Aqueous Flow Reactor and Vapour‐Assisted Synthesis of Aluminium Dicarboxylate Metal–Organic Frameworks with Tuneable Water Sorption Properties

Abstract Energy‐efficient indoors temperature and humidity control can be realised by using the reversible adsorption and desorption of water in porous materials. Stable microporous aluminium‐based metal–organic frameworks (MOFs) present promising water sorption properties for this goal. The development of synthesis routes that make use of available and affordable building blocks and avoid the use of organic solvents is crucial to advance this field. In this work, two scalable synthesis routes under mild reaction conditions were developed for aluminium‐based MOFs: (1) in aqueous solutions using a continuous‐flow reactor and (2) through the vapour‐assisted conversion of solid precursors. Fumaric acid, its methylated analogue mesaconic acid, as well as mixtures of the two were used as linkers to obtain polymorph materials with tuneable water sorption properties. The synthesis conditions determine the crystal structure and either the MIL‐53 or MIL‐68 type structure with square‐grid or kagome‐grid topology, respectively, is formed. Fine‐tuning resulted in new MOF materials thus far inaccessible through conventional synthesis routes. Furthermore, by varying the linker ratio, the water sorption properties can be continuously adjusted while retaining the sigmoidal isotherm shape advantageous for heat transformation and room climatisation applications.


Introduction
The broad family of porousm aterials finds widespread use in catalysis, [1] adsorptive separations [2] and ion exchange, [3] among many other applications. As ar elatively young branch of this family tree, [4] metal-organic frameworks (MOFs) are under evaluation for several real-life applications. They represent av ersatile group of compounds with record-breaking surface areas (> 7000 m 2 g À1 ) [5] and promising properties for sensing, [6] gas capturea nd separation [7,8] andh eat exchange. [9] Nevertheless, no large-scale applicationsh ave been implemented thus far. One challenge lies in the often hazardous and low-yielding synthesis conditions of MOFs. To overcome these challenges, synthesis protocols suitable for industrial scale-up have to be developed while taking into account pricing of the final product. [10,11] In this respect, ap articularly interesting MOF is aluminium fumarate, also known as Al-MIL-53-Fum (MIL = Material Institute Lavoisier). [12] It exhibits high porosity and stability, even under hydrothermals tress, [13] and therefore has sparked the interesto fi ndustrial researchers. [14] In particular, the sigmoidal water ad-/desorption curve without hysteresis demonstrated by Al-MIL-53-Fuma nd other Al-MOFs is attractive for applicationsi nh eat-exchange devices. [15][16][17][18][19] The patented synthesis of Al-MIL-53-Fum is ah ydrothermalb atch process that makes use of inexpensive and readily availables tarting materials (fumaric acid, NaOH, aluminium sulfate) and avoids hazard-  ous solvents such as dimethylformamide. [20] Similar,m ild synthesis conditions ( 100 8C) have been demonstrated for other Al-MOFs, [21][22][23][24] thereby avoiding pressure build-up. Also, synthesis methods based on extrusion, microwave-assisted heatinga nd starting from insoluble metal ion sources have been reported. [25][26][27] For the further scale-up of these MOFs, it can be desirable to move to continuous production in flow reactors as multiplying the output volume through multiple tubes in parallel enables more control compared with larger dimension batch reactors. [28,29] An alternative elegant approach to green and scalableM OF synthesisw ould be the conversion of non-salt precursors using no or minimal amounts of solvent. Severalo xide-based solvent-free syntheses or vapour-assisted methods have been reported, although only for MOFs based on divalent metal ions (e.g.,C u 2 + ,Z n 2 + ). [30][31][32] To date,v apourassisted synthesis of MOFs based on tri-and tetravalent metal ions (e.g.,F e 3 + ,A l 3 + ,Z r 4 + )h as required the use of metal salts. [33] In this contribution, we investigate the flow reactor (fr) and vapour-assisted (va) synthesis of Al-MIL-53-Fuma nd related mixed-linker Al-MOFs ( Figure 1) to find efficient scalable preparation methods for materials with improved water sorption properties. Salt andn on-salt aluminium precursors were used. The tested linker molecules are fumarica cid (H 2 Fum)a nd mesaconic acid (methylfumaric acid, H 2 Mes), and combinations thereof.A lthough H 2 Fum is industrially available, H 2 Mes can be derived from readily available citric acid. [34] The resulting MOFs crystallise either in as quare-or kagome-grid topologya so bserved in Al-MIL-53-Fum [35] and Al-MIL-68-Mes, [35] respectively. All materials were characterised for their structurala nd sorption properties.

