Tuning the Properties of MOF‐808 via Defect Engineering and Metal Nanoparticle Encapsulation

Abstract Defect engineering and metal encapsulation are considered as valuable approaches to fine‐tune the reactivity of metal–organic frameworks. In this work, various MOF‐808 (Zr) samples are synthesized and characterized with the final aim to understand how defects and/or platinum nanoparticle encapsulation act on the intrinsic and reactive properties of these MOFs. The reactivity of the pristine, defective and Pt encapsulated MOF‐808 is quantified with water adsorption and CO2 adsorption calorimetry. The results reveal strong competitive effects between crystal morphology and missing linker defects which in turn affect the crystal morphology, porosity, stability, and reactivity. In spite of leading to a loss in porosity, the introduction of defects (missing linkers or Pt nanoparticles) is beneficial to the stability of the MOF‐808 towards water and could also be advantageously used to tune adsorption properties of this MOF family.


Introduction
Metal-organic frameworks (MOFs) have attracted much attention in recent years due to their use in awide range of applications from gas storage and separations, [1,2] catalysis, [3][4][5] mechanics, [6,7] to drug delivery, [8] to name just af ew.M OFs are a subclass of coordination polymers constructed from metal oxo-clusters connected via organic linkers to form a3 Dp orous framework. [9] The wide library of both organic and inorganic components leads to an almostinfinite window to design MOF structuresa nd chemistries. [10] In particular,c ontrolling the properties of MOFs is of significant importance and many strategies have been used in this regard. Someofthese include the incorporation of mixed linkers, [11] mixedm etals [12] and linker functionalization [13] into the MOF structure. Am ore recent con-cept to improveM OF properties is to use defecte ngineering. [14][15][16] For instance, in Zr-based MOFs, such as UiO-66, [17,18] the removal of af ew linkers from the framework is not detrimental to its structural integrity, [19] whilst it improves its reactivity and catalytic activity. [16,20] The increased accessibilityo f metals ites as the linkersa re removed has been proposed as the mechanisma tt he origin of thesei mprovements. [21,22] More recently,t he tuning of the hydrophilicity of MOF crystalsv ia defecte ngineering for efficient oil/water separation has been demonstratedi nZ r-based UiO-66. [23] The enhancement in catalytic activity has also been achieved by preparing as o-called "composite MOF" or "metal@MOF" by metal impregnation [24] or metale ncapsulation. [25,26] Similarly,t uning of the catalytic properties of MFI-typez eolitesh as been suggested following the incorporationo fM oat defects ites within the structure. [27] In this case, metal incorporation equally rendered the zeolite more hydrophobic. In the case of MOFs, metal encapsulation has been considered to be more advantageous than metal impregnation in retaining the encapsulatedm etal insidet he MOF to preventl eaching during use. [26] Nevertheless metale ncapsulation maya lso be detrimental to the system since it may lead to metal deposition on the externals urface of the MOF or to a framework degradation due to the formation of large nanoparticles. [26] Apart from the extensively studied UiO-66,t here exist other zirconium-based MOFs with ad ifferent number of connectivity's and variable topologies. [17,28] One can cite NU-1000 that has an 8-connectivity with csq topology [29] and MOF-808c haracterized by a6 -connectivity with spn topology. [30] MOF-808 with hexanuclear zirconium is of particulari nterestb ecause of its lower node-connectivity that can provide more accessible open metal sites, especiallyi fm issing linker defects are engineeredi nto the framework. This MOF is constructed from zirconiumc lusters linked togethert hrought rimesic acid or ben-zene tricarboxylic acid (BTC). [31] Its micropore size distribution (pore size % 1.8 nm) and open metal sites were shown to be useful for some catalytic applications,s uch as the hydrogenation of ethyl levulinate to g-valerolactone, [18] the hydrolysis of nerve agent simulants, [31] and in the Meervein-Poondorf-Verley (MPV) reaction. [32] In this work defect engineering, metal encapsulation and both effects combineda re investigated in the case of MOF-808. The synthesised samples are first characterised in terms of structure, morphology, and texture. Water adsorption is used to follow variations in hydrophobicity of the samples andf urther changes in activity are probedv ia CO 2 microcalorimetry that allows for ad irect access to the adsorption enthalpy.T his investigationh ighlights the complex interplay between the synthesis, morphology,t exture, and chemistry of the materials which influence their sorption properties.

