Charting the Metal-Dependent High-Pressure Stability of Bimetallic UiO-66 Materials.

In theory, bimetallic UiO-66(Zr:Ce) and UiO-66(Zr:Hf) metal-organic frameworks (MOFs) are extremely versatile and attractive nanoporous materials as they combine the high catalytic activity of UiO-66(Ce) or UiO-66(Hf) with the outstanding stability of UiO-66(Zr). Using in situ high-pressure powder X-ray diffraction, however, we observe that this expected mechanical stability is not achieved when incorporating cerium or hafnium in UiO-66(Zr). This observation is akin to the earlier observed reduced thermal stability of UiO-66(Zr:Ce) compounds. To elucidate the atomic origin of this phenomenon, we chart the loss-of-crystallinity pressures of 22 monometallic and bimetallic UiO-66 materials and systematically isolate their intrinsic mechanical stability from their defect-induced weakening. This complementary experimental/computational approach reveals that the intrinsic mechanical stability of these bimetallic MOFs decreases nonlinearly upon cerium incorporation but remains unaffected by the zirconium: hafnium ratio. Additionally, all experimental samples suffer from defect-induced weakening, a synthesis-controlled effect that is observed to be independent of their intrinsic stability.

The permanent nanoporosity of metal–organic frameworks (MOFs) or porous coordination polymers makes these materials attractive for a large variety of applications.[1−7] Through the concept of reticular synthesis, this structural nanoporosity can be engineered extensively, both geometrically as well as in terms of chemical functionality, to design MOFs for targeted applications.[8−11] In the past decade, this endeavour has resulted in consciously designed MOFs that outperform competing materials in areas including gas/vapor adsorption and separation,[12−15] heterogeneous catalysis,[16,17] nanosensing,[18,19] and energy storage.[20−22] However, the porosity and rather weak coordination bonds present in MOFs also limit their mechanical, thermal, and chemical stability under operating conditions;[23,24] a stability that is typically further reduced during practical usage because of process-specific effects, such as pressure attrition.[25,26] A poor stability is prohibitive for certain synthetic or post-synthetic design strategies and shaping procedures[4,27−31] and hampers the large-scale deployment of MOFs.[29] Therefore, to bring MOF design to its full potential, it is crucial to identify to what extent the targeted incorporation of attractive functionalities in a MOF results in molecular-level alterations that impact their stability and whether the experimentally observed stability effects are intrinsic to the designed material or rather stem from the followed synthesis or activation protocol which may lead to structural defects or occluded molecules.[32,33] Herein, we systematically chart these intrinsic and synthesis-mediated stability effects for a series of monometallic and bimetallic UiO-66 materials by combining in situ high-pressure powder X-ray diffraction (PXRD) experiments with molecular dynamics (MD) simulations at operating conditions,[34,35] thereby yielding complementary insight in the mechanical stability window of these materials.

As UiO-66 exhibits a relatively high thermal,[36,37] mechanical,[28,34,38,39] and chemical stability,[23,36,40] the material can be easily subjected to post-synthetic modification strategies and has been proposed for a wide variety of applications,[3] including hydrogen storage[41] and heterogeneous catalysis.[42,43] The defect-free UiO-66(Zr) material is composed of 12-fold coordinated [Zr6(μ3-O)4(μ3-OH)4]12+ bricks, resulting in the intrinsic fcu topology shown in Figure a.[44] Experimental UiO-66 samples, however, often contain structural defects in which some of the ligands and even complete inorganic nodes can be missing from the structure,[36,45−50] the latter preferentially ordering into correlated nanodomains adopting the reo topology (see Figure b).[51] Despite the presence of these defects, which can be largely “ironed out” in UiO-66(Zr) by carefully optimizing the synthesis conditions,[32,47] its outstanding stability makes UiO-66(Zr) an ideal candidate for synthetic or post-synthetic design strategies without appreciably altering the structural integrity of the material.[4,41,52−54] To increase its catalytic activity while still retaining a substantial stability, for instance, mixed-metal UiO-66 materials in which the zirconium ions were partially replaced by either hafnium or cerium ions were designed.[55−59] However, this targeted increase in catalytic activity was observed to come at the expense of a substantial and nonlinear decrease in thermal stability with increasing cerium content, deteriorating from ∼435 °C for UiO-66(Zr) to 220 °C for UiO-66(Ce), thereby undesirably limiting the range of experimental conditions for which the material can be adopted.[60,61] Given the extraordinary research interest in catalytically active UiO-66-type materials, it is crucial to identify whether this reduced stability is an intrinsic effect of UiO-66(Zr:Ce) materials or whether it is rather due to the specific synthesis protocol and can therefore be mitigated by further optimizing the synthesis conditions.

