Synthesis of Chiral MOF‐74 Frameworks by Post‐Synthetic Modification by Using an Amino Acid

Abstract The synthesis of chiral metal–organic frameworks (MOFs) is highly relevant for asymmetric heterogenous catalysis, yet very challenging. Chiral MOFs with MOF‐74 topology were synthesised by using post‐synthetic modification with proline. Vibrational circular dichroism studies demonstrate that proline is the source of chirality. The solvents used in the synthesis play a key role in tuning the loading of proline and its interaction with the MOF‐74 framework. In N,N′‐dimethylformamide, proline coordinates monodentate to the Zn2+ ions within the MOF‐74 framework, whereas it is only weakly bound to the framework when using methanol as solvent. Introducing chirality within the MOF‐74 framework also leads to the formation of defects, with both the organic linker and metal ions missing from the framework. The formation of defects combined with the coordination of DMF and proline within the framework leads to a pore blocking effect. This is confirmed by adsorption studies and testing of the chiral MOFs in the asymmetric aldol reaction between acetone and para‐nitrobenzaldehyde.


Introduction
Chiralityp lays ak ey role in the understanding of biochemistry [1,2] and it remains al ong-standing enigma for explaining the origino fl ife because all chiral aminoa cids in proteins and enzymes exist only in their l form. [3] Producingc hiral compounds is essential for the pharmaceutical, [4] food, [5,6] agricultural [7] and biotechnological industries. [8] For the pharmaceutical industry,t he activity of ad rug is based on its specifici nteraction with the target made up of chiral fragments (e.g.,p roteins, nucleic acids, etc.). [9] Thus,t he biological and pharmacological influence of the drug is highly dependent on its chiral form. [9] The synthesis strategies of chiral molecules include chemicalt ransformation of enantiopurep recursors from the chiralp ool, [10] racemic resolution, [11] but also asymmetric synthesis. [12] Efficient homogeneous enantioselective catalysts are currently used in asymmetric synthesis, but there are stillc hallengesr elated to the catalysts ensitivityt oi mpurities, catalyst or ligand stability, the requirement of extreme conditions, as well as the unavailability of the ligand or the catalyst. [13][14][15] Alternatively,t he use of heterogeneousc atalytic systems enables the easy recovery of the catalystf rom the reactionm edium and its reuse in consecutivec ycles,t hereforer esulting in lowcost applications. In this context,c hiral metal-organic frameworks( MOFs) [16] have emerged as an ew class of potential asymmetricheterogeneous catalyst. [17][18][19][20] As traightforward approacht owardst he synthesis of chiral MOFsr elies on the appropriate combination of metal ions and enantiopure organic linkers. [18,[21][22][23][24][25] Although some efficient catalysts are designed by using this approach,t he synthesis of the chiral linkers often entails lengthy multi-step procedures. An alternativem ethodology focuses on the post-synthetic modification of achiral MOFs by either covalent binding of chiralg roups on the organic linkers [26] or by coordinating them to the metal ions. [27] For example, the achiral Cr-MIL-101 framework,n amely [Cr 3 O(H 2 O)F(bdc) 3 ]·4.5 H 2 O·0.15(H 2 bdc) where H 2 bdc = terephthalic acid, was converted to the isostructural chiralf orms( CMIL-1 and CMIL-2) by using ap roline derivative ligand, which is post-synthetically coordinated to the metal centres. [27] Recent studies also demonstrated the chirality transfer between chiral molecules and MOF crystallitest hrough weak or strong intermoleculari nteractions. [28] This synthetic strategy is known as chiral induction and can be achieved by using chiral templates, additives or solvents. [28][29][30][31][32] Consideringt he structuralv ariety of MOFs currently available, [20] as traightforward synthetic approachf or making chiral MOFs would imply the induction of chiralityi ne xisting network topologies. Using this synthetic strategy,ac hiral MOF-5 topologyw as prepared via the direct synthesis of MOF-5i nt he presenceo fl-o rd-proline and by using N-methylpyrrolidone (NMP) as solvent. [33] In the presence of l-proline, L-CMOF-5 is formed, whilst the addition of d-proline lead to D-CMOF-5. [33] The single-crystalX RD analysiso fb oths tructures shows that the benzene rings between adjacent Zn 4 Oc lusters are oriented at an on-908 dihedral angle (788 for L-CMOF-5 and 668 for D-CMOF-5) and the atropisomer-like bridging mode of the benzene-1,4-dicarboxylate linkers translates the chirality from one cluster to another. [33] Moreover,t he integrity and chirality of the structures was preserved due to the stabilising effect of the achiral NMP solvent molecules. [34] Similarly,the chiral induction was achieved in (Me 2 NH 2 )[In(thb) 2 ]·x DMF (H 2 thb = thiophene-2,5-dicarboxylica cid) using chiral alkaloids. [35] In another study,t he synthesis of chiral [Mn 3 (HCOO)(d-cam)] or [Mn 3 (HCOO)(l-cam)] was attained via the coordination of the chiral additive, d-o rl-camphoric acid (d-o rl-Hcam), to the metal sites of the framework. [31] Thes tudies so fari ndicatet hatt he