Synthesis
The general procedures to synthesize MOF-808 samples were adaptedf rom refs. [32] and [35].P ristine MOF-808 (noted MP) was synthesized using ZrOCl 2 .8H 2 Oa nd H 3 BTC with molar ratio 1:1a nd with as ynthesist imes of 7days. The structural defects were introduced by modifyingb oth reaction time and metal to linker ratio. The defective MOF-808( noted MD) wass ynthesized with ap recursor molar ratio of 3:1w ith as ynthesis times of 2days. All the detailso ft he synthesis are given in the Supporting Information (section 1).
The effect of metal encapsulation was investigated using Pt nanoparticles (Pt-NPs). The nanoparticles were synthetized using H 2 PtCl 6 .6H 2 Oa sP tp recursor and PVP as stabilizer The procedure was adapted from refs. [36,37,38],a nd is described in Supporting Information (section2). Prior to their use in the MOF synthesis, Pt nanoparticles (Pt-NPs) werec haracterizedi n depth (section2in Supporting Information), andT EM observations evidencedt hat the Pt nanoparticles possess an average diameter of 3.5 nm. The as synthetized Pt-NPs were incorporated during the synthesis of MOF-808t op roduce Pt-encapsulated pristine MOF-808 (Pt@MP)a nd Pt-encapsulated defective MOF-808( Pt@MD). All the details in terms of reactantq uantities and synthesis procedure are given in Supporting Information (section1 and Ta bleS1).T able1 summarizest he various synthetized samples together with some of their relevant characteristics for the present work.

Phase purity and crystallinity
The removal of DMF is shown to be completed after activation under vacuum to 373 K. This is confirmedb yFourier-transform infrared (FTIR) measurements ( Figure S4 in Supporting Information). Indeed, the vibrationb and corresponding to the CÀNv ibrationo fD MF located at 1256 cm À1 ,i sn ot observed in any of the synthesized samples. The X-ray diffraction (XRD)p atterns of all the samples (FigureS5i nS upporting Information) reveal that only the diffraction peaks corresponding to the cubic Fd3 m structure of MOF-808a re observed as confirmed by the simulated diffractogram. [32][33][34] It is worth noting that, in the case of Pt@MP and Pt@MD, no Bragg peaks of platinum are evidenced. This is linked to the nano-size characteristics of the Pt-NPs( Supporting Information, section 2). The instrumental broadening at high angles together with the quantity of Pt in the composite material make the detection of the Pt diffraction lines impossible. The presenceo fp latinum is nevertheless confirmed by TEM observations and by ICP analysis that gave aq uantity of Pt equal to 0.1 wt %a nd 0.04 wt %f or Pt@MP and Pt@MD, respectively.

Effect of synthesis conditions on crystal morphology
Scanning electron microscopy (SEM) was used to gain insight into the crystal morphology.A ss een in Figure 1, different synthesis conditions result in different crystal morphologies. Reactant stoichiometry strongly affects the crystal morphology. These effects can be observed by comparing the SEM images of MP (Figure 1a and a') with the ones of MD (Figure 1b  Both samples show regular octahedral crystal shapes.I nt he case of MP,t he crystal sizes are not uniforma lthough they seem dominated by slightly larger crystals of about 2 mm. Sample MD has smallerc rystal sizes with am ore uniform size around0 .8 mm. This may be linked to the shorter reaction time in MD synthesis that is not enough for the growth process to produce the larger crystals. Indeed, longerr eactiont imes would allow more nutrients in the solution to be incorporated into the growth phase as it has been observedi nz eolite synthesis. [39] Interestingly,M Ds hows less agglomerated crystals and this may be linked to the excessive amount of zirconium precursor in this particulars ynthesis reaction. Indeed, ap roposed mechanismf or MOF formation consists of an initiation stage through the deprotonation of organic linkers and dissociationo ft he metal salt followed by ac omplexation of the deprotonated organic linkers with the metal ions. [40] Whereas the organic linkers are provided directly as ar eactant, the metal precursors have to form the secondary building unit (SBU) in the synthesis process prior to combining with the organic linkers. The formation of the SBUi sr eported to be crucial for the MOF assembly process. [41] Thus, the excessive amount of zirconium precursor could increasethe probability of SBU formation that could in turn facilitatet he formation of the MOF crystals.

