Report

Heme-Like Coordination Chemistry Within Nanoporous Molecular Crystals

See allHide authors and affiliations

Science  26 Mar 2010:
Vol. 327, Issue 5973, pp. 1627-1630
DOI: 10.1126/science.1184228

Iron Exposure

The macrocyclic heme motif coordinates iron ions in proteins and plays a widespread role in biochemical oxidative catalysis. Bezzu et al. (p. 1627) prepared crystals in which analogous iron-centered macrocycles were aligned in pairs. The outer faces of the pairs exposed the iron ions to vacant cavities, where ligand exchange could take place; the inner faces were bound together by rigid bridging ligands lending the crystals structural integrity. The stability and high porosity of these crystals lend themselves to potential catalytic applications.

Abstract

Crystal engineering of nanoporous structures has not yet exploited the heme motif so widely found in proteins. Here, we report that a derivative of iron phthalocyanine, a close analog of heme, forms millimeter-scale molecular crystals that contain large interconnected voids (8 cubic nanometers), defined by a cubic assembly of six phthalocyanines. Rapid ligand exchange is achieved within these phthalocyanine nanoporous crystals by single-crystal–to–single-crystal (SCSC) transformations. Differentiation of the binding sites, similar to that which occurs in hemoproteins, is achieved so that monodentate ligands add preferentially to the axial binding site within the cubic assembly, whereas bidentate ligands selectively bind to the opposite axial site to link the cubic assemblies. These bidentate ligands act as molecular wall ties to prevent the collapse of the molecular crystal during the removal of solvent. The resulting crystals possess high surface areas (850 to 1000 square meters per gram) and bind N2 at the equivalent of the heme distal site through a SCSC process characterized by x-ray crystallography.

In biological environments, hemoproteins perform many important chemical tasks involving diatomic gases (1). These include oxidative catalysis by activating molecular oxygen under mild conditions (e.g., cytochrome P450), the storage and transport of oxygen (e.g., hemoglobin and myoglobin), and sensing of gas molecules such as CO and NO (e.g., guanylyl cyclase). Increasingly sophisticated synthetic models of the iron porphyrin in the heme reactive site have been used to understand this complex coordination chemistry and to prepare biomimetic catalysts (2). These model systems share at least some of the following structural features with the heme reactive site: an iron porphyrin, the differentiation of the two axial binding sites (proximal and distal), a proximal ligand to control gas binding, and steric protection of the distal binding site to prevent μ-oxo dimer formation. These synthetic models are designed to function in homogeneous solution, generally organic solvents rather than the environmentally benign aqueous media of hemoproteins.

Here, we demonstrate a heme model system based on a nanoporous crystal that possesses the basic requirements of these homogeneous models but with the added advantage that axial coordination chemistry at the iron cation is achieved by single-crystal–single-crystal (SCSC) transformations that can be characterized directly by x-ray diffraction. Furthermore, these crystals are compatible with aqueous media, and their nanoporous structure ensures rapid access of ligands to the axial binding sites, even by simple adsorption from the gas phase. Ultimately, the added convenience of these nanoporous crystals could lead to the development of practical industrial catalysts. In addition to Fe2+, we show that other transition metal cations (e.g., Co2+, Mn2+, and Ru2+) that are known to impart useful catalytic activity to porphyrins and phthalocyanines are incorporated successfully into the nanoporous molecular crystal (3).

Progress in the development of crystalline nanoporous materials (46), such as the metal-organic frameworks (MOFs), has produced numerous examples incorporating a metalloporphyrin component (717). However, despite the obvious motivation to prepare nanoporous heme models, no such material containing an iron porphyrin has emerged. It is likely that the high reactivity of the Fe2+ cation and its strong preference to be hexacoordinate results in competitive ligand binding that interferes with MOF formation. We reasoned that a nanoporous molecular crystal derived from an appropriate metallomacrocycle by simple crystallization would avoid this problem. Furthermore, the subsequent study of coordination chemistry at the axial sites of the Fe2+ cation would benefit from the absence of an extended framework containing transition metals, which could provide competitive binding sites for ligands. In this context, our interest was aroused by the serendipitous discovery of the molecular crystal of zinc 2,3,9,10,16,17,23,24-octa(2′,6′-di-iso-propylphenoxy)phthalocyanine, which contains very large (8 nm3) solvent-filled voids (18). The phthalocyanine macrocycle (i.e., tetraazatetrabenzoporphyrin) is closely related to porphyrin, and, importantly, the coordination chemistry of iron phthalocyanine is very similar to that of iron porphyrin (19). Hence, we prepared the analogous Fe2+ complex (Fig. 1) from 4,5-(2′,6′-di-iso-propylphenoxy)phthalonitrile and ferrous acetate by using a well-established synthetic procedure (20). Pleasingly, this complex gave a phthalocyanine nanoporous crystal (PNC), isomorphic with that of the Zn2+ complex (i.e., cubic symmetry, space group = Pn Embedded Imagen, a = 3.74 ± 0.03 nm, Z = 12), on diffusion of methanol into a concentrated chloroform solution. Even minor structural modifications to such molecules can alter markedly the subsequent packing arrangement within a molecular crystal. Therefore, it is remarkable that the PNC structure is obtained for many different metal complexes of this particular phthalocyanine derivative (Fig. 1A; M = Mg2+, Al3+, Ti4+, Mn2+, Fe2+, Co2+, Zn2+, Ru2+, or In3+), even as they encompass great variation in size, shape, type, and number of axial ligands (Fig. 1A; L). Indeed, the addition of the bulky ligand, t-butyl isocyanide (BuNC), to the Fe2+ (or Ru2+) complex before crystallization enhances PNC formation so that large crystals of several mm in diameter can be grown routinely (Fig. 1B). Therefore, we were able to perform and characterize SCSC transformations by the manipulation of single crystals.

