of Windsor, 401 Sunset Avenue, Windsor, Ont., Canada N9B 3P4
Received 5 July 2006; accepted 13 July 2006
Available online 28 July 2006
Inorganic Chemistry – The Next Generation.
Abstract
The behavior of several bi- or tri-dentate sp2-nitrogen-containing ligands with phosphorus(I) and arsenic(I) centers is investigated
using computational and, in some cases, experimental methods. Whereas the chelation of putative PnI cations (Pn = P, As) by 1,4-diaza-
butadienes (DAB) generate stable, isolable phosphenium or arsenium cations that contain PnIII centers, the corresponding reactions of
PnI cations with b-diketiminate (NACNAC) ligands do not yield similar planar heterocycles. The behaviors of the ligands 2,6-diimino-
pyridine and 2,20-bipyridine fall between the extremes delimited by the DAB and NACNAC ligands. The results of computational inves-
tigations of potential products provide a framework to rationalize the chemistry observed.
Ó 2006 Elsevier B.V. All rights reserved.
Keywords: Low oxidation state; Group 15; Diimines; Diketiminates; Nitrogen ligands; Computational chemistry; Phosphenium; Arsenium; Charge
transfer; Redox
1. Introduction
demonstrated the viability of stable, di-coordinate cationic
PI compounds [6]. In particular, Schmidpeter’s group
showed that salts containing ‘‘triphosphenium’’ cations,
2, which may be considered to consist of phosphine-stabi-
lized PI ions, can be obtained by the in situ reduction of
PCl3 in the presence of phosphine ligands, as depicted in
Scheme 1. Some corresponding arsenic analogues have
likewise been obtained using similar preparative
approaches [7–11].
In the last five years, the research groups of Dillon and
co-workers [12], Macdonald and co-workers [13–15] and
Woollins and co-workers [16] have shown that it is possible
to generate and isolate halide salts of triphosphenium ions
and their arsenic analogues in the absence of additional
reducing agents. When appropriate conditions are
employed in reactions using chelating diphosphine ligands
and PI3, it has proven possible to obtain analytically pure
iodide salts of the triphosphenium cations, as illustrated
in Scheme 2. In a similar vein, it has been demonstrated
that related PI halide salts can be generated by the treat-
The oxidation state of an element in a molecule can have
a significant influence on the structural features, bonding
and chemistry of that molecule. In recent years, many main
group chemists have been interested in the chemistry of
compounds containing main group elements in unusually
low oxidation states because such electron-rich species
often exhibit behavior that is manifestly different than com-
pounds containing the elements in more conventional oxi-
dation states [1]. In regard to the heavier group 15 elements
(pnictogens; Pn = P, As, Sb, Bi) we have been particularly
interested in compounds containing the elements in the +1
oxidation state [2]. While phosphinidenes and their heavier
PnI analogues, 1, are generally highly reactive, transient
species [3–5], in the 1980s, the seminal work of Schmidpeter
*
Corresponding author. Tel.: +1 519 253 3000x3991; fax: +1 519 973
7098.
E-mail address: cmacd@uwindsor.ca (C.L.B. Macdonald).
0020-1693/$ - see front matter Ó 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.ica.2006.07.047

330
B.D. Ellis, C.L.B. Macdonald / Inorganica Chimica Acta 360 (2007) 329–344
Scheme 1.
ment of PX3 (X = Cl, Br, I) with certain N-heterocyclic
carbenes, which exhibit comparable reactivity to phosphine
ligands [17].
Scheme 3.
Recently, Cowley and co-workers have investigated the
use of diimines as ligands for putative PnI cations gener-
ated by both of the synthetic protocols outlined above.
In the case of a 1,2-bis(arylimino)acenapthene (Ar-BIAN),
the reaction of PnCl3 (Pn = P, As) and SnCl2 in the pres-
ence of the BIAN ligand produced salts of the form [(Ar-
BIAN)Pn][SnCl5 Æ THF] and the reactions of Ar-BIAN
with PnI3 generated the corresponding triiodide salts,
[(Ar-BIAN)Pn][I3]; these reactions are illustrated in Scheme
3 [18]. Regardless of the identity of the anion, the metrical
parameters for the cations in these salts indicate that these
cations are best described as phosphenium [19] or arsenium
cations that contain PnIII centers generated with the con-
comitant 2-electron reduction of the Ar-BIAN ligand. In
this light, such products are analogous to those obtained
when phosphinidene transfer agents are treated with a-dii-
mines and related trapping agents [20]. Similar results are
obtained from the analogous reactions of the simpler dii-
mine 1,4-dimesityl-1,4-diazabutadiene (Mes-DAB) with
AsI3 or the AsCl3/SnCl2 mixture [21] and it is noteworthy
that 3 was the perhaps surprising result from the reaction
Scheme 4.
metrical parameters in this cation are much more consis-
tent with the presence of an AsI cation chelated by a
non-reduced Ar-DIMPY ligand. The overall T-shaped
arrangement about the As center is reminiscent of the so-
called ‘‘10-As-3’’ center in Arduengo’s unique family of
group 15 compounds (4, Pn = P, As, Sb; R = tBu or ada-
mantyl) containing the diketoamido ligand [23,24].
t
of R-DAB (R = Mes or Bu) with PCl3 and NEt3 [22].
In contrast, the reaction of AsI3 or AsCl3/SnCl2 with an
a,a0-diiminopyridine (Ar-DIMPY, Scheme 4) ligand gener-
ates salts containing the cation [(Ar-DIMPY)As]+; the
Given the importance of b-diketiminate (NACNAC)
ligands to transition metal [25] and main group chemistry
[26–28], several groups have attempted to coordinate such
ligands to phosphorus centers. In stark contrast to the
results outlined above for a-diimines, the reactions of
metallated NACNAC reagents with halogenated PIII spe-
cies results in the formation of a P–C bond to the unique
Scheme 2.

