Reactions of P4 and I2 with Ag[Al(OC(CF3)3)4]: From elusive polyphosphorus cations to subvalent P3I6+ and phosphorus rich P5I2+

Article abstract of DOI:10.1039/b103957c

Reactions of X₂ (X = Br, I), P₄ and Ag(CH₂Cl₂)[Al(OR)₄] [R = C(CF₃)₃] in suitable ratios to prepare naked polyphosphorus cations were carried out and led to products which sugges

Full text of DOI:10.1039/b103957c

View Article Online / Journal Homepage / Table of Contents for this issue
Reactions of P4 and I2 with Ag[Al(OC(CF3)3)4]: from elusive
polyphosphorus cations to subvalent P3I6؉ and phosphorus rich
P5I2؉ †
Ingo Krossing
University of Karlsruhe, Engesserstr. Geb. 30.45, 76128 Karlsruhe, Germany.
E-mail: krossing@chemie.uni-karlsruhe.de
Received 3rd May 2001, Accepted 5th November 2001
First published as an Advance Article on the web 29th January 2002
Reactions of X2 (X = Br, I), P4 and Ag(CH2Cl2)[Al(OR)4] [R = C(CF3)3] in suitable ratios to prepare naked
polyphosphorus cations were carried out and led to products which suggested the presence of these elusive cations
as intermediates. At temperatures above Ϫ30 ЊC to rt the initially formed cations decomposed the Al(OR)4Ϫ anion
giving, in two cases, the more stable fluoride bridged (RO)3Al–F–Al(OR)3Ϫ anion. When Br2 was used as the oxidising
agent the proposed intermediate phosphorus cation (P5ϩ?) reacted with the solvent CDCl3 by double insertion of a Pϩ
unit into the C–Cl bond giving Cl2P(CDCl2)2[(RO)3Al–F–Al(OR)3], 1. When I2 was used as the oxidiser the reaction
led to the marginally stable P3I6[(RO)3Al–F–Al(OR)3], 2 (X-ray). By using very mild conditions throughout (Ϫ80 ЊC)
ϩ
the primary product of the reaction of Ag(P4)2[Al(OR)4] and I2 was isolated: P5I2[Al(OR)4], 3, containing the P5I2
cation with a hitherto unknown C2v-symmetric P5 cage as structural building block. P3I6[Al(OR)4], 4, was directly
synthesised in quantitative yield starting from P2I4, PI3 and Ag(CH2Cl2)[Al(OR)4] in CH2Cl2 solution. P3I6ϩ is formed
ϩ
through the P2I5ϩ stage (31P-NMR). P3I6ϩ (av.: P2.33) is the first subvalent P–X cation (X = H, F, Cl, Br, I). P5I2
(av.: P0.6) is the first phosphorus rich binary P–X cation. They are the third and fourth example of a binary P–X
cation after the known PX4ϩ and P2X5ϩ cations. The observed reactions were fully accounted for by thermochemical
Born–Haber cycles based on (RI-)MP2/TZVPP ab initio, COSMO solvation and lattice enthalpy calculations (all
phases). The gaseous enthalpies of formation of several species were calculated to be (in kJ molϪ1): P5ϩ (913), P3I6
ϩ
(694), P5I2ϩ (792), P2I5ϩ (733), Ag(P4)2ϩ (784).
ϩ
ϩ
that finally led to the formation of salts of the P3I6 and P5I2
Introduction
cations. A preliminary account on P5I2ϩ has been given.10
Phosphorus, arsenic and fluorine are the only electronegative
non-metallic elements for which homopolyatomic elemental
cations are still unknown as “compounds in the bottle”. For P
and As this appears rare, since gaseous phosphorus and arsenic
cations are well investigated by mass spectrometric measure-
ments (up to P89ϩ),1,2 photo ionisation studies (E2–5ϩ, E = P, As)3
and theoretical investigations.1c,4 Many of the basic thermo-
dynamic properties of Enϩ (i.e. IP’s, ∆fH ) are known.5 (Radical-)-
Results
Syntheses and spectroscopic characterisation
Initial reactions of Li[Al(OR)4], P4 and iodine [R = C(CF3)3] in
various solvents in suitable ratios for the preparation of P5
ϩ
only led to an orange–red solid material, which, according to
the Raman spectra, did not contain the anion but rather was
a mixture of red phosphorus and P2I4. This was attributed to
the very hard and polarising Liϩ cation that, with the reaction
conditions employed, didn’t react with the soft phosphorus
iodides. However, when exchanging Liϩ for Agϩ and I2 for Br2
an ionisation occurred as shown by the NMR scale reaction
of Ag(CH2Cl2)[Al(OR)4], 5/4 P4 and 0.5 Br2 in CDCl3 [see
eqn. (1)]:
ϩ
Cations En with an even number, n, of elemental atoms were
shown to be considerably less stable than the diamagnetic
uneven species Enϩ.4 However, earlier work showed that clas-
sical approaches to salts of Enϩ (i.e. oxidation of E by MF5 or
F2S2O6, M = As, Sb) did not lead to success but rather to a
decomposition of the anions and formation of E–O and E–F
species.6 And, indeed, the strengths of the E–F bonds (484 and
490 kJ molϪ1 in EF3)5 are amongst the highest known in the
periodic table and therefore account for the decomposition
reactions. To circumvent the decomposition of the anions, we
employed a new generation of weakly basic anions of type
Ag(CH2Cl2)[Al(OR)4] ϩ 5/4 P4 ϩ 0.5 Br2
Cl2P(CDCl2)2[(RO)3Al–F–Al(OR)3] 1 ϩ
AgBr ϩ P4 ϩ ? (1)
Ϫ
Al(OR)4 [R = C(CF3)3], the syntheses of lithium and silver
salts of which was recently published by Strauss7 and us.8
Al[OC(CF3)3]4Ϫ also appeared to be an ideal spectator ion, since
[R = C(CF3)3]
ϩ
it stabilises the unusual D2h symmetric homoleptic Ag(P4)2
cation.9 Previously we showed8 that the Al[OC(CF3)3]4Ϫ anion,
which is stable in 35% HNO3, is one of the most weakly
coordinating anions known. Herein we present the results of
our efforts to oxidise the silver-P4 adducts9 with halogens
The unit cell of several of the uniform colourless crystals of
this reaction was determined and all of them showed the same
triclinic cell as 1 indicating that this was the main product
of reaction (apart from P4). The rt 31P-NMR of this solution
showed two lines: an intense line attributable to P4 and another
ϩ
weaker resonance we assign to the Cl2P(CDCl2)2 cation (144
† Electronic supplementary information (ESI) available: a drawing
of the single piece apparatus used for the reactions. See http://
www.rsc.org/suppdata/dt/b1/b103957c/
ppm, cf. 96 ppm in PCl4ϩPCl6). Presumably a very electrophilic
phosphorus cation [e.g. P5ϩ, cf. eqn. (2)] was initially formed
500
J. Chem. Soc., Dalton Trans., 2002, 500–512
This journal is © The Royal Society of Chemistry 2002
DOI: 10.1039/b103957c

