Proton transfer: Vs. oligophosphine formation by P-C/P-H σ-bond metathesis: Decoding the competing Br?nsted and Lewis type reactivities of imidazolio-phosphines

Article abstract of DOI:10.1039/d0dt03633a

Studies of the protonation and alkylation of imidazolio-phosphides and deprotonation of imidazolio-phosphines reveal a complex behaviour that can be traced back to an interplay of Br?nsted-type proton transfers and Lewis-type P-P bond formation reactions. As a consequence, the expected (de)protonation and (de)alkylation processes compete with reactions producing cyclic or linear oligophosphines. A careful adjustment of the conditions allows us to selectively address each reaction channel and devise specific synthesis methods for primary, secondary and tertiary imidazolio-phosphines, imidazolio-alkylphosphides, and cyclic oligophosphines, respectively. Mechanistic studies reveal that oligophosphines assemble in sequential P-P bond formation steps involving the condensation of cationic imidazolio-phosphines via σ-bond metathesis and concomitant elimination of an imidazolium ion. Imidazolio-phosphides catalyse these transformations. Computational model studies suggest that the metathesis proceeds in two stages via an initial nucleophilic substitution under expulsion of a carbene, and a subsequent proton transfer step that generates an imidazolium cation and provides the driving force for the whole transformation. As energy barriers are predicted to be low or even absent, different elementary steps are presumed to form a network of mutually coupled equilibrium processes. Cyclic oligophosphines or their dismutation products are identified as the thermodynamically favoured final products in the reaction network.

Full text of DOI:10.1039/d0dt03633a

Dalton
Transactions
View Article Online
View Journal
PAPER
Proton transfer vs. oligophosphine formation by
P–C/P–H σ-bond metathesis: decoding the
competing Brønsted and Lewis type reactivities of
imidazolio-phosphines†
Cite this: DOI: 10.1039/d0dt03633a
Mario Cicač-Hudi,a Christoph M. Feil,a Nicholas Birchall,a Martin Nieger b and
a
Dietrich Gudat
*
Studies of the protonation and alkylation of imidazolio-phosphides and deprotonation of imidazolio-
phosphines reveal a complex behaviour that can be traced back to an interplay of Brønsted-type proton
transfers and Lewis-type P–P bond formation reactions. As a consequence, the expected (de)protonation
and (de)alkylation processes compete with reactions producing cyclic or linear oligophosphines. A careful
adjustment of the conditions allows us to selectively address each reaction channel and devise specific
synthesis methods for primary, secondary and tertiary imidazolio-phosphines, imidazolio-alkylphos-
phides, and cyclic oligophosphines, respectively. Mechanistic studies reveal that oligophosphines assem-
ble in sequential P–P bond formation steps involving the condensation of cationic imidazolio-phosphines
via σ-bond metathesis and concomitant elimination of an imidazolium ion. Imidazolio-phosphides cata-
lyse these transformations. Computational model studies suggest that the metathesis proceeds in two
stages via an initial nucleophilic substitution under expulsion of a carbene, and a subsequent proton trans-
fer step that generates an imidazolium cation and provides the driving force for the whole transformation.
As energy barriers are predicted to be low or even absent, different elementary steps are presumed to
form a network of mutually coupled equilibrium processes. Cyclic oligophosphines or their dismutation
products are identified as the thermodynamically favoured final products in the reaction network.
Received 20th October 2020,
Accepted 6th November 2020
DOI: 10.1039/d0dt03633a
rsc.li/dalton
tants on elemental phosphorus or by deprotonating PH-func-
tionalised phosphines, may act as promoters. However, even if
Introduction
Construction and breakdown of the frameworks of P–P single the assembly and disassembly of complex Pn-architectures
bonds plays a key role in processes like the transformation of have been occasionally elucidated in some detail,1,2b many
elemental phosphorus into simple molecular phosphorus mechanistic issues remain still in the dark.
compounds or metal polyphosphides,1 both of which have
We have now found that “imidazolylidene-phosphinidenes”
relevance in academia and industry. The structural features I-X (Chart 1)5 provide a new and unexpected avenue to P–P
and relative stabilities of neutral,1c,2 anionic,1c,2,3 and even cat- bond formation reactions. These compounds have since their
ionic4 Pn-frameworks are meanwhile well known, and also the discovery attracted continuous attention, presumably because
initial stages in the gradual activation of white phosphorus are the synthesis from an N-heterocyclic carbene (NHC) and a
in principle understood.1c,d It is clear that phosphide anions, cyclic oligophosphine led to them being perceived as Lewis-
which can be created by the action of nucleophiles or reduc- complexes of elusive phosphinidenes.6,7 Particularly interest-
ing are P-functionalized specimens like I-H8,9 and I-Si,10
which offer a route to further derivatives through substituent
exchange.8b,10 Derivatization of a P–H bond was also reported
aInstitute of Inorganic Chemistry, University of Stuttgart, Pfaffenwaldring 55,
for II-H, which can be formally regarded as phosphinidene
complexes of CC-saturated imidazoline-2-ylidenes.11
70550 Stuttgart, Germany. E-mail: gudat@iac.uni-stuttgart.de
bDepartment of Chemistry, University of Helsinki, P.O. Box 55, 00014 Helsinki,
Finland
However, the fragmentation of I-X under transfer of a phos-
phinidene to another substrate, which would allow us to utilize
these compounds as a viable entry point in phosphorus(I)
chemistry, has to the best of our knowledge not yet been
reported. Whilst a solitary account on P–Ph-transfer from a Zn-
†Electronic supplementary information (ESI) available: Experimental pro-
cedures. Representations of NMR, IR, mass spectra, results of crystallographic
and computational studies. CCDC 1992146–1992153 and 1992155–1992158. For
ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/
D0DT03633A
This journal is © The Royal Society of Chemistry 2020
Dalton Trans.

View Article Online
Paper
Dalton Transactions
addressed by adaption of the reaction conditions, and give
mechanistic insight into an ostensible phosphinidene transfer
process that helps advance the general understanding of this
important class of reactions.
Results and discussion
Protonation and alkylation of primary imidazolio-phosphides
The pioneering study on imidazolio-phosphide protonation
reports on the reaction of a sterically congested derivative with
triflic acid (CF3SO3H).15 The choice of reactants was obviously
motivated by the desire to stabilize the cation 1a+, which was
presumed to be chemically fragile, using extremely bulky
N-aryl substituents and a weakly nucleophilic anion. We found
now that the analogues of 1a+ are readily accessible upon
addition of less shielded or even unprotected imidazolio-phos-
phides 4b–e to tetrafluoroboric or aqueous hydroiodic acid
(Scheme 1). The resulting salts are isolable in crystalline form
and are thermally stable regardless of the degree of steric pro-
tection. Their synthesis implies that cations 1+ tolerate even a
nucleophilic counter ion (I−) and water, and must thus be
viewed as chemically quite robust.
Chart 1 Molecular structures of I-X (X = Ph, CF3, H, SiMe3; R1 = alkyl,
aryl; R2 = H, Me), II-H (R1 = aryl) and imidazolio-phosphines 1a+–3a+
(Ar* = 2,6-bis(benzhydryl)-4-methyl-phenyl).
complex of I-Ph12 and the sporadic observation of related
transformations for the acyclic analogues of I-X13 indicate that
these processes are not totally unfeasible, they do not seem to
be a preferred reaction channel. This conjecture is rationalised
by recalling that the accepted representation of the bonding in
I-X through superposition of a leading canonical structure A
with large contributions from “ylidic” structures B1 and B2,5,6
which was recently validated by a computational study on
I–H,14 results in a P–C bond order between one and two and
implies a fairly robust nature of the NHC–PR skeleton. In line
with the contribution of the ylidic resonance structures, I-X
display rather a phosphide-like behaviour that is evidenced in
first reports on the protonation or alkylation to cationic
species like 1a+–3a+ 15 or the ability to bind as a ligand to two
metal centres or borane units,15,16 respectively. In order to
highlight this reactivity, we will denote compounds I-X from
now on as “imidazolio-phosphides”.
