MOLECULAR FRONTIERS JOURNAL

The synthetic scalability of water harvesting metal–organic frameworks (MOFs) is crucial for making these promising materials accessible and widely available for use in household devices. Herein, we present a facile, sustainable, and high-yield synthesis method to produce a series of water-harvesting MOFs, including MOF-303, CAU-23, MIL-160, MOF-313, CAU-10, and Al-fumarate. Using readily available reactants and water as the only solvent, we were able to synthesize these materials at the kilogram scale in a 200 L batch reactor with yields of 84–96% and space-time yields of 238–305 kg/day/m 3 under optimized reaction conditions. We also show that our procedure preserves framework crystallinity, porosity, and water-harvesting performance of the MOFs synthesized at scale.


INTRODUCTION
Water scarcity is a major global challenge, with over half the world's population experiencing shortages of this vital resource 1,2 . The temporal and spatial variation in the distribution of available water resources on this planet and the lack of transportation and supplement infrastructure make fresh water even more precious in some arid regions 3,4 . Fortunately, the use of sorbent-assisted atmospheric water harvesting devices has been proposed as a promising solution to address this issue. These devices use porous solid sorbents to trap water molecules at low relative humidity (RH) and release them with minimal energy expenditure [5][6][7][8] . Since this approach is neither spatially nor temporally restricted, the supplement of fresh water by the water harvesting device equipped with desirable sorbents can be achieved at any location and at any time of the year 7,9 . However, the development of suitable porous materials for use in these devices, particularly for household use and widespread accessibility, has been a challenge. In addition to having excellent water sorption properties, a desirable sorbent should be scalable, cost-effective, and reproducible in manufacturing in order to be practical for use in various complex realworld conditions [9][10][11] .
Microporous metal-organic frameworks (MOFs) have recently gained attention due to their high tunability, unprecedented variety, and intrinsic porosity, and have been successfully employed in desert water harvesting 5,6,10,[12][13][14][15] . While a significant body of work has been created in developing MOF sorbents with tailored capacity and isotherm shape, it has largely been limited to small-scale proof of concepts and there are few reports on the large-scale synthesis of water harvesting MOFs for industrial use [16][17][18] . On the other hand, it was proposed that the concept of green synthesis methods using water as a solvent can address the challenges of high production cost and environmental hazards in the industrial-scale synthesis of MOFs [19][20][21] . As such, our group has previously developed a green, scalable, and cost-effective method for the synthesis of aluminum-based MOFs on a kilogram scale using inexpensive and readily available starting materials 22,23 .
Prior to the analysis and characterization, the MOF samples were washed with H 2 O and ethanol, and evacuated at room temperature for 1 hour and at 120°C for an additional 12 hour. To analyze the linker composition of the MOF compounds with nuclear magnetic resonance (NMR) spectroscopy, the activated sample was fully hydrolyzed using a NaOD solution (10% in D 2 O). 1 H-NMR spectra were acquired on Bruker NEO-500 MHz spectrometers, and elemental analysis (EA) was performed on a PerkinElmer 2400 Series II CHNS elemental analyzer at the NMR facility and Microanalytical Laboratory of the College of Chemistry, University of California, Berkeley. Thermal gravimetric analysis (TGA) curves were taken using a TA Q500 thermal analysis system with a heating rate of 5°C/min under N 2 flow. Powder X-ray diffraction (PXRD) patterns were recorded using a Rigaku MiniFlex 6G equipped with a HyPix-400MF Hybrid Pixel Array detector and a normal focus X-ray tube with a Cu-source (λ = 1.54178 Å). Nitrogen sorption experiments were conducted using a Micromeritics Accelerated Surface Area and Porosimetry (ASAP) 2420 System. During the measurement, the sample was cooled to 77 K in a liquid nitrogen bath. Water vapor sorption experiments were carried out on a BEL Japan BELSORP-aqua or a Micromeritics 3Flex Surface Characterization Analyzer. The water vapor source was degassed through five freezepump-thaw cycles before the analysis. An isothermal water bath was employed to keep the sample temperature during the measurements.

MOF-303, Al(OH)(PZDC).