View Article Online
ϩ
which decomposed the anion and finally led to the observed
The obtained P3I6 salt was always contaminated with the
very intense Raman scatterer P2I4 which prevented its Raman
observation, however, we succeeded in the independent direct
synthesis of P3I6ϩ (see below).
ϩ
products. The formation of the Cl2P(CDCl2)2 cation in 1 is
then seen as a double insertion of a Pϩ unit (possibly from
P5ϩ) into a C–Cl bond (see Discussion).
We concluded that iodine should be a more suitable oxidising
agent than bromine since it is solid and reactions are, therefore,
slower. A thermochemical evaluation11,12 of eqn. (2) showed
We reinvestigated the reaction as proposed in eqn. (3) by low
temperature in situ NMR spectroscopy but using pure [Ag-
(P4)2][Al(OR)4]9 instead of the Ag(CH2Cl2)[Al(OR)4]/2 P4
mixture. When [Ag(P4)2][Al(OR)4]9 in CD2Cl2 was reacted with
3.5 equivalents of I2 at Ϫ78 ЊC one major product was observed
in the in situ 31P-NMR spectrum (Ϫ90 ЊC) recorded after 30
minutes: [P5I2][Al(OR)4] 3 [eqn. (5)] containing a C2v-symmetric
it to be favourable by 149 kJ molϪ1
:
Ag[Al(OR)4] (s) ϩ 5/4 P4 (s) ϩ 0.5 I2 (s)
P5[Al(OR)4] (s) ϩ AgI (s) (2)
∆rH = Ϫ149 kJ molϪ1
ϩ
P5I2 cation, the in situ 31P-NMR spectrum of which is shown
[R = C(CF3)3];
in Fig. 1.
Both pure PI3 and P2I4 are insoluble in CD2Cl2 at these tem-
peratures (31P-NMR) and, therefore, exchange with the cation
as in CS2 did not occur. Small amounts of solid P4 were also
visible in the 31P-NMR (not shown).16 A larger preparative
reaction, always kept at Ϫ80 ЊC, verified these conclusions
based on the mass balance, the Raman observation of pure
P5I2ϩ[Al(OR)4Ϫ] 3 in the soluble and PI3, P4 and traces of P2I4
in the insoluble material of this reaction as well as an elemental
analysis of 3. The balanced overall reaction with the observed
products is given in eqn. (5):
More polar solvents such as CH2Cl2 (cf. ε = 8.9 vs. 4.8 for
CHCl3)5c should also help to stabilise the ions (cf. the compu-
tational work below). Therefore a reaction as in eqn. (2) was
performed in CH2Cl2. The reaction temperature in this experi-
ment was always kept below Ϫ20 ЊC and immediately after con-
sumption of all visible amounts of iodine part of the orange
coloured solution was poured into a glass tube. All volatiles
were then removed in vacuo and the Raman spectrum of the
remaining orange material showed bands attributable to the
intact anion, Ag(P4)2ϩ, P2I4 but also new P–P stretches that
possibly stem from yet unknown phosphorus cations Pnϩ (n = 3,
5, 7).13 In further reactions according to eqn. (2) the presence of
larger amounts of P2I4 was reinforced by the isolation of orange
triclinic crystals which, according to their unit cell14 and
Raman spectrum, were P2I4. Therefore P2I4 appeared to be a
thermodynamic sink in this reaction. To exploit the formation
of P2I4 as an additional driving force the underlying thermo-
chemistry11 was examined in a suitable cycle [see eqn. (3)].
(5)
Ag[Al(OR)4] (s) ϩ 2 P4 (s) ϩ 3.5 I2 (s)
Solid pure 3 may be handled at rt for 1–2 hours, however,
upon warming the NMR tube overnight to Ϫ40 ЊC the P5I2
ϩ
P5[Al(OR)4] (s) ϩ 1.5 P2I4 (s) ϩ AgI (s) (3)
[R = C(CF3)3];
∆rH = Ϫ287 kJ molϪ1
signals vanish and, apart from other unassigned signals of
ϩ
lower intensity, those of P3I6 appear as major P-containing
peaks (cf. P3I6[Al(OR)4], 4, below). The observation of only
traces of P2I4 but large amounts of PI3 in the Raman spectrum,
at Ϫ80 ЊC, of the insoluble material of eqn. (5) shows that in the
earlier reactions P2I4, which was always found, is formed from
PI3 and P4 only upon warming. Initial PI3 formation is in
agreement with Tattershall’s17 earlier observation that P4 reacts
with I2 to immediately give PI3 which is then reduced by P4
giving P2I4. From the Raman spectrum18 of the very air and
moisture sensitive solid yellow 3, shown in Fig. 2, the presence
Indeed we found that eqn. (3) should be considerably more
favourable than a reaction according to eqn. (2) {∆rH [eqn. (3)]
= Ϫ287 kJ molϪ1 vs. ∆rH [eqn. (2)] = Ϫ149 kJ molϪ1}. Con-
sequently eqn. (3) was initially performed in CH2Cl2 and after
stirring the reaction mixture overnight at Ϫ30 ЊC all volatiles
were removed. To extract the very CS2 soluble by-product,
P2I4, the solid yellow–orange residue was extracted with CS2 at
rt15 and, to our surprise, much of the solid material was soluble
in this very non-polar solvent. From the concentrated CS2 solu-
tion of this reaction we obtained a mixture of single crystals.
Amongst orange P2I4 needles (unit cell determination14 and
Raman) a larger amount of yellow plate-like and extremely
sensitive crystals was found and the single crystal X-ray struc-
ture determination of these yellow plates showed them to be
P3I6[(RO)3Al–F–Al(OR)3], 2, containing the first subvalent
binary P–X cation (X = H, F, Cl, Br, I) with an average oxid-
ation state of P of 2.33. In the 31P-NMR of this CS2 solution
lines attributable to PI3 and P2I4 appeared at δ31P = 176 and 105
but were very broad (ν1/2 = 2100 and 1600 Hz) indicating rapid
exchange (with P3I6ϩ?). Repeating the 31P-NMR spectra at low
temperature (Ϫ30 and Ϫ70 ЊC) led to sharp lines (ν1/2 = 5 Hz)
with greatly enhanced signal-to-noise ratio at the expected
positions for PI3 and P2I4 at δ31P(Ϫ30 ЊC and Ϫ70 ЊC) = 176 and
105, but no additional lines due to a P-containing cation
emerged—indicating that the cation had precipitated or that a
fast exchange between the unknown species (P3I6ϩ?) and PI3,
P2I4 or the solvent was still occuring. The overall reaction with
all observed products is given in eqn. (4):
Ϫ
of the intact Al(OR)4 anion {796 and 745 cmϪ1, cf. 798
and 746 cmϪ1 in Ag(P4)2[Al(OR)4]9} as well as a P–I stretch at
168 cmϪ1 immediately follow and support the assignment of
the P5I2ϩ cation (see Discussion).
ϩ
With the awareness of the P3I6 cation we succeeded in
directly synthesising this species by reacting PI3, P2I4 and Ag-
(CH2Cl2)[Al(OR)4] according to eqn. (6).
Ag(CH2Cl2)[Al(OR)4] ϩ P2I4 ϩ PI3
P3I6[Al(OR)4], 4 ϩ AgI [R = C(CF3)3] (6)
After 12 h at Ϫ78 ЊC the obtained yellow–orange solution
(over an AgI precipitate) was investigated by variable tem-
perature 31P-NMR spectroscopy, shown in Fig. 3.
Only at very low temperatures (below Ϫ80 ЊC) was the reso-
lution of the spectra good enough to resolve P–P coupling.
P3I6ϩ was identified as main product (62% P), smaller amounts
of P2I5ϩ were also visible (28% P).19 After two weeks storage at
Ϫ30 ЊC the P2I5ϩ was no longer visible in the Ϫ90 ЊC 31P-NMR
ϩ
spectrum leaving P3I6 as the only observed P-containing
ϩ
Ag(CH2Cl2)[Al(OR)4] ϩ 2 P4 ϩ 3.5 I2
P3I6[(RO)3Al–F–Al(OR)3], 2 ϩ
PI3 ϩ P2I4 ϩ AgI ϩ ? (4)
[R = C(CF3)3]
species. This suggests that the formation of P3I6 proceeds
through P2I5ϩ. The formation of 4 as the only remaining P-
containing species according to eqn. (6) was verified in solution
(31P-NMR) and solid state (elemental analysis) on a larger
J. Chem. Soc., Dalton Trans., 2002, 500–512
501

View Article Online
Fig. 1 In situ 31P-NMR spectrum of P5I2ϩ in 3 at Ϫ90 ЊC. td = triplet of doublets, dt = doublet of triplets, tt = triplet of triplets.
1
coupling constant of JPP = 320.2 Hz (1 : 1 intensity ratio).
ϩ
In contrast, the P2I5 cation in P2I5AlI4 is only stable in the
solid state and NMR spectra of CS2 solutions of the salt
showed the signals of PI3 and AlI3.22 The solid state 31P-MAS-
22
NMR spectrum of P2I5AlI4 showed two broad resonances
centred at δ31P = ϩ114 and Ϫ142,22 confirming the assignment
of our solution values.20
To determine the activation energy for the exchange between
ϩ
the two different P atoms in P3I6 the coalescence temperature
was determined by a series of spectra ran at temperatures
between 183 and 298 K (see Fig. 4). Using the Eyring-
Fig. 2 Raman spectrum of solid 3. Bands marked with an asterisk
show small impurities due to the 100% A1 breathing modes of the very
intense Raman scatterers P4 (600 cmϪ1) and PI3 (284 cmϪ1).18
Fig. 4 Spectra of P3I6ϩ between 183 and 298 K; coalescence of P3I6ϩ at
243 K.
equation23 and the coalescence temperature of 243 K the free
energy of activation for the exchange of the two different
ϩ
phosphorus nuclei A and B in I2PA–PBI2–PAI2 (= P3I6ϩ) was
estimated as:
Fig. 3 NMR spectrum of the reaction of 1.0 P2I4 ϩ 1.0 PI3 ϩ 1.0
Ag[Al(OR)4] after 12 h at Ϫ78 ЊC in CD2Cl2 at Ϫ90 ЊC.
∆G#(PA
PB) = 38.9 kJ molϪ1
preparative scale, however, we were unable to grow single
crystals of 4. Concentrated crystallisation samples always dis-
proportionated giving P2I4 and P2I5[Al(OR)4]20 (Raman, unit
cell determination). Raman samples prepared from the yellow
solid obtained upon removal of all volatiles from the orange–
yellow CH2Cl2 solution of 4 decomposed rapidly in the beam of
the laser (1064 nm) even with wide focus and low laser energy.
Crystal structures
Details on the crystal structure solution and refinement are
included in Table 6 (see Experimental).
The Cl2P(CDCl2)2؉ cation in 1. Colourless block-like triclinic
crystals of 1 (space group P1) grew in a sealed NMR tube of a
NMR-scale reaction. Although the quality of the structure is
not very good,24 the cation—in contrast to the CF3 groups of
the anion—is well behaved with reasonable standard deviations
¯
ϩ
In CD2Cl2 solution the approximately C2 symmetric P3I6
cation (see X-ray) gave the expected doublet (δ31P = 89.2,
1Jpp = 385.5 Hz) and triplet (δ31P = Ϫ4.6, 1Jpp = 385.5 Hz) with a
ϩ
ϩ
2 : 1 intensity ratio while the approximately Cs symmetric P2I5
and shall therefore be briefly described (Fig. 5). Cl2P(CDCl2)2
may be derived from PCl4 by replacing two Cl atoms by
ϩ
cation21 gave two doublets at δ31P = 127.4 and Ϫ156.3 with a
502
J. Chem. Soc., Dalton Trans., 2002, 500–512