Like 4a,15 sterically less protected imidazolio-phosphides
4b–e undergo also alkylation with methyl iodide or Meerwein
salts to afford isolable salts of secondary cations (2b–e)+ or
(5c–e)+, respectively (Scheme 1). The selectivity and yield were
found to depend on the reaction conditions; the best results
were obtained when imidazolio-phosphide was added to a
solution of the electrophile in a polar solvent like acetonitrile.
The isolated salts were characterized with spectroscopic
and analytical data and, in several cases, with single-crystal
X-ray diffraction studies (Fig. 1 and ESI†). The crystals contain
arrays of well separated anions and cations. The coordination
geometry at the phosphorus atoms is pyramidal. There is no
evidence for P–H⋯X hydrogen bonding, but several specimens
display weak X⋯P contacts (2.91 to 3.37 Å for X = F and 3.44 to
3.85 Å for X = I) below the sum of the van der Waals radii
(P⋯F 3.36 Å, P⋯I 3.94 (ref. 19)). Since the halogens are
oriented roughly opposite to the P–C bond (X⋯P–C 153 to
174°), these contacts might be regarded as weak σ-hole inter-
actions.20 By analogy to 1a+,15 the P–Cimi distances in the sec-
ondary cations exceed those in imidazolio-phosphides 4b–e 8b,c
Cations 1a+–3a+ are analogues of tertiary carbenio-phos-
phines whose positive charge resides mainly on the
N-heterocyclic substituent,17 but their chemical properties
remain, in contrast to those of the tertiary congeners, largely
unexplored. We reasoned that cationic alkylphosphines like
2a+ and 3a+ might react with bases, like other secondary
phosphines, under deprotonation to afford rare imidazolio-
alkylphosphides,6b,18 and that the alkylation/deprotonation
sequence can offer an evidence-based mechanistic explanation
for previously reported8b substitution reactions. As we reveal
here, a thorough study of the acid/base chemistry of the repre-
sentatives of I–H and their protonation and alkylation pro-
ducts allowed us to prove these conjectures, but unveiled as
well more complex reactivity patterns that are characterized by
a rivalry between proton transfer and P–P bond formation pro-
cesses. The latter proceed under σ-bond metathesis with the
elimination of imidazolium cations and are to the best of our
knowledge the first reactions involving formal “R–P” transfer
from un-complexed imidazolio-phosphines.‡ Our investi-
gations show further how each reaction channel can be
Scheme 1 Synthesis of salts of primary ((1b–e)+) and secondary ((2b–
e)+, (5b–e)+) imidazolio-phosphines. Reagents and conditions: (i) 1.5
equivs. of aq. HI or aq. HBF4, toluene, −78 °C – r.t: (ii) 1 equiv. of MeI or
[R23O][BF4], MeCN, rt; R1 = 2,6-iPr2C6H3 (Dipp, 1b+–5b+), 2,4,6-Me3C6H2
(Mes, 1c+–5c+), iPr (1d+–5d+), Me (1e+–5e+).
‡A related reaction might have occurred upon treatment of I-Ph with BPh3, but
was not investigated further; see ref. 16a.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Dalton Transactions
Paper
Deprotonation of secondary imidazolio-phosphines
The acidic behaviour of secondary imidazolio-phosphines is
best illustrated by the outcome of a screening of the reactivity
of 5b[BF4] towards various bases. Analysis of the 31P{1H}
(Fig. 2) and 1H NMR spectra allowed us to establish in all reac-
tions the formation of mixtures of imidazolio-ethylphosphide
6b, cyclic oligophosphines (EtP)n (n = 3–5),2a and imidazolium
ion 7b+.
Analogous behaviour was also confirmed for (2b–e)[X] (X =
I, BF4) and led us to summarize the observed reactivity using
the generic equation shown in Scheme 2. The products can be
rationalized as arising from two competing reactions, viz. (i)
deprotonation of the starting materials to afford 6b and 8b–e
and (ii) cyclocondensation under P–P bond formation and
cleavage of (7b–e)[X] to produce (RP)n (R = Me, Et; n = 3–5). We
want to note that the combination of electrophilic alkylation
Fig. 1 Molecular structure of the cations of 1d[I] (left) and one of the
two crystallographically independent cations of 2c[BF4] (right) in the
crystal. Thermal ellipsoids of heavy atoms are drawn at the 50% prob-
ability level. The P-bound hydrogen and methyl substituents of 2c+ are
disordered over two positions, only one of which (with occupancy 0.75)
is shown.
but match the adjacent P–Calkyl distances (see Table 1). In view and subsequent base-induced deprotonation provides a viable
of this similarity, we address all P–C bonds in cations 2+ and mechanistic explanation for previously observed electrophilic
5+ as single bonds.
substitution reactions8b,18 at the primary imidazolio-phos-
The 31P NMR chemical shifts of cations (1b–e)+ (δ31P = phide 4b.
−165.7 to −176.1) and (2b–e)+ (δ31P = −115.2 to −95.3) are
The balance between both reaction channels is strongly
similar to those of primary or secondary methyl phosphines influenced by the choice of the base. Deprotonation to yield 6b
(δ31P −163.5 for MePH2 and −99.5 for Me2PH21) and the is generally preferred with anion bases like potassium tert-but-
1H NMR data indicate that secondary cations 2+ and 5+ with oxide (t-BuOK), potassium hexamethyldisilazide (KHMDS),
N–iPr- and N–Dipp-substituents are, like neutral phosphines, potassium hydride (KH), or lithium diisopropylamide (LDA).
at ambient temperature configurationally stable on the NMR The highest selectivity for 6b (x > 0.9) was achieved with
timescale. Based on their spectroscopic and structural fea- KHMDS, while the other anion bases gave less phosphide
tures, we address cations 1+, 2+ and 5+ as imidazolio-substi- (x = 0.70–0.84) at the expense of an increased amount of
tuted primary and secondary phosphines and consider them (EtP)n. On a preparative scale, t-BuLi reacted similarly to
analogues of the well-known tertiary α-cationic phosphines KHMDS when the reaction was carried out at low temperature,
that have received some attention in coordination chemistry and its application enabled isolation of the crystalline samples
and catalysis.17
of 6b and 8b in decent yields (58–65%). The molecular struc-
Table 1 Selected distances for imidazolio-phosphines (1a–e)[X], (2b,c)
[X], 3a[BF4], 5b[BF4], (19a,c)[BF4] and imidazolio-alkylphosphides 6b and
8b
R1
R2
R3
P–Cimi
P–Calkyl
1a[OTf]a
1b[I]
1b[BF4]
1c[I]
1d[I]
1e[I]
Arb
—
—
—
—
—
—
Me
Me
CPh3
Et
—
—
—
—
—
—
—
—
—
—
—
1.840(4)
1.845(7)
—
—
—
—
—
—
Dippc
Dippc
Mesd
iPr
1.816(4)
e
Fig. 2 31P{1H} NMR spectra of the reaction mixtures formed upon treat-
ment of 5b[BF4] with KHMDS (black trace), LDA (purple trace), NaBH4
1.825(2)
1.83(2)
Me
2b[I]
Dippc
Mesd
Arb
1.837(3)
1.828(3)
1.845(2)
1.828(2)
1.770(4)
1.757(4)
1.750(2)
1.845(2)
1.843(2)
1.805(4) (red trace) and Et3N (green trace) in MeCN (* = unidentified product).
e
2c[BF4]
3a[BF4]a
5b[BF4]
6bf
1.921(2)
Dippc
Dippc
e
Et
1863(4)
1.868(4)
1.855(2)
8b
19a[BF4]
19c[BF4]
Dippc
Dippc
Dippc
Me
Me
Et
—
Me
Me
1.823(2)
e
a Data from ref. 15. b Ar = 2,6-di-benzhydryl-4-methyl-phenyl. c Dipp =
2,6-di-isopropyl-phenyl. d Mes = 2,4,6-trimethyl-phenyl. e Unambiguous
determination unfeasible due to crystallographic disorder. f Data for
two crystallographically independent molecules.
Scheme 2 Generic equation describing the reaction of secondary imi-
dazolio-phosphines (2b–e)[X], 5b[X] with bases (X
N(SiMe3)2−, NiPr2−, t-BuO−, H−, BH4−, Et3N; 0 ≤ x ≤ 1).
= I, BF4; B =
This journal is © The Royal Society of Chemistry 2020
Dalton Trans.

