Thermodynamics Drives the Stability of the MOF-74 Family in Water.

The stability of functional materials in water-containing environments is critical for their industrial applications. A wide variety of metal-organic frameworks (MOFs) synthesized in the past decade have strikingly different apparent stabilities in contact with liquid or gaseous H2O, ranging from rapid hydrolysis to persistence over days to months. Here, we show using newly determined thermochemical data obtained by high-temperature drop combustion calorimetry that these differences are thermodynamically driven rather than primarily kinetically controlled. The formation reaction of a MOF from metal oxide (MO) and a linker generally liberates water by the reaction MO + linker = MOF + H2O. Newly measured enthalpies of formation of Mg-MOF-74(s) + H2O(l) and Ni-MOF-74(s) + H2O(l) from their crystalline dense components, namely, the divalent MO (MgO or NiO) and 2,5-dihydroxyterephthalic acid, are 303.9 ± 17.2 kJ/mol of Mg for Mg-MOF-74 and 264.4 ± 19.4 kJ/mol of Ni for Ni-MOF-74. These strongly endothermic enthalpies of formation indicate that the reverse reaction, namely, the hydrolysis of these MOFs, is highly exothermic, strongly suggesting that this large thermodynamic driving force for hydrolysis is the reason why the MOF-74 family cannot be synthesized via hydrothermal routes and why these MOFs decompose on contact with moist air or water even at room temperature. In contrast, other MOFs studied previously, namely, zeolitic imidazolate frameworks (ZIF-zni, ZIF-1, ZIF-4, Zn(CF3Im)2, and ZIF-8), show enthalpies of formation in the range 20-40 kJ per mole of metal atom. These modest endothermic enthalpies of formation can be partially compensated by positive entropy terms arising from water release, and these materials do not react appreciably with H2O under ambient conditions. Thus, these differences in reactivity with water are thermodynamically controlled and energetics of formation, either measured or predicted, can be used to assess the extent of water sensitivity for different possible MOFs.

Introduction

Metal–organic frameworks (MOFs) are a fascinating class of crystalline porous materials, constructed from the coordination of inorganic metal nodes and organic linkers.[1−3] Their potential functionalities, topologies, and physicochemical properties can be readily tuned by choosing appropriate components in the process of MOF assembly through the reticular chemistry approach.[4,5] As a result, a library of nanoporous MOF materials with ultrahigh specific surface areas, tailored pore sizes, and mechanical properties has been successfully synthesized, members of which have demonstrated great promise for various applications including gas storage and separation, sensing, drug delivery, catalysis, and energy storage.[6−10] However, though MOFs possess extremely high surface areas compared to zeolites and activated carbons, many show limited hydrothermal stability.[11−13] This significantly hinders their practical applications because H2O is commonly present in different forms in industrial processes. Thus, to prevent undesirable hydrolysis reactions, different strategies have been employed, including, but not limited to, constructing carboxylate MOFs with high valence metal cations,[14] building azolate frameworks with nitrogen-donor ligands,[15] and functionalizing pore walls with hydrophobic groups.[16] Despite great progress in fabricating MOFs with enhanced water stability, producing highly water/moisture stable MOFs still remains a daunting task. Therefore, it is imperative to explore the thermodynamic stability of MOFs in the presence of water.

Specifically, the present work seeks to answer whether water sensitivity for the MOF-74 family is driven by thermodynamics or controlled by kinetic factors. For initial thermodynamic analysis, we chose the MOF-74 family because of its wide recognition, particularly Mg-MOF-74 is known as one of the best candidates for the efficient postcombustion CO2 capture from water-containing flue gas generated from coal-fired power plants.[17,18] In contrast to other carboxylate MOFs, the MOF-74 family is very water-sensitive, thus it provides a good test of the importance of thermodynamic factors in determining MOF stability in the presence of water.