Figure 1

UiO-66(xZr:(6 – x)M) materials discussed in this manuscript. (a) The pristine and (b) the node-defective UiO-66 topologies with largest included spheres. (c) The different inorganic xZr:(6–x)M bricks consisting of x zirconium atoms on the one hand and 6 – x hafnium or cerium atoms on the other hand. The multiplicity of each inorganic brick is given between brackets. The shaded columns indicate those materials that are characterized experimentally.

To systematically elucidate the origin of this deteriorated stability and determine its dependence on the metal content of the inorganic brick, we here investigate a series of UiO-66 materials in which the zirconium cations are partially or completely replaced by either hafnium or cerium. This systematic experimental/computational procedure leads to 19 possible inorganic ZrCe6– and ZrHf6– bricks, depicted in Figure c. While bimetallic zirconium:cerium materials with varying metal composition were successfully synthesized,[60] these materials were observed to only contain the well-defined Zr6, Ce6 and Zr5Ce1 bricks in varying ratios.[62] However, as there is no fundamental reason why other bimetallic ZrCe6– bricks would be inaccessible during synthesis, all 19 bricks are retained in this study. Our in situ high-pressure PXRD data, obtained for the three monometallic MOFs and the two bimetallic MOFs with equal amounts of zirconium and either hafnium or cerium, demonstrate that the incorporation of cerium or, to a lesser extent, hafnium ions in UiO-66(Zr) decreases the material’s mechanical stability, confirming the earlier observed decrease in thermal stability upon cerium incorporation.[60,61] However, complementary MD simulations at operating conditions, which have been performed for all 19 UiO-66(xZr:(6–x)Ce) and UiO-66(xZr:(6–x)Hf) fcu materials containing one of the brick types in Figure c and, additionally, for the three reo-defective monometallic MOFs, unequivocally show that only the incorporation of cerium considerably reduces the intrinsic mechanical stability of UiO-66 in a gradual but nonlinear manner, an effect that is absent upon hafnium incorporation. The experimentally observed weaker mechanical stability of the hafnium-containing MOFs with respect to the UiO-66(Zr) sample is found to originate solely from the higher defect content of these samples and can therefore be mitigated by carefully optimizing the synthesis and activation procedures.[32,33]

The five experimental UiO-66 samples (see section S1 for the synthesis procedures) were characterized extensively (see section S2), confirming that MOF crystals consistent with the Fm3̅m space group were obtained.[44] Furthermore, it was verified that the bimetallic UiO-66(3Zr:3Hf) and UiO-66(3Zr:3Ce) samples show the targeted metal-to-metal ratio (see Table S2). For all materials, thermogravimetric analysis indicates that about 33% of the linker positions are unoccupied, except for the monometallic UiO-66(Zr) sample of ref (34), in which only 8.3% of the linker positions are unoccupied due to the optimized synthesis procedure (see Table S3). The in situ PXRD patterns, shown in Figure a for UiO-66(3Zr:3Hf) and Figures S13–S17 for all other samples, demonstrate that the diffraction peaks gradually broaden upon increasing pressure but remain consistent with the Fm3̅m space group for pressures up to 2 GPa, indicating that the crystallinity of the materials is reduced at elevated pressures without any phase transition. This loss of crystallinity is most pronounced for the bimetallic UiO-66(3Zr:3Ce) and the monometallic UiO-66(Ce) samples and remains present even after releasing the mechanical pressure again.

Figure 2

Experimental determination of the loss-of-crystallinity pressure at room temperature. (a). High-pressure powder X-ray diffraction patterns (λ = 0.410344 Å) obtained for the UiO-66(3Zr:3Hf) sample as a function of the applied pressure, with indication of the (111) peak used to determine the crystallinity. (b) Evolution of the crystallinity evaluated from the (111) Bragg reflection peak as a function of the applied pressure for the five experimental UiO-66 samples. Dotted lines are provided as a guide to the eye. (c) Experimental loss-of-crystallinity pressure extracted from the knee in panel b. The data for UiO-66(Zr) was obtained in ref (34).