View Article Online
Paper
Dalton Transactions
of initiating P–P bond formation. A first corroboration of this
conjecture was obtained during attempts to optimize the pro-
tocol for alkylating 4b–e. In particular, we noted that adding
imidazolio-phosphide to an excess alkylating agent gave high
yields of imidazolio-phosphines 2+ and 5+ while condensation
to furnish (7b–e)+ and (RP)n (R = Me, Et; n = 3–5) occurred with
high selectivity, or even exclusively, when the addition of the
reagents was reversed. We presume that the condensation is in
the first case inhibited by the immediate quenching of the imi-
dazolio-phosphide by the electrophile, and in the second case
promoted by the attack of a local excess of still unreacted 4b–e
in the cationic alkyl phosphine formed. The ability of 6b to cat-
alyse the condensation of 5b+ to yield (PEt)n and 7b+ was con-
firmed by independent control experiments (see the ESI†).
The progress of the reaction of 5b[BF4] with a catalytic
amount of 6b (17 mol%) at −35 °C was monitored with NMR
spectroscopy, further enabling some light to be cast on the for-
mation mechanism of the cyclic oligophosphines. We found
that cationic 5b+ is nearly instantly transformed into a mixture
of transient intermediates, which then decay slowly at −35 °C
and more rapidly and quantitatively at ambient temperature to
afford (PEt)n as the final products. Assessment of the 31P{1H}
NMR spectra (see Fig. 4 and the ESI†) allowed decomposition
of the signals of the transient species into two AX- and four
ABX-type patterns attributable to newly formed di- and tripho-
sphines. The diphosphines were identified as the two diaster-
eomers of 9+ (Chart 2) from the 31P and 1H,31P HMQC spectra,
which confirmed the presence of two ethyl-, a hydrogen- and
Fig. 3 Molecular structure of one of the two crystallographically inde-
pendent molecules of 6b in the crystal. For clarity, hydrogen atoms are
omitted and N–Dipp substituents are drawn using
a wire model.
Thermal ellipsoids are drawn at the 50% probability level. Only one of
the two positions of a disordered iPr group (occupancy 0.682(14)) is
shown.
tures in the crystal (Fig. 3 and ESI†) resemble those of known
imidazolio-phosphides. The P–Cimi distances (1.750(2) and
1.757(4) Å) match those in P–H- (1.752(1) to 1.772(1) Å (ref. 8))
or P–R-substituted congeners (R = Ph, CF3, 1.763(6) to 1.794(3)
Å (ref. 5b)), whereas the P–Calkyl distances (1.855(2) to 1.868(4)
Å) unexpectedly exceed that in cation 5b+ (1.805(4) Å). The
31P NMR chemical shifts of 6b and 8b (δ31P −61.4 and −39.9
in THF-d8) roughly match those of phenyl and CF3-substituted
analogues (δ31P −23.0 to −53.4 (ref. 8b)).
Cyclic oligophosphines (EtP)n became the major products
of the reaction of 5b[BF4] with sodium borohydride (NaBH4),
which was chosen as a more soluble hydride source than KH.
Whilst imidazolio-phosphide 6b was under these conditions
still detected as a minor component (x = 0.09), an exclusive for-
mation of (EtP)n (x = 0) occurred when the reaction was carried
out using an equimolar amount of triethyl amine (Et3N).
Reaction monitoring revealed that (EtP)3 and (EtP)4 formed in
this reaction and in all other reactions studied as kinetically
favoured cyclocondensation products. In contrast, (EtP)5 is a
secondary product that formed later but became the dominant
species in the final product mixture. The ring metathesis
process implied by these transformations was repeatedly
reported in the literature2a,22 and is driven by the decline in
ring strain with increasing ring size. Because the rate of ring
scrambling is slow compared to the consumption of the start-
ing material, we consider it a follow-up process that is not
related to the initial base reaction of 5b+.
In accord with the equation in Scheme 2, which implies
that the cyclo-condensation does not result in a net consump-
tion of the base, we established that the quantitative formation
of (EtP)n occurs as well upon exposure of 5b[BF4] to a sub-stoi-
chiometric quantity (approximately 10 mol-%) of Et3N.§ In
view of the stability of base-free solutions of imidazolio-phos-
phines, these findings clearly identify the cyclocondensation
as a base-catalysed process. Adopting this hypothesis, we
reasoned that an imidazolio-phosphide should also be capable
Fig. 4 Expansion of the 31P{1H} NMR spectra of a mixture of 5b[BF4]/6b
(5 : 1) in CD3CN (a) after 6 h of reaction time at −35 °C, (b) after 7 d at
ambient temperature. The red trace is the result of a simulation of the
low-frequency region as a superposition of the signals of the reactants,
(PEt)4, and two AX and three ABX signal patterns of transient di- and tri-
phosphines (see the ESI† for details). Note that the signal of 6b in (b) is
shifted because of a marked temperature dependence and split into
several lines as a consequence of H/D-exchange with the solvent. The
asterisk denotes the signal of an impurity present during the whole
reaction.
§We presume that even smaller quantities are sufficient, although this was not
practically tested.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Dalton Transactions
Paper
is obviously produced by the deprotonation of 5b+ at an early
stage, and transient 9+ and 10+ implies that the previously dis-
cussed route to (EtP)n is still viable. The available data allow
no decision if the new transient species (12 and 13) are
involved in an alternative or possibly competing pathway
leading to (EtP)n or merely in a reversible but ‘unproductive’
side reaction.¶
Chart 2 Proposed molecular structures of di- and triphosphine inter-
mediates 9+, 10+ (R1 = Dipp, R2 = Et), 12 (R = H), 13 (R = PHEt, 2 diaster-
eomers) and diphosphene-carbene adduct 11 (ref. 23) (Ter = 2,6-
Mes2C6H3).
Deprotonation of primary imidazolio-phosphines
A dichotomy between proton transfer and P–P bond formation
was also established for the base reactions of primary cationic
phosphines (1b–e)+. Condensation and cleavage of imidazo-
lium salts can be induced, as in the case of 2+ and 5+, by the
addition of catalytic quantities of Et3N or imidazolio-phos-
phides 4b–e (10–30 mol%),§ or by stockpiling the active reac-
tion mixtures with additional cationic phosphine. All reactions
proceed rapidly at ambient temperature to furnish mixtures of
soluble linear phosphines PnHn+2 (n = 1–3)2,24 and intractable
reddish precipitates. The composition of the solids could not
be elucidated, but their colour suggests the presence of Pn-
frameworks as in red phosphorus or insoluble polyphos-
phides. Added imidazolio-phosphide was not immediately
consumed in the condensation, but disappeared during
ageing of the reaction mixtures, presumably via an undi-
sclosed follow-up reaction.
Targeted deprotonation of a primary imidazolio-phosphine
occurred, as in the case of (2b–e)+, upon addition of cationic
phosphines to solutions of strong anion bases like KHMDS or
t-BuOK. The formation of varying amounts of phosphine (PH3)
as a by-product suggests, however, that condensation is not
completely suppressed.∥
As in the case of the secondary congeners, we can summar-
ize the two-sided reactivity of 1b–e[X] in a single generic
equation (Scheme 3). Note that the condensation yields in this
case no cyclic oligophosphines, but proceeds as a dismutation
to afford a mixture of “hydrogen-rich” linear phosphines and
“hydrogen-poor” insoluble polyphosphorus compounds of still
unknown composition (the formulation as neutral Pm was
made to ensure a correct stoichiometric balance and is not
intended to represent the actual composition). The dismuta-
tion consumes no base and can thus be initiated using a cata-
lytic amount of an imidazolio-phosphide or an external base
like Et3N.
an imidazolio-substituent on the P2 unit (see the ESI†). A
similar analysis of the signals attributed to the transient tri-
phosphines indicated that all phosphorus atoms are alkylated
and one carries an additional proton, and we assign these
species thus to the four possible diastereomers of 10+.
Monitoring the progress of the reaction over time revealed that
the concentration of 9+ increases and decays more rapidly than
that of 10+, whereas the amount of imidazolio-phosphide does
not vary significantly during the whole period.