chiral inductiond uring synthesism ight indeed be af acilea nd inexpensivea pproacht o synthesise chiral MOFs.I ta lsos hows that thec hemicali nteractionsb etweenc hirala dditives andt he MOFf rameworki sakey parameter. However, thee xamplesa bove also illustrate that the chemical functionalityo ft he chiral inductor matchesw ellt he chemical functionalitieso ft he organicl inkers,f or example, carboxylate groups.T he challenges till lies in thea bility to select thep roperc hirala dditivet hate nables thec hiralt ransferw hile retainingthe frameworkt opology. Forinstance, in theformation of CMOF-5, the chiral proline inductor determines the handedness, but the chirality stems from and is stabilised by the NMP guest molecules. [33,34] Thisa chiral solvent used for synthesis createsa ninternal stress on the framework and twisting of the organic linkersthat lead to ac hiral framework. [34] The aim of this work was to study and to understand if the chiral induction can be appliedt oo ther structural topologies and to gain further insight on the parameters governing the chiral induction process. It is focusedo nam aterial, which is recognised as the MOF with the highest density of open metal sites, namely [Zn 2 (dobdc)(H 2 O) 2 ] (H 4 dobdc = 2,5-dioxido-1,4-benzenedicarboxylic acid), known as Zn-MOF-74. [36] Zn-MOF-74 is intensively studied for gas storage [36] and separation [37] as well as in catalysis. [38][39][40][41][42][43][44][45] It has a3 Dh oneycomblike framework with 1D channels of ca. 11 , [46] built from organic linkers on the edges and ZnO 6 at the corners. The ZnO 6 octahedra form infinite helicalc hains via double oxygen bridges alongt he c axis, but of opposite handedness with P and M helical rods with a ratio P/M of 1:1( Figure 1). In recent work, we showed that by using chiral cinchonaa lkaloids during the crystallisation of Zn-MOF-74 does not influence the P/M ratio or induce chirality.T he bulkiness of the chiral molecule and its weakc oordination to the metal ions afforded achiral structural isomers with discrete secondary buildingu nits (SBUs). [46] Consequently, we turn our attention to chiral inductors which are smaller and would easily be accommodated in the channels of Zn-MOF-74, for example, the chiral amino acid l-o rd-proline (l-o rd-Pro).
Here, we discuss the synthesis and characterisation of the chiral materials obtained by using chiral induction with l-o r d-Pro.

Results and Discussion
The first approach to synthesise ac hiral Zn-MOF-74 topology aimed at using d-a nd l-Pro as chiral inductord uring the typical synthesis of Zn-MOF-74. [47] This molecule has the ability to interactw ith the MOF framework via hydrogen bonding with the carboxylate groups of the organic linkerso rt hrough coordination to the open sites on the Zn 2 + ions. [48] The size of d-o r l-Pro molecules is ca. 8.5 5.6 5.3 3 ,i ndicating that they can enter within the MOF-74's channels of ca. 11 .T herefore, we hypothesised that the interaction of l-o rd-Pro with Zn-MOF-74 may lead to ac hiral framework and that proline, asa uxiliary ligand,w ill lead to al ong-range chirality order.E arlier studies have shown that using l-Pro as modulator in the synthesis of Zr 4 + and Hf 4 + MOFs with UIO-66 topology affords access to large and high quality single-crystals that are free of defects [49] and that l-Pro can be incorporated as ad efect-capping structural element. [50,51] Recently,t he chiralityo fl-Pro modulated Zr 4 + -based MOFs was demonstrated in the aldol reaction between 4-nitrobenzaldehydea nd cyclohexanone. [52] Our group reported the synthesis of MOF-74 in the presence of bulky chiral alkaloids leadingt oanews tructuralt opology, HIMS-74, whichi sb uilt from binuclear secondary building units. [46] We attributed the formation of the binuclear SBUs to the N,O-coordination of the alkaloids to the Zn 2 + ions during the nucleation process, thereby inhibiting the formation of the chain-like SBUs, which are characteristico ft he MOF-74 topology.I ns ubsequent studies, we replaced the bulky chiral alkaloids with the smallerp roline molecule. Our hypothesis was that, unlike the reactions performed by using chiral alkaloids, the use of d-o rl-Pro duringt he synthesis of MOF-74 would lead to chiral materials because of the reduced size of the additive. Attempts to crystallise ac hiral MOF-74 topology with l-Pro via induction wereu nsuccessful, even when varying the synthetic parameters (molar ratio, solvent(s), temperature and scale of the reaction) or when using other amino acids (lvaline, l-alanine, l-tryptophan). The hydrothermal synthesis of MOF-74 in the presence of amino acids yielded powders that were amorphous as indicated by PXRD studies. This could be due to the low solubility of the amino acidsi nt he reaction mixture (DMF,N MP/H 2 O, DMF/H 2 O/EtOH). In the functionalisation of Zr-UIO-66 and Zr-UIO-67 MOF with l-Pro, HCl was added to ensure the dissolution of l-Pro by formation of hydrochloride salts of l-Pro that allowed the growth of MOF structures. [51] However,p erforming the crystallisation of MOF- 74 in the presence of l-Pro and HCl did not lead to any solid material, indicating that the presence of the acid inhibits the framework formation.