B.D. Ellis, C.L.B. Macdonald / Inorganica Chimica Acta 360 (2007) 329–344
331
Ca atom instead of chelation by the nitrogen atoms [29–31].
The adoption of the monodentate binding mode and avoid-
ance of P–N bond formation is also observed when reac-
tion conditions are used that are analogous to those that
have been employed successfully for the heterocycle synthe-
ses described above.
J = 6.9, 24H), 2.31 (s, 6H), 2.67 (septet, J = 6.9, 2H),
7.49 (d, J = 7.8, 4H), 7.69 (t, J = 7.8, 2H). m.p.: 215–
216 °C. Anal. Calc. for C28H40Cl5N2PSn (731.578): C,
45.97; H, 5.51. Found: C, 45.78; H, 5.45%.
2.3. Preparation of [(ArM-DAB)P][I3], 5[I3]
In light of the range of products observed from seem-
ingly similar reactions, herein we present computational
and some experimental results that elucidate important
aspects of the interactions between several chelates that
contain sp2-hybridized nitrogen centers and PI and AsI cen-
ters. It is anticipated that the insight afforded by such
investigations may facilitate the design of heterocycle form-
ing reactions involving low oxidation state reagents.
Solid ArM-DAB (0.352 g; 0.870 mmol) was added to a
solution of PI3 (0.358 g; 0.870 mmol) in CH2Cl2 (15 mL),
which immediately changed from reddish-orange to dark
red. The solution was left to stir for 16 h and then the vol-
atile components were removed in vacuo. Dissolution of
the red solid in a minimal amount of MeCN followed by
slow evaporation produced red crystalline material. Yield:
84% (0.597 g; 0.731 mmol). 31P{1H} NMR: 201.1 (s).
13C{1H} NMR: 13.9(s), 23.3 (s), 26.1(s), 29.5 (s), 125.0
(s), 125.8 (s), 132.9 (s), 144.3 (s), 145.5 (s). 1H NMR:
1.22 (d, J = 7.0, 24H), 2.31 (s, 6H), 2.67 (septet, J = 7.0,
2H), 7.48 (d, J = 7.9, 4H), 7.67 (t, J = 7.9, 2H). m.p.:
290–291 °C. Anal. Calc. for C28H40I3N2P (816.318): C,
41.20; H, 4.94. Found: C, 41.47; H, 4.93%.
2. Experimental
2.1. General methods
All work was carried out using standard inert-atmo-
sphere techniques. Phosphorus(III) chloride, phospho-
rus(III) iodide, and tin(II) chloride were obtained from
Strem; all solvents were obtained from Aldrich: all
reagents were used without further purification. The prep-
arations of 1,4-diaryl-2,3-dimethyl-1,4-diazabutadiene
(ArM-DAB) [32] and Ar-NACNAC [33] (Ar = 2,6-diiso-
propylphenyl) are described in the literature. Solvents were
dried on a series of Grubbs’-type columns [34] and were
degassed prior to use. NMR spectra were recorded at
room temperature in CD2Cl2 solutions on a Bruker
Advance 300 MHz spectrometer. Chemical shifts are
reported in ppm, relative to external standards (SiMe4
2.4. X-ray crystallography
Each crystal was covered in Nujol and placed rapidly
into the cold N2 stream of the Kryo-Flex low temperature
device. The data were collected using the SMART [35] soft-
ware on a Bruker APEX CCD diffractometer using a
graphite monochromator with Mo Ka radiation
˚
(k = 0.71073 A). A hemisphere of data were collected using
a counting time of 30 s per frame. The data were collected
at ꢀ100 °C. Details of crystal data, data collection and
structure refinement are listed in Table 1. Data reductions
were performed using the SAINT [36] software and the data
were corrected for absorption using SADABS [37]. The struc-
tures were solved by direct methods using SIR-97 [38] and
refined by full-matrix least-squares on F2 with anisotropic
displacement parameters for the non-H atoms using
SHELXL-97 [39] and the WinGX [40] software package and
thermal ellipsoid plots were produced using SHELXTL [41].
1
for H and 13C, 85% aq. H3PO4 for 31P). Coupling con-
stants are given in Hertz. Melting points were obtained
on samples sealed in glass capillaries under dry nitrogen
using an ElectrothermalÒ Melting Point Apparatus. Ele-
mental analysis was performed using a Perkin–Elmer
2400 C, H, N analyzer in the Centre for Catalysis and
Materials Research, Department of Chemistry and Bio-
chemistry, University of Windsor. It should also be noted
that the reactions appear to be quantitative according to
all spectroscopic data recorded.
2.5. Theoretical methods
2.2. Preparation of [(ArM-DAB)P][SnCl5], 5[SnCl5]
Calculations were preformed with the GAUSSIAN-98 suite
of programs [42]. Geometry optimizations have been calcu-
lated using density functional theory (DFT), specifically
implementing the B3PW91 method [containing Becke’s
three-parameter hybrid functional for exchange (B3,
including ca. 20% Hartree–Fock exchange) [43] combined
with the generalized gradient approximation for correla-
tion of Perdew and Wang (PW91) [44]] in conjunction with
the 6-31+G(d) basis set. The geometries were restricted to
the highest appropriate symmetry and the nature of each
stationary point was examined by vibrational frequency
analysis; in certain instances, the data obtained for the
C2v-symmetry structures, which are identified as transition
states, are reported because they provide the most appro-
Solid ArM-DAB (0.433 g; 1.070 mmol) was added to a
SnCl2 (0.203 g; 1.070 mmol) slurry in CH2Cl2 (20 mL), to
afford a yellow slurry. A solution of PCl3 (0.147 g;
1.070 mmol) in CH2Cl2 (10 mL) was slowly added and
the mixture became red. The mixture was left to stir for
16 h and then it was filtered through Celite and the volatile
components were removed in vacuo. Dissolution of the red
solid in a minimal amount of CH2Cl2 followed by slow
evaporation produced pink crystalline material. Yield:
88% (0.692 g; 0.946 mmol). 31P{1H} NMR: 200.7 (s).
13C{1H} NMR: 13.5 (s), 23.2 (s), 26.1(s), 29.4(s), 124.9(s),
125.8(s), 132.9(s), 144.3(s), 145.5(s). 1H NMR: 1.21 (d,

332
B.D. Ellis, C.L.B. Macdonald / Inorganica Chimica Acta 360 (2007) 329–344
described as containing PnIII centers, the analogous reac-
tions with the 2,6-diimino substituted pyridine ligand
appears to contain arsenic(I) centers. Given our interest
in the chemistry of and the factors affecting the stabiliza-
tion of low oxidation state main group fragments, these
results prompted us to elucidate the nature of the interac-
tion between the diimine ligands and the PnI centers and
to assess the viability of some other commonly used ligands
bearing imine-like moieties that contain sp2-hybridized
nitrogen centers.
Table 1
Summary of X-ray crystallographic data for compounds 5[SnCl5] and 5[I3]
Compound
5[SnCl5]
28H40Cl5N2PSn
731.53
173(2)
5[I3]
Empirical formula
Formula weight
Temperature (K)
Wavelength (A)
Habit, color
Crystal system
Space group
Unit cell dimensions
C
C28H40I3N2P
816.29
173(2)
˚
0.71073
0.71073
prism, pink
orthorhombic
Pnma
prism, red
monoclinic
C2/m
˚
a (A)
20.805(3)
15.709(2)
10.7872(14)
90
90
90
3525.6(8)
4
1.378
1.168
1488
2.13–27.50
ꢀ26 6 h 6 26,
ꢀ20 6 k 6 19,
ꢀ14 6 l 6 13
37062
19.044(2)
16.745(2)
11.6861(14)
90
120.005(1)
90
3227.2(7)
4
1.680
In light of the apparently odd results observed when the
b-dimine NACNAC ligands have been ligated to phospho-
rus centers [29–31], we wished to ascertain whether such
diimines were suitable ligands for PI centers, however,
given the reported reactivity with halophosphines, a minor
modification of the SnCl2 reduction protocol was man-
dated [47]. In this vein, a solution of [Li][Ar-NACNAC]
was added to an equimolar amount of SnCl2 in THF to
produce a cloudy, colorless mixture. After stirring for sev-
eral minutes, a solution of an equimolar amount of PCl3 in
THF was added to the reaction mixture. The mixture rap-
idly produced a pale yellow precipitate and a yellow solu-
tion. The supernatant contained no signals in the 31P
NMR spectrum. Similarly, the reaction of [Li][Ar-NAC-
NAC] with PI3 in THF resulted in the immediate formation
of a reddish solution and the formation of a copious
amount of orange precipitate. The solution contains no sig-
nals in the 31P NMR spectrum consistent with a neutral
compound of the form [(Ar-NACNAC)P]. Likewise, the
reaction of an equimolar mixture of [Ar-NACNAC]H
and SnCl2 in dichloromethane with a solution of PCl3 in
the same solvent produced a multitude of products. In each
case, the reactions appeared to produce numerous products
that have as yet eluded separation and identification, how-
ever, there is no evidence for the formation of a neutral
compound of the form [(Ar-NACNAC)P]; these reactions
are certainly not clean or quantitative.
˚
b (A)
˚
c (A)
a (°)
b (°)
c (°)
3
˚
V (A )
Z
Dcalc (g cmꢀ3
)
Absorption coefficient (mmꢀ1
)
2.973
1584
F(000)
h Range for data collection (°)
Limiting indices
1.73–27.50
ꢀ24 6 h 6 24,
ꢀ21 6 k 6 21,
ꢀ15 6 l 6 15
18204
Reflections collected
Independent reflections (Rint
Absorption correction
Refinement method
)
4178 (0.0492)
SADABS
3798 (0.0235)
SADABS
full matrix least
squares on F2
4178/0/185
1.202
R1 = 0.0421,
wR2 = 0.0884
R1 = 0.0712,
wR2 = 0.1061
1.277 and ꢀ0.487
full matrix least
squares on F2
3798/0/167
1.032
R1 = 0.0418,
wR2 = 0.0944
R1 = 0.0539,
wR2 = 0.1027
1.466 and ꢀ1.532
Data/restraints/parameters
Goodness-of-fit on F2
Final R indicesa [I > 2r(I)]
R indices (all data)
Largest difference maꢀp3
˚
peak and hole (e A
)
P
P
a
R1(F) = (jFoj ꢀ jFcj)/ jFoj for reflections with Fo > 4(r(Fo)).
P
P
2
2 1=2
wR2ðF 2Þ ¼ f wðjF 2oj ꢀ F c2jÞ = wðjF 2ojÞ g where w is the weight given
each reflection.
To confirm that our modified synthetic approach was
applicable for the preparation of PI centers, similar reac-
tions were undertaken with the a-diimine 1,4-diaryl-2,
3-dimethyl-1,4-diazabutadiene (ArM-DAB). Thus an equi-
molar mixture of ArM-DAB and SnCl2 in dichlorometh-
ane was stirred for several minutes and then a solution
containing an equimolar amount of PCl3 was added to pro-
duce a red solution. Multinuclear 1H and 13C NMR exper-
iments on the reaction mixture indicated the presence of a
singular, intact ArM-DAB ligand and the single peak at
200.1 ppm in the 31P NMR spectrum suggested the quanti-
tative formation of the phosphenium cation [(ArM-
DAB)P]+ (5). A crystalline solid was obtained by the slow
concentration of a dichloromethane solution of the resul-
tant solid and microanalysis was consistent with a salt of
the form 5[SnCl5].
priate models for structural and electronic structure com-
parisons. The electronic energies of the molecules have
been corrected by the unscaled, zero-point vibrational
energy (ZPVE). Single-point energies have been calculated
at the B3PW91/6-311+G(3df,2p)//B3PW91/6-31G(d) level
of theory. Population analyses were conducted using the
natural bond orbital (NBO) [45] method implemented in
GAUSSIAN-98. Drawings of the optimized structures of the
model compounds were made using SHELXTL and depictions
of the molecular orbitals were made using MOLDEN [46].
3. Results and discussion
3.1. Experimental investigations
Cowley recently demonstrated the differing behavior of
some diimine-based ligands with PI and AsI centers. In par-
ticular, whereas the a-diimines in Ar-BIAN or Mes-DAB
formed phosphenium or arsenium cations that are best
In a similar manner, the reaction of equimolar quantities
of ArM-DAB and PI3 in dichloromethane rapidly gener-
ates a red solution. Again, multinuclear NMR experiments
of the reaction mixture attest to the presence of a singular,