Altogether, these findings imply that 5+ reacts first via step-
wise chain elongation to yield linear di- and triphosphines,
which then cyclize to the final products. The absence of any
P1-building block other than 6b during the later stages of the
reaction suggests strongly that the imidazolio-phosphide is
engaged in the chain elongation process, and the persistence
of this species at the end of the reaction substantiates that this
engagement is catalytic. It should be noted that 9+ is a conge-
ner of the conjugate acid of zwitterion 11 (Chart 2) that has
been addressed as the NHC-adduct of a diphosphene and
reacts with H2O under cleavage of the carbene and hydration
of the resulting PP double bond rather than undergoing proto-
nation to an analogue of 9+.23 This different behaviour is pre-
sumably caused by the steric encumbrance imposed by the
bulky terphenyl groups in 11, which induce a marked weaken-
ing of the bond connecting P2 and the imidazole units.
Observations made during studies on the electrophilic
methylation of 4e prompted us to study also the suitability of a
primary imidazolio-phosphide as a catalyst for condensation
processes. By monitoring the titrations of 4b with 5b[BF4]
using 31P NMR spectroscopy, we found that the addition of a
sub-stoichiometric amount of the reagent resulted in the for-
mation of 6b along with a new product identified spectroscopi-
cally as imidazolio-diphosphide 12 (Chart 2). The reaction of
more than one equivalent of 5b+ with 4b gave rise to the for-
mation of detectable amounts of 9+ and 10+, which rapidly
decayed to produce (PEt)n (n = 3–5), and a yet-unobserved tri-
phosphine labelled as 13 (Chart 2, two diastereomers). The
signals of 12 and 13 decayed eventually, leaving (PEt)n and imi-
dazolio-phosphides 4b and 6b as the only persistent phos-
phorus-containing reaction products.
NMR-spectroscopy monitoring of the reaction of 1d[BF4]
with Et3N in acetonitrile at −60 °C revealed the presence of
transient intermediates beside the persistent products obser-
vable at room temperature. The simulation of the 31P{1H}
NMR spectra and comparison with literature data2 allowed us
to identify the prevailing intermediates as linear tetra- and
pentaphosphines and cyclopentaphosphine (PH)5 (see the
¶Another, yet undisclosed side-reaction is held responsible for the formation of
minor amounts of Et(H)P–P(H)Et (2 diastereomers) as
a further transient
The outcome of the titration experiments shows that the
cyclocondensation is mechanistically more complex when 4b
species.
∥Observation of PH3 as an exclusive product is attributable to the known liability
rather than 6b is used as a catalyst. The detection of 6b, which of P2H4 to react with strong bases under disproportionation, see ref. 2b.
This journal is © The Royal Society of Chemistry 2020
Dalton Trans.

View Article Online
Paper
Dalton Transactions
Computational studies and mechanistic considerations
To gain a deeper understanding of the P–P bond formation
processes, we set out to model the behaviour of the sterically
least protected imidazolio-phosphine 1e+ computationally.
DFT calculations were carried out at the B3LYP-D3BJ/def2-
tzvp//B3LYP-D3BJ/def2-svp level of theory, which reliably repro-
duces the experimentally observed structural features of 4e
and cations 1e+ and 5e+. Solvation effects were simulated using
a polarizable continuum model (PCM) with solvent parameters
for THF. Anions of salt-like reactants were neglected to cut
down the computational cost.
Starting with an account of the energetics of possible reac-
tions of 1e+ (or (15-P0)+, Scheme 4) with imidazolio-phosphide
4e (or 14-P0), we predict the degenerate proton transfer
between both reactants (eqn (1)n = 0, ΔG0298 = 0.0 kcal mol−1) as
thermodynamically less favourable than condensation to give
imidazolio-diphosphide 14-P1 (eqn (2)n=0, ΔG0298 = −5.8 kcal
mol−1, see Table 2). The negative Gibbs free energy of the
proton transfer from 1e+ to 14-P1 (eqn (1)n=1, ΔG2098 = −2.1
kcal mol−1) identifies the latter as a slightly stronger base than
4e. The products formed in such a step – diphosphine (15-P1)+
and 4e – may undergo further condensation to furnish an imi-
dazolio-triphosphide, 14-P2 (eqn (2)n=1, ΔG0298 = −6.4 kcal
mol−1), which is the thermodynamically preferred phos-
phorus-containing product of the reaction between 4e and two
cations 1e+. In a similar manner, further reaction of 14-P2 with
Scheme 3 Generic equation describing the reaction of primary imida-
−
zolio-phosphines (1b–e)[X], 5b[X] with bases (X = I, BF4; B = N(SiMe3)2
t-BuO−, Et3N; n = 1–3, 0 ≤ x ≤ 1).
,
ESI;† the composition of further minor species remains
unknown). The signals of higher oligophosphines (PnHm with
n > 3) broadened and decayed rapidly when the temperature
was increased. Triphosphine P3H5 remained visible for some
time at ambient temperature, but decomposed eventually to
produce P2H4 and PH3. The base-induced dismutation of
higher oligophosphines PnHm is well established,2 and we con-
sider its observation in the present case once more a follow-up
process that is unconnected with the initial reaction of 1d+.
Regardless of our failure to detect any imidazolio-substituted
oligophosphines, we presume that the observed Pn-frameworks
are formed in a similar manner to that in the condensation of
secondary imidazolio-phosphines.
1e+ under proton transfer (eqn (1)n=2, ΔG0298 = −5.5 kcal mol−1
)
and exergonic condensation (eqn (2)n=2
, = −13.6
ΔG0298
kcal mol−1) may afford tetraphosphide 14-P3. This species is
obviously slightly less basic than 4e (see Table 2), but the exer-
gonic chain extension (eqn (2)n=3, ΔG0298 = −7.1 kcal mol−1
)
remains still feasible. The formation of chains with more than
five phosphorus atoms was not investigated, but seems concei-
vable as well.
Taken together, the individual reaction steps discussed so
far describe in principle a reaction cascade for the assembly of
linear oligophosphines by sequential chain-elongation steps.
The Gibbs free energies of the gross reactions, which can be
represented as condensations of 4e with different amounts of
imidazolio-phosphine 1e+ (eqn (3), Scheme 5), indicate that
the formation of Pn-chains is thermodynamically favourable.
Scheme 4 Generic equations describing the proton transfer between
imidazolio-(oligo)phosphides 14-Pn and 1e+ (eqn (1)) and condensation
of the resulting products (15-Pn)+ and 4e to yield chain-extended imida-
zolio-oligophosphides (eqn (2)). Note that 4e and 1e+ can also be
denoted as 14-P0 and (15-P0)+, respectively. The calculated standard
Gibbs free energies ΔG0298 are listed in Table 2. For brevity, 1,3-dimethyl-
imidazolio-fragments are represented as “nhc(+)”.
Table 2 Standard Gibbs free energies ΔG0298 (in kcal mol−1) for the reactions in Schemes 4 and 5 computed at the PCM(THF)-B3LYP-D3BJ/def2-
tzvp//PCM(THF)-B3LYP-D3BJ/def2-svp level of theory
Reaction
n
(1)
(2)
m
(3)a
(4)a
(5)b
(6)
0
1
2
3
4
0.0
−5.8
−6.4
−13.6
−7.1
1
2
3
4
−5.8 (−5.8)
−12.2 (−6.1)
−25.8 (−8.6)
−32.9 (−8.2)
−7.9 (−7.9)
−17.7 (−8.8)
−25.5 (−8.5)
−32.6 (−8.1)
−2.1
−5.5
0.3
−23.7 (−6.0)
−32.1 (−6.7)
−54.2 (−21.6)
−37.6
−45.3
0.3
b
a Values in parentheses denote the average gain in ΔG0298 per P–P bond formed (ΔG0298, avg = ΔG0298/m). Values in parentheses denote ΔG0298 of
the ring closure step nhc(+)–(PH)y–PH2 → (PH)y+1 + nhcH(+)
.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Dalton Transactions
Paper
−12.7 kcal mol−1 associated with the generation of P2H4 and
P3H5, respectively, are more negative than the Gibbs free ener-
gies for ring closure steps yielding (PH)3 and (PH)4 (Table 2;
note that (PR)3,4 (R = Me, Et) are the kinetic products resulting
from the condensation of 2b+ and 5b+), but do not reach the
level predicted for the formation of (PH)5. It must be borne in
mind, however, that the dismutation energetics depends deci-
sively on the nature of the “hydrogen-poor” co-product. Since
the reaction becomes more exergonic when a more stable
phosphorus allotrope than P4 is formed, the values of ΔG2098
(dismut) represent only thresholds, and an absolute assess-
Scheme 5 Gross reactions describing the formation of linear imidazo-
lio-oligophosphides 14-Pn (eqn (3)) and –oligophosphines (15-Pm)
+ (eqn
(4)), cyclic oligophosphines (PH)m+1 (eqn (5)) and dismutation products ment remains unfeasible as long as the co-product cannot be
(P4 and P2H4/P3H5, eqn (6), m = 3, 4) by condensation of 1e+ or 1e+/4b,
respectively. Note that the phosphorus atom count in all reactions is m
+ 1. The calculated standard Gibbs free energies ΔG0298 are listed in
Table 2. For brevity, 1,3-dimethylimidazolio-subsitituents are denoted as
“nhc(+)”.
specified.