Effect of Pt-NPs encapsulation on crystal morphology
By comparing the SEM images of samples MP (Figure 1a and a') and Pt@MP ( Figure 1c andc '), it can be observed that the presence of Pt-NPs strongly influences the formation of MOF-808 crystals: they possess am ore regular shape and less roughness. However,t he crystal size distribution remains heterogeneous in both cases. In terms of crystal size, Pt@MP has approximately2 0% smaller average size than MOF-808 withoutp latinume ncapsulation MP.I tc an be proposed that platinum nanoparticles may act as seeds for MOF nucleation to occur.T he zirconiumi ons may arrange around the platinum nanoparticle before coordinatingw ith BTC and when the crystallization begins. This therefore leads to the platinum nanoparticles being encapsulated inside the MOF particles. As there are more nuclei, and sufficient linkers (1:1 ratio Zr:BTC), there will be more crystalsb ut with smaller sizes. Similar mechanisms have been reported in the case of Kegginp olyoxometa-late (POM) encapsulated in severalM OFs, such as Al-MIL-101-NH 2 and HKUST-1. [42,43] POM was also reported to assist the crystal formation of both Al-MIL-101-NH 2 and HKUST-1, acting as nucleation sites and finally encapsulated inside the cage. Other interesting features are observed when comparing sample MD (Figure 1b and b') with sample Pt@MD (Figure 1d and d'). In this case, the combinationo fe ffects (excessive reactant stoichiometry and Pt-encapsulation) eventually creates holes in most of the particles (see circles and arrow in Figure 1d andd ') andg enerates ah igh degree of heterogeneity in crystal size. Thoseh oles may be linked to (i)some Pt-NPs that leave the crystal or (ii)tot he free PVP introduced during the synthesis or (iii)toa ni ncomplete coverage of the MOF aroundthe Pt seeds.

Pt-NPs distribution in MOF-808
The presence of Pt-NPs inside the MOF-808 samples is evidenced using Scanning TEM (STEM)a ss hown in Figure 2. It can be observed that the particles are reasonably distributed inside the crystal and almostnoa gglomerates are visible.S ince only as mall contribution of Pt-NPs could be observed bulging out from MOF edges, it suggests that most of the Pt-NPs are encapsulated inside the MOF.T he distribution of the Pt nanoparticles in Pt@MP is rather uniform and follows an ormal distribution (Figure 2a') with an estimated mean size of around 4.5 nm. The fact that the mean size of the nanoparticles is slightly larger than their initial size prior to encapsulationi ndicates as light nanoparticle growth during the formationo f Pt@MOF-808. This may be linked to ap artial removal of the PVP duringt he MOF synthesis. It is worth noting that the size of the Pt-NPs exceeds the characteristicp ore size of the MOF-  808 (1.8 nm). [33] This is not unexpected given the initial size of the Pt-NPs. Such observations have already been made for both metal-impregnated and metal-encapsulated MOFs. [25] One can expect that this observation may result in local defects or deformation of the host framework.
Nevertheless,t he overall structural integrity of the MOF is not affected as can be observed from the sharp Bragg diffraction peaks ( Figure S5 in Supporting Information). In the Pt@MD sample, the Pt-NPs are also encapsulated inside the crystal but they seem more concentrated in thec entre of the crystal (Figure 2b). The average diameter of the Pt-NPs was estimatedi nt he same manner as for the Pt@MP sample and the particled istribution is given in Figure 2b'. Onec an observe that their mean value is slightly smaller than in Pt@MP. One possible explanation is that the PVP capping agent may be removed due to longerr eaction time for Pt@MP.I nc ontrast, some PVPi ss till remaining in the Pt@MD preventing the formation of larger Pt-NPs. Unfortunately,d ue to the encapsulation of Pt(PVP)-NPs inside the MOF crystal,t he peaks of PVP in the Pt@MOFss amples are not easily distinguishable in FTIR analysis( Figure S4 in Supporting Information).