View Article Online
AlO3 units and in P3I6ϩ[(RO)3Al–F–Al(OR)3]Ϫ the central
fluorine atom resides on a centre of inversion (see Fig. 7).
The structural parameters of the two anions are summarised
Ϫ
in Table 1 and compared to those of the Al(OR)4 anion [Ag-
(Cl2C2H4)3ϩ salt].8
The average Al–O bond lengths (Al–O–C bond angles) in
the fluoride bridged anions are shorter (wider) than those of the
homoleptic Al(OR)4Ϫ anion accounting for the greater electron
deficiency of the aluminium centres in 1 and 2. Therefore, and
due to the short Al–F bonds, we propose that the OR ligands
ϩ
ϩ
Fig. 5 Geometry of the Cl2P(CDCl2)2 cation in Cl2P(CDCl2)2
-
Ϫ
in the (RO)3Al–F–Al(OR)3 anions are more tightly bound
[(RO)3Al–F–Al(OR)3Ϫ], 1. Anisotropic thermal ellipsoids are drawn
at the 25% probability level. Selected bond lengths (Å) and angles (Њ):
P2–Cl3 1.937(3), P2–Cl4 1.938(3), P2–C100 1.818(7), P2–C101
1.835(9), C100–Cl5 1.721(9), C100–Cl6 1.769(9), C101–Cl1 1.756(9),
C101–Cl2 1.717(10); Cl3–P2–Cl2 110.0(1), C100–P2–Cl3 108.8(3),
C100–P2–Cl4 109.2(3), C101–P2–Cl3 110.2(3), C101–P2–Cl4 108.1(3),
C100–P2–C101 110.6(4).
Ϫ
than in the Al(OR)4 anion. The space filling representation
(see Fig. 8) of the anion also shows that the bridging fluoride
and all oxygen atoms are not accessible and therefore not
available for further decomposition reactions.
Solid state packing of 2. 2 forms a layer structure, as shown in
Fig. 9, that may formally be derived from a triclinic distorted
CsCl structure, i.e. the ellipsoidal (RO)3Al–F–Al(OR)3Ϫ anions
form a (distorted) primitive cubic lattice in which the cations
occupy all cubic interstices. The P3I6ϩ cations are linked via very
weak I–I contacts at 4.070(6) Å and form endless chains along
the crystallographic a-axis.
C(D)Cl2 groups to give an almost C2 symmetric cation with a
tetrahedrally coordinated central phosphorus atom. The P–Cl
bond lengths [av. 1.938(3) Å] are in the same range as those in
the PCl4ϩ cation (1.97 Å) and the bond angles around P2 range
from 108.1 to 110.6Њ. This structure is in very good agreement
with a MP2/TZVPP optimisation.25
Computational results
To verify the nature of the P5I2ϩ and P3I6ϩ cations and to under-
؉
The P3I6 cation in 2. The yellow extremely moisture and
oxygen sensitive plates of 2 crystallise in the triclinic space
stand the thermodynamics of the formation of the P–I cations,
¯
group P1 with two molecules in the unit cell. The structure of
the P3I6ϩ cation shown in Fig. 6 is almost C2 symmetric and may
ϩ
be described by replacing two iodine atoms of a central PI4
unit by phosphorus linked PI2 groups. The P–P bond lengths
within the P3 backbone are in the range of normal single bonds
[2.210(7) and 2.216(7) Å] and similar to the ones found in
P2I5AlI4 [2.218(13) Å] or P2I4 [2.230(3) Å]. The P–I bond
lengths range from 2.361(6) to 2.435(6) Å and are on average
2.393 Å which may be compared to the 2.409 Å (av.) found in
P2I5AlI4. Two iodine atoms (I1 and I6) with longer P–I dis-
tances are involved in intramolecular contacts to P1 and P3 at
d(I–P) = 3.621(6) and 3.674(6) Å. The coordination around the
central P2 atom is distorted tetrahedral as seen by the range of
bond angles of 105.1(2) to 114.9(2)Њ and the geometry of the
two PI2 units is very similar to that of the PI2 part in P2I5AlI4
[cf. d(P–I)av. = 2.398 Å, (I–P–I)av. = 105.3Њ in 2 and 2.420 and
103.8Њ in P2I5AlI4]. However, compared to P2I4 [d(P–I)av.
=
2.474 Å] the average P–I distances in 2 are shorter by 0.081 Å.
Ϫ
Fig. 7 Structure of the (RO)3Al–F–Al(OR)3 anion in 2. The oxygen
The [(RO)3Al–F–Al(OR)3]؊ anions. This bridged anion con-
sists of a linear central Al–F–Al unit [d(Al–F) = 1.760–1.761 Å].
Each Al atom is additionally coordinated by three OR ligands
with very short Al–O bond lengths of 1.671 to 1.719 Å. The
anions are staggered with respect to the orientation of the two
(light grey), aluminium (dark grey) and central fluorine atom(s) are
drawn with anisotropic thermal ellipsoids at the 25% probability level.
For clarity, all carbon (white) and the remaining fluorine (grey) atoms
are drawn as small isotropic circles of arbitrary scale. The geometry of
the (RO)3Al–F–Al(OR)3 anion in 1 is similar (see Table 1) and
therefore not shown.
Ϫ
ϩ
Fig. 6 Geometry and solid state contacts of the P3I6 cation in P3I6ϩ[(RO)3Al–F–Al(OR)3Ϫ], 2. Anisotropic thermal ellipsoids are drawn at the
25% probability level. Selected bond lengths (Å) and angles (Њ) are given in the figure, standard deviations are 0.006 Å for P–I, 0.007 Å for P–P and
0.2Њ for the bond angles.
J. Chem. Soc., Dalton Trans., 2002, 500–512
503

View Article Online
Table 1 Comparison of the structural parameters of the (RO)3Al–F–Al(OR)3Ϫ anions in 1, 2 and the Al(OR)4Ϫ anion [Ag(Cl2C2H4)3ϩ salt]
Ϫ
Parameter
1
2
Al(OR)4
d(Al–O) range/Å
d(Al–O)av./Å
(Al–O–C)av./Њ
d(Al–F) range/Å
d(Al–F)av./Å
1.681(4)–1.719(5)
1.692
152.8
1.761(3)–1.761(3)
1.761
176.6(2)
1.671(9)–1.700(9)
1.687
154.9
1.760(4)–1.761(4)
1.761
180
1.714(3)–1.736(3)
1.725
149.5
—
—
—
(Al–F–Al)/Њ
ϩ
ϩ
computed and experimental geometries of the P2I5 and P3I6
cations in Table 2.
Table 2 shows that the length of the P–P bond is sensitive
to the level of theory chosen and therefore is a good indicator
for the quality of the computation. The longest P–P bonds were
obtained by BP86/SVP (0.14 to 0.17 Å too long). Only MP2
(and to a lesser extent MPW1PW91) gave acceptable P–P
distances that were about 0.03 Å too long. A reason for this
discrepancy may be found in the possible interaction of the
occupied lone pair orbitals at the iodine atoms [= LP(I)] with the
empty (P–P)σ* orbital [= σ*(P–P)] as shown for P2I5ϩ in Fig. 10.
This electron transfer from LP(I)
σ*(P–P) should con-
siderably elongate the P–P bond but only slightly shorten the
P–I bonds. With the basis sets employed DFT and HF-DFT
theory appeared to overestimate the interaction, as shown in
Fig. 10, and therefore gave very long P–P bonds. Consequently
only the structural parameters of the species computed with
MP2 were collected in Fig. 11.
Fig. 8 Space filling representation of the (RO)3Al–F–Al(OR)3Ϫ anion.
as opposed to the homopolyatomic phosphorus cations, we
ϩ
ϩ
ϩ
fully optimised the geometries of P3I6 (C2), P5I2 (C2v), P2I5
ϩ
ϩ
(Cs), P2I4 (C2h), PI3 (Cs), P5 (C4v), P4 (T d), Ag(P4)2 (D2h),
I2 (D6h) and AgI (C1) at the BP86/SVP, B3LYP/TZVPP and
MP2/TZVPP levels. A careful analysis of the computed struc-
tural parameters revealed that both pure DFT (BP86) and
hybrid HF-DFT (B3LYP) were not capable of reproducing
the structural parameters of the experimental geometries. The
newer DFT MPWPW91 and, to a lesser extent, the HF-DFT
MPW1PW91 levels also failed to give satisfactory geometries.
However, at the (RI-)MP2/TZVPP level geometries with
structural parameters close to the experimental data were
obtained. This may be exemplified by a comparison of the
Fig. 10 Possible interaction leading to elongation of the P–P bond in
the DFT and HF-DFT optimisations.
Fig. 9 Solid state packing of 2. All carbon and fluorine atoms (apart from those in the Al–F–Al linkages) are omitted for clarity. The F(AlO3)2 units
are shown as polyhedra and the cations as ball and stick representations (P: light grey, I: dark grey).
504
J. Chem. Soc., Dalton Trans., 2002, 500–512

View Article Online
Table 2 Comparison of the computed and experimental geometries of the P2I5ϩ and P3I6ϩ cations. Distances are given in Å
MP2b
BP86c
B3LYPb
MPWPW91d
MPW1PW91d
a
Exp.(av.)
P3I6؉
P–P
2.213
2.415
2.381
2.379
2.247
2.430
2.414
2.373
2.355
2.497
2.469
2.454
2.328
2.464
2.447
2.422
2.315
2.462
2.440
2.417
2.270
2.434
2.420
2.388
I2Pϩ(PI2)2
I2Pϩ(PI2)2
P2I5؉
P–P
2.218
2.420
2.402
2.253
2.413
2.376
2.390
2.464
2.458
2.344
2.442
2.422
—
—
—
—
—
—
I3Pϩ–PI2
I3Pϩ–PI2
a Experimental data stem from the cations in 2 and in P2I5AlI4. b TZVPP basis set. c SVP basis set. d 6-311G(2df ) for P and SDD(spdf ) for I.
Fig. 11 Fully optimised geometries, atomic energies and zero point energies (= ZPE), thermal corrections40 to the free enthalpy at 298 K (= G),
solvation energies39 in CS2 (= CS2) or CH2Cl2 (= CH2Cl2) of several binary P–I species, Ag(P4)2ϩ, P5ϩ, P4, I2 and AgI at the (RI-)MP2/TZVPP level.
Computed partial charges are given in italics, shared electron numbers are shown in italics and underlined. All species included are true minima with
no imaginary frequencies.
ϩ
The quality of the data collected in Fig. 11 is further estab-
lished by comparison to the experimental geometries of P4,
P2I4, PI3, I2 and AgI: observed and calculated parameters differ
by a maximum of Ϫ0.020 Å and ϩ1.1Њ. The geometries of
This may be compared to the experimental ∆fH(P4 (g)) of
942 kJ mol .
Ϫ1 28 ∆fH(Ag(P4)2ϩ (g)) followed from the enthalpy of
reaction of Ag(P4)2ϩ (g) giving Agϩ (g) and 2 P4 (g) (∆rH = Ϫ341
kJ molϪ1) as:
ϩ
Cl2P(CHCl2)2 and CHCl3 were also optimised at the MP2/
TZVPP level.25,26
p∆fH(Ag(P4)2ϩ (g)) = ∆fH(Agϩ (g)) ϩ [2 × ∆fH(P4 (g))]
Ϫ 341 kJ molϪ1
= 1017 ϩ (2 × 54) Ϫ 341 = 784 kJ mol؊1
Computationally assessed thermochemical properties. For the
estimation of thermochemical properties of reactions of the
presented species and as an input for Born–Fajans–Haber cycle
calculations we derived the enthalpies of formation of several
species based on the accurate MP2/TZVPP calculations.
For applications see ref. 27. ZPE energies were included in
all cases. Known enthalpies of formation were taken from the
literature.5,11 [(g) = gaseous)]
ϩ
∆fH(P3I6 (g)) followed from the enthalpy of reaction of
Agϩ (g) ϩ 1.5 P2I4 (g) ϩ 0.5 I2 (g)
Ϫ212 kJ molϪ1) as:
P3I6ϩ (g) ϩ AgI (g) (∆rH =
∆fH(P3I6ϩ (g)) = [1.5 × ∆fH(P2I4(g))] ϩ ∆fH(Agϩ(g)) ϩ
[0.5 × ∆fH(I2(g))] Ϫ ∆fH(AgI(g)) Ϫ 212 kJ molϪ1
= [1.5 × Ϫ4.7] ϩ 1017 ϩ [0.5 × 62.4] Ϫ 135 Ϫ 212
= 694 kJ mol؊1
∆fH(P5ϩ (g)) followed from the reaction enthalpy of P3ϩ with
0.5 P4 (∆rH = Ϫ120 kJ molϪ1) as:
∆fH(P5ϩ (g)) = ∆fH(P3ϩ (g)) ϩ [0.5 × ∆fH(P4 (g))]
Ϫ 120 kJ molϪ1
ϩ
∆fH(P2I5 (g)) followed from the enthalpy of reaction of
= 1006 ϩ (0.5 × 54) Ϫ 120 = 913 kJ mol؊1
eqn. (g) in Table 5 (∆rH = ϩ45 kJ molϪ1) as:
J. Chem. Soc., Dalton Trans., 2002, 500–512
505