This work reports the first experimental study of the thermodynamic stability of water-sensitive MOFs using high-temperature drop combustion calorimetry. Unlike traditional room temperature acid or base solution calorimetry where the thermochemical parameters of MOFs are calculated from the measured heats of dissolutions,[19−21] the new methodology utilizes the heats of combustions at elevated temperature without any solvent. By using a specially designed dropping device developed initially for the study of actinide materials,[22] the activated samples can be kept in a totally water-free environment throughout the experiment. Importantly, drop combustion calorimetry allows the investigation of the thermodynamic properties of MOFs, which are not soluble or have limited solubility in acid or base solution at room temperature, hence dramatically extending the scope of MOF choice, including the MOF-74 family. Both methods have their merits. Complete solubility of the sample can be an issue in room temperature solution calorimetry, because incomplete dissolution or unexpected side reactions in the solvent may significantly affect the measured heat effect. Drop combustion calorimetry avoids this problem because of the high temperature and oxygen atmosphere, which convert organic–inorganic materials into metal oxide (MO), CO2, and H2O. However, unlike the large heat effects of drop combustion calorimetry, the heat effects of solution calorimetry are much smaller, and hence the presence of possible impurities can have less impact on the recorded heat. Thus, combustion calorimetry needs to be carried out very carefully and with well-characterized samples. This study paves the way for the thermochemical study of a much larger variety of MOFs, which will facilitate fundamental understanding and guide rational design and fabrication of MOFs with higher hydrothermal stability.

Results and Discussion

The isostructural Mg-MOF-74 and Ni-MOF-74 were synthesized according to the established protocols from the literature with details given in the experimental section. Powder X-ray diffraction (PXRD) was used to confirm the phase purity of the materials as shown in Figure A. Both samples have diffractograms well matching those of Mg-MOF-74 and Ni-MOF-74 reported in the literature.[23] The porosity was analyzed by N2 physisorption measurements (Figure B). As expected, both materials show typical type-I sorption isotherms with sharp N2 uptake at very low pressures, indicative of the microporous structure. The calculated Brunauer–Emmett–Teller (BET) surface areas are 1141 and 956 m2/g and give pore volumes of 0.49 and 0.47 mL/g for Mg-MOF-74 and Ni-MOF-74, respectively. The obtained surface areas are within the range reported in the literature and indicate high purity of the samples. For example, 940 m2/g was reported for Mg-MOF-74 and 941 m2/g for Ni-MOF-74.[24,25] The metal to ligand stoichiometry was confirmed by thermogravimetric analysis (TGA) in O2 (Figure S1). The weight loss was calculated from 250 to 600 °C and corresponds to 2:1 metal/ligand molar ratio within ±5 wt % uncertainty.

Figure 1

(A) PXRD patterns and (B) N2 sorption isotherms of Mg-MOF-74 and Ni-MOF-74.

The heats of combustion (ΔHcom) of the activated MOFs and their corresponding ligand, 2,5-dihydroxyterephthalic acid (H4DOBDC), were measured at 800 °C in O2. The MOF completely oxidizes to a MO (MgO or NiO), and gaseous CO2, and H2O. Using an appropriate thermochemical cycle (Table ) and the measured ΔHcom values (Table ), the enthalpy of reaction (ΔHr°) between MO (M = Mg or Ni) and H4DOBDC ligands forming the M-MOF-74 and H2O (liquid) at room temperature was determined. The obtained values of ΔHr° are 303.9 ± 17.2 and 264.4 ± 19.4 (kJ/mol of M) for Mg-MOF-74 and Ni-MOF-74, respectively. Mg-MOF-74 is slightly more energetically unstable than Ni-MOF-74. These positive values indicate that the formation of M-MOF-74 plus liquid H2O from oxide and linker is strongly endothermic, which directly means the reverse reaction of MOF-74 and H2O (eq ) is highly exothermic. The exothermic reaction suggests the metastability of the MOF with respect to hydrolysis