Experimental Section
Flow reactor set-up The flow reactor set-up is similar to the one recently described by the Stock group. Details regarding the reactor volume, flow rates and achievable temperatures are given in the Supporting Information (Section S2.1 in the Supporting Information). [36,37] The three syringes of the reactor are loaded with (i)ana queous 0.05 m aluminium sulfate solution, (ii)ana queous solution of linker mixtures (0.1 m)a nd KOH (0.3 m)a nd (iii)water,r espectively ( Figure 1, top). In at ypical procedure, only the precursor solutions (i and ii)a re initially pumped through the tubes and mixed via aq uadruple cross connector before passing to the reactor,w hich consists of ac oiled Teflon tube heated in an oil bath. Once the reactor is filled, the water syringe (iii)i su sed to push the remaining precursor solution and the product slurry out of the reactor.T he obtained product was centrifuged and washed with ethanol.

Vapour-assisted conversion process
Mixtures of aluminium nitride (AlN) and linker powders (1:2 ratio, 200 mg) were placed in a2 5mLs ealed glass bottle together with glass vials containing 1mLo fl iquid to generate vapours (Figure 1, bottom). After 48 hr eaction at 80 8Ci nap re-heated oven, the vapour and excess ligand were removed by heating the powder in vacuum at 200 8Cf or 2h.E ventually,t he material was calcined at 300 8Cf or 12 h.

Additionalinformation
Materials and methods, syntheses optimization details, structure refinements and fits, characterisation data ( 1 HNMR, elemental analysis, TGA, FTIR, SEM, N 2 sorption, PALS), and linker vapour pressure data can be found in the Supporting information. Deposition Number 1918969 contains the supplementary crystallographic data for this paper.T hese data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service www.ccdc.cam.ac.uk/structures.

Results and Discussion
Flow reactor (fr) synthesis

Single-linker fr-Al-MIL-53-Fum and fr-Al-MIL-68-Mes
For the synthesis optimization of fr-Al-MIL-53-Fum, different metal-to-linker ratios (2:1, 1:1, 1:2) and different reaction times, determined by the flow rate, were investigated at ar eaction temperature of 80 8C. A1 :1 metal-to-linker ratio and 15 min residence time in the flow reactor were identified as the optimal reaction conditions for the synthesis of fr-Al-MIL-53-Fum (Table S2.1 in the Supporting Information). For the synthesis of fr-Al-MIL-68-Mes, identical conditions (1:1 ratio, 15 min) were found to be optimal (Table S2.2 in the Supporting Information). The phase purity of both MOFs wasc onfirmed by PXRD (Le Bail fits Figures S4.1 and S4.2 in the Supporting Information), the composition was confirmed by 1 HNMR spectroscopy (Section S5 in the Supporting Information), elemental analysis( Section S6 in the SupportingInformation), thermogravimetric analysis (Section S7 in the Supporting Information), IR spectroscopy (SectionS8i nt he Supporting Information) and the morphology was investigated by electron microscopy (SectionS9i nt he Supporting Information).