To consistently quantify the mechanical resilience of the five samples, their remaining crystallinity at a given pressure is defined based on the broadening of the most prominent diffraction peak of the material, the (111) peak, following the approach outlined in ref (34) for UiO-66(Zr) (see section S1.5). As shown in Figure b, two crystallinity regions as a function of the applied pressure can be clearly distinguished for each of the five materials studied here. At low pressures, a rapid decrease in crystallinity with increasing pressure is observed, whereas this decrease is less pronounced or even absent at higher pressures where the solids can be considered as partially amorphous given the substantial broadening of the PXRD peaks with respect to atmospheric pressure.[34] The obtained loss-of-crystallinity pressure, that is, the critical pressure that separates the two regions, is reported in Figure c. Whereas an elevated loss-of-crystallinity pressure of ∼1400 MPa was observed earlier for UiO-66(Zr),[34] substituting the zirconium cations by either hafnium or cerium results in an appreciable decrease in mechanical stability. Unexpectedly, both the UiO-66(3Zr:3Hf) and UiO-66(Hf) samples show a reduced loss-of-crystallinity pressure of about 620 MPa despite the chemical similarity of the hafnium and zirconium cations. This effect is even more pronounced when substituting zirconium for cerium as the UiO-66(3Zr:3Ce) and UiO-66(Ce) samples remain stable only for pressures up to ∼210 MPa and ∼100 MPa, respectively, reminiscent of the earlier observed gradual decrease in thermal stability for these materials.[60,61]

While these experimental results demonstrate that substituting the zirconium cations in UiO-66(Zr) by either hafnium or cerium substantially reduces the material’s mechanical stability at elevated pressures, the origin of this effect remains unknown at this point. The reduced stability may be intrinsically linked to the hafnium- or cerium-containing inorganic bricks, thereby inherently limiting the conditions under which these materials can be used. Alternatively, it could originate from the increased amount of defects (∼33%) with respect to the zirconium-based material (∼8.3%), in which case the synthesis procedure could be optimized to reduce the amount of defects, similar to earlier extensive efforts to iron out the defects in UiO-66(Zr).[32,47]

Given the large amount of synthesis parameters that need to be optimized in order to decrease the defect content of our hafnium- and cerium-containing materials, experimentally isolating the intrinsic stability of these materials is not straightforward. Therefore, fully flexible and system-specific force fields are constructed for the 19 periodic UiO-66 materials containing the inorganic bricks depicted in Figure c to isolate the change in intrinsic stability due to the metal substitution. These force fields are derived from ab initio data using the QuickFF protocol for the covalent interactions,[63,64] similar to the approach followed in ref (39). As shown in section S4, good correspondence with both ab initio and experimental structural and mechanical properties is obtained, with a deviation in cell length that does not exceed 1.5 Å. Using these force fields, the loss-of-crystallinity pressure was derived for all 19 fcu materials by constructing the pressure-versus-volume equations of state based on MD simulations at 300 K, following the approach outlined in ref (65). As for other UiO-66-type materials,[35,39,66] these pressure equations of state consistently show two distinct regions as a function of the pressure (see section S4.3). In the elastic regime, near the equilibrium volume, the volume decreases linearly with increasing pressure, following Hooke’s law. At lower volumes, however, the equation of state deviates from linearity, indicating that the material is deformed plastically. The pressure necessary to bring the material to the critical volume separating both regimes is its computational loss-of-crystallinity pressure, as the symmetry of the crystallographic unit cell would be substantially reduced if the pressure was to be increased further (see section S4.4).

Figure shows the resulting loss-of-crystallinity pressures for all defect-free fcu structures (circles), indicating the intrinsic impact of metal substitution on the mechanical stability. For those materials in which the required metal ratio can be obtained by multiple inequivalent inorganic bricks (metal ratios 2:4, 3:3, and 4:2, see Figure c), the weighted average of the two inequivalent materials is reported. For the cerium-containing MOFs, a strong and nonlinear decrease in mechanical stability as a function of the cerium content is observed, although less pronounced than the experimental observations. As a result, the intrinsic effect of substituting zirconium by cerium in the inorganic bricks can already partially explain the nonlinear reduction in stability of the experimental UiO-66 samples upon increasing cerium content. In contrast, Figure demonstrates that the chemical similarity between zirconium and hafnium leads to bimetallic UiO-66(Zr:Hf) materials that have a comparable intrinsic stability as the monometallic UiO-66(Zr), indicating that extrinsic effects are fully responsible for the experimentally observed reduction in mechanical stability of the hafnium-containing samples.

Figure 3

Simulated loss-of-crystallinity pressures at 300 K for all UiO-66 materials as determined from the pressure equations of state, both in the pristine fcu (circles) and node-defective reo (squares) topologies. For those inorganic bricks with two configurations of the metal centers (see Figure c), the weighted average is shown, whereas the two independent results are included as semitransparant data points (please note the overlap for the systems with Zr:Hf ratios of 4:2, 3:3, and 2:4). Experimental results are indicated with stars.