Having demonstrated the thermodynamic feasibility of
various condensation processes, we also wanted to elucidate
viable kinetic pathways. To this end, we set out to identify the
transition states for key reaction steps, starting with the inter-
action between imidazolio-phosphide 4e and cationic phos-
phine 1e+. Probing various approaches of the reactants, we
The average gain in the free energy per P–P bond formed could locate the transition states for two reaction channels.
increases from ca. 6 kcal mol−1 for 14-P1 and 14-P2 to some The molecular geometry of TS 1 (Fig. 5) features a nearly sym-
8 kcal mol−1 for tetra- and pentaphosphines 14-P3 and 14-P4, metric P⋯H⋯P array and spawns the degenerate proton trans-
respectively.
fer between both reactants (eqn (1) with n = 0). A low Gibbs
An intriguing aspect turns up when we combine the for- free activation energy (ΔG‡298 = 7.1 kcal mol−1) for this process
mation of 14-Pn from 4e and 1e+ embodied by eqn (3) with the is in good accord with rapid proton scrambling at ambient
subsequent protonation of the resulting zwitterion by one temperature.
more molecule of 1e+. Imidazolio-phosphide 4e consumed in
The transition state TS2 embodies the energetically highest
the first reaction is then recovered and adopts thus the role of point on the reaction coordinate of a substitution process and
a catalyst, and the overall reaction can be represented accord- is noteworthy for two aspects. First, the coordination geometry
ing to eqn (4) as the condensation of imidazolio-phosphines at the reaction centre deviates from the common (pseudo) tri-
1e+ to yield oligophosphines (15-Pn)+ and an appropriate gonal–bipyramidal pattern in which the incoming and leaving
amount of 7e+. The Gibbs free energies indicate that the for- groups occupy axial positions, and appears rather as tetragonal
mation of di- and triphosphines (15-P2)+ and (15-P3)+ is associ- pyramidal with a phosphorus lone-pair adopting the axial and
ated with an even larger (by 2 to 3 kcal mol−1) driving force the exchanging substituents the adjacent basal positions.
than the assembly of zwitterions 14-P2 and 14-P3, whereas the Second, both exchanging groups (imidazolio-phosphide and
differences level out for cationic and neutral tetra- and penta- NHC) are not charged but neutral, rendering the whole
phosphorus species.
process more similar to a ligand displacement at a metal
Cyclic oligophosphines (PH)n (n = 3–5) can be formally complex25 than a typical substitution in an element-organic
obtained by an additional exergonic ring-closing condensation molecule.
of linear tri- to penta-phosphines (ΔG0298 = −6.0, −6.6 and
−21.6 kcal mol−1 for three- to five-membered rings, respect-
ively, Table 2), which makes the products thermodynamically
more stable than linear oligophosphines with the same
number of phosphorus atoms. The low Gibbs free energies for
ring-closure to (PH)3 and (PH)4 indicate obviously the ring
strain in the products,2,22 while the formation of (PH)5 benefits
from a strong driving force.
Last but not least, we looked also into the energetics of the
reactions producing P2H4 and P3H5. Assuming the formation
of P4 as a by-product, producing these species from 1e+ accord-
ing to eqn (6) is likewise thermodynamically favoured over
simple condensation to linear imidazolio-oligophosphines
Fig. 5 Computed
(PCM(THF)-B3LYP-D3BJ/def2-tzvp//PCM(THF)-
(Table 2). The driving force ΔG0298(dismut) for the creation of
the final products through formal dismutation of a linear pre-
cursor (15-Pm)+ (m = 4, 5) is measured by the difference in
B3LYP-D3BJ/def2-svp) molecular geometries of transition states TS1
(left) and TS2 (right) for different reactions between 1e+ and 4e. Selected
distances (in Å): TS1: H1–P1 1.422, P1–H3 1.757, H3–P2 1.756, P2–H2
ΔG0298 of eqn (6) and (4). The calculated values of −12.4 and 1.422; TS2: C1–P1 2.087, P1–P2 2.453, P2–C2 1.817.
This journal is © The Royal Society of Chemistry 2020
Dalton Trans.

View Article Online
Paper
At first glance, the coupling of 4e and 1e+ under the P–P
Dalton Transactions
bond formation and expulsion of NHC 16e associated with
TS2 (eqn (7), Scheme 6) is predicted to be distinctly endergonic
(ΔG0298 = +10.9 kcal mol−1). However, the strongly basic
carbene may abstract a proton from the coupling product
(15-P1)+ (eqn (9), ΔG2098 = −16.7 kcal mol−1) or a second mole-
cule of 1e+ (eqn (8), ΔG2098 = −18.8 kcal mol−1) without the
need to pass an energetic barrier. The last alternative is more
exergonic, and the concatenation of P–P-coupling and proton
transfer results in an overall exergonic condensation (eqn (10),
ΔG0298 = −7.9 kcal mol−1) representing the first chain-extension
step of the gross reaction described by eqn (4) with m = 1.
While a chain growth mechanism implied by eqn (10) is at
first glance in accord with our experimental findings, the
Fig. 6 Results of
a
relaxed potential energy surface scan (at the
B3LYP-D3BJ/def2-svp level) for the reaction between 16e and (15-P1)+
to give 1e+ and 4e using C(NHC)⋯P vector r1 as an independent variable.
(a) Plot of the reduced electronic energy E’ = E + 1293 Hartree (blue)
and its derivative dE‘/dr1 (red) vs. r1. (b) Computed molecular geometry
at an intermediate scan point (r1 = 2.427 Å, marked with a green square).
Gibbs free energy of activation (ΔG‡298(TS2) = 32.9 kcal mol−1
)
seems hardly compatible with an instantaneous reaction of
primary cationic phosphines (1b–e)+ under ambient con-
ditions (note that these species react obviously too fast to
render diphosphine intermediates observable by NMR). A solu-
tion to this dilemma emerged when we studied the possible
interactions between NHC 16e and cationic diphosphine 15-
P1. The simulation of different approaches of the reactants
allowed us to identify once again two reaction channels. The
first one represents the aforementioned proton transfer to the
NHC (eqn (9)). The second transformation reverses the reac-
tion (7), but differs from the route via TS2 in following a
“regular” pathway in which the NHC approaches from a direc-
tion opposite to the P–P bond to be cleaved (Fig. 6).
Intriguingly, our calculations depict both processes as barrier-
less transformations that do not require traversing a transition
state. Resorting to the concept of microscopic reversibility, the
“forward” reaction (7) of 1e+ and 4e to yield imidazolio-dipho-
sphine (15-P1)+ and 16e can then likewise be considered
barrier-less.
plified by eqn (10) and the analogous reactions of one mole-
cule of 1e+ with a homologue (15-Pn−1)+ generally as two-step
processes that are catalysed by imidazolio-phosphide 4e. The
first step involves the nucleophilic attack of 4e on a cationic
phosphine (1e+ or (15-Pn−1)+) under liberation of an NHC and
represents a Lewis-acid/base reaction mimicking the ligand
substitution in a metal complex. In the second step, the NHC
is sequestered by Brønsted-type proton transfer from 1e+ (note
that this reaction may compete for higher oligophosphines
(15-Pn)+ (n = 3, 4) with the deprotonation of the condensation
product). Both reaction steps require passing only low energy
barriers, or may even proceed without a barrier.