Aluminiumnitride as ar eactive precursor
Under mild conditions (80 8C), aluminium oxide will not react to form Al-MIL-53-Fum, even in the presence of water vapour, whereas aluminium nitride does (Table S3.1 in the Supporting Information). The higherr eactivity of the nitride is due to the softer AlÀNb onds, which are favourably replaced by the harder AlÀOb onds formed with dicarboxylate linkers. [38] At room temperature in moist air (80 %r elative humidity, RH), complete hydrolysis of aluminium nitride to aluminium hydroxide can be achieved, but only after prolonged reactiont imes (> 400 hf or micron-sized particles). [39] The conversion to Al-MOFs requires ah ighr elative humidity (94 %), yet is incomplete when only water vapour is present (Figures S3.2 and S3.4 in the SupportingI nformation). Likely,t he MOF forming at the surfaceo fm icron-sizeda luminium nitride particles hinders further conversion. [40] Indeed,s urface treatment of aluminium nitride with variousa cids that bind to Al 3 + (e.g.,p hosphoric acid, acetic acid) is known to delay or preventhydrolysis. [41] Solvent-free activation conditions Powder X-ray diffraction( PXRD) measured for the as-synthesized materialsr eveals the presence of the MOF,e xcess ligand and unreacted aluminium nitride ( Figure S3.4 in the Supporting Information). To avoid washing with organics olvents, a two-step activation treatment was optimized that takes advantage of the high thermal stability of Al-MIL-53 materials [42] (> 350 8C): (1) sublimation of the excess linker at 200 8Cu nder vacuum,f ollowedb y( 2) calcination at 300 8Ct or emovea dsorbed linker from the pores. Without calcination, no porosity is detected (Brunauer-Emmett-Teller,B ET,s urface area < 25 m 2 g À1 ), whereas after calcination the Type Ii sotherm expectedf or microporous materials is observed ( Figure S3.3 in the Supporting Information). In situ temperature-dependent PXRD shows the sublimation of crystalline linker around 200 8C. Also, the intensity of the framework reflection at approximately 13.38,i ndicative of electron density in the pores, thus linker molecules, gradually disappearsa round3 00 8C ( Figures S3.6 and S3.7 in the SupportingI nformation). Still, calcination steps bring additional energetic costs and should therefore preferably be performed by using low-grade waste heat, thus at temperatures 250 8C. [43] Formic acid vapour as synthesis modulator Formic acid has been used elsewheret om odulate the solvothermals ynthesis of Al-MIL-53-Fum,r esulting in improved isotherms and kinetics for water adsorption. [44] When appliedt o the vapour-assisted synthesis of Al-MIL-53-Fum, full conversion of aluminium nitride can be achieved through the addition of formic acidv apour to the reactiona tmosphere. Moreover,t he presence of formic acid vapour allowsM OF formation under lower relative humidity (79 %; Figure S3.8 in the Supporting Information). Only at race amount of nitrogen and formate ions (< 1%)i sf ound in the final product by elemental analysisa nd 1 HNMR spectroscopy,r espectively (Sections S5 and S6 in the Supporting Information). Formate is likely incorporated in the framework during synthesis, but it is removed upon thermal activation. [45] The organic content quantified by thermogravimetry matchest he expected aluminium fumarate chemical formula ([Al(OH)(Fum)]) and confirms full aluminium nitride conversion (Section S7 in the SupportingI nformation). IR spectroscopy reveals no significant difference between materialss ynthesized under solvothermal conditions and under vapour-assisted conditions in the presence or absence of formic acid ( Figure S3 In contrast, in the absence of formic acid vapour ( Figure S3.8 in the Supporting Information) or in solvothermalr eactions (flow reactor and batch synthesis),aproduct crystallising in the monoclinic space group P2 1 /c is obtained, even when formic acid is used as am odulator in solution. [44] va-Al-MIL-53-Fum has aB ET surface area of 592 m 2 g À1 ,w hich is much lower than the surface area of fr-Al-MIL-53-Fum (1000 m 2 g À1 )a nd that from the reported batch synthesis (1080 m 2 g À1 ). [12] However,t he reduced porosity is ascribed to thed ifferent pore geometries of the different crystal structures. The pore size in va-Al-MIL-53-Fum (Pnma)i sc ontracted in comparison to fr-Al-MIL-53-Fum (P2 1 /c). Positron Annihilation Lifetime Spectroscopy (PALS) measurements evidencet his difference in the pore dimensions. For va-Al-MIL-53-Fum (Pnma), free-volume elements with ad iameter of 3.5 are observed, whereas forf r-Al-MIL-53-Fum (P2 1 /c)amuch larger diameter of 5.9 is detected (Table S12.1 in the Supporting Information). These values are in good agreement with the size of the largest sphere that would fit in the pores, respectively,4 .1 and 5.8 ,c alculated by Monte Carlo sampling using Zeo ++. [46] Furthermore, simulations with RASPAi ndicateadifference in surface area of 32 % between va-Al-MIL-53-Fum (Pnma)a nd fr-Al-MIL-53-Fum (P2 1 / c), in line with the experimental data (41 %). [47] As for Al-MIL-53-Fum (P2 1 /c), [12] no framework flexibility is observed for va-Al-MIL-53-Fum ( Figure S3.6 in the Supporting Information). Still, alternative synthesis modulators to the corrosivef ormic acid are desirable for industrial productiona nd should be further investigated.

