Inside/Outside: Post‐Synthetic Modification of the Zr‐Benzophenonedicarboxylate Metal–Organic Framework

Abstract The Zr‐based metal–organic framework, Zr‐bzpdc‐MOF, contains the photoreactive linker molecule benzophenone‐4,4′‐dicarboxylate (bzpdc) which imparts the possibility for photochemical post‐synthetic modification. Upon irradiation with UV light, the keto group of the benzophenone moiety will react with nearly every C−H bond‐containing molecule. Within this paper, we further explore the photochemical reactivity of the Zr‐bzpdc‐MOF, especially with regard to which restrictions govern internal versus external reactions. We show that apart from reactions with C−H bond‐containing molecules, the MOF reacts also with water. By studying the reactivity versus linear alcohols we find a clear delineation in that shorter alcohol molecules (up to butanol as a borderline case) react with photoexcited keto groups throughout the whole crystals whereas longer ones react only with surface‐standing keto groups. In addition, we show that with the alkanes n‐butane to n‐octane, the reaction is restricted to the outer surface. We hypothesize that the reactivity of the Zr‐bzpdc‐MOF versus different reagents depends on the accessibility of the pore system which in turn depends mainly on the size of the reagents and on their polarity. The possibility to direct the post‐synthetic modification of the Zr‐bzpdc‐MOF (selective modification of the whole pore system versus surface modification) gives additional degrees of freedom in the design of this metal–organic framework for shaping and for applications.


Introduction
Metal-organic frameworks( MOFs) featurea no rdered structure with permanent porosity which is of high interest for av ariety of applications like catalysis, [1] gas storagea nd separation, [2,3] sensing, [4][5][6] energy storageand conversion [7] or biomedicine. [8,9] To fully exploit this potential, the structure andt he properties of aM OF need to be fine-tuned to the specific intended application in order to achieveo ptimum results. MOFs are wellp repared to meet such challenges, as their modular construction from metal oxide nodes ando rganic linker molecules allows the introduction of various properties by deliberately choosing the metal and the linker molecule. However,c ertain specific combinationso fp roperties are difficultt oe laborate due to the rather harsh synthesis conditions employed.E specially,s ome desired functionalities at the organic linker molecules may be too sensitive to withstand the synthesis environment. Therefore, other strategies have been developed to overcome such restrictions imposed by the use of labile or more sophisticated functional groups at the linker molecules during synthesis. One of the most strongly investigateda pproach to modify and functionalise MOFs is the post-synthetic modification (PSM) of the framework. Severalr outes like linker exchange,p ost-synthetic metalation or organic reactions at the linker molecules have been developed. [10][11][12][13] Especially,t he versatility of organic reactions to covalently modify the linker molecules has been widely studieda nd utilised to introduced ifferent functionalities to the framework after the MOF skeleton has been formed. Such PSM reactions are usuallyp erformedo ns pecific functionalities of the organic linker molecules like amino, alkyne, azide or halide groups or by click chemistry.Am ajor requirement to ensure the success of such post-synthetic treatments is the stability of the framework during the PSM. [10,11,[14][15][16] MOFs with Zr IV -based nodes are especially appropriatef or PSM due to their high stability towardst hermal and chemical stress, [17,18] making them ideal platformsf or introducing different functionalities by chemical transformations and modifications with the aim to adjust the properties of the material with as pecific application in mind. [13,16,[19][20][21] Recently,w eh ave introduced an ovel type of PSM reaction. This makes use of the specific reactivity of the benzophenone moiety.U pon excitation by irradiation with UV light, the ketyl radicals formed react with practicallya ny CÀHb ond of an organic molecule according to the reaction mechanism schemat-ically shown in Scheme1. [22,23] We have employed the linker molecule benzophenone-4,4'-dicarboxylic acid( H 2 bzpdc)t o synthesize the Zr-based metal-organic framework Zr-bzpdc-MOF. [24] The synthesis made use of the modulation method [25,26] and lead to ac ompound with formula Zr 6 O 4 (OH) 6 (HCO 2 ) 2 (bzpdc) 4 .S tock and co-workers have used this linker fort he constructiono ft he MOF CAU-8 with the formula Al(OH)(O 2 C-C 6 H 4 -CO-C 6 H 4 -CO 2 )a nd have studied its photochemicalb ehavior. [27,28] The Zr-bzpdc-MOF has at wo-dimensional layered structure, can be delaminatedt ot hin nanosheets and exhibits am oderate porosity of 650 m 2 g À1 .I ts high chemicala nd thermals tabilitym ake it highly suitable for PSM reactions. In ap reliminarys tudy,w eh ave converted this MOF in photochemical PSM reactions with decane and PEG and have observed drastically different wetting properties of the products. [24] Furthermore, we have initiated photochemical grafting-from polymerisation reactions of EDOT (3,4-ethylenedioxythiophene) to yield nanoporous electrically conducting MOF-PEDOT composites. [29] In all cases, the powder X-ray diffraction pattern of the material remained unchanged;a ccordingly,t he modification had only taken place on the surface of the MOF crystals. The central question which then arises is whether the photochemical PSM is in general only possible at the outer surfaceorwhether some characteristics of the organic molecule, like its size or its polarity,d ecide about whether the reaction takes place only at the surface or throughout the Zr-bzpdc-MOF crystal.
In the present study,w es ystemically investigate the photo-chemicalP SM of the Zr-bzpdc-MOF.F irst, we comparet he reactivity and the obtained products of the reactions of the linker molecule H 2 bzpdc,r eactede ither as af ree molecule or when immobilized in the framework. This part focusses on the reactions with simple molecules, namely methanol and ethanola s well as water.O nly few studies have reportedo nt he reactions of benzophenonei na queous media; [30][31][32][33] in the present case, these may be especially important,a si nt he laboratory the MOF samples are exposed to light and air humidity when no special precautions are taken.I nt he second part of the study, we have reacted the Zr-bzpdc-MOF with series of alkanesa nd alcohols of different chain length. We show that the PSM takes place either throughout the MOF crystals or only on the surface, depending on the polarity and on the chain lengths of the probe molecules. These results allow to design PSM strategies of this photoreactive MOF in view of different applications.