The formation of the simplest cyclic oligophosphine (PH)3
might in principle occur by ring closure of the cationic tripho-
sphine (15-P2)+ under cleavage of imidazolium ion 7e+, but our
failure to locate a suitable transition state makes this pathway
unlikely. Gratifyingly, we found that the target product can be
accessed via exergonic ring-closing fragmentation of tetraphos-
Similar pathways for the reaction of 4e with 1e+ were identi-
fied for selected chain growth steps yielding imidazolio-oligo-
phosphines (15-Pn)+ (n = 2–4) from 4e and appropriate precur-
sors (15-Pn−1)+. We therefore suggest describing the P–P bond
forming condensation of two cationic phosphines 1e+ as exem-
phide 17-P3 (eqn (11), Scheme 7, ΔG0298 = −14.9 kcal mol−1
)
arising from the deprotonation of the terminal PH2-unit of the
cation (15-P3)+ by a transient NHC. This reaction branch has a
similar driving force (ΔG0298 = −18.1 kcal mol−1) to the barrier-
less quenching of the NHC by 1e+ (eqn (8), ΔG2098 = −18.8 kcal
mol−1) but is kinetically less favoured (ΔG‡298 = +9.3 kcal
mol−1). The accumulation of the cyclotriphosphine as a major
product is thus deemed unlikely. Nonetheless, an analogous
reaction allows explaining the formation of alkylated cyclotri-
phosphines (PR)3 (R = Me, Et) from imidazolio-tetraphos-
phines nhc(+)–(PR)3–PHR, where deprotonation at the remote
Scheme 6 Elementary reaction steps involving nucleophile-induced
P–P bond formation under the expulsion of a carbene (eqn (7)) and
carbene protonation by 1e+ (eqn (8)) or (15-P1)+ (eqn (9)). Combining
eqn (7) and (8) adds up to a condensation of two molecules of 1e+, eqn
(10). 1,3-Dimethylimidazolio-fragments are denoted as “nhc(+)”.
Scheme 7 Ring-closing fragmentation of tetraphosphide 17-P3. For
brevity, 1,3-dimethylimidazolio-fragments are represented as “nhc(+)”.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Dalton Transactions
Paper
phosphorus atom (relative to the imidazolio-substituent) is the (see the ESI†) reveals that the coupling to yield a methylated
only feasible reaction mode.
analogue of 9+ (Chart 2, R1 = R2 = Me) and an NHC is less
Finally, we succeeded in elaborating the model reactions endergonic (ΔG0298 = 2.7 kcal mol−1) and the ensuing protona-
leading to P2H4 and P3H5 (Scheme 8). The key is the formation tion of the carbene by a further molecule of 2e+ is less exergo-
of specific stereoisomers of zwitterionic tri- and tetraphos- nic (ΔG0298 = −9.0 kcal mol−1) than the corresponding reactions
phides 18-R (R = H, PH2) with a phosphide unit in the of 1e+ and 4e (eqn (7) and (8), see Table 2). However, because
γ-position of the imidazolio group. These species have dis- both effects largely cancel each other out, the Gibbs free
tinctly lengthened Pα–Pβ bonds (Fig. 7a) and can, like 11 (ref. energy for the condensation of two secondary phosphines 2e+
23), be viewed as the charge-transfer complexes of elusive (ΔG0298 = −6.2 kcal mol−1) closely matches the value estab-
diphosphenes HPvPR (R = H, PH2) with 4e. Their reaction lished for the coupling of the primary homologues 1e+ (eqn
with imidazolio-oligophosphide 14-P2 (or higher congeners) (10), ΔG0298 = −5.8 kcal mol−1), and falls also below that of the
proceeds under formal dihydrogen transfer to the diphos- degenerate proton transfer between 2e+ and 8e. Moreover, the
phene unit (see Fig. 7b) to afford 4e, oligophosphines H2P– identification of a permethylated (R1 = R2 = Me) analogue of 12
PHR (R = H, PH2), and a new diphosphene that is assumed to as a most favourable P–P coupling product accessible from 2e+
be kinetically unstable and to decay in the follow-up reactions and 4e is in accord with the observed formation of 12 during
that were not analysed further.
the reaction of 2b with 4b (vide supra).
Oligophosphides 18-R can be obtained by NHC-induced
In summary, the computational studies suggest that the
deprotonation of phosphines (15-P2)+ and (15-P3)+, even if they complex reaction patterns of the primary and secondary imida-
are less stable than their tautomers 14-P2 (ΔΔG0298(18-H) = zolio-phosphines and the conjugate phosphides can be deli-
5.8 kcal mol−1) and 14-P3 (ΔΔG2098(18-PH2) = 7.7 kcal mol−1), neated into sequences of elementary steps in which the prota-
respectively. Alternatively, 18-H can form in a balanced (ΔG0298 gonists act either as Brønsted-acids or bases in proton transfer
≈ 0) barrier-less reaction through the NHC-induced fragmenta- events, or as Lewis acids or bases (or electrophiles and nucleo-
tion of an imidazolio-pentaphosphine (15-P4)+ (see the ESI†). philes) in reactions involving the formation or breakage of a
The generation of P2H4 and P3H5 from 18-R and 14-P2 requires P–P bond. All steps are characterized by low-energy transition
passing moderate kinetic barriers (ΔG‡298 = 17.6, and 15.5 kcal states, or may even proceed without energetic barriers. These
mol−1).
preconditions allow smooth reactions under ambient con-
A preliminary exploration of the reactions between second- ditions and the flexible concatenation of several elementary
ary cationic phosphine 2e+ and its conjugate phosphide 8e steps to more complex transformations, which finally sum up
to a diverse network that can be treated as one system of
mutually coupled dynamic equilibria. The outcome of the reac-
tion is in this setting controlled by thermodynamics, and the
composition of any equilibrium mixture can easily be modi-
fied by changing the stoichiometric boundary conditions, e.g.
by adding reagents. Moreover, the calculations allow us to also
identify the possible reaction pathways, which explain both the
sequential formation of linear imidazolio-oligophosphines
and the generation of cyclic oligophosphines (represented by
(PH)3 in the computational model) and P2H4/P3H5 as the pre-
ferred final products (Fig. 8).
Scheme 8 Model reaction describing the formation of P2H4 (R = H)
and P3H5 (R PH2) from transient zwitterions 18-R. 1,3-
Dimethylimidazolio-fragments are denoted as “nhc(+)”.
=
The computational modelling reveals some corollaries that
help understanding the factual reactivity of the primary and
secondary imidazolio-phosphines and -phosphides. First of
all, any direct encounter of the specimens of both types can
give rise to two different reaction modes, viz. (i) Brønsted-type
proton transfer and (ii) P–P bond forming condensation. The
second pathway is generally more favourable and proceeds in
two steps via an initial Lewis-type bond formation reaction
and ensuing protonation of the concurrently formed NHC by
an imidazolio-phosphine with the release of an imidazolium
ion (eqn (7)–(9), Scheme 6). The exergonic second step delivers
the energy input that propels the endergonic first step and
affords a net driving force for the whole process. Moreover, it
regenerates an imidazolio-(oligo)phosphide and enables thus
the reaction to proceed in an autocatalytic and repetitive
manner to form chains of P–P bonds. A practical consequence
of this autocatalysis is that the specific alkylation of imidazo-
Fig. 7 Computed
(PCM(THF)-B3LYP-D3BJ/def2-tzvp//PCM(THF)-
B3LYP-D3BJ/def2-svp) molecular structures of (a) 18-H, (b) the tran-
sition state of the reaction between 18-H and 14-P2 leading to the for-
mation of P2H4. Selected distances (in Å): 18-H: C1–P1 1.803, P1–P2
2.485, P2–P3 2.125; TS(18H + 14-P2): P1–P2 2.127, P3–P4 2.163, P1–H1
2.001; H1–P3 1.543, P2–H2 1.615, H2–P4 1.773.
This journal is © The Royal Society of Chemistry 2020
Dalton Trans.