To overcomet he challenges discussed above,wef ocusedo n as econd approach based on the post synthetic chiral induction of Zn-MOF-74. The material used was prepared by using a water-based room-temperature synthesis. Solvente xchange by using methanol( MeOH)a nd subsequentv acuum heating were carried out prior to the post synthetic modification in order to obtain the Zn-MOF-74 material containing open metal sites. Considering that the molecular size of d,l-Pro enables its diffusion within the pores of Zn-MOF-74, we anticipated that the chiral amino acid might coordinate to the Zn 2 + ions in am onodentate manner via its carboxylate group or bidentate, bridging neighbouring Zn 2 + ions while maintaining the six-fold coordination of the Zn 2 + ions (Figure2). Another expected interaction is the intermolecularh ydrogen-bonding of d,l-Pro with the carboxylate groups of the organic linker.
The reactiono fd,l-Pro with Zn-MOF-74 was performed in two solvents of different polarities: N,N'-dimethylformamide (DMF) commonly used when synthesising MOF-74 materials, and MeOH that is usuallye mployed for solvent exchange. We hypothesised that DMF,b eing as trongly coordinated solvent, may compete with the coordination of d,l-Pro to the Zn 2 + metal ions more than MeOH, whichi sk nown to have less binding strength towards metal ions as compared to DMF.T his was confirmed by calculating the binding energy values for MeOH,D MF,a nd proline to the Zn 2 + ions of aZ n-MOF-74 unit cell. Indeed, the binding energies for both prolinea nd DMF were lower than the one for MeOH (À150.78 kJ mol À1 ). Three differentb inding energies of l-Pro werec alculatedw hen considering all possible interactionso fl-Pro with the framework. For proline molecules retained in the pore via weak interactions, an energy of À202.37 kJ mol À1 was calculated, slightly higher than the DMF binding energy of À216.12 kJ mol À1 . The energies corresponding to proline coordinating to the metal ions are significantly higher,b oth for bidentate (À341.55 kJ mol À1 )a nd monodentate (À468.46 kJ mol À1 )c oordination.