Thermal and structural stability
The stability of the activated materials was followed using thermogravimetry analysis( TGA) and variable temperature XRD (VTXRD). The combination of these methods allows for a clearer understanding of the correlation between mass loss (speciesr emoval) ands tructure integrity.T GA analyses were carried out under synthetic air (Figure 3a)a nd under nitrogen flow (Figure3b). The analysiso fm ass decomposition was performed using the experiments carried out under air flow since it guarantees ac omplete oxidation of the samples. As can be seen in Figure 3a and 3b, below 373 K, av isible mass loss is observed for almosta ll samples (between 2a nd 5%). It can be attributed to the removal of water or other loosely bound species adsorbed on the external surfaceo ft he MOFs.F or the experiments performed under Air,i nt he case of Pt@MD, al arger mass loss is observed (around 8% at 373 K) and mayb ec orrelated to the larger externals urface area (see Ta ble 2i nn ext section). This is consistentw ith the SEM analysiss howingt he presence of holes in several Pt@MD crystals (Figure 1d and 1d').
It is noticeable that, whilet here is as mall plateau around 373 Kf or the non-defective samples, it is lacking in the case for MD and Pt-MD for which the mass continuously decreases with increasing the temperature. In the temperature region between 373 Ka nd 623 K, the mass loss can be attributed to the removal of coordinated -OH and H 2 Ot hat replacet he formic acid after activation. [31] Indeed, the formic acid was used during the synthesisasmodulator andwas removed during activation.N evertheless, we cannote xclude the possibility of incomplete formic acid replacement. Other work attributed these features to the loss of any remaining DMF solventt hat has not been completely exchanged and/ore vaporated. [28] However,i nt he present case FTIR analysiss trongly suggests that DMF has been completely removed from the samples ( Figure S4 in Supporting Information). The last notable region in the TGA curve is located between 723 Ka nd 873 K. It corresponds to the BTC linker decomposition. Quantifying the mass loss in this region can give information on the quantity of missing linkers in each sample. The methodology to estimate the missing linker concentration was adapted from ref. [28] and is explained in detail in Supporting Information. Assuming as imilar mechanism to UiO-66, at high temperature (about 673 K), the coordinated water and hydroxyls, that replace the formate, are removed. According to Moon et al., [31] this would lead to the activated MOF-808 with af ormula of Zr 6 O 8 [(C 6 H 3 )(COO) 3 ] 2 and am olar mass equal to 1095.6 gmol À1 . Assuming that the final residue is (ZrO 2 ) 6 (Mm = 739.32 gmol À1 ), [28] the theoreticall inker loss can be estimated. The obtainedr esultsare given in Ta ble 1. Onecan observe that  for all samples, the mass losses are lower than the theoretical ones for ap erfect MOF-808. This suggests that even the pristine materials MP may already contain inherent missing linker defects. It is noticeable that excessive zirconium ratio (MD and Pt@MD) results in ah igher concentration of missing linker defects. Although the mechanism of this effect on the formation of missing linkers is still unclear,a sd iscussed above,i ti sp ossible to propose that the rate of SBU formation is accelerated by the excessive amounto fz irconium precursor and increase the competition to assemble with am ore limited number of organic linkers. This in turn may result in an incomplete MOF structuref ormationc ontaining missing linker defects. VTXRD experiments were used to follow the structural changes upon heatinga nd the consequences of the removal of chemical species on the structural integrity.A ne xample of one VTXRD pattern is presented in Figure 3c for sample MP.I t is noticeable that the sharpB ragg peaks are maintainedu pt o around4 23 K, temperature at which the loss in crystallinity starts as evidenced by the enlargement of the diffraction lines. At 523 K, the increase of the scattered intensity at low 2q values may be the signature of the sample amorphization. In the literature, the temperature at whichthe structurald egradation occurs is reported at different values. Moone tal., [31] reported the structuralc ollapse of MOF-808a bove 523 Ku nder air,w hileP lessers et al., [32] reported structuralr etention up to 423 Ku nder vacuum with degradation occurring between 423 Ka nd 473 K. These latter values are consistent with the ones observed in this study.M oreover,t his temperature range is far below the temperature of the BTC linker removal,w hich occurs at around 773-873 K. This suggests that the removal of chemicals pecies below 673 Km ay be responsible for the structural degradation. By comparing the TGA results ( Figure 3b)w ith VTXRD patterns (Figure 3c), it can be proposed that structural degradationo ccurs in the temperature region where either coordinated H 2 Oa nd -OH or formic acid are removed.T his suggests that the removalo ft hese coordinated substances could be responsible for the loss of long-range order.A lthough the connectivity between the SBU to form a framework is maintained by BTC linkers and neither by formic acid nor H 2 Oa nd OH, the removal of those latter species may, nevertheless, destabilize the structure.