Results and Discussion
Synthesis of Zr-bzpdc-MOF The synthesis of Zr-bzpdc-MOF was adaptedf rom our already reported synthesis route [24] and leads to rhombic shaped crystals with edge length of about 80-100 mm, as shown in Figure 1. The materialw as extracted with acetone in aS oxhlet extractorf or 24 hours, dried andk ept under reducedp ressure. Essential properties, like crystallinity and sorption behaviour, are shown in Figure 2a nd Figure 3. With aB ET area of about 680 m 2 g À1 and at otal pore volume of about 0.3 cm 3 g À1 ,t he porosityp arameters calculated from the data shown are in very good agreement to the published values. [24] In order to ensure reproducibility and consistency of our studies, the synthesis approach was scaled up to obtain in one batch as ufficient amount of Zr-bzpdc-MOF for all post-synthetic modification reactions and the application of the variousc haracterisation methods.

Generala spects of the photochemical reactions
For every post-synthetic treatment, the sample was directly dispersed in the neat reactanta nd purged with argon before irradiation. Irradiated samples are designated by "UV", followed by the reactant (and possibly other relevant conditions), throughout the paper.Asuitable characterisation methodt o evaluatet he PSM of the Zr-bzpdc-MOF is solution-state NMR spectroscopy,p erformed on acid-digested samples. Further characterisation methods applied are powder X-ray diffraction and physisorption experiments.
In principle, photoexcited benzophenone can react in different ways, [31] as depicted in Scheme 2. Formation of the benzophenone diradical can be followed by the abstraction of an H atom from as ubstance RH and subsequent addition of Rt o Scheme1.Schematicmechanism for the photoreaction of the keto groupo fb enzophenone units in the framework of Zr-bzpdc-MOF: irradiation leads to the excitationoft he keto group resulting in ab iradicaloid triplet state which then reacts with aC ÀHb ond-containing molecule, resulting in the formation of a CÀCbond and the reduction of the keto group. [23]   the former keto carbon atom (path A). However,aradical electron of the photoexcited benzophenone can also becomed elocalized over an eighbouring aromatic ring, as exemplarily shown in path Bf or the reaction with water (see below) which leads to ah ydroxylation of the benzene ring (path B). [30,34] Both path Aa nd path Bc ould also occurw ith benzophenone moieties which are part of the framework. Another often observed reactiono fp hotoexcited benzophenones is the formation of a benzopinacolv ia dimerisation (path C), typical for example for the reaction with 2-propanol;asimilar reactions hould be possible with methanol. [35] However,t his does not appear to be possible in the MOF structure,d ue to the large distances between the keto groups (smallest distance of about4 .8 )a nd the geometric constraints of the framework.

