Hetero-interpenetrated metal–organic frameworks

Interpenetrated metal–organic frameworks (MOFs) comprise two or more lattices that are mutually entangled. Interpenetration can be used to tune the structures and pore architectures of MOFs to influence, for example, their stability or interactions with guest molecules. The interpenetrating sublattices are typically identical, but hetero-interpenetrated MOFs, which consist of sublattices that are different from one another, have also been serendipitously produced. Here we describe a strategy for the deliberate synthesis of hetero-interpenetrated MOFs. We use the cubic α-MUF-9 framework as a host sublattice to template the growth of a second sublattice within its pores. Three different secondary sublattices are grown—two of which are not known as standalone MOFs—leading to three different hetero-interpenetrated MOFs. This strategy may serve to combine different properties into one material. We produce an asymmetric catalysis by allocating separate roles to the interpenetrating sublattices in a hetero-interpenetrated MOF: an achiral secondary amine on one sublattice provides the catalytic activity, while the chiral α-MUF-10 host imparts asymmetry to aldol and Henry reactions. Interpenetration—in which two or more lattices are catenated—is common in metal–organic frameworks (MOFs). Now a deliberate synthesis of hetero-interpenetrated MOFs, with two distinct lattices, has been developed. It can combine the different properties of the two sublattices in one material, as demonstrated with chirality and catalytic activity, delivering an asymmetric catalyst.

Interpenetrated metal-organic frameworks (MOFs) comprise two or more lattices that are mutually entangled.Interpenetration can be used to tune the structures and pore architectures of MOFs to influence, for example, their stability or interactions with guest molecules.The interpenetrating sublattices are typically identical, but hetero-interpenetrated MOFs, which consist of sublattices that are different from one another, have also been serendipitously produced.Here we describe a strategy for the deliberate synthesis of hetero-interpenetrated MOFs.We use the cubic α-MUF-9 framework as a host sublattice to template the growth of a second sublattice within its pores.Three different secondary sublattices are grown-two of which are not known as standalone MOFs-leading to three different hetero-interpenetrated MOFs.This strategy may serve to combine different properties into one material.We produce an asymmetric catalysis by allocating separate roles to the interpenetrating sublattices in a hetero-interpenetrated MOF: an achiral secondary amine on one sublattice provides the catalytic activity, while the chiral α-MUF-10 host imparts asymmetry to aldol and Henry reactions.
Interpenetration 1,2 is a common phenomenon in MOFs, where entangled, interlocked sublattices exist in the same crystal.Methods to control interpenetration are important since interpenetration governs the size and chemical environment of the pores, the diffusivity of guest molecules and the framework stability [3][4][5][6][7][8] .The interpenetrating sublattices are typically identical, but hetero-interpenetrated MOFs comprise sublattices that are different from one another.Such frameworks are fascinating from the design and structural viewpoints, and they may exhibit properties that are not observed in the individual lattices on their own.For example, bringing together donor groups on one network with acceptor groups on another may result in new optical, magnetic and/ or electronic characteristics.Complementary functional groups may also be precisely placed in the pores of hetero-interpenetrated MOFs to deliver unique adsorption, catalysis and sensing applications.Although occasional examples of hetero-interpenetrated MOFs have been discovered serendipitously 3,[9][10][11][12] , their deliberate synthesis remains elusive.Recent computational studies have identified sublattices that are mutually compatible in silico.However, these hetero-interpenetrated MOFs have not been experimentally realized 13,14 .Under conventional synthetic protocols, one-pot methods using structurally related ligands will typically produce multivariate frameworks 15 , while ligands with different geometries produce either multicomponent MOFs 16 or mixed phases.In this light, the deliberate synthesis of hetero-interpenetrated MOFs is not straightforward.
To address this challenge, we formulated a two-step strategy towards the deliberate synthesis of hetero-interpenetrated MOFs.In this approach, a non-interpenetrated MOF acts as a host sublattice.A second, interpenetrating framework is grown in the pore space of this host by subsequent secondary growth.While the interpenetrating sublattice must be geometrically compatible with the host, it can be chemically distinct to deliver a hetero-interpenetrated framework.There is potential for this strategy to be disrupted by competing processes.For example, the second framework may grow outside the pores of the first in a separate phase, the incoming components may displace those of the original framework or the growth of the second framework may be hindered by mass transfer requirements.We herein Article https://doi.org/10.1038/s41557-023-01277-zThis was achieved using zinc(II) or cobalt(II) metal ions to build up the M 4 O clusters and ligands based on biphenyl-4,4′-dicarboxylate and its close analogues.
The concept of partial interpenetration is central to this work.In most interpenetrated MOFs, the number of interpenetrating lattices is an integer.However, in partially interpenetrated MOFs, different regions of the crystal are composed of different numbers of sublattices [17][18][19][20] .This results in a fractional value for the overall level of partial interpenetration, which can be expressed as a percentage (PIP).For a doubly interpenetrated MOF, the PIP corresponds to the occupancy of the second sublattice.In some reported cases, the PIP level is fixed 18,19 , while in others it can be controlled 17,20 .present a successful realization of this strategy that circumvents these potential limitations.