The synthesis of MOF-303 was adopted from our previous reports 22,34 . In a 200 L glass reaction vessel equipped with a heating jacket, a mixture of H 2 PZDC·H 2 O (3.48 kg, 20 mol) and NaOH (2.4 kg, 60 mol) was dissolved in 38 L deionized water. The resulting solution was stirred for 60 minutes until all the solids were completely dissolved and the solution cooled down to room temperature. Next, in a 20 L beaker, AlCl 3 ·6H 2 O (4.82 kg, 20 mol) was dissolved in 12 L deionized water and transferred to a 15 L glass material-feeding funnel. The 12 L aluminum chloride solution was slowly added to the 38 L linker solution in the vessel at a rate of 6 L per hour with vigorous stirring. The total addition time lasted 2 hours, and the white precipitate formed during the addition. Afterward, the temperature of the heating jacket was set to 120°C, thus heating the reaction mixture to 100°C. After refluxing for 6 hours and letting the reaction mixture cool down to 60°C, the white solid product was collected in a 20 L filtration funnel. For purification, the white powder was subsequently redispersed by stirring in 15 L aqueous 70% EtOH (v/v) solution, filtered off, and washed again with 30 L anhydrous EtOH at room temperature, followed by filtration and drying under air overnight. The obtained white powder was placed in a 120°C oven for 48 hours to yield the pure and desolvated product (3. CAU-23, Al(OH)(TDC). In a 200 L glass reaction vessel equipped with a heating jacket, a mixture of H 2 TDC (3.44 kg, 20 mol) and NaOH (2.4 kg, 60 mol) was dissolved in 38 L deionized water. The resulting solution was stirred for 60 minutes until all the solids were completely dissolved and the solution cooled down to room temperature. Next, in a 20 L beaker, AlCl 3 ·6H 2 O (4.82 kg, 20 mol) was dissolved in 12 L deionized water and transferred to a 15 L glass materialfeeding funnel. The 12 L aluminum chloride solution was slowly added to the 38 L linker solution in the vessel at a rate of 6 L per hour with vigorous stirring. The total addition time lasted 2 hours, and the white precipitate formed during the addition. Afterward, the temperature of the heating jacket was set to 120°C, thus heating the reaction mixture to 100°C. After refluxing for 6 hours and letting the reaction mixture cool down to 60°C, the white solid product was collected in a 20 L filtration funnel. For purification, the white powder was subsequently redispersed by stirring in 15 L aqueous 70% EtOH (v/v) solution, filtered off, and washed again with 30 L anhydrous EtOH at room temperature, followed by filtration and drying under air overnight. The obtained white powder was placed in a 120°C oven for 48 hours to yield the pure and desolvated product (3.50 kg, 84% based on the linker). EA for the activated sample of CAU-23: Calcd. for AlC 6  MIL-160, Al(OH)(FDC). In a 200 L glass reaction vessel equipped with a heating jacket, a mixture of H 2 FDC (3.12 kg, 20 mol) and NaOH (2.4 kg, 60 mol) was dissolved in 40 L deionized water. The resulting solution was stirred for 60 minutes until all the solids were completely dissolved and  (2,. In a 1 L glass round bottom flask, a mixture of H 2 PylDC (31 g, 0.2 mol) and NaOH (24 g, 0.6 mol) was dissolved in 0.4 L deionized water. The resulting solution was stirred for 10 minutes until all the solids were completely dissolved. Next, AlCl 3 ·6H 2 O (48.2 g, 0.2 mol) was dissolved in 0.1 L deionized water and added to the round bottom flask dropwise with vigorous stirring. The total addition time lasted 2 hours, and the white precipitate formed during the addition. Afterward, the reaction mixture was heated to 120°C. After refluxing for 6 hours and letting the reaction mixture cool down to room temperature, the white solid product was collected by centrifugation. For purification, the white powder was washed twice with deionized water and subsequently washed three times with EtOH. The white solid was air-dried overnight. Full activation of the MOF was conducted in vacuo at 120°C for 24 hours, yielding pure and desolvated product ( Al-fumarate, Al(OH)(Fum). In a 200 L glass reaction vessel equipped with a heating jacket, a mixture of fumaric acid (2.32 kg, 20 mol) and NaOH (2.4 kg, 60 mol) was dissolved in 26 L deionized water. The resulting solution was stirred for 60 minutes until all the solids were completely dissolved and the solution cooled down to room temperature. Next, AlCl 3 ·6H 2 O (4.82 kg, 20 mol) was dissolved in 24 L deionized water and divided into two equal portions. The first portion (12 L) was transferred to a 15 L glass material-feeding funnel and subsequently added to the 26 L linker solution in the vessel slowly at a rate of 12 L per hour with vigorous stirring. The procedure was then repeated for the second portion of the aluminum chloride solution. The total addition time lasted for 2 hours, and the white precipitate formed during the addition. Afterward, the temperature of the heating jacket was set to 110°C, thus heating the reaction mixture to 100°C. After refluxing for 6 hours and letting the reaction mixture cool down to room temperature, the white solid product was collected in a 20 L filtration funnel. For purification, the white powder was