In at ypical synthesis, Zn-MOF-74 was suspended in as olution containing the d-o rl-Pro in am olar ratio MOF to d,l-Pro of 1:1a nd stirred at ambient conditions for 48 h. The corresponding amount of prolinew ould fill up to ah alf the unit cell of MOF-74,w ith af ormula of Zn 2 (dobdc)(Pro) based on af ully loaded MOF with proline coordinated monodentate to the Zn 2 + ions via the carboxylate oxygen ( Figure 3). This was chosen in view of its potentiala pplication, such as asymmetric catalysis in which the diffusion within the MOF channels is a key parameter. Thus, we choose the Zn 2 + to Pro molar ratio of 2:1t op revent possible pore blocking. The recovered solid materials were analysed using complementary techniques to determinet heir purity,s tability and chirality.T here is ap erfect match between the PXRD patterns of Zn-MOF-74 and its derivatives preparedi nM eOH and in the presence of l-Pro, indicating that the overall framework topology is retained ( Figure 4). For the materials prepared in DMF,t he PXRD pattern reveals a new peak around 2 q = 9.28 that indicates ac hange in the framework. Notably,t he PXRD analysisa lso showst hat the type of compounds obtained in MeOH and DMF are the same,   Thermogravimetric analysis (TGA) combined with differential scanningc alorimetry (DSC) was carriedo ut on the modified MOFs and MOF-74 sample (Figures 5a nd S2).T he Zn-MOF-74 was characterised by TGA-DSC and it shows an initial ca. 14 % weightl oss at around1 00 8Cc orresponding to H 2 Om olecules as ar esult of the air exposure after activation. The decomposition of the framework is indicatedb yt he sharp exothermic peak observed in the DSC at ca. 365 8C( Figure 5). Synthesis of Zn-MOF-74-l-Proi nM eOH shows am uch larger initial weight loss of ca. 21 %a ccompanied by ab road DSC peak around 100 8Ct hat can be assigned to solventm olecules, both coordinated and adsorbed in the material. Above 100 8C, both TGA and DSC curvesa re similar to those obtainedf or the Zn-MOF-74. The Zn-MOF-74-l-Pros ample prepared in DMF shows ar elatively stable plateau until ca. 200 8C, as it is expected for the strong interactions established between DMF and the Zn-MOF-74 framework. The framework decompositioni ss lightly shifted to ah ighert emperature as compared to the pristine material, also in agreement with the PXRD datat hat indicate that some structuralc hanges occur.S imilar results are observed for the Zn-MOF-74-d-Pro prepared in MeOH and DMF ( Figure S2).
An uptake of ca. 1.2 % l-Pro was calculated for the sample prepared in MeOH,b ased on the DTG curve relatedt ot he broad peak in the region 260-325 8C, which corresponds to the removal of l-Pro from the framework ( Figure S3). Thisi ndicates that l-Pro is adsorbed and possibly bound throughw eak interactions in the Zn-MOF-74 framework. For the material prepared in DMF by using l-Pro, an uptake of ca. 6.9% was also calculated using the DTG curve ( Figure S3). In this case, the sharp exothermic peak at 306 8Ci nt he DSC curve indicates a monodentate coordinationo fl-Pro to Zn 2 + ,v ia one oxygen from the carboxylate group ( Figure 5). [54] The bidentatec oordination of l-Pro to the Zn 2 + ion is unlikely,a si tw ould give a DSC peak at highert emperature, ca. 345 8C, in agreement with as tronger binding effect. [55] FTIR spectra were recorded to further prove the presence of the l-a nd d-Prow ithin the framework of the synthesised ma-terials.S imilart ot he PXRD studies, independent of the handedness of the chiral amino acid, the materials have identical FTIR spectra (Figures S4 and S5). The higherc oncentration of l-a nd d-Pro in the materials obtained in DMF allows to identify the characteristic vibration modes, including the NH stretching at 3216 cm À1 and the CH 2 vibrations in the range 2800-3200 cm À1 ( Figure S5). [55] To probe the existence of chirality within the synthesised materials, we used the vibrational circular dichroism (VCD) technique. VCD measures the differential absorption of leftand right-handedc ircularly polarisedi nfrared light [56] and it is a powerful spectroscopicm ethodt oa nalyse the chiralitya tt he molecular level. VCD is at echniquet hat extends the regular circulard ichroism (CD) measurements into the infrared region. [43] It is av ery simple spectroscopicm easurement and can be used to obtain chiral information for moleculesi na ll phases.C onsidering that the VCD signals originate from vibrational transitions, it can be used to study all molecules, even if they lack an UV chromophore as neededf or CD. [57] So far,i t has been used to investigate small molecules, [38] synthetic polymers, [39] but also metal coordination polymers. [40][41][42][43] Furthermore, the quality of the VCD spectroscopic data is highly related to the sample homogeneity, [58] therefore our studies focused on post-modified materials synthesised using aZ n-MOF-74, which is highly homogeneous and has crystals with narrow size distribution. [46] The samples prepared in DMF show strong VCD signals in the region 1300-1700 cm À1 assignedt ot he coordinatedp roline molecules. The signala t1 609 cm À1 correlates very well with the IR band at 1600 cm À1 .T his band is assigned to the asymmetrics tretching of the carboxylate group of proline ( Figure 6). [59] Moreover,t here is ab and shift to al ower wavenumber when compared to the free proline, n as (COO À ) = 1626 cm À1 ,i na greement with the coordination of the carboxylate group to the Zn 2 + ions. [59] This also leads to as hift in the position of the VCDs ignals ( Figure 6) as compared the VCD spectra of l-a nd d-Pro (Figure S6). Thep resence of opposite signals originating from Zn-MOF-74-l-Pro andZ n-MOF-74-d-Pro indicates that the amino  acid retains its chirality upon coordination to the framework. For the samples prepared in MeOH,t he signals are relatively small when compared to those synthesised in DMF ( Figure S6). This correlates to the very low content of amino acid present in the MOFs.T he weak VCD signals for the samples prepared in MeOH match the ones of l-a nd d-Pro ( Figure S6) because proline is not coordinated to the metal ions as indicated by the broad peak around 300 8CinD SC ( Figure 5).