View Article Online
∆fH(P2I5ϩ (g)) = ∆fH(P3I6ϩ (g)) Ϫ [0.25 × ∆fH(P2I4 (g))] Ϫ
[0.125 × ∆fH(P4 (g))] ϩ 45 kJ molϪ1
= 694 Ϫ [0.25 × Ϫ4.7] Ϫ [0.125 × 54] ϩ 45 = 733 kJ mol؊1
ϩ
∆fH(P5I2 (g)) followed from the enthalpy of reaction of
eqn. (f ) in Table 5 (∆rH = Ϫ49 kJ molϪ1) as:
p∆fH(P5I2ϩ (g)) = ∆fH(P3I6ϩ (g)) ϩ ∆fH(P4 (g)) Ϫ
∆fH(P2I4 (g)) ϩ 49 kJ molϪ1
= 694 ϩ 54 Ϫ (Ϫ4.7) ϩ 49 = 792 kJ mol؊1
Discussion
Establishing the nature of P5I2؉ 31P-NMR and Raman spectra
:
One comment prior to discussion: It is not adequate to cal-
culate the 31P-NMR chemical shifts of the P–I cations by the
standard procedures built in many quantum chemical program
codes due to relativistic effects.29,30 Iodine substituents at tetra-
coordinate phosphorus atoms lead to a very pronounced rel-
ativistic upfield shift that may reach several hundred ppm.29,30
Fig. 12 Computationally assessed possible unsolvated and solvated
C2v-P5ϩ structures. Only one of several examined P5ϩ(CH2Cl2)n isomers
is schematically shown (n = 0, 1, 2).
units since a similar P5 cage is unknown. However, the electro-
positive Cp2M (M = Zr, Hf ) groups Z34 in Table 3 are good
ϩ
The connectivity of the novel P5 cage within P5I2 followed
ϩ
ϩ
substitutes for the PI2 unit in P5I2 and chemical shifts of PB
unequivocally from the coupling pattern and integration of the
first order 31P-NMR spectrum (Fig. 1). The C2v symmetric P5
cage of the cation is without precedence and was not found as
part of the many polyphosphides or organopolyphosphanes
known to date.31 However, it still remained to be established
that the single phosphorus atom PA bears two iodine atoms. A
P5ϩ cation with the same C2v cage structure could be ruled out
based on the calculated high MP2/TZVPP relative energies
(including solvation energies) and the calculated 31P-NMR
shifts for these species, even when the coordination of one or
two molecules of CH2Cl2 in several possible geometries was
1
and PC as well as the smaller J(PBC) coupling constants in
CpЉ2HfP4,34 Cp*2ZrP4 34 and P5I2ϩ match very well (Table 3).
Raman spectroscopy is also in agreement with the assign-
ment of 3 as P5I2[Al(OR)4]. Observed and calculated vibra-
tional frequencies are in very good agreement and are assigned
in Table 4.
The presence of the intact Al(OR)4Ϫ anion followed from the
characteristic bands at 796, 745, 318 and 234 cmϪ1 which are
very similar to those observed in Ag(P4)2[Al(OR)4]9 (Table 4).
Ϫ
Anion decomposition and (RO)3Al–F–Al(OR)3 formation
leads to new signals and a different pattern of the bands around
ϩ
assumed.32 Moreover C2v-P5 and its CH2Cl2 solvates are not
800 cmϪ1 35
. The presence of a PI2 moiety in 3 is evident from
true minima but transition states or saddle points with imagin-
the very intense 100% band at 168 cmϪ1 which is assigned to a
ary frequencies (see Fig. 12).33
ϩ
,
PI2 stretching mode (cf. PI4AlCl4: 169 cmϪ1, A1 mode of PI4
The dicoordinate formally positively charged single PA atom
100% relative intensity36). According to the MP2 calculation,
the position of the symmetric and antisymmetric PI2 stretches
at 169 and 170 cmϪ1 are indistinguishable with the 4 cmϪ1
resolution of the spectrum and, therefore, occur at the same
position which also may account for the high intensity of this
band. 10 of the 15 expected vibrational bands of the cation
were observed and all bands of the P5 cage are strongly mixed.
The symmetric breathing mode of the P5 cage (A1, 541 cmϪ1) is
ϩ
in C2v-P5 would be expected to resonate somewhere between
ϩ1000 and ϩ300 ppm,32 depending on the number, n, of
CH2Cl2 molecules in P5ϩ(CH2Cl2)n (n = 0, 1, 2, several isomers).
However, Table 3 shows that δ31P(PA in 3) is at considerably
higher field at Ϫ89.0 ppm. This is in the middle of the values
ϩ
observed in P3I6 (Ϫ4.6 ppm) with two electron-withdrawing
ϩ
PI2 units attached to PA and that of PA in P2I5 which bears a
third iodine atom which further shifts29 the signal to higher
slightly weakened if compared to the A1 mode of P4 (600 cmϪ1
)
field (to Ϫ142 or Ϫ156.2 ppm, see Table 3). Therefore the
but higher in energy than the P–P vibrations of red phosphorus
ϩ
ϩ
position of the electron-rich substituted PI2 unit in P5I2 at
δ31P = Ϫ89.0 is very reasonable (electron donating bicyclo-
butane-P4 moiety). Moreover the rather large 1JPP couplings of
the three PA atoms in the P–I cations are in the same range,
(highest energy band at 461 cmϪ1).
Stability and decomposition of the Al(OR)4؊ anion. Formation
of 1 and 2
ϩ
ϩ
cf. 278.5 Hz in P5I2ϩ, 320.2 Hz in P2I5 and 385.5 Hz in P3I6
Ϫ
(Table 3). The chemical shifts of the atoms of the naked P4 unit
Previous attempts to prepare P2I5ϩ salts by reaction of I3ϩMF6
in 3 can only be compared to metal–tetraphosphabicyclobutane
(M = As, Sb) with P2I4 in various solvents failed and led to
Table 3 Comparison of the 31P-NMR shifts of P5I2ϩ and P3I6ϩ with other relevant species
Z/Cation
δPA
δPB
δPC
δPD
1J(PAB
)
1J(PBC
)
1J(PAD
)
CpЉ2Hf34
—
—
Ϫ89.0
Ϫ4.6
Ϫ142
Ϫ156.2
117.5
93.3
168.2
—
—
—
Ϫ219.3
Ϫ214.0
Ϫ193.9
—
—
—
—
—
—
89.2
114
—
—
278.5
—
—
193.9
201.1
152.6
—
—
—
Cp*2Zr34
—
—
385.5
—
320.2
I2Pϩ
3
P3I6ϩ 4
P2I5AlI4 (MAS)22
P2I5[Al(OR)4]a
125.5
—
a From Fig. 2.
506
J. Chem. Soc., Dalton Trans., 2002, 500–512

View Article Online
Table 4 Experimental and non-scaled calculated vibrational frequencies of 3
b
νexp. a (% intensity)
νcalc.
Symmetry
Assignment
796 (21)
745 (20)
541 (32)
Within A1 at 541?
502 (20)
Within B1 at 448?
448 (10)
385 (30)
359 (sh)
354 (80)
Al–O {cf. 798 in Ag(P4)2[Al(OR)4]}
Al–O {cf. 746 in Ag(P4)2[Al(OR)4]}
νs “breathing mode”of the P5 cage
νas P2–P1–P3
νs P5 cage
νas P2,3,1 and I1,2
νas P2,3,4,5
νas P5 cage
νas P2,3,4,5
νs P2,3,4,5
νas P5 cage
544
537
505
450
444
390
357
350
314
A1
B1
A1
B2
B1
B1
A2
A1
A1
329 (70)
318 (sh)
234 (4)
Within A1 at 168?
168 (100)
126 (4)
87 (4)
n.o.
n.o.
Anion {cf. 322 in Ag(P4)2[Al(OR)4]}
Anion {cf. 234 in Ag(P4)2[Al(OR)4]}
νas PI2 unit
νs PI2 unit
P2–P1–P3 bend
PI2 bend
Cage deformation
Cage deformation
170
169
132
86
79
56
B1
A1
B2
A1
A2
B1
a n.o. = not observed. b MP2/TZVPP frequencies.
37
decomposition and formation of PF3, I2 and MI3.22 The charac-
583 31 kJ molϪ1
]
the subsequent fluoride ion abstraction
terisation of 3 and 4 is therefore remarkable and demonstrated
from the generated P5OR molecule close by appears likely. The
Ϫ
the stability of the Al(OR)4 anion in comparison to the
fluoride ion abstraction is additionally aided by C᎐C (in C F )
᎐
4 8
Ϫ
Ϫ
classical MF6 class of ions. Moreover, Al(OR)4 salts are
soluble in CH2Cl2 even at very low temperatures (Ϫ90 ЊC)
which allowed the variable temperature NMR studies to be
undertaken. However, at temperatures above Ϫ30 ЊC to rt
and P᎐O (in P Oϩ) double bond formation. In this step the
᎐
5
fluoride bridged anion was formed. Subsequently the P5Oϩ
ϩ
cation reacted with additional P5OR present giving the P5
cation and other likely stable phosphorus containing species
following decomposition reactions of the Al(OR)4 anion
[“O᎐POR”, cf. O᎐POH;38 that presumably dismutates to P2O3
Ϫ
᎐
᎐
ϩ
which led to the formation of the fluoride bridged (RO)3Al–
or P2O5, P(OR)3 and P4(or Pred)]. The P5 cation then inserted
ϩ
F–Al(OR)3Ϫ species indicated that cations such as P5ϩ or P5I2
into the C–Cl bond of CDCl3 (to give 1) or PI3 (to give 2).
and P3I6ϩ are very strong electrophiles. A possible hypothetical
but balanced mechanism for the formation of the fluoride
bridged anion in 1 and 2 and the subsequent insertion reactions
starting from P5ϩ is outlined in Scheme 1.
The (RO)3Al–F–Al(OR)3 anion has some similarity to the
Ϫ
Sb2F11 anion. For the latter species it was shown27 that the
Ϫ
fluoride ion affinity (= FIA) of 2 SbF5 (l) (giving Sb2F11Ϫ) is
greater than that of 1 SbF5 (l) (giving SbF6Ϫ) by 70 kJ molϪ1
.
Similarly it is clear that the FIA of 2 Al(OR)3 is greater than
that of Al(OR)3. Therefore it is likely that only the combined
power of 2 Al(OR)3 molecules is sufficient to abstract a fluoride
Ϫ
Ϫ
Ϫ
ion [to form (RO)3Al–F–Al(OR)3 and not two F–Al(OR)3
anions]. The structural parameters of the (RO)3Al–F–Al(OR)3
anion suggested that the fluoride bridged species is more stable
Ϫ
(towards electrophilic attack) than the parent Al(OR)4 ion,
Ϫ 27
just as Sb2F11Ϫ is more stable than SbF6
.
؉
Thermodynamics of the reaction of iodine with Ag(P4)2
,
stability and formation of Cl2P(CHCl2)2؉, P5I2؉ and P3I6؉
ϩ
In Scheme 1 we proposed that the Cl2P(CHCl2)2 cation was
ϩ
formed by the double insertion of the Pϩ unit of P5 into the
C–Cl bond of chloroform, i.e.
P5ϩ ϩ 2 CHCl3
Cl2P(CHCl2)2ϩ ϩ P4
(7)
∆rH = Ϫ208.2 (∆rG = Ϫ157.2) [MP2/TZVPP]
A reaction according to eqn. (7) is both exothermic and
exergonic and therefore likely to be the reason for the formation
ϩ
of the Cl2P(CHCl2)2 cation and also in agreement with the
31P-NMR spectroscopic observation of excess P4.
ϩ
ϩ
To understand the formation of P5I2 and P3I6 as opposed
ϩ
to that of P5 we assessed the underlying thermochemistry in
the gas phase and in solution (CS2 and CH2Cl2) by including
approximate free solvation energies with the COSMO model.39
All values given include the ZPE and thermal corrections to the
enthalpy (H ) or the free energy (G)40 at a temperature of 298 K
[eqns. (a)–(j) in Table 5].
Scheme 1 Hypothetical decomposition pathway leading to 1 and 2
[R = C(CF3)3].
The first step of this decomposition reaction is possibly the
abstraction of one OR ligand from Al(OR)4Ϫ. Since Al(OR)3
was shown to be a very strong Lewis acid8 and the Al–F bond
is the second strongest single bond in the Periodic table [BE =
ϩ
In the gas phase the enthalpy for the formation of P5
[eqn. (a)] is unfavourable by 104.5 kJ molϪ1. However, including
entropy, solvation energies and considering the (expected)
J. Chem. Soc., Dalton Trans., 2002, 500–512
507