The hetero-interpenetrated MUF-91 framework
Our first target was MUF-91, in which a [Zn 4 O(bpdc) 3 ] sublattice interpenetrates α-MUF-9 ([Zn 4 O(L1) 3 ]; Fig. 1b(i)).We first prepared α-MUF-9 using racemic H 2 L1 and then incubated it in a secondary growth solution comprising 4,4′-biphenyldicarboxylic acid (H 2 bpdc) and Zn(NO 3 ) 2 in DBF.We found 2-fluorobenzoic acid (FBA) to be a useful additive to suppress the formation of a standalone phase of [Zn 4 O(bpdc) 3 ].In addition, the cloudy supernatant was removed periodically throughout the secondary growth process to eliminate small crystallites that self-nucleate from H 2 bpdc and Zn(NO 3 ) 2 .The PIP level of MUF-91, which equates with the occupancy of the interpenetrating [Zn 4 O(bpdc) 3 ] sublattice, increases over time and can be determined quantitatively by single-crystal X-ray diffraction (SCXRD), powder X-ray diffraction (PXRD) and 1 H NMR spectroscopy (Fig. 1c).A crystallographic model was developed for the SCXRD datasets (Fig. 2b and Supplementary Tables 1 and 2).The occupancy of the secondary [Zn 4 O(bpdc) 3 ] sublattice was refined as a free variable to give a direct measure of the PIP level over the course of secondary growth.Restraints were used to ensure refinement stability and chemical correctness, and these were loosened as far as the data quality would allow.The P 4 3m space group of α-MUF-9 is consistently maintained.After 9 hours, the PIP level reaches 70%, which is the upper limit for the occupancy of the [Zn 4 O(bpdc) 3 ] sublattice (Fig. 1c).We could also gauge the PIP level over time by quantifying the changes in the PXRD patterns, and excellent agreement with the SCXRD data was observed.The intensity of the PXRD peak at 2θ = 5.2° decreases in intensity, and the peak at 7.3° increases (Supplementary Fig. 1).These results correspond to the (100) and (110) reflections that are, respectively, extinguished and enhanced by the growing [Zn 4 O(bpdc) 3 ] sublattice.These changes mirror those observed during the homo-interpenetration of MUF-9 (ref.17).In accord with the diffraction data, the 1 H NMR spectroscopic analysis of digested MUF-91 samples shows that the amount of bpdc relative to L1 increases with time.The PIP level deduced from the bpdc/L1 ratio measured by 1 H NMR spectroscopy on digested samples matches the PIP level given by the diffraction data for the first 9 hours of secondary growth (Fig. 1c).This confirms that bpdc does not simply displace L1 from the [Zn 4 O(L1) 3 ] host sublattice.After 9 hours, a shell of a different phase forms on the surface of the MUF-91 crystals, which is clearly distinguishable by PXRD (Supplementary Fig. 1) and optical microscopy (Supplementary Fig. 3).The growth of this [Zn 4 O(bpdc) 3 ] shell is only possible when secondary growth has occurred to a substantial extent and the occupancy of the interpenetrating lattice is already high.After this point, the bpdc/L1 ratio measured by 1 H NMR spectroscopy increases beyond the PIP level since the shell is composed of [Zn 4 O(bpdc) 3 ] with no L1.To optimize the occupancy of the interpenetrating lattice, we prepared MUF-91 starting from microcrystalline α-MUF-9 rather than large single crystals (Supplementary Fig. 6).3).The [Zn 4 O(bpdc-NH 2 ) 3 ] sublattice grows in over an incubation period of 9 hours to reach a PIP level of 74%.The phenyl rings of the bpdc-NH 2 ligand adopt an orthogonal conformation, which is impossible for L1 and thus allows the two sublattices in MUF-92 to be unambiguously differentiated by SCXRD (Fig. 2c).This indicates there is little, if any, displacement of the linkers from the [Zn 4 O(L1) 3 ] sublattice by bpdc-NH 2 , which was additionally verified by a control experiment (Supplementary Fig. 30).Beyond 9 hours, a shell layer comprising Zn(II) and bpdc-NH 2 grows around the crystals, which means the bpdc-NH 2 /L1 ratio deduced by NMR spectroscopy continues to rise while the PIP level of the MUF-92 core remains constant (Supplementary Fig. 9).When microcrystals of α-MUF-9 are used for secondary growth, the occupancy level of the [Zn ] sublattice is evidenced by SCXRD (Fig. 2d), 1 H NMR spectroscopy and atomic adsorption (AA) spectroscopy (which measured the Zn/Co ratio).As deduced by X-ray diffraction, large crystals of MUF-93 reach PIP levels of 70% over 60 hours of secondary growth (Fig. 1e and Supplementary Table 4), while starting from microcrystalline α-MUF-9 allows the [Co 4 O(bpdc) 3 ] framework to reach 80% occupancy over just 18 hours (Supplementary Fig. 16).These PIP values align with the 1 H NMR and AA spectroscopic results over the early stages of secondary growth (Fig. 1e and Supplementary Fig. 14), which indicates that there is little displacement of the zinc(II) ions by cobalt(II).This is corroborated by the site-specific anomalous scattering experiments detailed later.While the exchange of three out of four zinc(II) ions per node is possible by heating α-MUF-9 in highly concentrated solutions of cobalt(II) nitrate in DBF (Supplementary Section 5.2), low rates of metal exchange during the secondary growth of MUF-93 were ensured by using a low concentration of cobalt(II) nitrate.Once the occupancy of the [Co 4 O(bpdc) 3 ] sublattice reaches ~70% after 60 hours of secondary growth, the amounts of bpdc and Co rise beyond the PIP level due to the gradual displacement of the zinc(II) ions and L1 linkers of [Zn 4 O(L1) 3 ] by cobalt(II) and bpdc, respectively (Fig. 1e).
The spatial variation of PIP level within an individual MUF-93 specimen was probed by systematically collecting SCXRD datasets across the midpoint of a single crystal.The crystal was approximately ~180 µm in size, and synchrotron X-ray radiation with a spot size of approximately 10 µm (full-width at half-maximum) horizontally was used (Fig. 3 and Supplementary Fig. 13).Rastering of the crystal produced datasets from various spatial regions.As anticipated, we observed the highest PIP values (~72%) near both edges of the crystal since these regions are more accessible to the incoming components of the interpenetrating sublattice.The PIP levels drop to ~20% when the beam is directed at the centre of the crystal since the mass transfer limits the growth of the second lattice nearer the core.
Conventional single-crystal diffraction cannot reliably distinguish cobalt from zinc in MUF-93 because of their similar electron counts.This is exacerbated at low PIP levels due to high correlations between scattering factors, occupancy and atomic displacement parameters.However, tuning the X-ray wavelength to be near the respective absorption edges for cobalt and zinc enhances their anomalous dispersion and, in principle, permits the discrimination and quantification of these ] sublattice is highest towards the edges of the crystal and lowest at its centre, consistent with secondary growth being more prevalent at the crystal surface than its centre.Note that the number of unit cells actually present in the X-ray beam at each data collection point is many orders of magnitude greater than illustrated.