View Article Online
Paper
Dalton Transactions
Fig. 8 Computed (PCM(THF)-B3LYP-D3BJ/def2-tzvp//PCM(THF)-B3LYP-D3BJ/def2-svp) minimum energy reaction pathways illustrating (a) the
+
stepwise assembly of linear imidazolio-oligophosphines (15-Pn)+ (n = 2–4) by the condensation of (15-Pn−1
)
with 1e+ in the presence of 4e as a
catalyst, (b) ring closure and dismutation reactions leading to (PH)3 (in green) and P2H4/P3H5 (in dark/light red) respectively. Dashed lines denote the
transformations that may proceed according to the relaxed potential energy scans without energy barriers.
lio-phosphides to produce secondary cationic phosphines and natural driving force arising from a coupled proton transfer is
the deprotonation of the latter to yield P-alkylated imidazolio- here unavailable and the reaction needs to be coerced by pro-
phosphides are only feasible when the coexistence of both viding a suitable external energy input.
reactants is avoided, e.g. by rapid quenching of the initial reac-
tants with an efficient alkylating agent or a strong base.
As a further conclusion, we note that adding an excess of
NHC will shift the equilibria of eqn (7)–(9) to the side of the
Our DFT studies provide also some comments on the role imidazolio-phosphides. Because parallel trends apply as well
of NHCs in the reaction network. First of all, the substitution to the related reactions of oligophosphines (15-Pn)+, NHCs
step compares to the known reaction patterns of tertiary carbe- should be viable reagents for the disassembly of polypho-
nio-phosphines,17,26 and its course backs thus the operative sphine frameworks into imidazolio-phosphides as ultimate
definition15 of imidazolio-phosphines as NHC-adducts of fragments. The validity of this forecast has in fact been proved
elusive phosphenium ions. Despite this analogy, the reactivity long before in the known syntheses of imidazolio-phosphides
of the primary and secondary imidazolio-phosphines differs in from NHCs and cyclic oligophosphines.6,12 Building on this
a qualitative sense from that of their tertiary analogues, as the insight, we can now consider these reactions as reciprocally
Dalton Trans.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Dalton Transactions
Paper
related to the condensation of the imidazolio-phosphines reagent was chosen because of its ability to serve as a source of
studied in this work: both types of transformations embody in both a nucleophilic hydride and an electrophilic aluminium
essence a shift of the same coupled equilibria to one side or centre for sequestering the released carbene. The NMR-spec-
the other, allowing one to annul the synthesis of imidazolio-phos- troscopic analysis of the volatiles collected after distillation
phides by quenching the NHC with a suitable acid. The reported confirmed the formation of secondary phosphines (Me2PH,
degradation of an imidazolio-phenylphosphide by triphenylbor- Me(Et)PH) as main products. The samples contained neither
on16a suggests that Lewis-acids may have the same effect.
organic by-products traceable to imidazolio-substituents nor
boron- or aluminium-containing impurities, thus confirming
that the sequestering strategy succeeded, but produced minor
amounts (6–14%) of diphosphines. We envisage that these by-
products originate from a dismutation between different phos-
phorus-containing species or from a SET reaction between a
transient NHC and a phosphorus electrophile,27 but further
studies are needed to clarify this issue. Considering that the
imidazolio-phosphides used can be produced directly from
elemental phosphorus,8c their conversion into secondary (di)
phosphines represents a new approach for the preparation of
simple organophosphorus compounds that bypasses the use
of environmentally problematic1c,d chlorophosphines.
Dialkylated imidazolio-phosphines
Imidazolio-monoalkylphosphides 6b and 8b react smoothly
with alkyl halides or Meerwein salts to afford imidazolio-di-
alkylphosphines 19[BF4] (Scheme 9) that are readily isolated as
crystalline salts. Depending on the choice of reagents, pro-
ducts with like and different alkyls can be prepared.
Crystallographic studies of 19a[BF4] and 19c[BF4] reveal that
the phosphorus atoms in the cations (Fig. 9) exhibit a similar
pyramidal coordination and show comparable P–Cimi and
P–Calkyl distances to their secondary congeners (Table 1).
Dimethylated 19a[I] was also obtained directly from primary
imidazolio-phosphide 4b, by adding first an excess of methyl
iodide and then, after monoalkylation to 2b+ was completed,
adding Hünig’s base to produce imidazolio-methylphosphide
8b that was immediately trapped by the leftover alkylating
agent. The formation of an appreciable quantity (≈25%) of
(MeP)n (n = 3–5) indicates, however, that the unwanted conden-
sation of transient 2b+ is not completely suppressed.
Conclusions
Primary and secondary imidazolio-phosphines and conjugate
imidazolio-phosphides exhibit a diverse chemical behaviour
due to their ability to act as Brønsted as well as Lewis acids or
bases, respectively. The reactivity patterns associated with both
modes show both similarities and divergences. Similarities are
observed in the reactions of imidazolio-phosphides with
Brønsted acids or electrophiles to give primary or secondary
cationic phosphines, respectively; divergence is exemplified in
the behaviour of the cationic phosphines, which react with
Brønsted bases under attack at the P–H- bond and with Lewis
bases under attack at the P–C(imidazole) bond. Mutual reac-
tions between the imidazolio-phosphines and -phosphides
proceed as P–C/P–H σ-bond metatheses under the formation
of a new P–P bond and release of a P-free imidazolium ion,
and can be delineated into an initial nucleophilic (Lewis-type)
substitution involving P–P bond formation under the liber-
ation of an NHC and an ensuing (Brønsted-type) protonation
of the carbene. DFT studies suggest that this last step provides
the net driving force for the whole process, and that its kinetic
barrier is not caused by a high degree of kinetic inertness of
the P–C bonds in the phosphines but rather an unfavourable
energy balance of the initial substitution step. The observed
reactivity patterns and proposed reaction mechanism back the
perception of imidazolio-phosphines as NHC adducts.15,17,26
The energetically favourable interplay of nucleophile-
induced P–P bond formation and P–H-deprotonation using
strong bases enables primary/secondary imidazolio-phos-
phines to react with a sub-stoichiometric amount of a conju-
gate phosphide in cascade-like sequences of σ-bond meta-
theses. The final products are imidazolium salts and imid-
azole-free polyphosphorus species, and the net reaction quali-
fies as the phosphide-catalysed polycondensation of the cat-
ionic phosphines. The reaction cascades embody the systems
Anticipating that cationic imidazolio-dialkylphosphines
display a similar liability towards nucleophilic displacement of
the NHC unit to that of diphenylated derivatives,17 we explored
the reactions of 19a[BF4] and 19c[BF4] with LiAlH4. This
Scheme 9 Alkylation of P-alkyl imidazolio-phosphides.
Fig. 9 Molecular structure of the cation of 19c[BF4] in the crystal. For
clarity, hydrogen atoms are omitted and N-Dipp substituents are using a
wire model. Thermal ellipsoids are drawn at the 50% probability level.
The P-alkyl substituents are disordered over two positions, only one of
which (with occupancy 0.676(4)) is shown. The selected distances are
listed in Table 1.
This journal is © The Royal Society of Chemistry 2020
Dalton Trans.

View Article Online
Paper
Dalton Transactions
of coupled dynamic equilibria, which revert the well-known
syntheses of imidazolio-phosphides6 from NHCs and cyclic oli-
gophosphines. Due to their exergonic nature and low kinetic
barriers, these processes compromise the establishment of
regular acid/base equilibria between the cationic phosphines
and their conjugate bases, or between one of these species and
an external acid or base. Hence, synthetically useful appli-
cations of the (de)protonation of imidazolio-phosphines or
-phosphides using external bases or acids are only feasible
under reaction conditions suppressing any direct interaction
of both phosphorus-containing reactants.
Even if the application of phosphorus-heteroatom σ-bond
metatheses for the assembly of polyphosphorus frameworks is
known,28 the reactions reported here are to the best of our
knowledge the first documented examples involving carbon as
the heteroatom. Moreover, they provide a rare occasion to
obtain insight into the assembly of (RP)n-chains, illustrating a
growth mechanism by sequential attachment of P1-building
blocks that involves no free phosphinidenes or phosphenium
ions but is best described as a substitution-like process
mediated by a nucleophilic catalyst. If one considers that the
reported reactions of imidazolio-phosphines relate closely to
the dismutation of other (poly)phosphines and the assembly
and disassembly of bond networks in elemental phosphorus
or polyphosphides,1b,2 the mechanistic details revealed in this
work can improve the understanding of the base- or nucleo-
phile-induced growth and disintegration of P–P bonded frame-
works in a more general sense, and thus help further advan-
cing a basically important topic in phosphorus chemistry.