B.D. Ellis, C.L.B. Macdonald / Inorganica Chimica Acta 360 (2007) 329–344
333
intact ArM-DAB ligand and the 31P NMR spectrum con-
tains only a single peak at 201.1 ppm. The slow concentra-
tion of an acetonitrile solution of the solid provided
crystalline material having an elemental composition con-
sistent with the formula 5[I3].
In the case of each salt, the material obtained after
recrystallization was suitable for analysis by single-crystal
X-ray diffraction experiments. A summary of the experi-
mental crystallographic details are compiled in Table 1;
depictions of the molecular structures of both salts are pre-
sented in Figs. 1 and 2; and selected metrical parameters
for the salts are listed in the captions of each figure. The
pentachlorostannate salt 5[SnCl5] crystallizes in the space
group Pnma with one half of a cation and an anion in
the asymmetric unit. The phosphorus atom sits on a crys-
tallographic mirror plane that relates the two halves of
the cation and the tin atom and three of the chlorine atoms
are likewise situated on the mirror plane relating the two
parts of the anion. The triiodide salt 5[I3] crystallizes in
the spacegroup C2/m with one half of a cation and two
quarters of anions in each asymmetric unit. Again, the
phosphorus atom resides on a crystallographic mirror
plane that relates the two halves of the cation and all of
the iodine atoms are located on special positions in the unit
cell.
Fig. 1. Thermal ellipsoid plot (30% probability surfaces) of [(ArM-
DAB)P][SnCl5]. Hydrogen atoms are omitted for clarity. Selected
distances (A) and angles (°): P(1)–N(2) 1.667(3), N(2)–C(3) 1.377(4),
C(3)–C(3A) 1.373(6), N(2)–C(5) 1.449(4), Sn(1)–Cl(1) 2.3782(16), Sn(1)–
Cl(3) 2.4093(14), Sn(1)–Cl(2) 2.3411(9), Sn(1)–Cl(4) 2.3278(16), N(2)–
P(1)–N(2A) 88.88(17), C(3)–N(2)–P(1) 115.1(2), C(3A)–C(3)–N(2)
110.43(16), Cl(2)–Sn(1)–Cl(1) 90.05(4), Cl(4)–Sn(1)–Cl(3) 91.31(6),
Cl(1)–Sn(1)–Cl(3) 176.36(7), Cl(2)–Sn(1)–Cl(3) 88.25(3), Cl(4)–Sn(1)–
Cl(3) 91.31(6), Cl(4)–Sn(1)–Cl(2) 117.92(3), Cl(2)–Sn(1)–Cl(2A) 124.11(5).
˚
While the metrical parameters of each of the anions in
these salts compare well with the average values obtained
from the Cambridge Structural Database (CSD) [48] and
do not warrant further discussion, the identities of the
anions unambiguously demonstrate that a redox reaction
has taken place: the SnII in the starting material has clearly
been oxidized to SnIV in the trigonal bipyramidal [SnCl5]ꢀ
anion and the production of triiodide anions is consistent
with the presence of I2 and Iꢀ in the other reaction mixture.
It should be noted that there are no unusually short dis-
tances between the cations and the anions in either salt.
For the purposes of this work, the metrical parameters
of the cations in each salt are important for comparison
to the computed structures detailed below and can aid in
the assessment of the nature of the interaction between
the diimine ligand and the pnictogen center. Regardless
of the identity of the anion, the metrical parameters of
the cation are indistinguishable. For example the P–N dis-
˚
tance in 5[SnCl5] is 1.667(3) A and the distance in the trii-
˚
odide salt is 1.668(3) A; similarly, the C–N and C–C
˚
distances within the heterocycle are 1.377(4) A and
˚
˚
˚
1.373(6) A, respectively, in 5[SnCl5] versus 1.375(4) A and
1.376(7) A for 5[I3]. The corresponding angles within the
heterocyclic cations in each salt are also indistinguishable
from each other: the N–P–N angles are 88.88(17)° and
88.90(19)° in 5[SnCl5] and 5[I3], respectively; the P–N–C
angles are 115.1(2)° and 115.1(2)°, respectively; and the
N–C–C angles are 110.43(16)° and 110.43(18)°, respec-
tively. Overall, the metrical parameters of the cations in
these salts compare excellently with the average values
obtained from the CSD for the other unsaturated five-
membered ring phosphenium cations that have been
Fig. 2. Thermal ellipsoid plot (30% probability surface) of [(ArM-
DAB)P][I3]. Hydrogen atoms are omitted for clarity. Selected distances
˚
(A) and angles (°): P(1)–N(2) 1.668(3), C(3)–N(2) 1.375(4), C(3)–C(3A)
1.376(7), N(2)–C(10) 1.456(4), I(2)–I(1) 2.9207(7), I(3)–I(4) 2.9435(9),
N(2)–P(1)–N(2A) 88.90(19), C(3)–N(2)–P(1) 115.1(2), N(2)–C(3)–C(3A)
110.43(18), I(1)–I(2)–I(1A) 180.000(19), I(3)–I(4)–I(3A) 180.0.