Table 1

Thermochemical Cycle Used to Calculate the Enthalpy of Reaction at Room Temperature (ΔHr°) between MO (M = Mg or Ni) and H4DOBDC Ligand Forming the M-MOF-74 and H2O via High-Temperature Drop Combustion Calorimetry at 800 °C in O2a

reactions used in the thermochemical cycle	enthalpy
M2[C8H2O6](s,25°C) + 6.5O2(g,800°C) → 2MO(s,800°C) + 8CO2(g,800°C) + H2O(g,800°C)	ΔH1 = ΔHcom(MOF-74)
C8H6O6(s,25°C) + 6.5O2(g,800°C) → 8CO2(g,800°C) + 3H2O(g,800°C)	ΔH2 = ΔHcom(H4DOBDC)
MO(s,25°C) → MO(s,800°C)	ΔH3 = H800° – H25°
H2O(l,25°C) → H2O(g,800°C)	ΔH4 = H800° – H25°
2MO(s,25°C) + C8H6O6(s,25°C) → M2[C8H2O6](s,25°C) + 2H2O(l,25°C)	ΔHr° = 2ΔH3 + ΔH2 – ΔH1 – 2ΔH4

ΔH1 and ΔH2 are experimentally measured and their values are listed in Table , ΔH3 and ΔH4 are the heat contents of MO and H2O at 800 °C and their corresponding values can be found in ref (35).

Table 2

Thermodynamic Data Obtained by High-Temperature Drop Combustion Calorimetry Measurements at 800 °C in O2

compound	formula	ΔHcom (kJ/mol formula)	ΔHr° (kJ/mol of M)	H800° – H25° (kJ/mol)
Mg-MOF-74	Mg2[C8H2O6]	–3079.9 ± 16.44	303.89 ± 17.21
Ni-MOF-74	Ni2[C8H2O6]	–2989.9 ± 18.76	264.38 ± 19.44
H4DOBDC	C8H6O6	–2394.0 ± 5.12
MgO	MgO			34.24
NiO	NiO			39.73
H2O	H2O			73.30
ZIF-1	C6H6N4Zn		19.9 ± 2.5[37]
ZIF-4	C6H6N4Zn		22.1 ± 2.7[37]
ZIF-7	C14H10N4O2.24Zn		27.2 ± 3.9[37]
ZIF-8	C8H10N4Zn		27.1 ± 1.9[37]
qtz-Zn(CF3Im)2	C7H6N4F3Zn		27.1 ± 1.0[41]
SOD-Zn(CF3Im)2	C7H6N4F3Zn		42.7 ± 1.1[41]

The release of H2O in the formation reaction and its consumption during hydrolysis will make a small entropic TΔS contribution to the free energy of reaction. The enthalpic contribution especially at room temperature and with liquid water as a reference state is by far the larger term, so the free energy of reaction is dominated by ΔH. The persistence of a phase under potential reaction conditions is governed by an interplay between thermodynamic driving forces (negative free energy of reaction) and kinetic barriers. In general, the larger the thermodynamic driving force the more likely the system is to overcome any potential kinetic barrier to reaction. Thus, given the very large exothermic driving forces for hydrolysis, it is not surprising that the two MOF-74 samples degrade in the presence of water.

In a fully dehydrated state, the metal nodes in the MOF-74 structure are coordinated with the five oxygen atoms of carboxyl and hydroxy groups of the ligand, leaving one coordinatively unsaturated metal site, which was originally occupied by water or solvent molecules. The formed structure with etb topology creates unimodal ∼1.1 nm hexagonal pores composed of square-pyramidal metal-oxide clusters forming the vertices of the pores. The exposed coordinatively unsaturated metal atoms, which serve as Lewis acid sites, are known to strongly bind with Lewis bases, such as H2O and CO2 molecules with remarkably high uptake at room temperature and low partial pressures,[26,27] which is supported by our thermodynamic results. This means that the interaction of water molecules with MOF-74 is strongly thermodynamically driven and not simply a matter of kinetics.