Article
https://doi.org/10.1038/s41557-023-01277-zmetals at specific crystallographic sites (Supplementary Fig. 18) [23][24][25] .On this basis, we developed a method for using anomalous dispersion to differentiate metals in crystalline materials.For MUF-93, we calculated the differences in reflection intensities between datasets collected just below the cobalt(II) absorption edge at 7,500 eV (where the in-phase anomalous scattering contribution (f′) by cobalt is significant) and a high-resolution dataset collected at 17,440 eV (where there is little anomalous scattering) on the same crystal.The reflection intensity differences obtained in this way arise partly from the difference in anomalous scattering by the cobalt and thus can be used to locate and quantify the cobalt sites.Similarly, we used 9,670/17,440 eV difference datasets, which maximize zinc anomalous dispersion, to pinpoint the zinc sites.Figure 4 illustrates the datasets obtained in this way, showing differences between peaks near the unit cell centre (crystallographically identical metal atom sites of the [Zn 4 O(L1) 3 ] host sublattice) and peaks near the unit cell corners (crystallographically identical metal atom sites of the interpenetrating [Co 4 O(bpdc) 3 ] sublattice).After a secondary growth time of 60 hours, the occupancy of [Co 4 O(bpdc) 3 ] in MUF-93 reaches its maximum.At this point, a distinct peak for cobalt appears near the corner of the unit cell in the difference datasets due to the cobalt ion in the [Co 4 O(bpdc) 3 ] sublattice (Fig. 4a).No signal for cobalt can be detected near the midpoint of the unit cell, which demonstrates that cobalt(II) ions do not displace zinc(II) ions from the [Zn 4 O(L1) 3 ] host over the 60 hours of secondary growth.Displacement of the zinc(II) ions in the [Zn 4 O(L1) 3 ] sublattice becomes evident only after a much longer reaction time (Fig. 4b).
Two illustrative control experiments were also performed.First, a 7,500/17,440 eV difference dataset on a crystal of α-MUF-9 in which the zinc(II) ions had been partially replaced by cobalt(II) showed a single peak for cobalt near the midpoint of the unit cell and no sign of an interpenetrating lattice (Supplementary Fig. 19c).Second, as expected, a 9,670/17,440 eV difference dataset for homo-interpenetrated β-MUF-9 revealed two equally strong peaks for the two independent zinc sites in the unit cell (P 4 3m space group assumed; Supplementary Fig. 19d).
The anomalous dispersion experiments clearly show that the two sublattices in MUF-93 are distinct from each other, and thus the framework can be genuinely described as being hetero-interpenetrated.This observation underscores the power of the two-step strategy employing secondary growth since any attempt to synthesize MUF-93 directly from a mixture of H 2 bpdc, Zn(NO 3 ) 2 and Co(NO 3 ) 2 would result in a mixed-metal multivariate material.
With a catalyst loading of around 3% (defined as the molar ratio of catalytic L2 units to aldehyde), MUF-101 catalyses the aldol reaction between 2-chloro-5-nitrobenzaldehyde (1) and acetone (Fig. 5).High-performance liquid chromatography (HPLC) revealed the reaction product (2) to have an enantiomeric excess (e.e.) of −13.9% when catalysed by (R)-MUF-101 (Supplementary Fig. 22 and Supplementary Table 7).Importantly, the enantioselectivity was reversed when (S)-MUF-101 was used as the catalyst (Supplementary Table 8).This confirms that the preferred handedness of the reaction product is induced by the chirality of the MUF-10 host sublattice.To verify this, we grew a [Zn 4 O(L2) 3 ] sublattice inside the racemic α-MUF-9 host to produce the MUF-99 framework.As expected, MUF-99 is catalytically active but does not confer asymmetry on 2 due to the racemic host environment.As an additional test of the activity of MUF-101, we found that it catalyses the Henry reaction between 1 and nitromethane (Supplementary Scheme 4).The reaction product (3) has an e.e. of −9.4% when catalysed by (R)-MUF-101, the sign of which is reversed when (S)-MUF-101 is the host sublattice (Supplementary Table 8).