Fig. 1

(A) The molecular structures of the phthalocyanine complexes (M = metal cation Mg2+, Al3+, Ti4+, Mn2+, Fe2+, Co2+, Zn2+, Ru2+, or In3+) and a selection of axial ligands (L) that are compatible with the formation of the PNCs. (B) Examples of crystals (PNC[cBuNC-Fe-vBuNC]) formed by slow diffusion of methanol into a chloroform solution. The cubic crystal on the right is 4 mm in diameter.

The nanoporous structure of the PNCs (Figs. 2 and 3) resembles Schoen’s I-WP triply periodic minimal surface (21) in which free volume is unequally partitioned between two interpenetrating labyrinths by a non–self-intersecting, two-sided surface. The larger labyrinth is here termed “void” and is composed of the 8-nm3 voids inside the cubic assembly of six phthalocyanines and the interconnecting channels located at each corner of the assembly; the smaller labyrinth is termed “cavity” and is composed of the narrow interconnecting channels that lie between the assemblies. For each phthalocyanine complex, the PNC structure differentiates the two axial binding sites so that one faces into the void, here denoted v, and the other faces into the cavity, denoted c. It is notable that within PNC[cBuNC-Fe-vBuNC] the ligand in the relatively narrow cavity (max diameter ~ 1.3 nm) is forced out of axial linearity by steric congestion, whereas the ligand in the more spacious 8-nm3 void is free to adopt a linear configuration (Fig. 3B).

Fig. 2

(A) The nanoporous structure of a PNC as represented by Schoen’s I-WP triply periodic minimal surface with the transition metal cations (e.g., Fe2+) denoted as M. The void side of the surface is shaded yellow, and the cavity side is shaded gray. (B) Schematic representation, based on a space-filling model of the crystal structure, of a cross section through the unit cell of a PNC composed of the phthalocyanine complex where M is the metal cation, Lv is the ligand in the void, and Lc is the ligand in the cavity. The void enclosed by the hexa-phthalocyanine assemblies is shaded yellow, and the cavity between these assemblies is shaded gray. (C) A cross section through a cavity brick wall showing the role of wall ties in maintaining stability. The location of the analogous bidentate ligand molecular wall tie within the cavity of the PNC is indicated in (B).

Fig. 3

(A) A perspective representation of the crystal structure of PUNC[vN2-Fe-cpdic-Fe-vN2], with the phenoxyl substituents removed and the carbon atoms of the pdic molecular wall ties colored green for clarity. The phenyl ring of the pdic ligand is disordered over two orientations because of the cubic symmetry of the crystal. The large square represents the boundary of the unit cell (a =3.7 nm), and the blue cube represents one of the two evacuated nanovoids, which is located at the center of the unit cell; the other is located in eight 1-nm3 portions at each corner of the unit cell. (B) The SCSC coordination chemistry used to make the PUNCs [vN2-Fe-cpdic-Fe-vN2], [vN2-Fe-cbipy-Fe-vN2], and [Co-cbipy-Co]. Gray and green, C; blue, N; pink, Co; gold, Fe; and red, O.