For the materials synthesised in DMF,T GA and DSC analysis indicateamonodentate coordination of prolinev ia one oxygen of the carboxylate group. Computational methods were used to determine the most favourable coordination of proline to the Zn 2 + ions within the framework of Zn-MOF-74-l-Pro, which indeed strongly supports the monodentateb inding. The calculations also show that the bidentate coordination of proline to the open sites of two neighbouring Zn 2 + ions would cause as train on the framework that, in turn, would lead to at wisting of the linker molecule in order to accommodate the coordinated prolinem olecule ( Figure S7). This is a much less stable framework than the initial one. Although the monodentate coordination of proline to Zn 2 + is nicely demonstrated by both experimental and theoretical studies,the PXRD analysisi ndicates clearly that structuralc hanges have occurred. The calculated PXRD pattern,w hen adding metal ions in the channels of MOF-74, indicates that the peak observed at 2 q = 9.28 corresponds to the presence of Zn 2 + ions in the middle of the hexagonal channels of the Zn-MOF-74 framework. It suggests that defects are formed during the post-synthetic modification of Zn-MOF-74 and some Zn 2 + ions have migrated within the framework. This meanst hat the post-synthetic modification of the MOF in DMF has led to the formation of defects. By using 1 HNMR analysiso ft he digested MOFs,t he ratio of ca. 1:1.7 linker to l-Pro was calculated for the sample prepared in DMF ( Figure S8). This was not possible for the samples synthesised in MeOH most likely due to the low amount of amino acid. Since the initial 1:1r atio of linker to amino acid is changed significantly,itisc lear that the post-syntheticallym odified MOFs have missing linker defects due to the dissolution of the MOF during the loading process. This was further observed when analysing the sample morphology as SEM images show that the integrity of the crystalsi sh eavily damaged ( Figure 7).
To further confirm the coordination of proline and to investigate the structuralc hanges observed, solid-state 13 CC PM AS NMR studies were carriedo ut for the samples prepared by using l-Pro in DMF and MeOH.T he 13 CC PM AS NMR spectra of Zn-MOF-74-l-Pro obtained in MeOH ( Figure S9) shows four resonances at 174.0, 156.8, 127.7 and 126.5 ppm, corresponding, to C1, C3, C4 and C2 of the dobdc 4À linker,r espectively.I n line with the TGA-DSCd ata, there is no methanol signal (expected at % 50 ppm) and, due to the low amount of amino acid, there were no l-Pro signals. This is not the case forZ n-MOF-74-l-Pro synthesised in DMF that exhibits slightly broader resonances fort he organic linker (173.9, 157.5, 127.4 and 126.8 ppm), as well as ab road resonance at 165.7 ppm assigned to DMF.T he C3 peak of the linker is most affected by guest adsorption, suggesting perhaps an NH···O(C3) hydrogen bondingi nteraction between prolinea nd the MOF or ad istortion of the coordination environmento ft he Zn so that the ZnÀO(C3) bond is lengthened. The remainings ignals can be attributed to l-Pro and the two inequivalent CH 3 groups of DMF.T he carboxylate region contains two signals for C6 of l-Pro at 181.3 and 178.7 ppm, suggesting two molecular conformations that can only be due to the partial racemisationo fl-Pro. Thisw as not expected due to the mild synthesis conditions used and racemisation of l-Pro upon PSMw as previously reported only for the post-synthetic thermal deprotection of DUT-32-NHPro. [60] The other resonances for l-Pro are also split with the C2 signal at 62.0 and 60.5 ppm, C5 at 50.2 and 48.1 ppm, and C3 and C4 at 37.9, 33.7, 31.4, 29.6 and 28.7 ppm.