Textural properties
The texture of the synthesized materials was characterized by means of nitrogen physisorption at 77 Ki no rder to evaluate the surface area and porosity.T he adsorption-desorption isotherms are presented in Figure4aa nd b. According to the last IUPAC classification, [44] they are between type I(b)a nd type IV(b) in shape which suggests they are characteristico fm icroporouss ystems having an arrow pore size distribution that are filled and unfilled at the same relative pressure. These observations are consistent with literature on MOF-808. [33,34,45] The defective and/or impregnateds amples present an additional uptake in the very last part of the isotherm (p/p 0 > 0.9) and this step is more obvious in the case of Pt@MD. It may be related to the roughness of the crystal surface, the presence of holes (Figure 1d and 1d') and to the particles izes that are all responsible for an increase in the external surfacea rea. Such behaviour was reported in defective and impregnatedM OF 808 but also in pristineM OF 808. [45][46][47][48] The surface area (BET method)a nd the pore volume at p/p 0 = 0.7 were calculated for each sample and the values are tabulated in Table 2. The external surface area calculated using the t-method is also given in Ta ble 2a nd represented in Figure 4c together with the BET areas. The pristine material MP shows the highest BET area. The values obtained in this study are consistent with ones published in the literature which vary between 1205 m 2 g À1 and 2060 m 2 g À1 for pristine MOF-808 samples. [33,34,46,48,49] The effects of different reactants toichiometries can be studied by comparing MP and MD. Both BET areas and pore volumea re slightly lower for the defective sample MD. This is as urprising result since it has been observed in other MOFs systemss uch as UiO-66, [16] that the presence of missing linkers tendst oi ncrease the surface area and the pore volume. However,i nt he present case the shorter reaction time in the case of MD may have led to an incomplete formation of someo f the crystals.
Pt-encapsulation inside the pristine sample MP also induces as light reduction in BET area. It can be correlatedt ot he fact that metal encapsulation may block someo ft he pores. [25] Such an effect has been observedf or Pd-modified MOF-808 [47] and imidazole-MOF-808. [45] Nevertheless, the small reduction in surface area could equally be linked to the difference in densities and in the quantity of Pt that contributes to the nitrogen sorption. Finally,t he case of Pt@MD is typical as it highlights a combination of effects. The textural properties (BET area and pore volume)o ft his sample are the lowest. This could be causedb yt wo factors:( i) Pt-nanoparticlest hat occupy the pores as in the case of Pt@MP and (ii)incomplete formation of the crystal as hypothesized above (Figure 1d and d').
The pore size distributions were determined using QSDFT kernels of nitrogen adsorption on carbon slit and cylindrical pores.T he half pore width distribution is given in Figure 4d. There are no major differences between the samples, that all possess am ean half pore size around 7 .T his value correspondsw ellw ith the meana pertured iameter size of 14 given in in the literature. [33,34] Activityassessment Water adsorption has been proposed asauseful method to explore the chemistry of defective MOFs. [16] Moreover it offers insight into the stability of MOFs towardw ater which is of utmost importance for being industrial applications of MOFs. [50] The water adsorption isotherms of the various samples are presentedi nF igure 5a and b. The complete set of adsorption-desorptioni sotherms are given for all the samples in Figure S7 in Supporting Information.