View Article Online
Table 5 MP2/TZVPP enthalpies and free energies40 of reaction (kJ molϪ1) in the gas phase as well as CS2 and CH2Cl2 solution (COSMO model).39
The values in parentheses include the formation of solid AgI
Reaction
∆rH(gas)
∆rG(CS2)
∆rG(CH2Cl2)
(a) Ag(P4)2ϩ ϩ 3.5 I2
(b) Ag(P4)2ϩ ϩ 3.5 I2
P5ϩ ϩ 2 PI3 ϩ 0.25 P4 ϩ AgI
ϩ104.5 (Ϫ92.4)a
Ϫ19.9 (Ϫ216.9)a
Ϫ205.8
Ϫ184.8
Ϫ188.6
ϩ48.7 (Ϫ131.2)b
Ϫ20.3 (Ϫ200.2)b
Ϫ101.2
Ϫ116.6
Ϫ123.6
ϩ14.5 (Ϫ149.4)c
Ϫ36.8 (Ϫ200.7)c
Ϫ75.8
Ϫ98.1
Ϫ113.2
P5I2ϩ ϩ 4/3 PI3 ϩ 5/12 P4 ϩ AgI
(c) P5ϩ ϩ 2 PI3
(d) P5ϩ ϩ I2
P3I6ϩ ϩ P4
ϩ
P5I2
ϩ
ϩ
(e) P4 ϩ PI2
P5I2
P3I6ϩ ϩ P4
P3I6ϩ ϩ AgI
(f ) P5I2ϩ ϩ P2I4
Ϫ49.0
Ϫ38.7
Ϫ25.7
(g) P2I4 ϩ PI3 ϩ Agϩ
Ϫ197.8 (Ϫ394.7)a
Ϫ20.3
Ϫ140.9 (Ϫ320.8)b
Ϫ35.8
Ϫ115.2 (Ϫ278.8)c
Ϫ32.5
(h) P2I5ϩ ϩ P2I4
P3I6ϩ ϩ PI3
ϩ
(i) P3I6
(j) P3I6
P2I5ϩ ϩ 0.25 P2I4 ϩ 0.125 P4
P2I5ϩ ϩ 1/3 PI3 ϩ 1/6 P4
45.4
53.8
30.9
29.2
27.1
25.3
ϩ
a The enthalpy for the process AgI (g)
AgI (s) is 196.9 kJ molϪ1
.
b The enthalpy for the process AgI (solv)
AgI (s) in CS2 is Ϫ179.9 kJ molϪ1
.
c The enthalpy for the process AgI (solv)
AgI (s) in CH2Cl2 is Ϫ163.6 kJ molϪ1
.
ϩ
precipitation of solid AgI (s) favours the P5ϩ side and eqn. (a) is
now exergonic by Ϫ131.2 to Ϫ149.4 kJ molϪ1. P5ϩ formation is
aided by solvation effects and more favourable in polar solvents
Eqn. (g) shows that the formation of P3I6 starting from
phosphorus iodides and Agϩ is thermochemically possible.
Fig. 2 suggested that P3I6 in 4 was formed through P2I5
i.e. PI3 reacted faster with Ag(CH2Cl2)[Al(OR)4] than P2I4 and
the P2I5 formed reacted then with P2I4 to give P3I6 and PI3.
Eqn. (h) is in agreement with this proposal and the small
exergonic reaction energy of only Ϫ32.5 kJ molϪ1 (in CH2Cl2)
and the low solubility of P2I4 in cold CH2Cl2 account for the
slow P3I6ϩ formation [eqn. (6) above]. P3I6ϩ has a dynamic solu-
tion structure and coalesces at only 243 K (∆G# = 38.8 kJ
molϪ1). The transition state for the process that makes the
ϩ
ϩ
,
ϩ
such as CH2Cl2. In Scheme 1 we proposed that P3I6 was
formed by the double insertion of the Pϩ unit of P5ϩ into the P–
I bond of PI3. Eqn. (c) shows that this reaction is exergonic and
therefore may be the reason for the formation of 2. According
to eqn. (b), starting with the same stoichiometry as eqn. (a), the
direct formation of P5I2ϩ is exergonic. P5I2ϩ formation may pro-
ϩ
ϩ
ϩ
ceed through the P5 stage by reaction with excess I2 [eqn. (d)]
ϩ
or alternatively through a PI2 stage by insertion of the latter
ϩ
ϩ
into the P–P bond of P4 [eqn. (e)]. The PI2 cation may have
two different P atoms in P3I6 equivalent at rt is not the D3h
formed from Agϩ and PI3 which in turn originated from P4
[from Ag(P4)2ϩ] and I2. In agreement with this proposal it was
shown that P4 and I2 initially form PI3 and only then P2I4 (with
excess P4)17 and we observed PI3 formation in eqn. (5) above.
However, solely based on thermodynamics it can not be decided
whether eqn. (d), eqn. (e) or yet another pathway is actually
occurring. It may be noted though that in the case of formation
of the Cl2P(CDCl2)2ϩ cation in 1 a PX2ϩ insertion pathway [X =
Br, I; cf. eqn. (e)] must be excluded since the resulting cation
would have to incorporate a PBr2ϩ unit which clearly is not the
case. Therefore only the intermediate participation of a phos-
phorus cation such as P5ϩ explains the formation of 1 [eqn. (7)]
and so P5I2ϩ formation as shown in eqn. (d) appears more likely,
although eqn. (e) may be an alternative approach to synthesise
symmetric three-membered ring cyclo-P3I6 owing to the high
ϩ
relative energy of this species [Erel.(cyclo-P3I6ϩ) = ϩ251 kJ molϪ1
ϩ
vs. C2-P3I6 at the MP2/TZVPP level]. Less symmetric start
geometries always led to the C2 symmetric ground state and
therefore it is most likely to assume a bimolecular exchange
process for this reaction, i.e.:
2 I2PA–PBI2–PAI2
{P6I12
}
I2PA–PAI2–PBI2ϩ etc.
ϩ
2ϩ #
However, different or additional exchange processes such as
those outlined in eqns. (i) and (j) are also likely to be involved,
especially since the experimental ∆G# and ∆G(CH2Cl2) of eqns.
(i) and (j) are on the same order of magnitude (38.8 vs. 25 to
27 kJ molϪ1).
In contrast to P2I5 (in P2I5AlI4) the P3I6 cation was previ-
ously never observed with other anions such as AlI4Ϫ. Therefore
we investigated the stability of 2, 4 and a hypothetical P3I6AlI4.
Since gaseous disproportionation reactions according to eqns.
(i) and (j) are still endothermic by 45 to 54 kJ molϪ1 this process
has to be governed by lattice potential enthalpies. Therefore the
solid state disproportionation [eqn. (8)] was assessed with three
different anions, A, by a Born–Fajans–Haber calculation11
ϩ 10,41
ϩ
P5I2
.
In the in situ NMR reaction we observed that P5I2
ϩ
ϩ
ϩ
.
had already decomposed at Ϫ40 ЊC with formation of P3I6
Eqn. (f ) gives a reasonable explanation for this reaction: at this
temperature P2I4, formed from PI3 and P4 in the precipitate,
becomes slightly soluble in CH2Cl2 and therefore is available
ϩ
ϩ
for a reaction with P5I2 giving P3I6 and P4 (or Pred since P2I4
catalyses this transformation42). Scheme 2 summarises these
findings.
ϩ
which showed that P3I6 is only stable with the large (RO)3-
Al–F–Al(OR)3Ϫ anion. With Al(OR)4Ϫ its solid state stability is
slightly unfavourable.
P3I6[A] (s)
P2I5[A] (s) ϩ 0.25 P2I4 (s) ϩ 0.5 Pred (s) (8)
[A] = (RO)3Al–F–Al(OR)3 ∆rH(solid) = ϩ2.1 kJ molϪ1
[A] = Al(OR)4Ϫ ∆rH(solid) = Ϫ0.9 kJ molϪ1
[A] = AlI4 ∆rH(solid) = Ϫ10.9 kJ molϪ1
Since P4 was never observed in solution a transformation
of the P4 molecule into red phosphorus (as catalysed by phos-
phorus iodides)42 was assumed which additionally favoured the
right hand side of eqn. (8) by 8.85 kJ molϪ1 (i.e. ∆rH for 1/4 P4
(s)
Pred (s) is 17.7 kJ molϪ1).42 Eqn. (8) provided evidence for
ϩ
Scheme 2 The reaction of Ag(P4)2ϩ with I2 in CH2Cl2 solution: Likely
reaction pathways. All free energies are given in kJ molϪ1. (s) = solid.
our observation that the P3I6 cation is only marginally stable
and that—with many attempts—we were unable to obtain
508
J. Chem. Soc., Dalton Trans., 2002, 500–512