subsequently redispersed by stirring in 15 L aqueous 70% EtOH (v/v) solution, filtered off, and washed again with 30 L anhydrous EtOH at room temperature, followed by filtration and drying in air overnight. The obtained white powder was placed in a 120°C oven for 48 hours to yield the pure and desolvated product ( CAU-10, Al(OH)(IPA). In a 200 L glass reaction vessel equipped with a heating jacket, a mixture of isophthalic acid (3.32 kg, 20 mol) and NaOH (2.4 kg, 60 mol) was dissolved in 26 L deionized water. The resulting solution was stirred for 60 minutes until all the solids were completely dissolved and the solution cooled down to room temperature. Next, Al 2 (SO 4 ) 3 ·18H 2 O (6.67 kg, 10 mol) was dissolved in 24 L deionized water and divided into two equal portions. The first portion (12 L) was transferred to a 15 L glass material-feeding funnel and subsequently added to the 26 L linker solution in the vessel slowly at a rate of 12 L per hour with vigorous stirring. The procedure was then repeated for the second portion of the aluminum sulfate solution. The total addition time lasted 2 hours, and the white precipitate formed during the addition. Afterward, the temperature of the heating jacket was set to 130°C, thus heating the reaction mixture to 100°C. After refluxing for 6 hours and letting the reaction mixture cool down to room temperature, the white solid product was collected in a 20 L filtration funnel. For purification, the white powder was subsequently redispersed by stirring in 15 L deionized water, filtered off, and washed again with 3 × 15 L anhydrous EtOH at room temperature, followed by filtration and drying in air overnight. The obtained white powder was placed in a 120°C oven for 48 hours to yield the pure and desolvated product

RESULTS AND DISCUSSION
The synthesis scale, condition, and yield of a series of waterharvesting MOFs are summarized in Table 1. In up-scale MOF synthesis, important considerations include linker and metal availability, purity of chemicals, raw material costs, the toxicity of reagents, reaction time, reaction yield, and safety 20,43 . As shown in "Experimental Methods and Materials" section, all six MOFs were prepared from commercially available and relatively cheap linkers and aluminum salts. While linkers purchased in bulk usually have a lower purity (90~95%), it should be noted that if impurities do not react with Al 3+ in aqueous solution, it will not affect the quality of the MOF product. On the other hand, minimizing the amount of organic solvent used N,N-Dimethylformamide (DMF) is essential when transferring syntheses from the small laboratory to the technical scale, as they are toxic and harmful to the environment and account for a large proportion of the manufacturing cost 16,20,44 . Consequently, we chose water as the solvent and used enough base to fully deprotonate the linker to increase its solubility.
To overcome the low yield and long incubation time associated with traditional solvothermal synthesis of MOFs, we developed a reflux-based synthesis method with vigorous stirring to allow sufficient reaction equilibrium in a faster manner (6 hours). This method also allows for higher concentrations compared to unperturbed solvothermal synthesis, resulting in a twofold increase in space-time-yield 34 . More importantly, while the solvothermal synthesis is not scalable, the reflux method is highly scalable and can be adapted for use in an industrial setting. As such, we first optimized the synthesis conditions in a small laboratory setting (100 mL round bottom flask with oil bath) 22 , and then adapt the procedure to a scale-up setting (200 L reaction vessel with heated jacket) for further optimizations (Figure 1a).
During the optimization of the reaction parameters for the synthesis of six aluminum-based MOFs with rod SBUs [Al(L)(OH), where L is the deprotonated linker such as PZDC 2− , TDC 2− , and FDC 2− ], we found that using exactly three equivalents of NaOH was key to maximizing the yield. Two equivalents of NaOH were used for the full deprotonation of the dicarboxylic acid linker, and one equivalent was used for the formation of the SBU. It is worth noting that, although the formation of high crystalline product can be achieved in a shortperiod of time (1 h), we kept the mixture under reflux for 6 hours in kilogram-scale synthesis with a large solvent volume to achieve full porosity and water uptake and minimize undesirable hysteresis in the water isotherm. Using these optimized conditions, we were able to produce approximately 3 kg of activated MOF per batch with a high yield and good water uptake (Figure 1b), which are comparable to results from gram-scale synthesis reported in the literature 6,10,16,26,35 .
The crystallinity of the scaled-up MOF products was investigated by PXRD measurements (Figure 2a). The measured values were in good agreement with the simulation data for each MOF. It should be noted that all materials, after washing, showed high crystallinity, with significant peaks that could be well indexed ( Figure S1-S6, Supporting Information). This suggests that the scale-up method, with a suitable refluxing time, does not result in a loss of crystallinity for the MOF materials.