At low relative pressure, where the adsorption occurs at the most energetic sites, the strength of the interaction can be estimatedf rom the Henry's constanta nd this can equallyb ea n estimation of the potential reactivity.T he Henry'sc onstants were calculated following previous works [16] and the obtained values are summarized in Figure 5c.I tc an be observed that the highest Henry's constant is obtainedf or MD sample and this can be attributed to the existence of missing linker defects, providing more accessible open metal sites whichi sc onsistent with previousw orks. [16] The presence of Pt-nanoparticles strongly reduces the activity of the samples and the combined effect of encapsulationa nd defect( Pt@MD) leads to the sample having the lowest interaction with water.This may suggest that the metal encapsulation blockst he access to the defect sites thus rendering the sample more hydrophobic with respectt ot he samples withoutP t. Thist rend is similar to that observedd uring metal encapsulation in zeolites, as mentioned above. [27] The first plateau region in the water adsorption isotherm (p/ p 0 between 0.02 and 0.2 in Figure5aa nd 5b) corresponds to the completion of surfacec overage. Pore filling occurs for p/p 0 valuesa round 0.2 to 0.6;w hen significant variations in uptakes are evidenced. Interestingly,t he slope of the pore filling step is steeper for the two samples with encapsulated than for the samples withoutp latinum that evidence more gradual slopes. Ae xplanation for such ab ehaviour is probably am ore rapid water cluster formation for samples containing Pt. [51] Furthermore, Figure 5b shows ap ronounced hysteresis in the water isotherms.I ti sk nown that water can lead to instability in many MOFs ande ven fors o-calleds table MOFs,w ater can lead to pronouncedr eversible structural changes. [52] Thus the observed hysteresis can be explained by ar ehydroxylation of the sample or by reversible structural modifications during water adsorption. One can comparet he pore volume measured with nitrogen at 77 Ka nd water vapor at 298 K. To convert the amount adsorbed at the plateau (p/p 0 = 0.6-0.8)i nto ap ore volume,i ti s commonlya ssumed that the pores are filled with the condensed adsorptive in the bulk liquid state (Gurvich rule). [53] The pore volumes calculated at p/p 0 = 0.7 are given in Figure 5d and it can be observed that volumes obtained with water are systematically lower than those obtainedw ith nitrogen and the difference is smallerf or Pt-containing samples. This tendency could be explained by (i)the specific mechanismsa ssociated to water adsorption such as clustering or even structural contraction, [52] (ii)the presence of hydrophobic domainsw ithin the pores or (iii)the deteriorationo ft he MOF structure with water.I ndeed, water adsorption in MOF-808h as been reported to strongly deteriorate their structure. [33] In ordert oq uantify this effect in the samples of this study,n itrogen sorption isotherm was performedo nt he same batch after the water sorption cycle. The remaining BET area and pore volumec alculated from those isotherms can be used to monitor the extent of the deterioration of the samples.T he obtained values are presented in Figure 5e.I tis noticeable that both the BET area and pore volumef ollow the same trends which means that the un-damagedp art of the sample keeps its porosity.M oreover,o ne can observe that, while the pristine material is strongly damaged after water adsorption, the deteriorationi sl ower for the modified samples;t he less damageds ample being the defective and Pt encapsulated sample Pt@MD.
Activityassessment-CO 2 adsorption microcalorimetry at 303 K CO 2 is an ideal probe,d ue to its quadrupole moment, to explore the surface activity of porousm aterials. The CO 2 adsorption isotherms at 303 K for all studied samples and their corresponding adsorption enthalpies are presented in Figure 6.