334
B.D. Ellis, C.L.B. Macdonald / Inorganica Chimica Acta 360 (2007) 329–344
reported previously [22,49–51], thus the descriptions of
these cations as containing phosphorus(III) is certainly
justified.
optimized structures of [(NACNAC)P], [(DIMPY)P]+ and
[(NACNAC)As] are not minima on the potential energy
hypersurface when they are constrained to C2v-symmetry.
True minima are observed for these model compounds
when they are optimized in Cs-symmetry; the structures
thus obtained are depicted in Fig. 5 and will be discussed
below. In regard to the C2v structures, one important fea-
ture distinguishes the models containing DAB, DIMPY
and BIPY ligands from the models including the NAC-
NAC ligand. Whereas the relatively short Pn–N distances
The nature of the anions in the phosphenium salts
described above and the other salts described in the intro-
duction indicates that redox processes occur in the forma-
tion of the heterocycles that are completely consistent with
the processes postulated to occur in the formation of phos-
phine- or carbene-stabilized PnI salts [8,12,13,15,17]. In the
case of a-diimines, a further intermolecular redox process
involving oxidation of PnI to PnIII and reduction of the
ligand to its doubly-reduced form has been suggested to
generate the observed pnictogenium cations [18,21]. In
the case of the diimino-pyridine ligand, while the first redox
reaction generates an AsI center and the corresponding
anions, the intermolecular redox process does not appear
to occur [21]. Finally, the b-diketiminate ligand does not
appear to chelate to putative PnI centers under conditions
analogous to those employed for the other imine-contain-
ing ligands [47].
˚
˚
(P: 1.685–1.757 A; As: 1.798–1.850 A) in the cationic a-dii-
mine systems that suggest the presence partial multiple-
bonding, the Pn–N distances in the optimized neutral
˚
[(NACNAC)Pn] models are extremely long (P: 1.854 A;
˚
As: 1.955 A) and are beyond the values typically observed
for single bonds. The contrast between the P–N bond dis-
tances suggests significant differences in the nature of the
bonding between the ligands and the group 15 element.
The contrasting behavior is highlighted by the Cs-symme-
try optimized structures of [(NACNAC)Pn] which, while
retaining very long Pn–N bonds, adopt boat-like confor-
mations with a clearly pyramidal carbon environment.
Whereas the difference between the planar and non-pla-
nar [(NACNAC)Pn] structures are drastic, the planar
Cs-symmetry structure of the [(DIMPY)P]+ model is dis-
torted only slightly away from C2v-symmetry, as illustrated
in Fig. 5, and the imaginary frequency corresponding to
this deformation is only ꢀ75 cmꢀ1. The metrical parame-
ters of this cation suggest that it is probably best under-
stood as being composed of an intermolecularly imine-
coordinated phosphenium cation, as particularly evidenced
by the dramatically different N–Cimine bond distances of
3.2. Computational investigations
In an attempt to understand the clearly different chelating
behavior of the various diimine ligands and PnI centers, we
performed a series of DFT calculations and NBO analyses
on the model compounds of the Pn-containing compounds
chelated by the ligands DAB, NACNAC, DIMPY and
BIPY. Models of the free ligands and their doubly-reduced
forms are also presented for comparative purposes. Addi-
tional calculations were undertaken on the dications
[(NACNAC)Pn]2+ to assess the viability of b-diketiminate
chelate of PnIII cations. In each model, all substituents have
been replaced with hydrogen atoms and the highest-reason-
able point symmetry was enforced; important computa-
tional results are compiled in Table 2 and more complete
data are found in the Supporting information.
The C2v-symmetry optimized geometries of the phos-
phorus- and arsenic-containing model compounds are
depicted in Figs. 3 and 4, respectively, and the computed
values of important metrical parameters are provided.
The only phosphorus compound for which analogues have
been structurally characterized experimentally (with sub-
stituents other than hydrogen) is the cation [(DAB)P]+;
given the simplicity of the model used, the agreement
between the metrical parameters of the calculated and
experimental structures (vide supra) is excellent and attests
to the suitability of the computational method employed
herein. Furthermore, the calculated structure of [(DAB)P]+
is consistent with those computed previously using different
theoretical methods [52]. Similarly, there is good agreement
between the computed metrical parameters of the models
[(DAB)As]+ and [(DIMPY)As]+ and those reported for
their corresponding crystallographically characterized ana-
logues [21,51].
˚
˚
1.320 A and 1.291 A and the corresponding Cimine–Ca dis-
˚
˚
tances of 1.403 A and 1.430 A, respectively.
To assess the structural changes that occur to the ligands
upon coordination to the pnictogen centers and to evaluate
the postulate that the ligands become doubly-reduced in
certain instances, we have determined the optimized geom-
etries of the free ligands (Fig. 6) and their doubly-reduced
forms (Fig. 7). In both figures, only the C2v-symmetry
structures are presented because they are the most appro-
priate models for comparison to the pnictogen-containing
heterocycles. It should be noted that for DAB and BIPY,
the C2v-symmetry structures are first-order transition states
and structures and metrical parameters for the true C2h-
symmetry minima are presented in the Supporting infor-
mation; the true minimum optimized structures of
[BIPY]2ꢀ and [DIMPY]2ꢀ were not determined as these
are not likely to exist as isolated dianions.
Because large changes in the angles within the ligands in
their uncomplexed and complexed forms are unavoidable,
the examination of such metrical parameters is not as help-
ful as is the comparison of bond distances. Several impor-
tant observations are gleaned from the comparison of the
bond distances of the ligands and those of their group 15
complexes. Perhaps the most noteworthy is that the
imine-like N–C distances in the complexes involving the
There are several noteworthy findings with regard to the
structures presented in Figs. 3 and 4. Most importantly, the

B.D. Ellis, C.L.B. Macdonald / Inorganica Chimica Acta 360 (2007) 329–344
335
Table 2
Selected computed energies for the model ligands and pnictogen complexes
Symmetrya Nimag
Corrected energyc, Etotal (a.u.) Relative energy HOMO (eV) LUMO (eV) HOMO–LUMO (eV)
b
Models
(kJ molꢀ1
)
Ligands
Oxidized
DAB
C2h (1)
C2h (2)
C2v
0
0
1
0
0
0
1
ꢀ188.0087479
ꢀ188.0053223
ꢀ187.997689
ꢀ226.7216236
ꢀ435.0009775
ꢀ495.1845832
ꢀ495.1735031
0
ꢀ7.82
ꢀ7.65
ꢀ7.70
0.02
ꢀ7.32
ꢀ6.74
ꢀ6.78
ꢀ2.18
ꢀ2.42
ꢀ2.35
3.80
ꢀ2.17
ꢀ1.79
ꢀ1.81
5.64
5.23
5.36
3.78
5.15
4.95
4.97
8.99
29.04
[NACNAC]ꢀ
DIMPY
BIPY
C2v
C2v
C2h
C2v
0
29.09
Reduced
[DAB]2ꢀ
C2v
C2v
C2v
C2v
0
0
2
3
ꢀ187.8128015
ꢀ226.2856646
ꢀ434.8818511
ꢀ495.0533583
7.09
8.34
5.48
5.16
7.45
8.77
5.90
5.74
0.36
0.43
0.42
0.58
[NACNAC]3ꢀ
[DIMPY]2ꢀ
[BIPY]2ꢀ
Complexes
P species
[(DAB)P]+
[(NACNAC)P]
C2v
C2v
Cs
0
1
0
1
0
0
ꢀ529.1543655
ꢀ568.0027854
ꢀ568.0124268
ꢀ776.1639187
ꢀ776.1638783
ꢀ836.3003994
ꢀ12.33
ꢀ3.50
ꢀ3.95
ꢀ10.28
ꢀ10.28
ꢀ9.65
ꢀ7.51
ꢀ1.96
ꢀ1.87
ꢀ6.93
ꢀ6.87
ꢀ6.30
4.82
1.53
2.08
3.35
3.41
3.36
25.31
0
[(DIMPY)P]+
[(BIPY)P]+
C2v
Cs
C2v
As species
[(DAB)As]+
[(NACNAC)As]
C2v
C2v
Cs
0
1
0
0
0
ꢀ2423.677708
ꢀ2462.543142
ꢀ2462.549112
ꢀ2670.6982
ꢀ11.90
ꢀ3.55
ꢀ3.91
ꢀ10.01
ꢀ9.43
ꢀ7.56
ꢀ1.87
ꢀ1.81
ꢀ6.77
ꢀ6.31
4.35
1.68
2.10
3.24
3.12
15.67
0
[(DIMPY)As]+
[(BIPY)As]+
C2v
C2v
ꢀ2730.827911
Dications
[(NACNAC)P]2+
C2v
0
0
ꢀ567.362344
ꢀ2461.895212
ꢀ19.12
ꢀ18.58
ꢀ14.08
ꢀ14.03
5.04
4.55
[(NACNAC)As]2+ C2v
a
Point group symmetry.
b
c
Number of imaginary vibrational frequencies.
Etotal = Ecalculated + ZPVE.
neutral ligands (DAB, DIMPY, and BIPY) are longer than
those observed for the free ligands. In stark contrast, the
N–C distances in the optimized [(NACNAC)Pn] complexes
(for both the C2v and Cs models) are almost identical to
that found in the free anion [NACNAC]ꢀ. Similarly, while
responding parameters in the [(BIPY)Pn]+ complexes. For
˚
example, while the imine N–Cimine distance of 1.303 A in
+
˚
[(DIMPY)As] is between the values of 1.273 A in DIMPY
and 1.331 A in [DIMPY]2ꢀ, the N–Cpyridyl distance of
1.385 A in [(DIMPY)As] is identical to that predicted
˚
+
˚
there are obvious and significant contractions of the Cimine
–
for the [DIMPY]2ꢀ model. Overall, the examination of
the metrical parameters in each of these models suggests
that the DAB ligand is the most completely reduced upon
complexation to a putative PnI center, followed by DIMPY
and then BIPY. In sharp contrast, the bond distances in the
[NACNAC]ꢀ anion appear to be largely unaffected by
complexation to the electron-rich group 15 center.
Ca bonds for each of the complexes of the neutral ligands,
there is only a slight contraction observed for the com-
plexes of [NACNAC]ꢀ. Thus, once again, the metrical
parameters highlights the considerably different behavior
of the b-diketiminate ligand.
In regard to the metrical parameters calculated for the
species containing DAB, DIMPY and BIPY ligands, it is
also noteworthy that the distances observed for the
[(DAB)Pn]+ models are more similar to those calculated
for the doubly-reduced [DAB]2ꢀ dianion than those for
the free ligand. For the DIMPY and BIPY complexes,
many of the metrical parameters appear to fall between
those found in the neutral and doubly-reduced ligands,
with the [(DIMPY)Pn]+ complexes having parameters clo-
ser to those of the doubly-reduced ligand than are the cor-
In order to quantify the nature of the interaction
between the various ligands and the pnictogen atoms, we
performed NBO population analyses of each of the model
compounds and pertinent results are compiled in Tables 3
and 4. With regard to the discussion of the metrical param-
eters described above, it is noteworthy that the NBO
Wiberg bond indices (WBI) mirror the trends suggested
by the analysis of the bond distances. For example,
whereas the Wiberg bond indices for the N–C bonds and

336
B.D. Ellis, C.L.B. Macdonald / Inorganica Chimica Acta 360 (2007) 329–344
Fig. 3. Optimized structures (C2v-symmetry) of the phosphorus-containing species [(DAB)P]+, [(NACNAC)P], [(DIMPY)P]+, and [(BIPY)P]+. Selected
˚
bond distances (A) and angles (°) are indicated.
Fig. 4. Optimized structures (C2v-symmetry) of the arsenic-containing species [(DAB)As]+, [(NACNAC)As], [(DIMPY)As]+, and [(BIPY)As]+. Selected
˚
bond distances (A) and angles (°) are indicated.