The composition of the six MOFs was analyzed using NMR spectroscopy and elemental analysis. After activation, five of them show only one singlet peak from the corresponding dicarboxylate linker in the range of 6.3-7.2 ppm in their 1 H digestion NMR ( Figure S7-S10 and Figure S12, Supporting Information). CAU-10 showed three peaks from its IPA linker in the range of 7.4-8.1 ppm in its 1 H digestion NMR ( Figure S11, Supporting Information). In all six 1 H NMR spectra, the peaks of washing solvent EtOH were not present, indicating that MOFs were fully activated, and no solvent remained. These results confirm the high purity of all MOFs synthesized at scale. In addition, the good agreement between the calculated and found elemental compositions of all MOF samples (section "Experimental Methods and Materials") indicates that all impurities were removed and no water or EtOH was left in the activated MOFs. Next, the permanent porosity of the aluminum MOF series was investigated using nitrogen sorption analysis (Figure 2b). The Brunauer−Emmett−Teller (BET) surface areas and specific pore volumes of the compounds ranged from 1379 to 654 m 2 /g and 0.47 to 0.23 cm 3 /g, respectively ( Figure  S13-S24, Supporting Information). These results are comparable to literature data, 6,10,16,26,35 and demonstrate that all scaled-up MOF samples had optimal BET surface areas, pore diameters, and pore volumes (Table S1, Supporting Information). The thermal stability of these MOFs was also evaluated using TGA, and they showed no weight loss up to ∼400°C under nitrogen atmospheres (Section S5, Supporting Information).
Finally, we measured the water vapor sorption isotherms of the scaled-up MOF products at 25°C (Figure 2c). All samples showed a consistent S-shaped profile with different steps and water uptake capacities (Table S2, Supporting Information). MOF-303 had an inflection point at 12% RH and the highest gravimetric water capacity of 39 wt% at 40% RH among all six MOFs. This can be attributed to its highest pore volume among the MOFs ( Figure S25, Supporting Information). CAU-23 had an inflection point at 26% RH and water capacity of 33 wt% at 40% RH, which was due to the hydrophobic nature of TDC linker ( Figure S26, Supporting Information). MIL-160 had the lowest inflection point at 7% RH and gravimetric water capacity of 36 wt% due to its hydrophilic pore environment ( Figure S27, Supporting Information). MOF-313 had an inflection point at 13% RH and a gravimetric water capacity of 34 wt%, as for MOF-303 ( Figure S28, Supporting Information). CAU-10 had an inflection point at 17% RH and a gravimetric water capacity of 32 wt% because of its smaller pore volume established by bulkier IPA linker ( Figure S29, Supporting Information). Al-fumarate had an inflection point at 27% RH and the gravimetric water capacity of 39 wt% ( Figure S30, Supporting Information). For most samples, only a small degree of hysteresis was observed between the adsorption and desorption curves, indicating that the optimal conditions (e.g., reaction time, reaction temperature, etc.) minimized the number of defective sites within the crystal lattice. These results show that the water isotherm behavior of all six aluminum-based MOFs prepared at kilogram scale is comparable to that reported at gram or milligram scale in the literature, and our synthesis method leads to high-quality water harvesting MOF materials at scale.

CONCLUSIONS
In this work, we demonstrated that a series of important water harvesting MOFs (MOF-303, CAU-23, MIL-160, MOF-313, CAU-10, and Al-fumarate) can be synthesized under industrially suitable, green conditions using inexpensive, commercially available linkers. Our facile and robust synthesis method led to production of MOFs at the kilogram scale without compromising framework crystallinity, porosity, or water-harvesting properties. These results demonstrate the feasibility of commercializing MOFs as water-harvesting sorbents and will contribute to the widespread adoption of water-harvesting technologies in the future.

ACKNOWLEDGMENTS
We acknowledge the financial support from Defense Advanced Research Projects Agency (DARPA) under contract HR0011-21-C-0020. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of DAR-PA. ZZ thanks members of the Yaghi Group: Dr. Nikita Hanikel for water isotherm measurements, and Drs. Ha L. Nguyen, Wentao Xu, and Daria Kurandina for helpful discussions. We acknowledge the College of Chemistry Nuclear Magnetic Resonance Facility for resource instruments, which are partially supported by NIH S10OD024998, and staff assistance from Dr. Hasan Celik. We thank Dr. Nick Settineri for his assistance in measuring powder X-ray diffraction patterns. ZZ thanks for the financial support through a Kavli ENSI Graduate Student Fellowship.