Considering that the chiral Zn-MOF-74 materials are highly relevant for applications in asymmetricc atalysis or enantioselectiver ecognition, the porosity of the materials wass tudied using nitrogena dsorption (Figure8). The results show that proline is preferentially introduced in the framework, resulting in ad ecrease of the nitrogen adsorption.T his was also observedi np reviousr eports for proline-modulated Zr-MOFs. [47,52,61] For DUT-67-Pro, the surface area is decreased after proline binds to the Zr-cluster by solventa ssisted ligand incorporation (SALI), exchanging ap osteri the acetate ions. [52] Notably,f or the Zn-MOF-74-l-Pro materialo btained in DMF, our studies show that l-Pro is presenta sadefect-capping structurale lement and not stuck within the pores. This is likely due to the presence of defects in the originalM OF structure, that allow coordination of the amino acid molecule. Similarly, for the application of Zr-MOFs in aldol asymmetric catalysis, the functionalisation with proline is achieved on defected MOFs,w ith reduced cluster connectivity. [52] In the adsorption isotherm of the Zn-MOF-74-l-Pro synthesised in DMF,t here is also am icropore to mesopore transition in the material after modification, shown by ah ysteresis in the region 0.6-1.0 p/p 0 and the increased micropore diameter as compared to the pristineZ n-MOF-74 (Table S1). Indeed, both TGA and 1 HNMR analysisa lso indicate that defects have been introduced in Zn-MOF-74 during the post-synthetic modification,w here dobdc 4À or Zn 2 + are missing in the framework. In the material synthesised in MeOH there is no indication of defects, and at ype Ia dsorptioni sotherm is obtained. This corresponds to am icroporous structure, with as urface area of 264 m 2 g À1 ,m ore than fourfold lower than Zn-MOF-74 (1168 m 2 g À1 ). Zn-MOF-74-l-Pro samples synthesised in MeOH have little or no additional defects after PSM as confirmed by the size of the pore diameter (Table S1)a nd additional electron microscopy imaging (Figure S10).
To prove the pore blocking effect,Z n-MOF-74-l-Pro was tested in the asymmetric aldol reactionb etween acetonea nd 4-nitrobenzaldehyde (pNBA) (Scheme 1). The aldol reaction is usually catalysed by coordination compounds via preformed stable enolates, but also by organocatalysts through an enamine mechanism. [62] In particular, l-Pro is awell-known asymmetric homogeneous catalyst for aldol reactions. [63,64] It is reported that the [Zn(l-Pro) 2 ]c omplex is aw ater-stable catalystt hat catalyses enantioselectively the aldol reaction. [65] Therefore, the choice of reactants, along with their corresponding b-aldol product, is based on their good fit within the Zn-MOF-74 pores. The catalytic tests were also used to confirm the presence of proline within the MOF-74 pores and not at the surface of the MOFs prepared. If prolinew ould have been attached at the MOF-74 surface, the aldol reaction would proceed in an enantioselective manner because the framework itself is am ild Lewis acid that can only influence the rate of the reaction. The role of Lewisa cid catalysts was previously investigated by addition of chloride salts (e.g.,Z nCl 2 )t ot he l-Pro-catalysed aldol reaction of cyclohexanone and pNBA. [66] This lead to improved conversion and stereoselectivity without influencing the formation of the preferred enantiomer. [66] Indeed, the catalytic testing of the pristine MOF-74 shows no significant ee values (Figure 9) similar to the use of MIL-101, another MOF containing mild Lewis acid sites used as catalyst for the asymmetric aldol reaction between acetonea nd pNBA that gave no ee. [27] As expected, upon blocking the catalyst's pores,t he incorporated chiral amino acid is not accessible and thus the asymmetrica ldol reactiond oes not occur enantioselectively.
Initially,Z n-MOF-74-l-Prow as synthesised and catalytically tested by using non-activated Zn-MOF-74i nD MF/ac (4:1, v/v) because DMF is similar to DMSO,t he solvent employed in the asymmetrica ldol betweena cetonea nd pNBA by using l-Pro. [64,67] The choice of solventi sr elated to the common use of DMF in MOF synthesis and using as-synthesised materials would reduce the number of steps for catalystp reparation. To confirm that the structureo btained was the same, the material was characterised by PXRD and TGA ( Figure S11). Indeed, the same PXRD pattern was obtained when compared to that of the materialm odified in DMF by using MOF-74w ith exposed open metal sites ( Figure 4). Thus, the activations tep does not influence the formation of the defected structure observed previously for the chiral Zn-MOF-74-l-Proprepared in DMF.Obviously,t he catalytic testing of this material in the asymmetric aldol reaction as am odel reactions hows very low ee (Figure 9) due to blocking of the pores. Ap revious study of the same aldol reactioni nD MF catalysed by CMIL-1, am odified MIL-101 with proline-basedl igands coordinated to the open chromium metal sites, shows fair enantioselectivity of 66 %, after 36 h (Figure 9).
