Tuning Hydrophilicity of Aluminum MOFs by a Mixed‐Linker Strategy for Enhanced Performance in Water Adsorption‐Driven Heat Allocation Application

Abstract Water adsorption‐driven heat transfer (AHT) technology has emerged as a promising solution to address crisis of the global energy consumption and environmental pollution of current heating and cooling processes. Hydrophilicity of water adsorbents plays a decisive role in these applications. This work reports an easy, green, and inexpensive approach to tuning the hydrophilicity of metal–organic frameworks (MOFs) by incorporating mixed linkers, isophthalic acid (IPA), and 3,5‐pyridinedicarboxylic acid (PYDC), with various ratios in a series of Al−xIPA‐(100−x)PYDC (x: feeding ratio of IPA) MOFs. The designed mixed‐linkers MOFs show a variation of hydrophilicity along the fraction of the linkers. Representative compounds with a proportional mixed linker ratio denoted as KMF‐2, exhibit an S‐shaped isotherm, an excellent coefficient of performance of 0.75 (cooling) and 1.66 (heating) achieved with low driving temperature below 70 °C which offers capability to employ solar or industrial waste heat, remarkable volumetric specific energy capacity (235 kWh m−3) and heat‐storage capacity (330 kWh m−3). The superiority of KMF‐2 to IPA or PYDC‐containing single‐linker MOFs (CAU‐10‐H and CAU‐10pydc, respectively) and most of benchmark adsorbents illustrate the effectiveness of the mixed‐linker strategy to design AHT adsorbents with promising performance.

Torr). N2 isotherms were measured by a volumetric sorption analyzer (Micromeritics 3flex) at liquid N2 temperature (−195.8 °C). The BET surface area (SBET) was calculated using the Brunauer-Emmett-Teller equation from adsorption isotherm curve, and the total pore volume (VP) was measured by a single point method at P/P° = 0.95. Furthermore, the micropore surface area (Smicro) and the micropore volume (VP,micro) were determined by the t-plot method. To obtain the pore size distribution (PSD) of the micropore of samples, Ar isotherm was acquired at liquid Ar temperature (−185.8 °C) using ASAP 2020 equipment. The PSD were fitted by its Ar adsorption isotherm curve with the cylindrical model of Horvath-Kawazoe equation with Ar-oxide model.
The thermogravimetric curve (TG) was obtained with the Scinco TGA i-1000 model under 30 cm 3 min −1 N2 with a 5 °C min −1 ramping rate after exposure to water vapor at 80% relative humidity in a chamber controlled by the NH4SO4 salt solution.
Water sorption measurements. Water sorption isotherms of samples were measured by an intelligent gravimetric analyzer (IGA, Hiden Analytical Ltd.). The IGA was automatically operated to finely control the water vapor pressure (RH of 1-95%) and temperature (20-100 °C). Before measuring the water adsorption experiments, the samples were degassed at 150 °C for 6 h under high vacuum (< 10 −6 Torr). Multiple cycles of water adsorption-desorption were carried out using a thermogravimetric analyzer (TGA, DT Q600, TA Instruments, Universal V4.5A) connected with a humidity generator. The humidity of gas stream was controlled using two mass flow meter, and a humidified nitrogen gas flow was passed through a thermogravimetric chamber. During the water adsorption-desorption cyclic measurement, the weight of dehydrated sample was determined by first adsorption cycle after activated at 150 °C for 1 h. After then, the adsorption profiles were measured at 30 °C in humid nitrogen at 35% of RH, while desorption data were collected at 70 °C in 4.8% of RH nitrogen flow at a cycle time of 1.5 h.
Water desorption kinetic. The desorption experiments of the adsorbents were conducted by the same TGA instrument (DT Q600, TA instruments, Universal V4.5A). Before the experiments, all samples were allowed to adsorb water to saturation under humid flow gas of relative humidity of 60% for 90 minutes. Then, the samples were heated from 30 °C to 150 °C under various heating rates of 4, 6, 8, 10, and 12 o C min −1 under humid flow gas mentioned above. The sample weight for the measurement was about 10 mg.

Molecular simulations.
The initial experimental structural model of the hydrated KMF-2 was geometry optimized by Density functional theory (DFT) calculations maintaining fixed the cell parameters. These DFT calculations were carried out within the Gaussian Plane Waves method as implemented in CP2K package within Generalized Gradient Approximation (GGA) with the Perdew-Burke-Ernzerhof (PBE) exchange functional. [1] The energy cutoff for the plane waves is set at 600 Ry. Goedecker, Teter, and Hutter (GTH) pseudo-potentials approximation [2] and double zeta basis sets (DZVP) [3] were employed while DFT-D3 van der Waals dispersion corrections were considered with a cut-off radius of 10 Å. [4] The Density Derived Electrostatic and Chemical (DDEC) net atomic charges [5] were further computed using the DFT optimized electron density as an input (see the cif file of the DFT optimized structure incorporating the charges). The textural properties including theoretical free pore volume, pore limiting diameter and N2-accessible surface area were computed using a geometric method as implemented in the Zeo++ software package.
Grand-Canonical Monte Carlo (GCMC) simulations were performed to predict the water adsorption isotherms at Tads = 30 °C. The simulation box was made of 12 conventional unit cells (2×2×3) maintaining the atoms fixed in their initial positions. The interactions between the guest water molecules and the MOF structure were described by a combination of site-to-site LJ contributions and Coulombic terms. The Lennard-Jones (LJ) parameters for all atoms of the MOF framework were taken from the generic UFF forcefield. [6] Following the treatment adopted previously for MIL-160, [7] the hydrogen atoms of both the μ-OH moieties interact with water via only a Coulombic term. The TIP4P/2005 model [8] was used as a microscopic model to represent the water molecule. Short-range dispersion forces were truncated at a cutoff radius of 12 Å while the cross-term LJ parameters were calculated by means of the Lorentz-Berthelot combination rule.
The long-range electrostatic interactions were handled using the Ewald summation technique. For each point of the adsorption isotherm, 4x10 8 MC cycles were considered to ensure the convergence. In order to gain insight into the configurational distributions of the water in the MOF, additional data were calculated at different pressure including the hydrogen bond networks and the radial distribution functions (RDF) of the intermolecular atomic pairs of the water and the MOF framework as well as water-water.