The original solvent of recrystallization within the PNCs [~240 methanol molecules (~20 mass %) per unit cell, as measured in deuterated chloroform by 1H nuclear magnetic resonance spectroscopy of the dissolved PNC] can be rapidly and reversibly exchanged with other solvents that do not dissolve the phthalocyanine derivative, such as acetone, hexane, or water, without loss of crystalline order, whereas attempted exchange with toluene, chloroform, or tetrahydrofuran dissolves the PNCs. Rapid exchange of the axial ligands by a SCSC transformation is achieved by the simple addition of a drop of the new ligand to the solvent in contact with a crystal. In each case, the displaced ligand remains in the contact solvent, and the reactions are driven either by a higher concentration of the new ligand or by its greater affinity with the metal cation. For example, the BuNC ligand at the void binding site of PNC[cBuNC-Fe-vBuNC] could be partially replaced (70%) by pyridine (py) to give PNC[cBuNC-Fe-v(BuNC)0.3(py)0.7] or quantitatively displaced by N-methylimidazole (Meim), with concurrent exchange of the cavity BuNC ligand by methanol originating from the solvent, to give PNC[cMeOH-Fe-vMeim]. The ligand exchange at the cavity binding site suggested the possibility that bidentate ligands of an appropriate length (~1 nm) might bind simultaneously to two metal cations across the cavity, thus forming a bridge between adjacent hexa-phthalocyanine assemblies. This outcome was achieved with surprising ease by the SCSC addition of either 4,4′-bipyridyl (bipy) or 1,4-phenylenediisocyanide (pdic) to PNC[cBuNC-Fe-vBuNC] to give PNC[vBuNC-Fe-cbipy-Fe-vBuNC] or PNC[vN2-Fe-cpdic-Fe-vN2], respectively. The assignment of N2 as axial ligand in the latter PNC was unexpected but is supported by infrared spectroscopy (figs. S3 and S4) and has firm precedents in the coordination chemistry of ruthenium porphyrins (22, 23). Indeed, the addition of pdic to PNC[cBuNC-Ru-vBuNC] gives the analogous structure, PNC[vN2-Ru-cpdic-Ru-vN2]. This ligand likely originates from the cryogenic stream of nitrogen used to maintain the crystals at 150 K during x-ray diffraction (XRD) analysis, replacing the original vBuNC ligand. The following PNCs were prepared to show the generality of SCSC ligand exchange: PNC[Zn-vpy], by addition of pyridine to PNC[Zn-vH2O]; PNC[cH2O-Mn-vim], by addition of imidazole to PNC[cH2O-Mn-vH2O]; and PNC[vpy-Co-cbipy-Co-vpy], by addition of bipy to PNC[Co-vpy]. That rapid SCSC axial ligand exchange is achieved within the PNCs with much the same ease as solution-based coordination chemistry can be explained by the lack of structural reorganization required to accommodate even large ligands within the void or cavity. Although axial coordination chemistry within some porphyrin-containing MOFs was inferred by their catalytic activity (13, 17), direct structural characterization of axial ligand exchange on a metallomacrocycle within a MOF or other nanoporous crystal has been elusive.

As is usually observed for solvent-containing molecular crystals (i.e., clathrates), most of the PNCs suffer structural collapse during the rapid loss of included solvent on exposure to the atmosphere, which is evident by the loss of reflectivity from their macroscopic appearance. However, the PNCs that contain the rigid bidentate ligands, bipy or pdic, within the cavity retain their sheen on prolonged exposure to a stream of nitrogen, a process that ensures the removal of included solvent as determined by thermal gravimetric analysis (TGA). XRD analysis confirmed the retention of the crystal structures and shows that for PNC[vBuNC-Fe-cbipy-Fe-vBuNC] the vBuNC ligand is replaced by N2. Similarly, for PNC[vpy-Co-cbipy-Co-vpy] the removal of included solvent is accompanied by the loss of the pyridine ligand from the void. These phthalocyanine unsolvated nanoporous crystals (PUNCs) are made by two consecutive SCSC transformations (e.g., PNC[cBuNC-Fe-vBuNC] PNC[vBuNC-Fe-cbipy-Fe-vBuNC] PUNC[vN2-Fe-cbipy-Fe-vN2]), and it is notable that the metal-metal distance across the cavity is reduced during this process (Fig. 3B). This example of crystal engineering to prevent the collapse of the PNC structure during the removal of solvent is closely analogous to the use of wall ties to stabilize cavity brick walls in architectural engineering (Fig. 2C). It should be emphasized that the molecular wall tie bridges two phthalocyanines to form a dimeric complex rather than forming an extended framework; therefore, the PUNCs are still molecular crystals rather than MOFs. It is of interest that both types of molecular wall tie provide a pathway for electronic conjugation between the Fe2+ cations that may be useful for the stabilization of mixed oxidation states, which have been implicated in the recently reported low-temperature oxidation of methane by a iron phthalocyanine dimer (24).