View Article Online
single crystals of 4. Rather 4 disproportionated giving P2I5[Al-
(OR)4]20 and P2I4. To obtain clean solid 4 one therefore has to
quickly remove all volatiles of the diluted CH2Cl2 solution and
the thus obtained solid 4 appears to be metastable against a
disproportionation as in eqn. (8).
Cl2P(CDCl2)2ϩ[(RO)3Al–F–Al(OR)3Ϫ], 1, which therefore is
seen as the major product of this reaction. NMR data of
the solution: 13C-NMR (63 MHz, CDCl3, 25 ЊC): δ = 120.8
(q, CF3, JCF = 291.2 Hz); 31P-NMR (101 MHz, CDCl3, 25 ЊC): δ
= ϩ144 [s, Cl2P(CDCl2)2ϩ, 10% of P content] Ϫ500 (s, P4, 90%
of the P content).
Conclusion
Reaction leading to P3I6؉[(RO)3Al–F–Al(OR)3؊], 2
A series of reactions was performed that finally led to salts of
Ag(CH2Cl2)[Al(OR)4] (0.624 g, 0.522 mmol) was weighed into a
single piece apparatus44 and P4 (1.72 ml of a 0.605 M solution
in CS2, 1.044 mmol) was added to the solid. All volatiles
were removed in vacuum (slightly brownish residue). Solid I2
(0.465 g, 1.832 mmol) was weighed into the second bulb of
the apparatus, CH2Cl2 (10 ml) condensed onto the P4/Ag-
(CH2Cl2)[Al(OR)4] mixture and subsequently all the iodine was
sublimed onto the phosphorus containing side. The resulting
orange solution over yellow precipitate was stirred overnight at
Ϫ30 ЊC. All volatiles were removed in vacuum (theoretically
expected weight of the residue: 1.174 g, found: 1.132 g) and the
yellowish residue was extracted with 10 ml of CS2 at rt. From
this yellow CS2 solution 0.501 g of a mixture of single crystals
was obtained: orange needles of P2I4 (unit cell determination)
and yellow extremely sensitive plates of P3I6ϩ[(RO)3Al–F–Al-
(OR)3Ϫ], 2 (X-ray). A Raman spectrum of this mixture only
showed the bands of P2I4 (which in comparison to 2 is a very
good Raman scatterer18 and therefore prevents the observation
ϩ
ϩ
the P5I2 and P3I6 cations. In some of the reactions the inter-
mediate presence of naked polyphosphorus cations (e.g. P5ϩ) as
concluded from the reaction products and ab initio calculations
appeared likely. The phosphorus atoms in P3I6ϩ have an average
oxidation state of 2.33 and this species is the first subvalent
phosphorus P–X cation (X = H, F, Cl, Br, I) and a derivative
of the as yet unknown subhalide P3I5 (only P3F5 was briefly
ϩ
characterised,43 all other halides P3X5 are unknown). This P3I6
cation in 2 is only marginally stable in the solid state. The
ϩ
P5I2 cation is the first example of a binary phosphorus rich
P–X cation (X = halogen, H, organyl) and shows that this
class of cations is accessible if provided with a suitable robust
counterion such as Al(OR)4Ϫ. Classical (oxidative) approaches
to the PnImϩ (n ≥ 2) cations were earlier shown to be impossible
since reactions always led to a complete fragmentation of the
Ϫ
counterions MF6 (M = As, Sb). Moreover these classical
routes only allow the isolation of oxidised but no subvalent
species. This highlights the usefulness of the silver metathesis
route employed here in combination with the new generation of
spectator ions as exemplified by Al(OR)4Ϫ. However, the
decomposition of the latter anion in solution at temperatures
above Ϫ30 ЊC to rt showed that even more stable counterions
ϩ
of the weak P3I6 bands). NMR data of the CS2 solution:
13C-NMR (63 MHz, CS2–CD2Cl2, 25 ЊC): δ = 122.2 (q, CF3,
JCF = 290.6 Hz); 27Al-NMR (78 MHz, CS2–CD2Cl2, 25
ЊC): δ = 34.0 (s, ν1/2 = 28 Hz); 31P-NMR (101 MHz, CS2–CD2Cl2,
25 ЊC): δ = 176 (ν1/2 = 2100 Hz), 105 (ν1/2 = 1600 Hz); 31P-NMR
(101 MHz, CS2–CD2Cl2, Ϫ30 ЊC): δ = 175.5 (ν1/2 = 5 Hz), 104.5
(ν1/2 = 5 Hz); 31P-NMR (101 MHz, CS2–CD2Cl2, Ϫ70 ЊC):
δ = 175.7 (ν1/2 = 4 Hz), 103.9 (ν1/2 = 4 Hz).
Ϫ
than Al(OR)4 are needed. The fluoride bridged (RO)3Al–F–
Ϫ
Al(OR)3 anion appears to be a suitable candidate for this
purpose. Currently we are investigating possible routes to the
direct synthesis of silver salts of the latter anion.
NMR reaction leading to P5I2؉[Al(OR)4؊], 3
Experimental
Ag(P4)2ϩ[Al(OR)4Ϫ] (0.151 g, 0.114 mmol) was weighed into
an NMR tube connected to a valve. I2 (0.101 g, 0.399 mmol)
was sublimed onto the solid at 77 K after which 0.9 ml of
CD2Cl2 was condensed onto the mixture. The NMR tube
was sealed and then placed in a dry ice–isopropanol bath
and activated with ultrasound at Ϫ78 ЊC for about 10 minutes.
The initial 31P-NMR spectra were run 30 minutes later and the
13C- and 27Al-NMR spectra after storage at Ϫ80 ЊC one week
later (no decomposition visible in the 31P-NMR). 13C-NMR
(63 MHz, CD2Cl2, Ϫ90 ЊC): δ = 122.4 (q, CF3, JCF = 290.1 Hz);
27Al-NMR (78 MHz, CD2Cl2, Ϫ90 ЊC): δ = 39.5 (s, ν1/2 = 27 Hz);
31P-NMR (101 MHz, CD2Cl2, Ϫ90 ЊC): δ = 168.2 (2P, dt,
All manipulations were performed using grease free Schlenk
or dry box techniques and a dinitrogen or argon atmosphere.
A drawing of the employed “Single Piece Apparatus” has been
deposited as ESI. All apparatus were closed by J. Young valves
and the solvents were rigorously dried over P2O5 and degassed
prior to use and stored under N2 on molecular sieves (4 Å).
Yellow phosphorus was sublimed prior to use and dissolved in
CH2Cl2 or CS2 giving stock solutions which were manipulated
by syringe techniques. The silver aluminate Ag(CH2Cl2)[Al-
(OR)4] and Ag(P4)2)[Al(OR)4] were prepared according to the
literature.8,9 Raman spectra were recorded on a Bruker IFS 66v
spectrometer equipped with the Raman model FRA106 and
were obtained in the back scattering mode from solid samples
sealed in a 5 mm NMR tube (1064 nm irradiation, 4 cmϪ1 reso-
lution). NMR spectra of sealed samples were run on a Bruker
AC250 spectrometer and were referenced against the solvent
(1H, 13C) or external H3PO4 (31P) and aqueous AlCl3 (27Al).
1JP2,3–P1 = 278.5 Hz, 1JP2,3–P4,5 = 152.6 Hz), Ϫ89.0 (1P, tt, 1JP1–P2,3
=
278.5 Hz, 2JP1–P4,5 = 26.7 Hz), Ϫ193.9 (2P, td, 1JP4,5–P2,3 = 152.6 Hz,
2JP4,5–P1 = 26.7 Hz). Upon overnight warming to Ϫ40 ЊC
ϩ
the P5I2 signals vanished and, apart from other unassigned
signals of lower intensity, those of P3I6ϩ appeared as the major
P-containing peaks [δ31P(Ϫ80 ЊC) = 89.2 (2P, d, 1Jpp = 385.5 Hz),
Ϫ4.6 (1P, t, 1Jpp = 385.5 Hz)].
Reaction leading to Cl2P(CDCl2)2؉[(RO)3Al–F–Al(OR)3؊], 1
Synthesis of P5I2؉[Al(OR)4؊], 3
Ag(CH2Cl2)[Al(OR)4] (0.101 g, 0.087 mmol) was weighed into
an NMR tube and Br2 (0.047 ml of a 0.942 M solution in
CS2, 0.0435 mmol) and P4 (0.41 ml of a 0.266 M solution
in CS2, 0.109 mmol) were added at rt with the exclusion of
light. The brownish suspension was exposed to ultrasound
for 5 minutes. All volatiles were removed in vacuum and the
greyish–brown residue was dissolved in 0.5 ml of CDCl3 giving
a brownish solution over little precipitate that appeared to be
AgBr. The sample was sealed, NMR spectra were recorded
and after two months a large amount of uniform colourless
single crystals had formed. The unit cell of at least 10 crystals
was determined and all of them were shown to be
Ag(P4)2ϩ[Al(OR)4Ϫ] (1.020 g, 0.765 mmol) was weighed into
a two bulbed vessel incorporating a sintered glass frit and
stopped by J. Young valves. I2 (0.697 g, 2.746 mmol) was sub-
limed onto the solid at 77 K after which 5 ml of CH2Cl2 were
condensed onto the mixture. The apparatus was placed in a dry
ice–isopropanol bath until the solvent had thawed and was then
stored in a Ϫ80 ЊC freezer and every 30 minutes heavily shaken
for about one minute (10 times). After four days at Ϫ80 ЊC the
yellow solution over a yellow–orange precipitate was filtered at
Ϫ80 ЊC. All volatiles were then quickly removed at about 0 ЊC
(expected weight of the material: 1.717 g, found: 1.737 g) and
J. Chem. Soc., Dalton Trans., 2002, 500–512
509