B.D. Ellis, C.L.B. Macdonald / Inorganica Chimica Acta 360 (2007) 329–344
337
Fig. 5. Optimized structures (Cs-symmetry) of the phosphorus- and arsenic-containing species [(DIMPY)P]+, [(NACNAC)P], [(NACNAC)As]. Selected
˚
bond distances (A) and angles (°) are indicated.
ꢀ
˚
Fig. 6. Optimized structures (C2v-symmetry) of the free ligands DAB, [NACNAC] , DIMPY, and BIPY. Selected bond distances (A) and angles (°) are
indicated.
the C–C bonds in [(DAB)Pn]+ are almost identical to those
in the [DAB]2ꢀ dianion, the corresponding values for the
neutral [(NACNAC)Pn] models are much closer to those
observed in the free [(NACNAC)]ꢀ anion. While the indi-
ces for the [(BIPY)Pn]+ models fall between those of the
neutral and doubly-reduced ligands, as anticipated on the
basis of the metrical parameters, it is somewhat surprising
that the indices for the N–C and C–C bonds in the [(DIM-
PY)Pn]+ are remarkably similar to those in the doubly-
reduced [DIMPY]2ꢀ dianion.
The Wiberg bond indices of the Pn–N bonds provide
some initial insight into the magnitude of the interactions
between the various ligand and the group 15 element. Most
notably, the WBIs for the Pn–N bonds are greatest for the
[(DAB)Pn]+ models, which exhibit the shortest bonds
between those elements. The next highest indices are
observed for the bonds between the pnictogen atoms and
the pyridyl nitrogen atoms in the DIMPY ligands; the val-
ues for these indices are very similar to those calculated for
the [(BIPY)Pn]+ models and suggest that such cations may

338
B.D. Ellis, C.L.B. Macdonald / Inorganica Chimica Acta 360 (2007) 329–344
Fig. 7. Optimized structures (C2v-symmetry) of the doubly-reduced ligands [DAB]2ꢀ, [NACNAC]3ꢀ, [DIMPY]2ꢀ, and [BIPY]2ꢀ. Selected bond distances
˚
(A) and angles (°) are indicated.
Table 3
Selected computed data for the model ligands
Model
Symmetrya q(N)b
N–Cimine WBIc Cimine–Ca WBIc p HOMOd (eV) p LUMOd (eV) p HOMO–LUMO (eV)
Oxidized
DAB
C2h (1)
C2h (2)
C2v
ꢀ0.62
ꢀ0.59
ꢀ0.58
ꢀ0.81
1.9256
1.9433
1.9506
1.6575
1.0487
1.0512
1.0276
1.315
1.0366
1.0402
1.0343
ꢀ8.19
ꢀ8.97
ꢀ8.60
0.02
ꢀ7.37
ꢀ6.74
ꢀ6.78
ꢀ2.18
ꢀ2.42
ꢀ2.35
4.47
ꢀ2.17
ꢀ1.79
ꢀ1.81
6.01
6.55
6.25
4.45
5.20
4.95
4.98
[NACNAC]ꢀ
DIMPY
BIPY
C2v
C2v
ꢀ0.58 (ꢀ0.36)e 1.9377
C2h
C2v
ꢀ0.46
ꢀ0.41
1.3802
1.3863
Reduced
[DAB]2ꢀ
C2v
ꢀ1.14
ꢀ1.02
1.2941
1.8007
1.5346
1.339
1.1819
1.4436
7.09
6.74
5.48
5.16
9.51
10.82
6.58
2.42
4.08
1.10
1.97
[NACNAC]3ꢀ C2v
[DIMPY]2ꢀ
[BIPY]2ꢀ
C2v
C2v
ꢀ0.85 (ꢀ0.69)e 1.5374
ꢀ0.54
1.0989
7.13
a
Point group symmetry.
b
c
NBO charge on the dicoordinate imine N atoms.
NBO Wiberg bond indices for the specified bonds.
Frontier orbitals involving only the p-system in the ligands.
NBO charge of the Npyridyl atom.
d
e
be reasonable synthetic targets. The Pn–N bond indices for
the [(NACNAC)Pn] models are considerably smaller than
those of the cationic a-diimine models and suggest that
the long P–N bonds may be considerably weaker than
those in the other models.
Two other quantities obtained from the NBO analyses
may provide insight into the nature of the ligand–Pn inter-
action and are worthy of discussion. Firstly, the almost
negligible positive charges (0.24–0.3) on the Pn atoms in
the [(NACNAC)Pn] models (Table 4) demonstrate that
the pnictogen centers in these compounds are considerably
more electron-rich than are the Pn atoms in any of the
other model compounds (q(Pn) > 0.87 in each case). Such
an observation is consistent with the trends observed for
the metrical parameters which suggest that the NACNAC
ligand does not become reduced significantly.
More importantly in the context of intermolecular
ligand reduction, the population of one of the valence p
orbitals on the Pn center can be used to gauge (admittedly
in a rough manner) of the amount of electron density trans-
ferred from the pnictogen to the ligand. Using the assump-
tion for such an analysis that a singlet PnI cation suitable
for ligation should have a population of 2.0 electrons in
two valence orbitals and that the interaction of the ligands
will generate the r-framework of the molecule (with one of
the filled valence orbitals becoming a ‘‘lone pair’’ of elec-
trons), the most obvious transfer of electron density should
be manifested in the population of the valence p orbital of