The standard test for the confirmation of permanent nanoporosity in a solid is reversible nitrogen adsorption at 77 K following the application of a vacuum to remove adsorbed molecules. The unsolvated crystals of PNC[cBuNC-Fe-vBuNC] show minimal nitrogen uptake (<1 mmol g−1) consistent with the prior collapse of structure during removal of solvent (Fig. 4). In contrast, PUNCs [vN2-Fe-cbipy-Fe-vN2], [vN2-Fe-cpdic-Fe-vN2], and [Co-cbipy-Co] all show significant nitrogen adsorption at low partial pressures (11 to 13 mmol N2 at p/po < 0.2), which confirms the success of the molecular wall-tie strategy in stabilizing these molecular crystals toward the complete evacuation of included molecules. Brunauer, Emmett, Teller (BET) surface areas in the range of 850 to 1000 m2 g−1 and nanopore volumes in the range of 0.40 to 0.46 ml g−1 can be calculated from these isotherms (e.g., Fig. 4). The surface area, total pore volume, and pore size of the PUNCs all exceed those of other nanoporous molecular crystals (25, 26), zeolites (27), and porphyrin-based MOFs (717). Nitrogen adsorption studies show that nanoporosity is retained indefinitely under ambient conditions and for samples previously heated to 400 K. For each PUNC, TGA shows a mass reduction at 500 K attributable to the loss of the wall ties, which is likely to coincide with the loss of nanoporosity.

Fig. 4

The nitrogen adsorption isotherms of PUNC[vN2-Fe-cpdic-Fe-vN2] (solid circles) and an unsolvated (i.e., collapsed) sample of PNC[vBuNC-Fe-cBuNC] (open circles) collected at 77 K. A BET surface area of 1002 m2 g−1 and a total pore volume of 0.46 ml g−1 can be calculated for PUNC[vN2-Fe-cpdic-Fe-vN2] from these data.

The symmetry-driven self-assembly of the PNCs via crystallization of a phthalocyanine complex, itself made from a symmetry-driven reaction of an easily prepared organic compound, provides a simple and readily scalable (28) method of constructing sophisticated and versatile nanoporous structures. It has been demonstrated that the formation of the PNCs is compatible with a wide selection of transition metal cations, including those that impart useful catalytic activity. It is highly probable that this range of cations can be extended to many more of the 70 elements, some in different oxidation states, that are known to bind to the phthalocyanine macrocycle (fig. S1) (29). Initial studies show that the related macrocycle 2,3,9,10,16,17,23,24-octa(2′,6′-di-iso-propylphenoxy)-1,4,8,11,15,18,22,25-octaazaphthalocyanine has very similar PNC-forming properties (30) so that the reactivity of the transition metal center within these materials may also be tuned by modification of the macrocycle structure. Hence, the present study has only begun to explore the scope and potential of the SCSC coordination chemistry that can be performed within the PNCs and PUNCs.

The differentiation of the void and cavity axial binding sites of the phthalocyanine within the PNCs is analogous to the differentiation of the distal and proximal axial sites within hemoproteins (1). Thus, the molecular wall tie within the cavity could perform a similar role to that of a heme proximal ligand (e.g., a histidine or cysteine residue) in the manipulation of the reactivity of the metallomacrocycle in addition to its task of maintaining the structural stability of the molecular crystal. The observation by XRD analysis of a diatomic gas molecule (N2) at the distal axial site within the PUNCs supports the analogy. This differentiation of the two axial sites by crystal engineering is achieved without the considerable effort that is required to prepare synthetic heme models with differentiated binding sites for use in homogeneous organic solutions (2). Following the recent XRD characterization of a transient reaction intermediate in a MOF (31), the PNCs offer the opportunity to probe intermediates by using similar in situ XRD experiments for a range of metal-centered catalytic reactions, including those of heme proteins. An additional attractive feature of the PNCs for such studies is their long-term hydrolytic stability following the exchange of included organic solvent for water, which will enhance the similarity with the natural environment of hemoproteins and may lead to practical heterogeneous catalysts for use in aqueous media.

Supporting Online Material

www.sciencemag.org/cgi/content/full/327/5973/1627/DC1

Materials and Methods

Figs. S1 to S5

References and Notes

References and Notes

  1. Materials and methods are detailed in supporting information available on Science Online.
  2. It is anticipated that large-scale preparation of octa(2′,6′-di-iso-propylphenoxy)phthalocyanine and its metal complexes could be achieved readily by processes similar to those that produce many thousands of tons of phthalocyanine derivatives per annum for use as colorants, electronic materials, and homogeneous catalysts for the desulfurization of crude peteroleum (29). In addition, we have found that multigram batches of microcrystalline PUNC can be obtained by rapid recrystallization of the phthalocyanine complex in the presence of the bidentate ligand.
  3. This work was supported by Engineering and Physical Sciences Research Council grant no. GR/F019114. We thank F. Tuna (University of Manchester) for help with the synthesis of PNC[Co-cPy], H. Nowell (Diamond) and B. Kariuki (Cardiff University) for technical support during crystallographic data collection, and K. D. M. Harris and G. J. Hutchings for valuable discussions. Crystallographic parameters for the PNCs and PUNCs are available free of charge from the Cambridge Crystallographic Data Centre (CCDC) under deposition numbers CCDC 761407 to 761422.
View Abstract

Stay Connected to Science

Navigate This Article