The interactions between adsorbate-adsorbent can be well monitored at very low coverage and give indicationsc oncerning the adsorbent affinity towards this probe. [54] In this region, the interaction of an adsorbate molecule with an energetically homogenous surface will give rise to ac onstantc alorimetric signal. On the other hand, ad ecreasing signal will be observed in the case of interactions between an adsorbate molecule with an energetically "heterogenous" surface. [53,55] The energetically heterogenouss urface may arise from the pore size distribution and/or varying surfacec hemistries. [56] In the case of the samples discussed in this work, all of the CO 2 enthalpies show decreasing enthalpies up to ac overage of around2mmolg À1 ,s uggestingt hat specific sorptions ites are available for the CO 2 probe. The CO 2 adsorption isotherms ( Figure 6a)s hows two groups for the samples with or without the Pt nanoparticles. The MOF-808 samples without the Pt-NPs show higher uptakes than those loaded with the Pt-NPs. Again, this could be linked to pore blocking or to the difference in the sample densities. Unlike in nitrogen adsorption at 77 Ka nd water adsorption at 298 K, stepped isothermsa re not observed for the measurement of CO 2 adsorption at 303 K. This is correlated with the differenta dsorption mechanisms in sub-critical states (nitrogen at 77 Ka nd water at 298 K) and close to the super-critical state (CO 2 at 303 K). In the former state, the adsorbed liquid can reach relative pressures where pore filling can occur,whereas for CO 2 ,adsorption is essentially restricted to monolayere dification.
Comparing the samples without the Pt-NPs,t he defective sample MD shows lightly higher uptakes with respectt ot he MP sample. This slight difference between the samples is less visiblei nt he calorimetry data. Indeed, the adsorption energies at zero coverage obtained by linear interpolationl ead to adsorptione nergies equal to 57 kJ mol À1 for MD and 56 kJ mol À1 for MP.T hisd ifference could be related to the number of defects, presumablym issing linkersa sd iscussed above.H owever, one must underline that those energy differencesa re quite moderate with respectt ot he energy differences for samples containing the Pt-nanoparticles. The CO 2 isothermso btained with the Pt-containing samples are quite similar. This could be www.chemeurj.org the result of adsorption at similars ites. However, the initial adsorptione nergies are significantly different. With sample Pt@MP,t he initial adsorptione nthalpy is the highest observed in this sample set. This highlightst he potentialr oleo ft he Pt nanoparticles that strongly interactw ith the CO 2 .H owever,t he calorimetric signals rapidlyd rop indicating that the number of active sites that govern this strong interaction is limited. [57] Surprisingly,P t@MD has the lowest initial adsorption enthalpy. This lowest energy is in accordance with the results of the Henry's constant obtained from water adsorption.I ndeed, one would expect that the Pt would provide strong adsorption sites. One hypothesis, discussed above for water,m ay be that the metal nanoparticles are situated at the MOF defects ites. A second explanation is that the polymer protecting agent (PVP) initially present on the surfaceo ft he Pt-nanoparticles is not completely removed during the synthesis of the Pt@MD and can, thus, somehowc over the active sites in this material (explainingt he lower enthalpy), and some of the porosity (explainingt he lower uptakes in the water and N 2 isotherms).

Conclusions
In this study,t he synthesis and characterization of defective and Pt-encapsulated MOF-808 was presented.S tructurald efects such as missing linkersw ere introduced by using an excess of metal precursors in the reactionm ixture. Pt-nanoparticles have been successfully encapsulated in both pristinea nd defective MOFs. It is found that synthesis conditions (reaction time, reactants toichiometry and encapsulating Pt-nanoparticles) influence morphology, textural properties and affinity towards water and CO 2 .
The BET area of pristine MOF-808( MP) is highert han defective and Pt-encapsulated MOF-808 (Pt@MP). This indicates the roles of morphologya nd pore blockingo nt he textural properties of MOF-808. Interestingly,P t-encapsulation on the pristine MOF-808s eems to enhance the crystallinity of MOF-808 since well-defined octahedral crystal with smooth surface topography are observed.
Water adsorption,u sed to explore the hydrophilicity of the samples, demonstrate that the highest Henry's constant is obtained for defective MOF-808s ample (MD). The presence of missing linker defects that provide more accessible open metal sites, are proposed to be responsible for this effect.Interestingly,t he MOF-808s tability towardsw ater is systematically enhanced by the presence of missing linkers and/orPtencapsulation. The sample that combines both effects (missing linkers and Pt encapsulationi nP t@MD) presents ar emaining porosity after water adsorption close to 60 %( as compared with af ew %f or pristine sample.).