Besides the intrinsic mechanical stability of the defect-free materials, the decreased mechanical stability of the experimental hafnium- and cerium-containing samples could also originate from the increased presence of defects with respect to the UiO-66(Zr) sample. For the latter, the synthesis procedure has been extensively optimized.[32] We demonstrated earlier that reducing the inorganic brick coordination in UiO-66(Zr) from 12 to 11 linkers, corresponding with the defect content of our experimental UiO-66(Zr) sample, leads to a reduction in the loss-of-crystallinity pressure from 1.8 GPa to 1.2–1.5 GPa, coinciding with the experimental loss-of-crystallinity pressure of UiO-66(Zr) in Figure .[39] Given that the cerium- and hafnium-containing UiO-66 samples are even more defective, we have derived additional force fields for the monometallic UiO-66(Zr), UiO-66(Ce), and UiO-66(Hf) structures containing node defects. These node-defective materials exhibit the reo topology postulated by Cliffe et al. and contain the experimentally determined amount of linker vacancies (see Figure b), although the same linker-to-node ratio could in theory also be reached by linker vacancies alone.[51] The pressure-versus-volume equations of state for these node-defective reo materials reveal loss-of-crystallinity pressures that are consistently about 0.7–1 GPa lower than the intrinsic mechanical stability observed for the defect-free fcu materials. This weakening is largely independent of the metal content of the inorganic brick and can hence be considered as an orthogonal effect to the intrinsic stability.

As demonstrated in Figure , this procedure leads to a predicted mechanical stability that approaches the experimental values for all samples. To obtain quantitative agreement between the experimental and computational stabilities reported in Figures c and 3, it would be necessary to depart from our assumption that all inorganic bricks in a given material are identical but rather take into account the possibility that different types of inorganic bricks can coexist, as observed in ref (62). However, as the coordination bonds between the inorganic bricks and the organic ligands form very local nucleation points for mechanical instability in these materials, the stability of bimetallic materials with multiple inorganic brick types can be assumed to be defined by the stability of the weakest inorganic brick type, which immediately follows from our results in Figure . As a result, Figure reveals that the reduced mechanical stability of the experimental UiO-66(3Zr:3Hf) and UiO-66(Hf) samples is not an intrinsic effect but only stems from defect-induced weakening. In addition, defect-induced weakening also further decreases the already lower intrinsic stability of the UiO-66(Ce) and UiO-66(3Zr:3Ce) samples.

In conclusion, we have demonstrated through high-pressure in situ PXRD experiments that incorporating hafnium or cerium in the inorganic bricks of UiO-66(Zr) substantially decreases the mechanical stability of these MOFs. This effect was found to be the most pronounced for the cerium-containing materials, following the same nonlinear trend as a function of the cerium content as their thermal stability.[60,61] To elucidate the origin of this decreased stability and isolate whether this effect is intrinsic or rather due to the different synthesis protocols for the various samples, complementary MD simulations have been performed using ab initio-based flexible force fields. These simulations indicate that the nonlinear decrease in mechanical stability of the cerium-containing samples is an intrinsic effect of the inorganic brick, whereas the zirconium:hafnium ratio does not affect the intrinsic stability of the hafnium-containing MOFs. Furthermore, the increased presence of defects in the UiO-66(Ce), UiO-66(3Zr:3Ce), UiO-66(Hf), and UiO-66(3Zr:3Hf) samples, a result of the highly optimized synthesis protocol for UiO-66(Zr), yields an additional weakening of these experimental samples. This defect-induced weakening, which was found to reduce the stability of the materials by 0.7–1 GPa independent of the content of the inorganic brick, fully accounts for the reduced stability of the hafnium-containing samples and further decreases the stability of the cerium-containing samples in good agreement with the experimental results. This combined experimental/computational approach therefore demonstrates that carefully optimizing the synthesis protocol may lead to designed UiO-66(Zr:Hf) MOFs with an extraordinary mechanical stability similar to UiO-66(Zr), whereas the intrinsic stability of bimetallic UiO-66(Zr:Ce) MOFs rapidly decreases with increasing cerium content, independent of the synthesis protocol. To further bring these bimetallic MOFs to practical applications in, for example, heterogeneous catalysis or sorption, it will be essential to extend these design principles to also account for the lifetime of these materials during the process.