Because of such a high water affinity, Mg-MOF-74 undergoes structural changes or chemical transformations when exposed to water at ambient temperatures.[28−30] Specifically, it has been demonstrated that though Mg-MOF-74 exhibits superior adsorption capacity for CO2 in dry flue gas, the CO2 capacity was dramatically decreased to 16% of the initial capacity after exposure at 70% relative humidity (RH).[26] Furthermore, after 1 day of exposure at 75% RH, the surface area of Mg-MOF-74 decreased by 74% and after 14 days was only 6 m2/g.[31] This may suggest some irreversible decomposition on water exposure. Although the XRD pattern may still show some retention of the original MOF diffraction peaks, the loss of surface area strongly supports partial structural collapse and amorphization. It has been determined experimentally that Ni-MOF-74 is somewhat more water stable than the Mg-MOF-74.[32] In addition, Jiao et al.[33] observed that a small amount of Ni doping into Mg-MOF-74 can increase the percentage of surface area retained after water exposure compared to that of pure Mg-MOF-74. Density functional theory calculations demonstrate that H2O molecules bind more strongly with Mg (73.2 kJ/mol) open metal sites than with Ni (60.6 kJ/mol).[34] This persistent difference is confirmed by our thermodynamic results and 39.5 kJ/mol energetic difference between two MOFs indicates the higher hydrolytic stability of Ni-MOF-74 compared to its magnesium counterpart.

Based on molecular dynamic simulations and different experimental observations, the initial stage of MOF-74 hydrolysis involves M–OH bond formation.[12,29] This has been demonstrated by employing a reactive force field simulation for Zn-MOF-74.[36] Upon interaction with the active electrophilic metal centers, the dissociation of water molecules into H+ and OH– species is influenced by the radius of the metal ion, OH– combines with M2+ whereas H+ protonates the oxygen atoms on the ligand. For this reason, we have also calculated the ΔHr assuming metal hydroxide as the product instead of oxide after the hydrolysis (Tables S1 and S2). The data show that ΔHr is 378.7 ± 17.4 (kJ/mol of Mg) for Mg-MOF-74 and 294.0 ± 19.5 (kJ/mol of Ni) for Ni-MOF-74, showing that the hydrolysis of MOF-74 forming M(OH)2 is even more exothermic compared to the formation of the oxide, and maintaining the difference between Mg and Ni.

A more complete picture of the thermodynamic stability of MOFs relative to hydrolysis can be obtained by combining these new results with previously published data on MOF formation enthalpies using solution calorimetry (Table ).[20,37−41] Compared to M-MOF-74, the ΔHr of MOF plus water forming oxide and ligand is much less exothermic for a series of zinc-containing zeolitic imidazolate frameworks (ZIFs), ranging from −42 to −19.5 kJ/mol of Zn.[37,40] This small exothermic effect, partially compensated by negative entropy terms arising from water consumption, implies the high stability of ZIFs with respect to water and indeed their hydrolytic stability has been confirmed by numerous literature reports.[42]

Such a large enthalpy difference (200–250 kJ/mol of M) between ZIFs + H2O and (Mg,Ni)-MOF-74 + H2O can be used to evaluate the magnitude of water sensitivity for different MOFs. For divalent MOs, one can find that, with increasing electronegativity of the metal atom, the stability of oxides against hydrolysis increases. The standard formation enthalpy of Mg(OH)2 from MgO plus H2O(l) is −37.4 kJ/mol and for Ni(OH)2 is −14.8 kJ/mol,[43] hence NiO is somewhat less unstable in water than MgO. Because carboxylate MOFs are composed of M–O moieties, this trend can be simply extended to explain the hydrolytic stability of MOFs as well. Therefore, Ni-MOF-74 is less hydrolytically unstable than Mg-MOF-74. It should be noted that the large enthalpy values obtained from the thermochemical cycles are specific to the overall hydrolysis reaction indicating that these MOFs will eventually undergo hydrolysis under the presiding thermodynamic driving force, perhaps slowly at room temperature or at a much higher rate at elevated temperatures.