B.D. Ellis, C.L.B. Macdonald / Inorganica Chimica Acta 360 (2007) 329–344
339
Table 4
Selected electron population analysis data for the model pnictogen complexes
Model
Symmetrya
Occupancy of Pn p(p)b
q(Pn)c
q(N)c
Pn–N WBId
N–Cimine WBId
Cimine–Ca WBId
P species
[(DAB)P]+
[(NACNAC)P]
C2v
C2v
Cs
0.893
1.905
1.408
1.046
1.18
0.24
0.27
0.98
ꢀ0.86
ꢀ0.79
ꢀ0.81
ꢀ0.76
ꢀ0.62e
ꢀ0.79
ꢀ0.72
ꢀ0.62e
ꢀ0.59
1.0496
0.6690
0.6900
0.6129
0.8738e
0.7498
0.4560
0.8786e
0.8559
1.2281
1.4854
1.4579
1.5475
1.5450
1.3278
1.3489
1.2318
[(DIMPY)P]+
C2v
Cs
1.040
1.278
0.99
0.87
1.4377
1.6682
1.2945
1.1714
[(BIPY)P]+
C2v
1.1486
1.2079
As species
[(DAB)As]+
[(NACNAC)As]
C2v
C2v
Cs
0.916
1.924
1.473
1.076
1.22
0.28
0.30
1.04
ꢀ0.87
ꢀ0.80
ꢀ0.81
ꢀ0.76
ꢀ0.62e
ꢀ0.60
0.9995
0.6316
0.6486
0.5624
0.8411e
0.8011
1.2614
1.4955
1.4751
1.5763
1.5085
1.3352
1.3472
1.2179
[(DIMPY)As]+
[(BIPY)As]+
C2v
C2v
1.318
0.90
1.1649
1.3083
Dications
[(NACNAC)P]2+
[(NACNAC)As]2+
C2v
C2v
0.592
0.565
1.52
1.61
ꢀ0.86
ꢀ0.88
1.0119
0.9499
1.2979
1.3286
1.4200
1.4131
a
Point group symmetry.
b
NBO population of the valence p orbital on the Pn atom of suitable symmetry to interact with the p-system of the ligand.
NBO charge on the dicoordinate P atom.
NBO Wiberg bond indices for the specified bonds.
c
d
e
NBO charges and Wiberg bond indices involving the Npyridyl atom.
p-type symmetry (p(p)). In this light, several observations
on the ligands. MOLDEN depictions of all of the frontier
orbitals for each of the ligands are supplied in the Support-
ing information, however, given the discussion regarding
the population of the Pn p(p) orbital, the composition of
the vacant orbitals involving the p-system on each free
ligand are presumably the most relevant. The lowest unoc-
cupied molecular orbitals involving the p-system are
depicted in Fig. 8 and illustrate an important difference
between the neutral ligands and the [NACNAC]ꢀ chelate.
Whereas the LUMO in each of the neutral ligands is of
the correct symmetry (B1 in C2v) to interact with a filled
p(p) orbital on the PnI center, the lowest vacant orbital
involving the p-system on the [NACNAC]ꢀ model does
not have the appropriate symmetry (A2 in C2v) to interact
in the same way. In that system, the first vacant orbital
of B1 symmetry is the LUMO + 6 orbital, which is signifi-
cantly higher in energy than the LUMO (5.67 eV versus
4.47 eV) and which is based primarily on the unique a-car-
bon atom in the ligand.
are noteworthy. Perhaps most obviously, the occupancy
of the valence p(p) for both the [(DAB)Pn]+ and [(DIM-
PY)Pn]+ models are considerably lower than 2.0 (ranging
from 0.893 to 1.076 electrons) and are consistent with sig-
nificant transfer of electron density from the Pn atom to
the ligand. While the [(BIPY)Pn]+ models are consistent
with somewhat less transfer of electron density (P: 1.278;
As: 1.318), the magnitudes of the transferred electron den-
sity is significantly larger than those calculated for phos-
phine-stabilized PnI centers [13,15]. Probably the most
important observation is that the occupancy of the Pn
p(p) orbitals in the C2v-symmetry [(NACNAC)Pn] models
(P: 1.905; As: 1.924) suggest that virtually no electron den-
sity is transferred to the p-system of the ligand! The adop-
tion of the non-planar structure reduces the occupancy of
the corresponding orbitals in the Cs-symmetry [(NAC-
NAC)Pn] models and suggest one reason for the adoption
of the non-planar structure.
For phosphine-stabilized PnI cations, the transfer of
electron density was considered in terms of hyperconjuga-
tion (or back-bonding) between the p(p) orbital and the
ligand p-system. For the systems in this work, the greater
magnitude of the donation suggests that it may be more
appropriate to call the interaction Pn–ligand electron
transfer in the case of the DAB ligand, however, the under-
lying concept is the same in both instances (it is just much
more difficult to reduce a phosphine than a diimine). To
understand how the ligands do or do not remove electron
density from the supposed PnI cations, one must examine
the shape and composition of important molecular orbitals
In regard to the composition of the orbitals of the neu-
tral ligands depicted in Fig. 8, one other point should be
noted. Because of its relatively localized nature, population
of the LUMO in the DAB model will have a dramatic and
obvious effect on the metrical parameters of the resultant
structure. In contrast the LUMOs of the BIPY and
DIMPY are considerably more delocalized and thus the
changes in individual bond distances in the ligands upon
population of the LUMO are anticipated to be more sub-
tle. For this reason, population analyses of the type pre-
sented herein are a valuable tool for the examination of
such systems.

340
B.D. Ellis, C.L.B. Macdonald / Inorganica Chimica Acta 360 (2007) 329–344
Fig. 8. Lowest unoccupied molecular orbitals involving the p-systems on the model ligands DAB, BIPY, DIMPY and [NACNAC]ꢀ in C2v symmetry.
While it appears that the differing nature of the vacant
orbitals explains the contrasting behavior of the neutral
ligands there does not appear to be any obvious correlation
between the apparent ‘‘amount’’ of reduction observed for
the various ligands and the HOMO–LUMO energy differ-
ences (Table 2) or the energy differences between the high-
est occupied and lowest unoccupied orbitals of p-type
symmetry (Table 3). However, it may be important to note
that the overall energy of the LUMO in DAB is the lowest,
followed by that of DIMPY, then BIPY and finally [NAC-
NAC]ꢀ, thus the apparent magnitude of ligand reduction
may be related to that energy. This correlation also fits
with the anticipated behaviors of the ligands: it should be
easier to reduce the neutral four p-electron system in
DAB than to reduce the conjugated aromatic systems in
the DIMPY and BIPY and all of these should be much eas-
ier to reduce than the anionic ligand that already contains
six electrons in its p-system.
The composition of the filled molecular orbitals on the
ligands can also provide insight into the differing nature
of the interaction between Pn and the neutral and anionic
ligands. Whereas the HOMOs for each of the neutral
ligands corresponds to ‘‘lone pair’’ orbitals in the r-sys-
tems of the molecules, the HOMO for the [NACNAC]ꢀ
model involves the p-system and the largest lobe is situated
on the a-carbon atom (this atom also has a relatively large
negative charge of ꢀ0.559), as illustrated in Fig. 9. In this
light, the observed monodentate ligation to phosphorus
centers through the Ca atom should not be considered sur-
prising or unexpected. Furthermore, given that the HOMO
of the [NACNAC]ꢀ ligand is of B1 symmetry, it is more
reasonably suited to donate electron density into vacant
p(p) orbitals – the shape of this occupied orbital explains
why this ligand is able to interact successfully through
the nitrogen atoms with electron-poor centers from group
13 [53–56] and group 14 [57,58].
both the phosphorus- and arsenic-containing models are
virtually identical in every case, only those of the arsenic-
containing species are presented. For example, selected
frontier orbitals of [(DAB)As]+ are presented in Fig. 10.
The appearances of these orbitals, in particular the ‘‘lone
pair’’ of HOMO ꢀ 2 and the p-systems in the HOMO
and HOMO ꢀ 1 orbitals, are entirely consistent with those
calculated previously for group 15 [52,59,60] and group 14
N-heterocyclic carbenoids [61,62], and the results suggest
that the description of cations of the form [(DAB)Pn]+ as
containing PnIII centers is not unreasonable.
Selected frontier orbitals of the model [(DIMPY)As]+
are depicted in Fig. 11. These orbitals are particularly
informative in that they show quite clearly the r-type ‘‘lone
pair’’ on As in HOMO ꢀ 1 and, while there is some delo-
calization in the HOMO, the largest lobe is obviously
located on the As center and the orbital is most consistent
with it being a p-type ‘‘lone pair’’ on As. Such ‘‘lone pairs’’
are exactly consistent with the orbitals found for Ardu-
engo’s ‘‘10-Pn-3’’ low-coordinate, hypervalent pseudo-trig-
onalbipyramidal systems [24]. In this light, the description
of this cation as containing AsI is likely the most appropri-
ate in spite of the transfer of some electron density from the
Pn center to the ligand.
Not surprisingly, although the situation is similar to that
in [(DIMPY)As]+, the orbitals are somewhat more compli-
cated for the [(BIPY)As]+ system, shown in Fig. 12. Many
of the orbitals are attributable to the more extensive p-sys-
tem in BIPY, however, the HOMO appears to be consis-
tent with a p-type ‘‘lone pair’’ on As and HOMO ꢀ 5 is
clearly a r-type ‘‘lone pair’’ on As. While some Pn–N
p-delocalization is apparent in HOMO ꢀ 4, it is consider-
ably less than that observed, for example, in the HOMO
of [(DAB)As]+. Overall, the orbitals in [(BIPY)As]+ are
most consistent with the presence of an AsI center in the
molecule. Such an interpretation is also in keeping with
the ‘‘trapped singlet carbon’’ analogue [(BIPY)C] reported
by Weiss and co-workers [63]. It is also worthy of note that
and the energy difference between the HOMO and LUMO
Finally, an examination of some of the orbitals in the
pnictogen complexes provides further insight into the elec-
tronic structure of these species. Because the orbitals for
Fig. 9. Highest occupied molecular orbitals (HOMOs) for the model ligands DAB, BIPY, DIMPY and [NACNAC]ꢀ in C2v symmetry.

