The desire to improve on existing catalytic processes and find new catalytic systems continues to drive synthetic chemists to create novel ligand systems. For example, ongoing efforts to find post-metallocene polymerization catalysts for alkenes (Gibson & Spitzmesser, 2003
) have led to the introduction of a large class of salicylaldiminate and pyrrolaldiminate ligands (see, for example, Yoshida et al.
; Tsurugi & Mashima, 2006
; Cui et al.
; Yang et al.
). The combination of an anionic pyrrolate moiety with other donors is, however, less common. In an effort to explore routes to new N
- and N
-chelate ligands based on pyrrolates, we have recently introduced the families of pyrrole–phosphine oxide [see (I) in the scheme as an example] and pyrrole–phosphine ligands [see (II) in the scheme as an example] (Broomfield, Boschert et al.
; Broomfield, Wright et al.
). These ligands are attractive as the nature of the groups on the P atom [phenyl in both (I) and (II)] can be varied at a late stage of the synthesis by appropriate choice of the phosphine starting material. The introduction of other groups onto the pyrrole ring is another potential method for tuning the properties of the ligand system. Exploring the reactivity of pyrrole–phosphine oxide ligands with early transition metals has shown that the phosphine oxide acts as a well behaved ligand with these metals. We report here on efforts to use 2-(diphenylphosphinomethyl)pyrrole, (II)
, with early transition metals.
Reaction of (II)
in tetrahydrofuran (THF) was carried out with the intention of forming the bis-ligand complex (III)
). However, crystallization of the reaction mixture led to the isolation of crystals with an approximately stoichiometric mixture of cocrystallized (I)
, denoted (I)/(II)
(Fig. 1). Synthesis of (II)
is complicated by the reactivity of the phosphine with oxygen, and it is possible that the metal complex is not involved in formation of the cocrystals. The occupancy of O1 was refined freely, establishing that the site is largely occupied [occupancy = 0.861 (6)].
Figure 1 The structure of (I)/(II), showing the atom-numbering scheme. Displacement ellipsoids are drawn at the 50% probability level. H atoms except H1 and H2 have been omitted for clarity. The dashed line indicates a hydrogen bond.
The molecular geometry of (I)
here (Table 1) is very similar to that observed when the same molecule crystallizes in the absence of (II)
(Broomfield, Wright et al.
). The geometric data observed for (I)
are also in the range anticipated for this molecule. More interesting are the intermolecular interactions between (I)
(Fig. 2 and Table 2). Hydrogen bonding between two molecules of (I)
leads to solid-state dimer formation. This is in contrast with the supramolecular arrangement seen in crystals of (I)
alone, where infinite one-dimensional chains are formed by hydrogen bonding between adjacent molecules (Broomfield, Wright et al.
). The O1
[symmetry code: (i) −x
+ 1, −z
] separation observed here [2.821 (4) Å] is slightly shorter than that seen in the infinite chain arrangement [2.868 (3) Å]. In addition to this pairing hydrogen-bond interaction, each molecule of (I) also hydrogen bonds to one molecule of (II)
, with an O1
N2 distance of 2.790 (3) Å. The result of this hydrogen-bond framework is the formation of an overall zero-dimensional 2:2 supramolecular structure.
Table 1 Selected geometric parameters (Å, °) for (I)/(II)
Figure 2 The hydrogen-bond arrangement in (I)/(II). Displacement ellipsoids are drawn at the 50% probability level and dashed lines indicate hydrogen bonds. Primed atoms are generated by the symmetry operation (−x, 1 − y, − (more ...)
Table 2 Hydrogen-bond geometry (Å, °) for (I)/(II)
A search of the Cambridge Structural Database (CSD, Version 5.31, 20 November 2009 release; Allen, 2002
) reveals five structures which consist of independent molecules of a phosphine and its oxide, viz.
refcodes AZUDII (Atikinson et al.
), EXUDAC (Mohamed et al.
), QEMFAP (Chekhlov, 2000
), SIXJEO (Carriedo et al.
) and UJAPEA (Hitchcock et al.
). All show statistical interspersion of the phosphine oxide in the structure, in marked contrast with the pattern seen here. The presence of the strongly hydrogen-bonding amine group and the resulting hydrogen-bond network may account for the greater order seen in the present structure.
In contrast with the lack of reaction between (II)
, reaction of (II)
did lead to the formation of a metal-containing species (see scheme
). However, analysis of the difference Fourier map following data collection revealed that C3 carries two H atoms and the ligand is neutral overall, giving complex (IV)
(Fig. 3). The C—C distances in the nitrogen-containing ring are fully in agreement with the locations of the H atoms (Table 3). The coordination geometry of the metal atom is approximately octahedral, with the N1—Ti1—P1 bite angle significantly smaller than 90°, presumably due to the constraint imposed by the ligand backbone. The mechanism for the reaction is likely to involve initial coordination of (II)
to the metal, yielding (V)
, followed by an acid-mediated rearrangement (see scheme). This type of rearrangement has previously been reported when pyrroles are reacted in the presence of the strong Lewis base B(C6
(Guidotti et al.
; Focante et al.
; Kehr et al.
Figure 3 The structure of (IV), showing the atom-numbering scheme. Displacement ellipsoids are drawn at the 50% probability level and H atoms are shown as small spheres of arbitrary radii.
Table 3 Selected geometric parameters (Å, °) for (IV)
There are two structures in the CSD which contain a rearranged pyrrole ligand without the presence of a porphyrin ring or secondary bonding interactions, viz.
refcodes WELTIQ (DuBois et al.
) and XITLES (Kreickmann et al.
). WELTIQ contains rhenium, and has an Re—N bond distance of 2.171 (6) Å, while XITLES is a tungsten complex, with a W—N distance of 2.136 (3) Å. The Ti—N distance obtained in the current work, 2.174 (5) Å, is therefore somewhat longer than might initially be anticipated on the basis of the smaller size of the Ti centre compared with Re and W. In the case of Ti—P bonds, a search of the CSD gives a range of 2.254–2.904 Å for 226 database entries, with a mean value of 2.584 (4) Å. The value observed for (IV)
[2.6428 (17) Å] is above the average but is not atypical. Presumably, the chelating nature of the ligand in (IV)
is responsible for the length of the bonds observed here.
In summary, whilst ligand (II)
shows good promise in forming complexes with later transition metals, early metals reveal additional ligand reactivity which may have to be modulated in order to obtain the desired complexes successfully. Efforts to control this reactivity are ongoing, for example, by blocking the 2-position on the pyrrole ring with a bulky alkyl to prevent ligand rearrangement.