To gain insight in the solvent effect in the post-synthetic modification with proline, another polar aprotic solventw as chosen, that is, THF,w hich has al ower binding strengtht han DMF.T he synthesis of Zn-MOF-74-l-Pro was performed in THF/ Scheme1.The asymmetric aldol reaction between acetone and 4-nitrobenzaldehyde(pNBA) to obtain the chiral b-aldolp roduct. The reaction can also undergowater elimination to form the a,b-unsaturatedp roduct. Figure 9. Enantioselectivity towards the (R)-b-aldol product in the asymmetric aldolr eactionb etween acetone and pNBA by using 30 mol %l oading of Zn-MOF-74-l-Pro in 4:1, v/v,THF/acetone (black) or DMF/acetone (grey). Literature reported ee values for using 10 mol %l oadingofC MIL-1. [27] 5mol % loadingof[ Zn(l-Pro) 2 ](blue)w as tested in H 2 O. [65] acetone( 4:1, v/v)a nd the resulting material characterised by using PXRD and TGA (FigureS13). PXRD analysiss hows the incipient formationo ft he peak at 2 q = 9.28,w hich is very pronounced in the case of Zn-MOF-74-l-Pro materials prepared in DMF.T he reduced intensity of this peak for the Zn-MOF-74-l-Pro in THF indicates that the structure does not suffer major structuralc hanges. Also the ca. 13 %i nitial weight loss around 100 8Co nt he TGA curve indicates that this materiali sag ood candidate for catalysis, with more accessible active sites. (Figure S11). Indeed, the testing of the catalyst, in THF,s hows improved ee. This can be attributed to the catalytic activityo f proline molecules incorporated in the modified MOF,a nd not to the freelyp roline in solutionw hichg ives almost 70 % ee (Figure 9). Unlike our catalyst, in the presence of CMIL-1, the reaction in THF does not occur,e ven after 48 h. [27] This can possibly be attributed to the differencesb etween CMIL-1 and Zn-MOF-74-l-Pro, such as the prolinebinding within the framework.
In order to obtain af air comparison to the CMIL-1 catalyst, the subsequent catalytic tests of Zn-MOF-74-l-Pro were done using 10 mol %l oading. The study focused on the materials synthesised in MeOH using activated MOF-74 starting materials and the catalytic testing was done in DMF,T HF,A CN andn eat acetone( Figure S12). The pore blocking effect for these catalysts should not be observed as the pores of these catalysts are more accessible according to the porosity studies ( Figure 8 and Ta ble S1). The homogeneous catalystw as initially tested and showed moderate to good performances in all solvents, with the highest enantioselectivityi nD MF.H owever,t he heterogeneous proline modified MOF-74 catalysts yieldedl ess than 5% of the b-aldol product ( Figure S12). We attribute the low performance of the catalystt ot he low amount of proline present in the catalyst. It shouldn ot be related to pore accessibility since the integrity of the framework is retained after reaction as confirmed by FTIR andP XRD studies ( Figures S13 andS 14).

Conclusions
The incorporation of prolinew ithin the Zn-MOF-74 framework was achieved by post-synthetic modification.C ombined experimentala nd theoretical studies show that this synthetic approach enables the introduction of both chirality and defects in the MOF framework. Vibrational circular dichroism studies demonstrate not only the binding of the l-Prot ot he Zn-MOF framework but also that l-Pro retains its chirality upon its incorporation within the MOF framework. The monodentatec oordination of l-Pro was determined by calculatingt he binding energies of proline with the Zn 2 + metal ions, whichw as also confirmed by the experimental studies. The presence of defects as well as the strong coordination of proline and DMF to the Zn 2 + within the framework reduces pore's accessibility,a s indicated by the nitrogen adsorption studies and the low enantioselectivity obtained in the aldol reaction between acetone and 4-nitrobenzaldehyde. To the best of our knowledge, this is the first study reporting on the facile synthesis of MOF-74 derivatives containing both chiral centres andd efects. Despite the low catalytic activity of the synthesised materials, we believe that this study has broughtf urther insighto nt he potential of PSM as ag eneral and effective methodf or the synthesis of chiral MOFs.