B.D. Ellis, C.L.B. Macdonald / Inorganica Chimica Acta 360 (2007) 329–344
341
sequence of having a non-Huckel aromatic total of eight
¨
electrons in the p-system (HOMO ꢀ 5, HOMO ꢀ 4,
HOMO ꢀ 1 and HOMO are all filled). In this context,
the unfavorability of the planar [(NACNAC)Pn] models
is not surprising.
Distortion of the [(NACNAC)As] model from planarity
to the Cs-symmetry boat conformation changes the appear-
ances (and energies – see Table 2) of the frontier orbitals, as
shown in Fig. 14. While the appearance of the r-type ‘‘lone
pair’’ orbitals in HOMO ꢀ 2 does not change drastically,
the appearance of the HOMO is instructive. The bending
of the As atom and the Ca atom toward each other allows
for the in-phase overlap of the lobes on these atoms. To
some extent, such overlap (and those in the other distorted
orbitals) explains the dramatically lower p(p) population in
the Cs-symmetry model. In any event, it must still be
emphasized that the energy difference between the HOMO
and LUMO are still considerably smaller for this model
than for any of the other models described above and such
molecules may be so reactive as to continue to elude exper-
imental isolation.
Because the most obvious problems with the [(NAC-
NAC)Pn] species are clearly associated with the presence
of eight potential p-electrons in the six-membered ring gen-
erated by the chelation of the six p-electron [NACNAC]ꢀ
fragment with the two p-electron PnI cation, we reasoned
that the [NACNAC]ꢀ ligand would likely produce a more
stable complex if it chelates to a PnIII cation bearing no
p-electrons. As illustrated by the data in Tables 2 and 4,
the resultant dications of the form [(NACNAC)Pn]2+
should be considerably more stable than their neutral ana-
logues. The optimized structures for the cations are shown
in Fig. 15 and the metrical parameters are clearly indicative
of ligand–Pn bonding that is very different from that in
[(NACNAC)Pn]. For example, the Pn–N distances in
[(NACNAC)Pn]2+ are almost identical to those in the
[(DAB)Pn]+ models and the N–C and C–C distances are
consistent with removal of electron density from the
HOMO of the [NACNAC]ꢀ ligand.
Fig. 10. Selected frontier orbitals for the model [(DAB)As]+.
Fig. 11. Selected frontier orbitals for the model [(DIMPY)As]+.
in the [(BIPY)As]+ models (see Table 2) suggests that such
compounds may have enough stability for their experimen-
tal isolation.
For the model [(NACNAC)As], it is instructive to
examine the orbitals of both the unfavorable C2v form in
addition to those of the Cs form. The orbitals for the
C2v-symmetry compound are presented in Fig. 13. Several
interesting features are observed upon examination of these
orbitals. While the HOMO ꢀ 2 orbital is similar in appear-
ance to the other r-type ‘‘lone pair’’ orbitals found in the
other models, the HOMO is actually quite different in
detail than those of any of the complexes described above.
In particular, the HOMO in [(NACNAC)As] is clearly
anti-bonding in regard to the pnictogen and nitrogen atoms
whereas the corresponding orbitals are either bonding (in
[(DAB)As]+ and [(DIMPY)As]+) or non-bonding
([(BIPY)As]+) in the other models. Perhaps more impor-
tantly, the HOMO ꢀ 5 also shows that the p-system in this
model is not completely delocalized and; this is likely a con-
Fig. 12. Selected frontier orbitals for the model [(BIPY)As]+.

342
B.D. Ellis, C.L.B. Macdonald / Inorganica Chimica Acta 360 (2007) 329–344
Fig. 13. Selected frontier orbitals for the C2v-symmetry model [(NACNAC)As].
Fig. 14. Selected frontier orbitals for the Cs-symmetry model [(NACNAC)As].
nions. Cowley and co-workers’ recent report of the isola-
tion of salts of the form [(R-NACNAC)P-Cl][SO3CF3]
lend credence to these predictions because the cations in
their isolated systems are the mono chloride adducts of
these putative [(R-NACNAC)P]2+ dications [47].
4. Conclusions
Experimental investigations in this work and those of
other groups have demonstrated that the chelation of puta-
tive PnI centers by diimines is a viable and very clean
method for the synthesis of Pn-containing heterocycles.
In the case of the reaction of 1,4-diaryl-2,3-dimethyl-1,4-
diazabutadiene with either PI3 or a mixture of PCl3 and
SnCl2, the products are salts of phosphenium cations,
which contain PIII centers, as one would expect on the basis
of the reported behavior of related systems. In contrast,
when the analogous reactions using the similar b-diketimi-
nate ligand are attempted, many products appear to be
formed and the reactions are certainly not clean or
quantitative.
The computational investigations reveal that whereas
ligands containing a-diimine-like fragments (DAB,
DIMPY, BIPY) are predicted to chelate effectively with
PnI cations to produce planar complexes, the b-diketimi-
nate anion [NACNAC]ꢀ form relatively unstable com-
plexes which adopt boat-like conformations. The nature
Fig. 15. Optimized structures (C2v-symmetry) of [(NACNAC)P]2+ and
[(NACNAC)As]2+
indicated.
.
Selected bond distances (A) and angles (°) are
˚
In a similar vein, selected frontier orbitals for [(NAC-
NAC)As]2+ are depicted in Fig. 16 and these are much
more similar to those of the [(DAB)As]+ model in Fig. 10
than to those of either of the neutral models for [(NAC-
NAC)As]. It is clear that the HOMO is bonding in terms
of the Pn–N interaction, which looks very similar to that
of the [(DAB)As]+ model and, perhaps more importantly,
the appearance of HOMO ꢀ 5 demonstrates that the p-sys-
tem is completely delocalized in this case. The difference in
HOMO and LUMO energies is quite large for these dica-
tions and the preponderance of the data suggest that salts
of [(NACNAC)Pn]2+ are viable synthetic targets that
should be obtainable through judicious choice of countera-

B.D. Ellis, C.L.B. Macdonald / Inorganica Chimica Acta 360 (2007) 329–344
343
Fig. 16. Selected frontier orbitals for the C2v-symmetry model [(NACNAC)As]2+
.
of the interactions between the ligands and the pnictogen
atoms are also clearly dependent on the nature of the
ligand and, in particular, on the symmetry and composi-
tion of the LUMO associated with the p-system on the
ligands. For [(DAB)Pn]+, the metrical parameters and pop-
ulation analyses are consistent with the presence of the
ligand in its dianionic form and implies the oxidation of
PnI to PnIII. The metrical parameters for the models con-
taining pyridyl groups, namely [(DIMPY)Pn]+ and
[(BIPY)Pn]+ are more consistent with a smaller transfer
of electron density to the ligand and can still be considered
most appropriately as containing PnI centers. This con-
forms to the understanding that imines should be easier
to reduce than pyridyl groups and suggests that many
ligands containing these fragments may be employed for
analogous heterocycle synthesis.
In contrast, the interaction of the six p-electron system
on [NACNAC]ꢀ with the two p-electrons on a PnI cation
produces unstable complexes and it is proposed that such
anionic ligands can be more successfully employed in the
stabilization of PnIII centers. The large HOMO–LUMO
energy difference and orbital structure consistent with
known pnictogenium species suggest that the correct selec-
tion of counteranion should allow for the isolation of the
dicationic [(NACNAC)Pn]2+ salts.
Appendix A. Supporting information
A summary of computational results and pictures of cal-
culated structures and molecular orbitals are supplied. This
information is available free of charge via the Internet at
www.elsevier.com. CCDC 610976 and 610977 contain the
supplementary crystallographic data for the structures pre-
sented in this paper. These data can be obtained free of
charge from The Cambridge Crystallographic Data Centre
via www.ccdc.cam.ac.uk/data_request/cif. Supplementary
data associated with this article can be found, in the online
version, at doi:10.1016/j.ica.2006.07.047.
References
[1] C.L.B. Macdonald, B.D. Ellis, in: R.B. King (Ed.), Encyclopedia of
Inorganic Chemistry, Low Oxidation State Main Group, Wiley,
Hoboken, NJ, 2005.
[2] B.D. Ellis, C.L.B. Macdonald, ACS Symp. Ser. 917 (2006) 108.
[3] F. Mathey, Angew. Chem., Int. Ed. Engl. 26 (1987) 275.
[4] K. Lammertsma, M.J.M. Vlaar, Eur. J. Org. Chem. (2002) 1127.
[5] K. Lammertsma, Top. Curr. Chem. 229 (2003) 95.
[6] A. Schmidpeter, Heteroat. Chem. 10 (1999) 529.
[7] S.F. Gamper, H. Schmidbaur, Chem. Ber. 126 (1993) 601.
[8] R.J. Barnham, R.M.K. Deng, K.B. Dillon, A.E. Goeta, J.A.K.
Howard, H. Puschmann, Heteroat. Chem. 12 (2001) 501.
[9] M. Driess, H. Ackermann, J. Aust, K. Merz, C. Von Wullen, Angew.
Chem., Int. Ed. 41 (2002) 450.
[10] H. Ackermann, J. Aust, M. Driess, K. Merz, C. Monse, C. Van
Wullen, Phosphorus, Sulfur Silicon Relat. Elem. 177 (2002) 1613.
[11] B.D. Ellis, C.L.B. Macdonald, Phosphorus, Sulfur Silicon Relat.
Elem. 179 (2004) 775.
[12] J.A. Boon, H.L. Byers, K.B. Dillon, A.E. Goeta, D.A. Longbottom,
Heteroat. Chem. 11 (2000) 226.
[13] B.D. Ellis, M. Carlesimo, C.L.B. Macdonald, Chem. Commun.
(2003) 1946.
[14] B.D. Ellis, C.L.B. Macdonald, Inorg. Chem. 43 (2004) 5981.
[15] B.D. Ellis, C.L.B. Macdonald, Inorg. Chem. (2006) in press.
[16] P. Kilian, A.M.Z. Slawin, J.D. Woollins, Dalton Trans. (2006) 2175.
Acknowledgements
We thank the Natural Science and Engineering Re-
search Council (NSERC) of Canada for funding and schol-
arships (B.D.E.). We thank the Canada Foundation for
Innovation, the Ontario Innovation Trust, the Ontario
Research and Development Challenge Fund for support
of the University of Windsor’s Centre for Catalysis and
Materials Research.