View Article Online
Table 6 Crystallographic data and refinement details
the apparatus immediately transferred into a glove box. Soluble
yellow P5I2ϩ[Al(OR)4Ϫ], 3 [0.963 g (0.700 mmol), expected:
1.052 g; yield: 92%] and 0.611 g insoluble material (expected:
0.665 g) were isolated while 0.140 g were not accessible within
the flask (total: 0.963 g ϩ 0.611 g ϩ 0.140 g = 1.714 g; expected:
1.717 g). Raman spectra of 3 (see Table 4) and the insoluble
material (P4, PI3 and traces of P2I4) were recorded immediately
after sample preparation. A 31P-NMR sample of yellow 3 in
Compound
1
2
Crystal size/mm
Formula
Formula weight
0.3 × 0.4 × 0.5
C26H2Al2Cl6F55O6P
1752.91
0.05 × 0.3 × 0.3
C24Al2F55I6O6P3
2337.51
Crystal system
Space group
Triclinic
P1
Triclinic
¯
¯
P1
a/Å
b/Å
c/Å
α/Њ
12.781(3)
13.445(3)
16.561(3)
88.45(3)
69.56(3)
82.99(3)
2646.3(9)
2
10.375(2)
10.757(2)
26.591(5)
95.07(3)
92.84(3)
98.48(3)
2917.6(10)
2
2.661
3.521
0.832/0.873
51.90
175
18169
10614
2630
0.1850
0.676
0.0744
0.1636
1.088
CD2Cl2 gave the same spectrum as the one observed in the in
Ϫ
situ reaction above. Elemental analysis for P5I2ϩ[Al(OR)4
]
(C16Al1F36I2O4P5): found I 18.7, calc. I 18.5%.
β/Њ
NMR scale synthesis of P3I6؉[Al(OR)4؊], 4
γ/Њ
V/Å3
Ag(CH2Cl2)[Al(OR)4] (0.110 g, 0.095 mmol), PI3 (0.040 g,
0.097 mmol) and P2I4 (0.054 g, 0.095 mmol) were weighed
into a NMR tube attached to a valve. 1.0 ml of CD2Cl2 was
condensed onto the mixture at 77 K and the tube was flame
sealed in vacuo. Upon warming an orange solution and AgI
precipitate formed at low temperatures. The mixture was
heavily shaken for 15 minutes after which all visible phos-
phorus iodides were consumed. The tube was than stored
at Ϫ78 ЊC until the NMR spectra were recorded 12 h later
(P2I5ϩ and P3I6ϩ present). Another NMR run after 14 d storage
at Ϫ30 ЊC showed that only P3I6ϩ was present (NMR data given
in text).
Z
ρ(calc)/Mg mϪ3
2.200
0.625
µ/mmϪ1
Max./min. transmission
2θ/Њ
T /K
0.678/0.746
51.84
170
Reflections collected
Reflections unique
Reflections observed (4σ)
R(int)
20542
9396
6929
0.0415
1.087
GOOF
Final R (4σ)
Final wR2
0.1298
0.3729
1.339
Largest residual peak/e ÅϪ3
Synthesis of P3I6؉[Al(OR)4؊], 4
squares refinement. All atoms were refined anisotropically.
The P3I6 cation in 2 is disordered over two positions (0.865
ϩ
PI3 (0.205 g, 0.498 mmol), P2I4 (0.286 g, 0.502 mmol) and
CH2Cl2 (5 ml) were added into one bulb of a single piece
apparatus.44 This mixture was cooled to Ϫ78 ЊC when Ag-
(CH2Cl2)[Al(OR)4] (11.3 ml of a 0.0443 M solution in CH2Cl2,
0.500 mmol) were slowly added. An orange solution and AgI
precipitate formed at low temperatures (Ϫ78 ЊC). The stirred
mixture was allowed to reach room temperature within 30 min-
utes and after an additional 30 minutes all visible phosphorus
iodides were consumed. The apparatus was then stored for 4
days at Ϫ30 ЊC after which a 31P-NMR sample of the solution
was prepared from a special outlet. This NMR spectrum
to 0.135 occupancy) and only the major position is discussed
in the text. The anions in 1–2 exhibit rotational disorder of
the CF3 groups as well as the entire C(CF3)3 groups. Therefore
several fluorine and carbon atoms were split over two positions
resulting in site occupation factors of 20–40%. Moreover a
series of about 233 (1)/177 (2) SADI and FREE restraints had
to be used in both structures to assign reasonable structural
parameters to the CF3 groups. However, the cations in 1–2
were well behaved. The solid state packing of 2 was drawn with
Diamond 2.1 (K. Brandenburg, Crystal Impact GbR, 1998).
Relevant data concerning crystallography, data collection and
refinement details are compiled in Table 6.
ϩ
showed P3I6 to be the only P-containing species present and,
therefore, the mixture was filtered. The insoluble precipitate
(AgI, found: 0.120 g, calc.: 0.117 g) was washed several times
(back condensing of CH2Cl2) and then all volatiles of the
filtrate were quickly removed in vacuo leaving a yellow powder
of 4 (0.724 g, 79%). Raman samples prepared from this powder
rapidly decomposed in the beam of the laser with all conditions
checked. NMR data and analysis of yellow 4 in CD2Cl2:
CCDC reference numbers 162636 and 162637.
See http://www.rsc.org/suppdata/dt/b1/b103957c/ for crystal-
lographic data in CIF or other electronic format.
Computational details
13C-NMR (63 MHz, CD2Cl2, 25 ЊC): δ = 121.5 (q, CF3, JCF
=
Initial calculations and the calculation of the thermal contribu-
tions to the enthalpy and free energy were performed with
Gaussian98.45 For P the 6-311G(2df ) basis set was used and
for I the SDD effective core potential augmented with one set
of uncontracted s, p, d and f polarisation functions [= SDD-
(spdf )]. Since it was realised that MP2 geometries were needed
to give an accurate description of the species in question all
further computations were done with the program TURBO-
MOLE.46 The geometries of all species were optimised at the
(RI-)MP2 level47 with the triple ζ valence polarisation (one d
and one f function) TZVPP basis set.48 The 46 core electrons of
I were replaced by a quasi-relativistic effective core potential.
Approximate solvation energies (CS2 and CH2Cl2 solution with
εr = 2.63 and 8.92) were calculated with the COSMO model49
at the BP86/SV(P) (DFT-)level using the MP2/TZVPP
geometries. Frequency calculations were performed for all
species and all structures represent true minima without
imaginary frequencies on the respective hypersurface. For
thermodynamic calculations the zero point energy and thermal
contributions to the enthalpy or the free energy at 298 K
have been included.40 For selected species a modified Roby–
Davidson population analysis has been performed using the
MP2/TZVPP electron density.
292.1 Hz); 27Al-NMR (78 MHz, CD2Cl2, 25 ЊC): δ = 36.0 (s,
ν1/2 = 22 Hz); 31P-NMR (101 MHz, CD2Cl2, 25 ЊC): δ = 61 (ν1/2
1000 Hz) 31P-NMR (101 MHz, CD2Cl2, Ϫ90 ЊC): δ = 89.2
=
1
1
(d, JPP = 385.5 Hz, 2 P), Ϫ4.6 (t, JPP = 385.5 Hz, 1 P). C16-
Al1F36I6O4P3: found I 41.4, calc. I 41.8%.
X-Ray crystal structure determinations
Data collections for X-ray structure determinations were per-
formed on a STOE IPDS diffractometer using graphite-
monochromated Mo-Kα (0.71073 Å) radiation. Single crystals
were mounted in perfluoroether oil on top of a glass fibre and
then brought into the cold stream of a low temperature device
so that the oil solidified. Crystals of 2 given to cooled (Ϫ20 ЊC)
dry perfluoroether or hydrocarbon oil decomposed very rapidly
and from the fluorinated oil one had about 45 to 60 seconds
to select a crystal, put it on top of the glass fibre and bring it
onto the diffractometer. Of the approximately 15 crystals tested
the one reported gave the best data set. All calculations were
performed on PC’s using the SHELX97 software package.50
The structures were solved by direct methods and successive
interpretation of the difference Fourier maps, followed by least-
510
J. Chem. Soc., Dalton Trans., 2002, 500–512