Determination of the isosteric heat of adsorption
To assess the COP of a working pair, knowledge of the enthalpy of adsorption is of prime importance. [7] As described in more detail in Ref. [7], isosteric heats of adsorption were estimated by the Clausius-Clapeyron equation and compensated by the virial equation. [9] The isosteric enthalpy of adsorption was calculated from the isotherms at multiple temperatures by Eq. (S1): where ΔadsHW is the isosteric enthalpy of adsorption, R is the universal gas constant, p and T represent temperature and pressure, respectively, and W is the volume of water (liquid) adsorbed per volume of adsorbent (crystalline densities are used for the conversion). [7] Furthermore, the virial equation was also employed to calculate the isosteric heat of adsorption (Qst) from isotherms measured at different temperatures. [9] Ln = ln + ( where v is quantity adsorbed, and ai and bi are empirical parameters. The virial-type equation was applied compared with the Clausius-Clapeyron equation, as shown in Figure S12. A set of temperature-independent parameters, which lead to direct evaluation of Qst can be derived by fitting Eq. (S2) continuously using adsorption isotherms obtained at different temperatures. [9] = − ( Water uptake / mL (liq) g -1

Evaluation of activation energy for desorption (Ed)
Activation energy of desorption (Ed) is derived from Kissinger equation. [10] d(ln where Td (K) is temperature at the highest desorption rate recorded through derivative thermogravimetry (DTG) curve, β (K min −1 ) is heating rate, R is ideal gas constant (J K −1 mol −1 ).   Thermodynamic calculations. The thermodynamic evaluation of adsorption cooling and heat pump cycles were calculated using the methodology previously reported by De Lange et al. [7] The coefficient of performance (COP) is used to evaluate the energy efficiency of the cooling and heat pump cycle from a thermodynamic perspective. The energy analysis allows the determination of the COP, which is a ratio of useful heating or cooling energy output provided to work required.
Calculation details were provided as follows

Detailed procedures of thermodynamic evaluation for AHT application
For calculation procedures, a characteristic curve needs to be constructed to transfer the loading from two dependent variables (p, T) to one, i.e., the adsorption potential, A, which is defined as the molar Gibbs free energy of adsorption with an opposite sign, defined as: where P° is the temperature-dependent vapor pressure of the adsorbate of choice. The amount adsorbed should be expressed as the volume occupied by the adsorbed phase. As the density of the adsorbed phase is often not known, the liquid phase density is used as an approximation: where q is the mass adsorbed, W is the volume liquid adsorbed, and ρliq wf is the liquid density of the same adsorbate and E is the characteristic energy of adsorption. [11] If temperature invariance is assumed, all measured adsorption data should collapse onto one single "characteristic curve". An adsorption-driven heat pump cycle can be used for either heating or cooling. As the working pair (sorbent-sorbate) is known, only four distinct temperature levels need to be defined to determine the COP for either application. [7] Notably, the desorption temperature was varied to investigate the required desorption temperature. The COP is defined as the useful energy output divided by the energy required as input. For heating, this becomes: where Qcon is the energy released during condensation, and Qads is the energy released during adsorption. Both have a negative value, as energy is withdrawn from the adsorption cycle.
Qregen is the energy required for regeneration of adsorbent. In this case, a positive quantity as energy is added to the system. For cooling, the coefficient of performance COPC becomes: where Qev is the energy withdrawn by the evaporator. Notably, the COPH should have a value between 1 and 2, and the COPC is per definition not larger than unity. The specifics on how to exactly calculate these energetic contributions are explained in detail elsewhere. [7] Finally, it must be noted that the specific heat capacity is assumed to be 1 J/(g · K), which is an average value for a variety of MOF materials. In fact, the actual value of this quantity has a negligible effect on calculated COP values.               d Values of crystal density, working capacity, heat from evaporator, and heat storage capacity for MIP-200, [12] Co-CUK-1, [13] CAU-10-H, [14] MOF-303, [15] SAPO-34, [7] MIL-160, [16] CAU-23 [17] and KMF-1 [18] were calculated by characteristic curves and data taken from references. CAU-23 AlCl3·6H2O b 2,5-Thiophenedicarboxylic acid 84 9.81 [17] a Material cost was calculated by using the required amounts of metal precursor and ligand precursor for obtaining 1g of adsorbent excluding not only the prices related to starting materials such as solvent, modulator, and additives but also the operation and utility cost for hydrothermal reaction, filtrate, and drying etc. b The actual metal precursor is solution of AlCl3. For calculation, the cheapest solid salt (AlCl3·6H2O) is selected.