Recognition of the insight gained may also aid developing new
rational pathways to functionalise imidazolio-phosphines.
A. Rossin and M. Peruzzini, Chem. Rev., 2010, 110, 4178;
(c) M. Scheer, G. Balázs and A. Seitz, Chem. Rev., 2010, 110,
4236; (d) J. E. Borger, A. W. Ehlers, J. C. Slootweg and
K. Lammertsma, Chem. – Eur. J., 2017, 23, 11738.
2 (a) M. Baudler and K. Glinka, Chem. Rev., 1993, 93, 1623;
(b) M. Baudler and K. Glinka, Chem. Rev., 1994, 94, 1273.
3 H. G. v. Schnering and W. Hönle, Chem. Rev., 1988, 88, 243.
4 T. Köchner, T. A. Engesser, H. Scherer, D. A. Plattner,
A. Steffani and I. Krossing, Angew. Chem., Int. Ed., 2012, 51,
6529.
5 Reviews: (a) T. Krachko and J. C. Slootweg, Eur. J. Inorg.
Chem., 2018, 2734; (b) A. Doddi, M. Peters and M. Tamm,
Chem. Rev., 2019, 119, 6994.
6 (a) A. J. Arduengo, H. V. R. Dias and J. C. A. Calabrese,
Chem. Lett., 1997, 26, 143; (b) A. J. Arduengo,
J. C. Calabrese, A. H. Cowley, H. V. R. Dias, J. R. Goerlich,
W. J. Marshall and B. Riegel, Inorg. Chem., 1997, 36,
2151.
7 For an earlier report describing the synthesis of a ben-
zannulated cyanophosphinidene heterocycle without
making reference to the carbene-phosphinidene picture
see: A. Schmidpeter, W. Gebler, F. Zwaschka and
W. S. Sheldrick, Angew. Chem., Int. Ed. Engl., 1980, 19,
722.
8 (a) K. Hansen, T. Szilvási, B. Blom, S. Inoue, J. Epping and
M. Driess, J. Am. Chem. Soc., 2013, 135, 11795;
(b) A. M. Tondreau, Z. Benkő, J. R. Harmer and
H. Grützmacher, Chem. Sci., 2014, 5, 1545; (c) M. Cicač-
Hudi, J. Bender, S. H. Schlindwein, M. Bispinghoff,
M. Nieger, H. Grützmacher and D. Gudat, Eur. J. Inorg.
Chem., 2016, 649.
9 For
a
4-lithio-derivative, see: Y. Wang, Y. Xie,
M. Y. Abraham, R. J. Gilliard, P. Wei, H. F. Schaefer,
P. v. R. Schleyer and G. H. Robinson, Organometallics, 2010,
29, 4778.
Conflicts of interest
There are no conflicts to declare.
10 A. Doddi, D. Bockfeld, T. Bannenberg, P. G. Jones and
M. Tamm, Angew. Chem., Int. Ed., 2014, 53, 13568.
11 O. Lemp, M. Balmer, K. Reiter, F. Weigend and C. von
Hänisch, Chem. Commun., 2017, 53, 7620.
12 T. Krachko, M. Bispinghoff, A. M. Tondreau, D. Stein,
M. Baker, A. W. Ehlers, J. C. Slootweg and H. Grützmacher,
Angew. Chem., Int. Ed., 2017, 56, 7948.
13 (a) L. Weber, B. Quasdorff, H.-G. Stammler and
B. Neumann, Chem. – Eur. J., 1998, 4, 469; (b) L. Weber,
U. Lassahn, H.-G. Stammler and B. Neumann, Eur. J. Inorg.
Chem., 2005, 4590.
14 Z. Kelemen, R. Streubel and L. Nyulászi, RSC Adv., 2015, 5,
41795.
Acknowledgements
The authors acknowledge financial support from the German
Research Foundation (DFG) through grant no. GU 415/17-1.
We further thank B. Förtsch for elemental analyses, and
J. Trinkner and Dr W. Frey (both from Institute of Organic
Chemistry, University of Stuttgart) for the recording of mass
spectra and the collection of X-ray data sets, respectively. The
computational studies were supported by the state of Baden-
Württemberg through bwHPC and the German Research
Foundation (DFG) through grant no INST 40/467-1 FUGG
(JUSTUS cluster).
15 L. Liu, D. A. Ruiz, F. Dahcheh and G. Bertrand, Chem.
Commun., 2015, 51, 12732.
16 (a) A. J. Arduengo, C. J. Carmalt, J. A. C. Clyburne,
A. H. Cowley and R. Pyati, Chem. Commun., 1997, 981;
(b) O. Lemp and C. v. Hänisch, Phosphorus, Sulfur Silicon
Relat. Elem., 2016, 191, 659.
Notes and references
1 (a) B. M. Cossairt, N. A. Piro and C. C. Cummins, Chem. 17 (a) Y. Canac, C. Maaliki, I. Abdellahab and R. Chauvin, New
Rev., 2010, 110, 4164; (b) M. Caporali, L. Gonsalvi,
J. Chem., 2012, 36, 17; (b) M. Alcarazo, Acc. Chem. Res.,
Dalton Trans.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Dalton Transactions
Paper
2016, 49, 1797; (c) Y. Canac, Chem. – Asian J., 2018, 13,
1872.
18 During writing of this manuscript, we became aware of a
report on the synthesis of an imidazolio-alkyl-phosphide
D. Scheschkewitz, V. Chandrasekhar and A. Jana, Chem.
Sci., 2018, 9, 4235; (b) D. Dhara, S. Das, P. Kalita, A. Maiti,
S. K. Pati, D. Scheschkewitz, V. Chandrasekhar and A. Jana,
Dalton Trans., 2020, 49, 993.
through a similar approach: J. E. Rodriguez Villanueva, 24 (a) P. Junkes, M. Baudler, J. Dobbers and D. Rackwitz, Z.
M. A. Wiebe and G. G. Lavoie, Organometallics, 2020, 39,
3260.
19 S. Alvarez, Dalton Trans., 2013, 42, 8617.
Naturforsch., B: Anorg. Chem., Org. Chem., 1972, 27, 1451.
25 Regarding the analogy between main group elements and
transition metals, see: P. P. Power, Nature, 2010, 463, 171.
20 (a) P. Politzer, J. S. Murray and T. Clark, Phys. Chem. Chem. 26 K. Schwedtmann, R. Schoemaker, F. Hennersdorf,
Phys., 2010, 12, 7748; (b) P. Politzer, J. S. Murray,
G. V. Janjic and S. D. Zaric, Crystals, 2014, 4, 12;
A. Bauzá, A. Frontera, R. Weiss and J. J. Weigand, Dalton
Trans., 2016, 45, 11384.
(c) P. Politzer, J. S. Murray, T. Clark and G. Resnati, Phys. 27 Z. Dong, C. Pezzato, A. Sienkiewicz, R. Scopelliti, F. Fadaei-
Chem. Chem. Phys., 2017, 19, 32166. Tirani and K. Severin, Chem. Sci., 2020, 11, 7615.
21 L. Maier, P. J. Diel and J. C. Tebby, in CRC Handbook of 28 (a) K.-O. Feldmann and J. J. Weigand, J. Am. Chem. Soc.,
Phosphorus-31 Nuclear Magnetic Resonance Data, ed.
J. C. Tebby, CRC Press, Boca Raton, FL, 1991, ch. 6.
22 M. Baudler, J. Hahn and E. Clef, Z. Naturforsch., B: Anorg.
Chem., Org. Chem., 1984, 39, 438.
2012, 134, 15443; (b) R. Schoemaker, K. Schwedtmann,
A. Franconetti, A. Frontera, F. Hennersdorf and
J. J. Weigand, Chem. Sci., 2019, 10, 11054; (c) C. Taube,
K.
Schwedtmann,
M.
Noikham,
E.
Somsook,
23 (a) D. Dhara, P. Kalita, S. Mondal, R. S. Narayanan,
K. R. Mote, V. Huch, M. Zimmer, C. B. Yildiz,
F. Hennersdorf, R. Wolf and J. J. Weigand, Angew. Chem.,
Int. Ed., 2020, 59, 3585.
This journal is © The Royal Society of Chemistry 2020
Dalton Trans.