View Article Online
fluorescence and dilute the amount of Raman scatterer present
(= number of scattering P5I2 cations per volume). Therefore the
Acknowledgements
ϩ
intensity of the signals in the spectrum is low. For comparison:
the 100% peak in the Raman spectrum of 3 spans only 0.1 Raman
units while P4 and PI3 with the same conditions (laser power, sample
sealed in NMR tube) have intensities of 3.0 to 5.0 Raman units for
their 100% bands. This shows that already very small impurities of
P4, PI3 or other intense Raman scatterers show in the obtained
spectrum.
We thank Prof. H. Schnöckel and Prof. J. Passmore for valuable
discussions and advice, Dipl. Chem. G. Stößer and Dipl. Chem.
J. Bahlo for recording the Raman spectra and H. Berberich and
Dr. E. Matern for recording the many low temperature NMR
spectra. Financial support from the German science foundation
DFG and the Fond der Chemischen Industrie are gratefully
acknowledged.
19 Excess P2I4 is insoluble at Ϫ90 ЊC and, therefore, invisible in the
31P-NMR at Ϫ90 ЊC.
20 We could independently verify this assignment by directly
synthesising and characterising P2I5[Al(OR)4]. We found that
P2I5ϩ[Al(OR)4]Ϫ is rapidly formed from 2 PI3 and Ag(CH2Cl2)-
[Al(OR)4], however, this is subject to ongoing research and will be
published elsewhere.
21 S. Pohl, Z. Anorg. Allg. Chem., 1983, 498, 20.
22 C. Aubauer, G. Engelhardt, T. M. Klapötke and A. Schulz, J. Chem.
Soc., Dalton Trans., 1999, 1729.
References and notes
1 (a) T. P. Martin, Z. Phys. D: At., Mol. Clusters, 1986, 3, 211;
(b) R. Huang, H. Li, Z. Lin and S. Yang, J. Phys. Chem., 1995,
99, 1418; (c) R. B. Huang, Z. Y. Liu, P. Zhang, Y. B. Zhu, F. C. Lin,
J. H. Zhao and L. S. Zheng, Chin. J. Struct. Chem., 1993, 180;
(d ) Z. Y. Liu, R. B. Huang and L. S. Zheng, Z. Phys. D: At., Mol.
Clusters, 1996, 38, 171.
23 M. Hesse, H. Maier and B. Zeeh, Spektroskopische Methoden in der
Organischen Chemie, 4th edn., 1991, Georg Thieme Verlag,
Stuttgart–New York, p. 96.
2 J. A. Zimmerman, S. B. H. Bach, C. H. Watson and J. R. Eyler,
J. Phys. Chem., 1990, 95, 98.
3 L.-S. Wang, B. Niu, Y. T. Lee and D. A. Shirley, J. Chem. Phys.,
1990, 93, 6318.
24 Although many crystals were tried and three full data sets were
recorded at 170 K none of them gave a fully satisfactory result and
the agreement factors were never better than 12.98%. However, the
crystal system and space group are well determined and the thermal
ellipsoids are normal. All trace electron density (about 1.0 to 1.3 e
4 (a) Pn and Pnϩ, n = 1–8, R. O. Jones and D. Hohl, J. Chem. Phys.,
1990, 92, 6710; (b) Pn and Pnϩ, n = 1–11, R. O. Jones and G. Seifert,
J. Chem. Phys., 1992, 96, 7564; (c) En and Enϩ, n = 1–11, E = P, As,
R. O. Jones and P. Ballone, J. Chem. Phys., 1994, 100, 4941.
5 See for example: (a) D. D. Wagman, W. H. Evans, V. B. Parker,
R. H. Schumm, I. Halow, S. M. Bailey, K. L. Churney and R. L.
Nuttal, J. Phys. Chem. Ref. Data, Suppl., 1982, 11, 2; (b) S. G. Lias,
J. E. Bartmess, J. F. Liebman, J. L. Holmes, R. D. Levin and
W. G. Mallard, J. Phys. Chem. Ref. Data, Suppl., 1988, 17, 1;
(c) CRC Handbook of Chemistry and Physics, 76th edn., editor-in-
chief D. R. Lide, CRC Press, Boca Raton, FL, 1986; (d ) http://
www.nist.gov/chemistry.
Å
Ϫ3) appeared in the vicinity of the 18 CF3 groups which suggested
manifold rotational disorder as the reason for the bad agreement
factors.
25 The Cl2P(CHCl2)2ϩ cation was fully optimised in C2 symmetry at the
MP2/TZVPP level and all structural parameters of the cation in 1
could be reproduced in the gas phase within ϩ0.02 Å and 1Њ, i.e. P–
Cl = 1.953 Å, P–C = 1.830 Å, C–Cl = 1.751, 1.758 Å, Cl–P–Cl =
110.9Њ, C–P–C = 110.4Њ. For energies see ref. 26.
ϩ
26 For the geometry of Cl2P(CHCl2)2 see ref. 25; total energy:
6 J. D. Corbett, in Progress in Inorganic Chemistry, vol. 21, editor
S. J. Lippard, John Wiley, New York, 1976, pp. 129–158.
7 S. M. Ivanova, B. G. Nolan, Y. Kobayashi, S. M. Miller, O. P.
Anderson and S. H. Strauss, Chem. Eur. J., 2001, 7, 503.
8 I. Krossing, Chem. Eur. J., 2001, 7, 490.
9 (a) I. Krossing, J. Am. Chem. Soc., 2001, 123, 4603; (b) I. Krossing
and L. van Wüllen, Chem. Eur. J., 2002, 8, 700.
Ϫ3176.10253 a.u., ZPE: 0.04672, G: Ϫ0.00544. CHCl3 (C3v):
d(C–H) = 1.087 Å; d(C–Cl) = 1.763 Å, total energy: Ϫ1417.71307
a.u.; ZPE = 0.02027, G = Ϫ0.01092.
27 S. Brownridge, H. D. B. Jenkins, I. Krossing, J. Passmore and
H. K. Roobottom, Coord. Chem. Rev., 2000, 197, 397.
28 J. A. Zimmerman, S. B. H. Bach, C. H. Watson and J. R. Eyler,
J. Chem. Phys., 1990, 95, 98.
10 I. Krossing and I. Raabe, Angew. Chem., 2001, 113, 4544.
11 Data employed for the Born–Fajans–Haber cycle calculations (all
enthalpies in kJ molϪ1): (a) Lattice potential enthalpies were taken
from the literature (AgI: 890.3)5c or estimated by Jenkins and
Passmore’s volume based modified Kapustinskii equation:12
Ag[Al(OR)4] = 364, P5[Al(OR)4] = 345, P3I6[Al(OR)4] = 334,
P2I5[Al(OR)4] = 340, P3I6[(RO)3Al–F–Al(OR)3] = 311, P2I5[(RO)3-
Al–F–Al(OR)3] = 314, P2I5[AlI4] = 398, P3I6[AlI4] = 386. The
29 It was shown convincingly that iodine substituents at tetra-
coordinate phosphorus atoms lead to a very pronounced upfield
shift due to relativistic spin–orbit coupling mediated to the nucleus
by an effective Fermi contact mechanism.30 Therefore only
relativistic 31P-NMR shifts give meaningful answers as may be
illustrated by the calculated non relativistic (NR) and relativistic (R)
ϩ 30
chemical shift of PI4
which differ by over 700 ppm: δ31PNR
=
ϩ211 and δ31PNR = Ϫ520! At present no routine for the standard
calculation of relativistic NMR shifts is implemented in the current
program codes.
Ϫ
Ϫ
thermochemical volume of Al(OR)4 is 725 Å3, Agϩ (4 Å3), AlI4
ϩ
ϩ
(244 Å3). The volumes of P5 [≈120 Å3, with S3N2ϩ(97) < P5
<
)
S3Cl3ϩ(146)],12 P2I5ϩ [=261 Å3, from V(P2I5AlI4) = 50521 and V(AlI4
30 M. Kaupp, C. Aubauer, G. Engelhardt, T. M. Klapötke and
O. Malkina, J. Chem. Phys., 1999, 110, 3897.
Ϫ
above], P3I6 [≈330 Å3, with V(P3I6ϩ) ≈ 1.5*V(P2I4)] and (RO)3Al–
ϩ
F–Al(OR)3 [=1129 Å3, from V(P3I6[(RO)3Al–F–Al(OR)3]) = 1459
31 A search of the chemical abstracts data base with Scifinder and the
Cambridge structural database CSD with the program Conquest
did not reveal a precedent for this P5-cage. However, similar C2v
symmetric MP4 units (i.e. M = Zr, Hf ) formally containing a
P42Ϫ unit are known (see ref. 34).
Ϫ
and V(P3I6ϩ) above] were assessed/estimated; (b) Enthalpies of
sublimation were taken from the literature (P4: 54.4, I2: 62.4) or were
assessed [P2I4: ∆Hsubl(P2I4) = 83.0; by computing the enthalpy of
reaction for the process: 0.5 P4 (g) ϩ 2 I2 (g)
P2I4 (g); ∆rH =
ϩ
Ϫ156.7 at the MP2/TZVPP level. ∆fH(P4 (g)), ∆fH(I2 (g)) (see below)
and ∆fH(P2I4 (s)) = Ϫ87.7 are known]; (c) Enthalpies of formation
were taken from the literature (Agϩ (g): 1017, IϪ (g): Ϫ188.6, P4
32 The chemical shifts of the naked or solvated P5 cation can be
meaningfully computed since it bears no heavy iodine atoms.
33 If not restrained by symmetry, the calculations always lead to C4v-
ϩ
ϩ
(g): 54.4, I2 (g): 62.4, P3 (g): 1006, AgI (g): 135) or assessed (P2I4
P5ϩ. Species containing C4v-P5 (and solvates) are true minima
(g): Ϫ4.7, MP2/TZVPP see above; P5ϩ (g): 913; see text).
12 H. K. Roobottom, H. D. B. Jenkins, J. Passmore and L. Glasser,
Inorg. Chem., 1999, 38, 3609.
without imaginary frequencies and are lower in relative energy.
34 O. J. Scherer, M. Swarowsky and G. Wolmershäuser, Angew. Chem.,
Int. Ed. Engl., 1988, 27, 694.
35 I. Krossing, Habilitation thesis, University of Karlsruhe, 2002.
36 C. Aubauer, M. Kaupp, T. M. Klapötke, H. Nöth, H. Pietrowski,
W. Schnick and J. Senker, J. Chem. Soc., Dalton Trans., 2001, 1880.
37 J. E. Huheey, Inorganic Chemistry, 4th edn., Harper and Collins,
New York, 1993.
13 We optimised the known4 global minima of the Pnϩ cations (n = 3, 5,
7) at the BP86/SVP, B3LYP/TZVPP and MP2/TZVPP levels and
also calculated the vibrational frequencies and Raman intensities
(only BP86) of these species. The observed new P–P stretches (480
to 630 cmϪ1) fell into a range calculated to be typical for these
cations.
38 H. Schnöckel and S. Schunck, Phosphorus, Sulfur, Silicon Relat.
Elem., 1988, 39, 89.
14 Zdirad Zak and M. Cernik, Acta Crystallogr., Sect. C, 1996, 52,
290.
39 Solvation energies were obtained from the MP2/TZVPP geometries
using the COSMO model at the BP86/SVP level and dielectric
constants for CS2 and CH2Cl2 of 2.63 and 8.92 respectively.
40 Thermal and entropic contributions to the enthalpy and free energy
were obtained by fully optimising all the species in question with
Gaussian98W at the semiempirical PM3 level. All these species also
represented true minima without imaginary frequencies at the PM3
level. ZPE’s were taken from the MP2 calculation. Since statistical
15 Similarly CS2 was used to remove excess S8 in highly electrophilic
and oxidising sulfur cation or S2Nϩ chemistry.
16 We showed that solid P4 is also visible in a conventional solution
NMR spectrometer due to the high local symmetry of the cubic
solid P4 (see ref. 9b).
17 B. W. Tattershall and N. L. Kendall, Polyhedron, 1994, 13, 1517.
Ϫ
18 As observed earlier, large counterions such as Al(OR)4 lead to
J. Chem. Soc., Dalton Trans., 2002, 500–512
511

View Article Online
thermodynamics is only little influenced by the employed geometries
and the entropic contributions mainly arise from the moment of
inertia of a species, it is believed that errors associated with this
procedure are very small if used consistently.
a fine sintered glass frit and the product side also included an
additional valve that allowed the direct preparation of NMR
samples of the reaction solution.
45 Gaussian 98, Revision A.3, Gaussian, Inc., Pittsburgh, PA, 1998.
46 TURBOMOLE, Version 5; (a) R. Ahlrichs, M. Bär, M. Häser,
H. Horn and C. Kölmel, Chem. Phys. Lett., 1989, 162, 165;
(b) M. v. Arnim and R. Ahlrichs, J. Chem. Phys., 1999, 111,
9183.
47 F. Weigend and M. Häser, Theor. Chim. Acta, 1997, 97, 331.
48 (a) A. Schäfer, H. Horn and R. Ahlrichs, J. Chem. Phys., 1992, 97,
2571; (b) A. Schäfer, C. Huber and R. Ahlrichs, J. Chem. Phys.,
1994, 100, 5829.
41 In agreement with this we succeeded in the preparation of P5Br2ϩ by
reacting PBr3, P4 and Ag(CH2Cl2)[Al(OR)4] in CH2Cl2 (X-ray,
Raman, NMR). However, this is subject to ongoing research and
will be published elsewhere. A preliminary account appeared in
Angew. Chem. (see ref. 10).
42 A. F. Holleman, E. Wiberg and N. Wiberg, Lehrbuch der
Anorganischen Chemie, 101st edn., Walter de Gruyter, Berlin–New
York, 1995, p. 1255.
43 D. Solan, U. S. C. F. S. T. I. PB Rep., 1969, (PB-187819), p. 247;
From US Gov. Res. Dev. Rep., 1970, 70, 61.
44 This U-shaped apparatus incorporates two bulbs stopped with
J. Young valves. Both bulbs are connected by a glass tube including
49 A. Klamt and G. Schürmann, J. Chem. Soc., Perkin. Trans. 2, 1993,
799.
50 G. M. Sheldrick, SHELX97, Program for crystal structure
determination, University of Göttingen, Germany, 1997.
512
J. Chem. Soc., Dalton Trans., 2002, 500–512