344
B.D. Ellis, C.L.B. Macdonald / Inorganica Chimica Acta 360 (2007) 329–344
[17] B.D. Ellis, C.A. Dyker, A. Decken, C.L.B. Macdonald, Chem.
Commun. (2005) 1965.
[18] G. Reeske, C.R. Hoberg, N.J. Hill, A.H. Cowley, J. Am. Chem. Soc.
128 (2006) 2800.
[19] A.H. Cowley, R.A. Kemp, Chem. Rev. 85 (1985) 367.
[20] F. Mathey, in: M. Regitz, O.J. Scherer (Eds.), Multiple Bonds and
Low Coordination in Phosphorus Chemistry, Phosphinidenes, Thi-
eme Verlag, Stuttgart, 1990, p. 33.
[21] G. Reeske, A.H. Cowley, Chem. Commun. (2006) 1784.
[22] D. Gudat, A. Haghverdi, H. Hupfer, M. Nieger, Chem. Eur. J. 6
(2000) 3414.
[23] A.J. Arduengo III, C.A. Stewart, F. Davidson, D.A. Dixon, J.Y.
Becker, S.A. Culley, M.B. Mizen, J. Am. Chem. Soc. 109 (1987) 627.
[24] A.J. Arduengo III, C.A. Stewart, Chem. Rev. 94 (1994) 1215.
[25] L. Bourget-Merle, M.F. Lappert, J.R. Severn, Chem. Rev. 102 (2002)
3031.
[26] L. Bourget-Merle, Y.X. Cheng, D.J. Doyle, P.B. Hitchcock, A.V.
Khvostov, M.F. Lappert, A.V. Protchenko, X.H. Wei, ACS Symp.
Ser. 917 (2006) 192.
[42] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb,
J.R. Cheeseman, V.G. Zakrzewski, V.G. Montgomery Jr., R.E.
Stratmann, J.C. Burant, S. Dapprich, J.M. Millam, A.D. Daniels,
K.N. Kudin, M.C. Strain, O. Farkas, J. Tomasi, V. Barone, M.
Cossi, R. Cammi, B. Mennucci, C. Pomelli, C. Adamo, S. Clifford,
J. Ochterski, G.A. Petersson, P.Y. Ayala, Q. Cui, K. Morokuma, P.
Salvador, J.J. Dannenberg, D.K. Malick, A.D. Rabuck, K. Raghav-
achari, J.B. Foresman, J. Cioslowski, J.V. Ortiz, A.G. Baboul, B.B.
Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R.
Gomperts, R.L. Martin, D.J. Fox, T. Keith, M.A. Al-Laham, C.Y.
Peng, A. Nanayakkara, M. Challacombe, P.M.W. Gill, B. Johnson,
W. Chen, M.W. Wong, J.L. Andres, C. Gonzalez, M. Head-Gordon,
E.S. Replogle, J.A. Pople, GAUSSIAN-98, Revision A.11.1, Gaussian,
Inc., Pittsburgh, PA, 2001.
[43] A.D. Becke, J. Chem. Phys. 98 (1993) 5648.
[44] J.P. Perdew, Y. Wang, Phys. Rev. B 45 (1992) 13244.
[45] A.E. Reed, L.A. Curtiss, F. Weinhold, Chem. Rev. 88 (1988) 899.
[46] G. Schaftenaar, J.H. Noordik, J. Comput.-Aided Mol. Des. 14 (2000)
123.
[27] H.W. Roesky, S.S. Kumar, Chem. Commun. (2005) 4027.
[28] H.W. Roesky, ACS Symp. Ser. 917 (2006) 20.
[29] P.B. Hitchcock, M.F. Lappert, J.E. Nycz, Chem. Commun. (2003)
1142.
[30] P.J. Ragogna, N. Burford, M. D’Eon, R. McDonald, Chem.
Commun. (2003) 1052.
[31] N. Burford, M. D’Eon, P.J. Ragogna, R. McDonald, M.J. Ferguson,
Inorg. Chem. 43 (2004) 734.
[32] H. Tom Dieck, M. Svoboda, T. Greiser, Z. Naturforsch. B 36B
(1981) 823.
[47] D. Vidovic, Z. Lu, G. Reeske, J.A. Moore, A.H. Cowley, Chem.
Commun. (2006) in press.
[48] F.H. Allen, Acta Crystallogr., Sect. B B58 (2002) 380.
[49] I.A. Litvinov, V.A. Naumov, T.V. Griaznova, A.N. Pudovik, A.M.
Kibardin, Dokl. Akad. Nauk SSSR 312 (1990) 623.
[50] M.K. Denk, S. Gupta, A.J. Lough, Eur. J. Inorg. Chem. (1999)
41.
[51] C.J. Carmalt, V. Lomeli, B.G. McBurnett, A.H. Cowley, Chem.
Commun. (1997) 2095.
[52] D. Gudat, Eur. J. Inorg. Chem. (1998) 1087.
[33] J. Feldman, S.J. McLain, A. Parthasarathy, W.J. Marshall, J.C.
Calabrese, S.D. Arthur, Organometallics 16 (1997) 1514.
[34] A.B. Pangborn, M.A. Giardello, R.H. Grubbs, R.K. Rosen, F.J.
Timmers, Organometallics 15 (1996) 1518.
[53] C.M. Cui, H.W. Roesky, H.G. Schmidt, M. Noltemeyer, H.J. Hao,
F. Cimpoesu, Angew. Chem., Int. Ed. 39 (2000) 4274.
[54] N.J. Hardman, B.E. Eichler, P.P. Power, Chem. Commun. (2000)
1991.
[35] SMART: Molecular Analysis Research Tool, Bruker AXS Inc., Mad-
ison, WI, 2001.
[55] M.S. Hill, P.B. Hitchcock, R. Pongtavornpinyo, Dalton Trans.
(2005) 273.
[36] SAINT-Plus: Data Reduction and Correction Program, Bruker AXS
Inc., Madison, WI, 2001.
[37] SADABS: An Empirical Absorption Correction Program, Bruker AXS
Inc., Madison, WI, 2001.
[38] A. Altomare, M.C. Burla, M. Camalli, G. Cascarano, C. Giacovazzo,
A. Guagliardi, M.A.G.G., G. Polidori, R. Spagna, SIR-97: A New
Program for Solving and Refining Crystal Structures, CNR-IRMEC,
Bari, 1997.
[56] M.S. Hill, P.B. Hitchcock, Chem. Commun. (2004) 1818.
[57] Y.Q. Ding, H.W. Roesky, M. Noltemeyer, H.G. Schmidt, P.P.
Power, Organometallics 20 (2001) 1190.
[58] P.B. Hitchcock, J. Hu, M.F. Lappert, J.R. Severn, Dalton Trans.
(2004) 4193.
[59] R.R. Sauers, Tetrahedron 53 (1997) 2357.
[60] R.J. Boyd, N. Burford, C.L.B. Macdonald, Organometallics 17
(1998) 4014.
[39] G.M. Sheldrick, SHELXL-97, Universitat Go¨ttingen, Go¨ttingen, Ger-
many, 1997.
[40] L.J. Farrugia, J. Appl. Crystallogr. 32 (1999) 837.
[41] G.M. Sheldrick, SHELXTL: Structure Determination Software Suite,
Bruker AXS Inc., Madison, WI, 2001.
[61] C. Boehme, G. Frenking, J. Am. Chem. Soc. 118 (1996) 2039.
[62] C. Heinemann, T. Mueller, Y. Apeloig, H. Schwarz, J. Am. Chem.
Soc. 118 (1996) 2023.
[63] R. Weiss, S. Reichel, M. Handke, F. Hampel, Angew. Chem., Int.
Ed. 37 (1998) 344.