This methodology can be generalized to investigate the thermodynamics of condensation/polycondensation reactions of various organic molecules (aldol condensation, Claisen condensation, and so on). Note, vitally important peptide bond formation is also a condensation reaction between two amino acids with the release of water molecules and, therefore, the developed method can be also employed to analyze thermodynamics in forming higher peptides from primary amino acids.

In summary, the thermodynamic stability of two MOF-74 materials in water was experimentally analyzed using a new methodology of high-temperature drop combustion calorimetry. These results, in conjunction with the previously obtained thermochemical data for other MOFs by room temperature solution calorimetry, define a general trend of energetics, indicating that the hydrolytic stability of the MOF-74 family is thermodynamically driven and not simply a matter of kinetics. This study provides a fundamental understanding to design and synthesize MOFs with high water stability for various emerging applications.

Experimental Methods

Synthesis of Mg-MOF-74

0.712 g of Mg(NO3)2·6H2O and 0.167 g of 2,5-dihydroxyterephthalic acid were dissolved in a mixture of 67.5 mL of dimethylformamide (DMF), 4.5 mL of ethanol, and 4.5 mL of water under sonication in a 250 mL screw cap jar. The reaction jar was capped tightly and heated in an oven at 125 °C for 26 h. After cooling to room temperature, the solvent was carefully decanted from the product and replaced with methanol. The solvent exchange was carried out six times over the next three days. The solvent was removed in a dynamic vacuum at 220 °C for 12 h.

Synthesis of Ni-MOF-74

0.602 g of Ni(NO3)2·6H2O and 0.12 g of 2,5-dihydroxyterephthalic acid were dissolved in a mixture of 45 mL of DMF, 3 mL of ethanol, and 3 mL of water under sonication in a 250 mL screw cap jar. The reaction jar was capped tightly and heated in an oven at 110 °C for 60 h. After cooling to room temperature, the solvent was carefully decanted from the product and replaced with methanol. The solvent exchange was carried out six times over the next three days. The solvent was removed in a dynamic vacuum at 220 °C for 12 h.

Characterization

The crystal structure and purity of MOFs were characterized by PXRD with a Bruker D8 (AXS) Advance diffractometer, employing Ni-filtered Cu Kα radiation at 40 kV and a detector current of 40 mA. The N2 sorption experiments were performed using a Micromeritics ASAP 2020 instrument at 77 K and specific surface areas were calculated from the BET equation. Prior to gas sorption experiments, the powders were degassed at 493 K overnight under 10–6 Torr vacuum. The TGA was conducted using a Labsys evo Setaram instrument at a ramping rate of 10 °C/min in a 30 mL/min oxygen flow.

High-Temperature Drop Combustion Calorimetry

The drop combustion enthalpies (ΔHcom) of MOFs were obtained with a Calvet-type twin calorimeter (AlexSYS Setaram Inc.) in an empty quartz crucible under continuous 70 mL min–1 oxygen flow at 800 °C. To prepare the dropping pellets, the MOF samples were degassed at 220 °C for 12 h under 10–6 Torr vacuum to remove the adsorbed water. Then, the sample containing a sealed tube was transferred into an Ar-filled glove box with <1 ppm H2O and O2 concentration, respectively. It is very important to use completely dehydrated MOF samples because a small amount of adsorbed water can affect the combustion reaction. Because of the low density of MOFs, for each experiment around 2 mg of the samples were weighed inside the glove box using a microbalance with an accuracy of 1 μg. Subsequently, the weighed pellets were placed inside the specially designed 3D printed dropper, which allows transferring the sample from the glove box to the calorimeter without exposing samples to air. For each sample, 4–6 dropping measurements were conducted and an average value was used for the thermodynamic calculations. The heat effect of benzoic acid was used to determine the calibration factor of the instrument.