Novel Al-MIL-53-Mes material
By replacingf umaric acid with mesaconic acid, the optimized vapour-assisted synthesis conditions with formic acid as modulator yield va-Al-MIL-53-Mes. The MIL-53 type structure is formed under these conditions, in contrast to the MIL-68 type product from flow and batch reactor syntheses (Figure 1, right). The crystal structure was confirmedb yR ietveld refinement ( Figure 2). Al-MIL-53-Mes is an ew microporous MOF material that crystallises in the orthorhombic space group Pnma, has aB ET surface area of 527 m 2 g À1 and am icropore volume of 0.169 cm 3 g À1 (theoretical value:0 .210 cm 3 g À1 ). The formation of Al-MIL-53-Mes illustrates the potential of vapour-assisted conditions to obtain new materials not accessible under solvothermalr eaction conditions.

Mixed-linker MOFs with tuneable properties
The optimised fr-and va-synthesis conditions for the singlelinker MOFs were used for the synthesis of mixed-linker materials. For the fr-syntheses, both linkers were dissolved in aqueous KOH. For the va-synthesis, physical mixtures of the linker powders were used. The ratio of fumarica nd mesaconic acid was varied from 0t o1 00 %i n1 0% steps for both approaches to obtain mixed-linker MIL-53 and MIL-68 MOFs. IR spectroscopy shows the incorporation of both linkers in the framework (SectionS8i nt he Supporting Information). As for the singlelinker MOFs, elemental analysis shows only small impurities from the precursor,s ulfur and nitrogen in the fr-and va-products, respectively (SectionS6i nt he Supporting Information). 1 HNMR spectroscopy confirms the absence of formate ions in the activated va-products (Section S5 in the Supporting Information). Allf r-materials and va-materials are large aggregates of crystallites smallert han 1 mm, with the crystallites from vasynthesis having am ore elongated shape and al arger size (Section S9 in the Supporting Information). The fraction of each linker was quantified by 1 HNMR spectroscopy after dissolving the activated MOF.F or the fr-products, the linker ratio is close to the one in the precursor solution (< 3% deviation), meaningt here is no preferential linker incorporation (Figure 3I). The composition of the va-products also followst he  ratio in the linker powder mixture (< 10 %d eviation). However, as light preference is observed for mesaconate or fumarate, respectively,b elow and above 65 %m esaconatec ontent ( Figure 3II). At constant temperature and with excesss olid linker present,t he partial pressure of both H 2 Fum and H 2 Mes is independento ft heir ratio in the solid phase. Based on thermogravimetricm easurements and the Knudsen effusion method, an approximately 4.5 times higher vapour pressure was determined for mesaconic acid at the reactiont emperature (Section S11i nt he Supporting Information). For an equilibrium reaction, products with ac onstantl inker ratio are expected as long as each linker maintains its saturation vapour pressure (i.e.,a sl ong as solid linker is present). Nevertheless, as the incorporated linker ratio is not constant but varies with the solid linker mixture composition, it appears that the va-synthesis of mixed-linkerM OFs is not an equilibrium reactionb ut rather under kinetic control. In other words, the rate of linker sublimation, which scales with the linker fraction in the reaction mixture,d etermines the compositiono ft he mixed-linker MOFs in the va-route. Lastly,t he organic content in the materials was quantified by thermogravimetry.F or the va-materials, the obtained values match very well the theoretical values, whereas for the fr-materials they are generally lower,s uggesting the slight presenceo f( hydr)oxide impurities (Tables S7.1 and S7.2 in the Supporting Information).

fr:the linker ratio directs the topology
The fr-products exhibit aM IL-53 type structure at mesaconate contentsb elow 40 %, and aM IL-68 type structure above 60 % mesaconate. Thus, the linker presenti nt he highest concentration determines the resulting framework topology in the frroute. Between 40 %a nd 60 %m esaconate, mixed phases are observed( Figure 3III). In situ PXRD measurements in batch reactors show that for all linker ratios the final products crystallise directly from the precursor solution,w ithoutt he formation of transientphases (Figure4).

va:M IL-53 topology at all linker ratios-solid solutions
Only the square-grid MIL-53 structure was observed for the vaproducts,i ndependento ft he linker ratio. The lattice parameters b and c extracted from the positiono ft he (0 11)r eflection by Pawley fits change linearly with increasing mesaconate content ( Figure 3IV), whereas a remains constant (6.62 AE 0.03 ). According to Vegard'sl aw, [48] the mixed-linker MOFs obtained via va-synthesis can thus be considered solid solutions. Conversely, the cell parameters of the fr-products remained constant ( Figure 3III). However,a sm entioned before,t he va-Al-MIL-53 materials crystallisei nt he orthorhombic space group Pnma whereas for the fr-Al-MIL-53 materials the best fit is observed in the monoclinic space group P2 1 /c.