Probing the samples with CO 2 shows slightd ifferences upon defect introduction and/or Pt encapsulation. In general,t he increasedn umber of defects leads to slightly higher CO 2 adsorption energies. Addition of Pt into the sample with fewer defects (Pt@MP) leads to increased interactions with CO 2 .T he inverse observation is made for the sample containing the most defects (Pt@MD) which may suggest that the metal shields (or even anneals) the defect sites in the MOF.
To summarize, defect engineering of metal-organic frameworks can be achieved in the raw MOF-808m aterial which leads to expected correlations of energy vs. number of defects. The introduction of metal nanoparticles leads to ad ecorrelation which could be the result of metal shieldingt he MOF defects ites.

Experimental Section
Characterization techniques:P rior to their characterization, the samples were activated at 100 8Cu nder vacuum. The structure of both platinum nanoparticles and MOF-808 series were investigated using XRD. The measurements were carried out using aP analytical Empyrean instrument working with ac opper source (l = 1.54 ). The patterns were recorded from 2q between 38 and 508.T he same apparatus was used to perform XRD measurements upon heating in order to follow the thermal stability of the samples. The heating was carried out under an itrogen atmosphere using an Anton Paar XRK 900 furnace. Combined with TGA analyses, XRD measurements allowed some insight to be obtained on the chemical species that are responsible for the structural degradation of MOF-808.
Fourier-transform infrared (FTIR) was carried out using an ATRF TIR (PerkinElmer UATR Two). Data were collected from wavenumber 450 to 4000 cm À1 .T he samples were placed in ah older and measured without mixing with KBr. The morphology of the MOF-808 series was studied using Scanning Electron Microscopy (SEM). The SEM measurements were performed with aZ eiss Gemini 500 instrument from CP2M platform (FR 1339-CNRS-AMU). The voltage was set between 0.5 to 5kVt o have optimum conditions leading to ag ood compromise between the suppression of the charge effect and lateral resolution. The working distance between camera and sample was set at 1.8 mm or 0.9 mm to enhance the resolution Scanning Transmission Electron Microscopy (STEM) and high resolution TEM were used to characterize the Pt-encapsulated samples. The measurements were performed using aJ eol 2100F working at 200 kV and aG atan Digital Micrograph with aD igiscan II system. The grids for STEM measurement were prepared by dispersing a small amount of sample in methanol followed by sonification and decantation. Afterwards, af ew drops were deposited on the dedicated copper grid and left in oven at 363 Kfor 1h our to evaporate the methanol solvent.
Platinum loading in impregnated samples was measured by using Induced Coupled Plasma (ICP) analysis with a7900 ICP-MS (Agilent) equipped with aS PS4 autosampler (Agilent). Thermal stability of the synthesized materials was studied by using thermogravimetric analysis (TGA). The analyses (TGA) were performed with aT AQ 500 instrument using platinum crucibles under synthetic air flow of 100 mL min À1 and another set of experiments performed under nitrogen flow of 100 mL min À1 .T he analyses were initiated by applying an isotherm at room temperature (298 K) for 15 minutes followed by heating at ar ate of 5Kmin À1 up to 973 K. Texture was analysed using nitrogen sorption at 77 K. The measurements were done using aB ELSORP Max 1. The available surface area, pore volume, external surface and pore size distribution were deduced from the measurements. Prior to the measurements, the samples were activated to 373 Ku nder vacuum for 17 hours. Water vapor adsorption at 298 Kw as performed using aB ELSORP Max 1. Adsorption equilibrium was assumed when the variation of the cell pressure was 0.5 %f or am inimum period of 300 seconds.
Prior to the measurements, the samples were activated to 373 K under vacuum for 17 hours. CO 2 adsorption microcalorimetry was carried out at 303 K using a lab-built instrument. [56] AT ian-Calvet type microcalorimeter, equipped with thermopile with around 900 chromel-alumel thermocouples, was used to directly measure the adsorption energies. Manometry is used to measure the adsorption isotherm using a stepwise introduction of the carbon dioxide. In the initial regions of loading, errors in the enthalpies are of the order of AE 1kJmol À1 . In these experiments, at hermal equilibration time was set at 200 minutes between each gas dose. Prior to the measurements, the samples were activated to 373 Ku nder vacuum for 16 hours.