Experimental Section
Materials and synthesis:A ll chemicals were purchased from Sigma Aldrich, except l-a nd d-Proline which were purchased from TCI Chemicals, and used without further purification. The digestion of the MOF samples was performed following ar eported procedure. [47] Zn-MOF-74:S ample was prepared using ar eported procedure [68] and activated following ar eported procedure. [54] The MOF was heated under vacuum from room temperature to 80 8C, from 80 8C to 100 8C, from 100 8Ct o1 50 8C, from 150 8Ct o2 00 8C, and from 200 8Ct o2 65 8C, at ac onstant rate of 4 8Cmin À1 ,w ith the temperature held at 1h at the end of each ramp;e xcept for at 265 8C, at for the temperature was held for 12 h. The same procedure was used to treat all MOFs prior to N 2 adsorption measurements. The General method for the catalytic testing:E ach catalytic reaction was performed by mixing 4-nitrobenzaldehylde (0.076 gram; 0.5 mmol;1 .0 equiv.), proline (0.017 gram;0 .15 mmol;3 0.0 mol %) and acetone (1 mL;1 3.61 mmol;2 7.2 equiv.) in 4mLo fo rganic solvent, except for the neat testing. The reaction was carried out at room temperature. After 20 hours of reaction time, the reaction mixture was quenched with an aqueous saturated ammonium chloride solution (5 mL), the catalyst was removed by gravitational filtration and the reaction mixture was extracted with ethyl acetate (3 10 mL). The combined organic fractions were dried on magnesium sulphate, which was separated from the reaction mixture by gravitational filtration. Solvent evaporation of the filtrate followed by drying under vacuum at temperatures up to 100 8Cr esulted in as olid. This solid was dissolved in chloroform-d and separated by TLC-chromatography eluting with hexanes/ethyl acetate (3:1). The target compound was scraped off the support and eluted with ethanol using am icrocolumn. The ee was determined in duplo by chiral-phase HPLC analysis on aYMC column. Physical characterisation:I nfrared spectra (4000-400 cm À1 ,r esol. 1cm À1 )w ere recorded on aV arian 660 FTIR spectrometer using a KBr module. VCD spectra of the rotated solid samples, prepared as KBr pellets, were obtained on aB ruker Vertex 70 FTIR spectrometer equipped with aB ruker PMA50 VCD module. Powder X-ray diffraction (3-608,2 .58 min À1 )m easurements were carried out on a Rigaku Miniflex X-ray Diffractometer using Cu Ka radiation (l = 1.5406 ). Thermogravimetric analysis (35-5008,5Kmin À1 )a nd differential scanning calorimetry were performed using aN ETZSCH Jupiter STA4 49F3 instrument. The measurements were carried out under flow of air (10 mL min À1 )a nd protective argon (10 mL min À1 ). N 2 adsorption isotherms were measured at 778 Ko n aT hermo Scientific Surfer instrument after stepwise evacuation from room temperature to 80 8C; from 80 8Ct o1 00 8C; from 100 8C to 150 8C; from 150 8Ct o2 00 8C, and from 200 8Ct o2 65 8Ca ta constant rate of 4 8Cmin À1 ,w ith the temperature held for 1h at the end of each ramp, except for at 265 8C, at which all samples were held for 12 h. 1H NMR spectra were recorded with Bruker Bruker AMX 400.1 MHz spectrometer.D MSO was used as solvent, and the 1H NMR spectra were referenced to the residual solvent signal. The morphology of the samples with sputtered gold was studied by using Scanning Electron Microscopy (SEM, FEI Verios 460 scanning electron microscope) operated at 5kV. Solid-state NMR spectra were recorded using aB ruker Avance III spectrometer equipped with aw ide-bore 9.4 Ts uperconducting magnet (at Larmor frequencies of 400.13 and 100.9 MHz for 1 Ha nd 13 C, respectively). Samples were packed into standard 4mmm agic angle spinning (MAS) rotors and spectra were recorded with aM AS rate of 12.5 kHz. 13 CNMR spectra were recorded with cross polarisation (CP) from 1 Hu sing ac ontact pulse (ramped for 1 H) of 1msa nd high-power (n 1 %100 kHz) TPPM-15 1 Hd ecoupling was applied during acquisition. Signal averaging was carried out for 2048 transients with ar ecycle interval of 3s.C hemical shifts are reported in ppm relative to TMS using l-alanine (CH 3 d = 20.5 ppm) as as econdary solid reference.
[69À72] The used functional SCAN + rVV10 [73] is av ersatile van der Waals density functional developed by combining the Strongly Constrained Appropriately Normed (SCAN) meta-GGA semi-local exchange-correlation functional with the rVV10 non-local correlation functional. [74,75] SCAN is comparable to hybrid functionals at the cost of about only 3t imes PBE.