Thermal stability
All materials show high thermals tability in air (> 350 8C). For the fr-materials, the decomposition temperature,c alculateda s the inflection point of the wt %-T curve upon decomposition, are in the range 425-465 8C( Figure 3V), in line with the values observed by temperature-dependent PXRD ( Figure S2.4 in the Supporting Information). For the va-Al-MIL-53s amples, ac ontinuousd ecrease in decomposition temperature from approximately 470 to 395 8Ci so bserved with increasingm esaconate content ( Figure 3VI). Between 65 %a nd 100 %m esaconate, the decompositiont emperature drops more rapidly.

Sorption properties and water cycling
Water and nitrogen sorption measurements werec arried out to investigate the effect of the crystal structure and the fraction of bulky andh ydrophobic mesaconate linker on the sorption properties. The mixed-linker fr-materials with MIL-53 type structures how as pecific surface area comparable to Al-MIL-53-Fum (approx.1 000 m 2 g À1 ), [12] yet slowly decrease with increasingm esaconate content ( Figure 5). The surfacea rea of the mixed-linkerf r-Al-MIL-68 samples is slightly higher,a se xpected from literature dataf or Al-MIL-68-Mes (1040 m 2 g À1 ). [35] Especially at 60 %a nd 80 %m esaconate, the materials exhibita surprisingly but reproducibly high surfacea rea of up to nearly 1400 m 2 g À1 and asubstantially higher decomposition temperature. With respecttot he water uptake capacity,the fr-materials show similar performance (42 AE 5wt% at 60 %R H), independent of the crystal structure (Figure 5c). However,t he influence of the increasing mesaconate content is clear from the shape of the water sorption isotherm. The sigmoidal isotherms show as harp uptake at as pecific relative pressure ( Figure 6). This 'adsorption edge', defined as the RH value at the inflection point of the isotherm, shifts linearly from 30 to 50 %R Hw ith increasing mesaconate content (Figure 5a).
Similar to the single-linker materials, mixed-linker va-Al-MIL-53 materials, crystallising in the orthorhombic space group Pnma,a re porous but display lower surfacea reas than their monoclinic fr-counterparts. Although ac ontinuous decrease in surfacea rea would be expected with increasing mesaconate content,am inimum is found at 65 %m esaconate (Figure 5f). This compositionc orresponds to the mosth ydrophobic material as indicatedb yt he highest adsorption edge at 59 %R H (Figure 5b). At am esaconate content lower than 65 %, the expectedd ecreaseo ft he adsorption edge and increase in water uptake capacity is observed (Figure 5d). At am esaconate content higher than 65 %, the water uptake capacity decreases and the isotherms gradually lose their sigmoidal shape, making it impossible to determine the adsorption edge ( Figure 6).
The material sorptionp roperties determine its application potentiala nd which fields can be targeted. When compared with best-in-class water adsorbents, the va-andf r-materials are competitive as they show comparable water uptakes. Further,t he adsorption edge of the fr-Al-MIL-53 materials (30-38 %R H), fr-Al-MIL-68 materials (46-50 %R H) and va-Al-MIL-53 materials (42-59 %R H) is comparatively higher ( Figure 7), [49,50]   and covers not only the desired range for heat transformation applications( 5-40 %R H), buta lso the desired range for room climatisation (40-60 %RH). [51] Conclusion Twop otentially scalables ynthesis methods were developed to obtain single and mixed-linker aluminium dicarboxylateM OFs under mild conditions. Depending on the synthesis conditions, the crystallisation can be directed to different structure types, yieldingm aterials with tuneablew ater sorption properties. These results will hopefully fosterf urtherr esearch in the integration of MOFs in heat-exchange or room climatisation devices. The discovery of anovel compound through vapour-assisted synthesis indicates the opportunities in solvent-free MOF synthesis and processing. Figure 7. Comparison of fr-and va-Al-MOFs (experimental data, coloured markers) to best-in-class water adsorbents (literaturevalues, black marker). [49,50] The water uptake at ar elative pressure of 0.6 is plotted against the adsorption edge of the water adsorption isotherm measured at 25 8C. If the isotherm does not show aS -shape, the relative pressure correspondingtoh alf of the uptake is used instead. The adsorption edge desired range for several applicationsis also indicated.