Outlook
The rational design of hetero-interpenetrated MOFs has inherent challenges since the component sublattices must be geometrically and chemically compatible, phase separation is possible and the exchange of the metal ions and linkers may occur.Here we have taken advantage of the templating effect of a host lattice to promote the growth of interpenetrating sublattices in a second, discrete step.These hetero-interpenetrated MOFs retain the make-up of the individual sublattices since there is no detectable exchange of components between them.In certain cases, methods to produce the interpenetrating sublattices have not been previously reported.
We are optimistic that this deliberate strategy regarding hetero-interpenetrated MOFs will open new perspectives on the field of framework chemistry.Specific functional properties can emerge when different and complementary sublattices are coupled to one another.Here we show how the two principal roles of an asymmetric catalyst can be assigned to the different sublattices: the host framework provides a chiral environment for the second, catalytically active, sublattice.The enantioselectivity is dictated by the handedness of the host sublattice in a way that is reminiscent of the active site in enzymes where the chirality of the catalytic pocket influences the reaction stereochemistry.

Online content
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41557-023-01277-z.

Fig. 3 |
Fig. 3 | The variation in PIP across an individual crystal of MUF-93 as determined from multiple SCXRD datasets.An illustration of a crystal of MUF-93 is presented showing the regions pinpointed by the synchrotron X-ray beam.The occupancy level of the interpenetrating [Co 4 O(bpdc) 3] sublattice is highest towards the edges of the crystal and lowest at its centre, consistent with secondary growth being more prevalent at the crystal surface than its centre.Note that the number of unit cells actually present in the X-ray beam at each data collection point is many orders of magnitude greater than illustrated.

Fig. 4 |
Fig. 4 | The differentiation of cobalt and zinc sites in hetero-interpenetrated MOFs using the anomalous dispersion of X-rays.Slices of the (101) plane of the anomalous electron density difference maps derived from X-ray diffraction datasets collected at 17,440 eV and 7,500 eV are presented.Peaks near the midpoint of the unit cell correspond to cobalt occupying the erstwhile Zn 4 O cluster sites of the host sublattice, and peaks near the corner of the unit cell correspond to cobalt occupying Co 4 O cluster sites of the interpenetrating Article https://doi.org/10.1038/s41557-023-01277-z