Photochemical reactions in water
We first studied photochemicalr eactions in water.T he reactivity towards water is of specialc oncern,b ecause as part of the MOF framework, all the benzophenonem oieties are exposed, and even air humiditym ight lead to ar eaction under light exposure.A ctually,t here are only very few studies in the literature for the reactiono fb enzophenone moieties in aqueous media. [30][31][32][33] For the studies of the photochemical reactivity towards water,w eu sed the Zr-bzpdc-MOF and, for comparison, the free linker molecule H 2 bzpdc to obtain fundamental insights into possible differences in photoreactivity.
We investigated the photochemical reactivity of Zr-bzpdc-MOF in water at ap Ho f7and for comparison, of the free linker moleculeH 2 bzpdc in water of pH 7a nd pH 10. The pH valuesw erea djusted with sodium hydroxide and ap Hm eter.  The MOF was of course present as ad ispersion;t he free diacid formed ad ispersion at pH 7a nd was dissolved at the higher pH value. The mosto bvious effect of the photochemical reaction is ac olour change from colourless to dark yellow upon irradiation.T his was observed for irradiated Zr-bzpdc-MOF and for H 2 bzpdc under basic conditions. For H 2 bzpdc irradiated in water at an eutral pH value, no colour change could be observed which can be ascribed to the insolubility of the free acid and the inaccessibilityo ft he molecules.
More detailed resultsc oncerning the photochemical reaction can be obtained by NMR spectroscopici nvestigations on solutions. For this purpose, the MOF samples were digested in [D 6 ]DMSO with hydrofluorica cid whereas the irradiated benzophenoned icarboxylic acid samples were dissolved in [D 6 ]DMSO.T he crucial regions of the NMR spectra of irradiated samples of H 2 bzpdc and Zr-bzpdc-MOF are shown in Figure 2 (full spectra are shown in the Supporting Information, Section 1.1, Figures S1-S10).
PXRD patterns of the pristineZ r-bzpdc-MOF and the irradiated sample in water ( Figure 2) show ac omparable crystallinity and there are no indications for major changes in the crystal structure. 1 HNMR spectra in Figure 2s howt he characteristic multiplet of the aromatic hydrogen atoms of the benzophenone moieties. The additional signal at about 8.11ppm for dissolved Zrbzpdc-MOF samples belongs to the formic acidm olecules which are an integral part of the MOF structure.
After irradiationo fH 2 bzpdc under basic conditions andZ rbzpdc-MOF in water at pH 7a na dditional protons ignal appears at 8.03-8.04 ppm ( Figure 2) in the 1 HNMR spectra, indicating the formation of an o-hydroxybenzophenone moiety. That the reactiono ccurs according to path Ba nd not to path A( see Scheme 2) is supported by the 13 CNMR spectra, where the signal at about 195 ppm, characteristico ft he keto carbon atoms, remains mainlyu nchanged. The differences and the additional signals in the area of 160-170 ppm could be assigned to the aromatic carbon atoms that are affectedb yt he hydroxylation. Based on the 1 HNMR spectra, ac omparison of the intensities of the additional signals to the signals of those of the other aromatic protons allows the estimation that about 70 %o ft he linker molecules have become hydroxylatedi n case of H 2 bzpdc and about 50 %i nc ase of Zr-bzpdc-MOF,r espectively.F or the MOF samples, physisorption measurements were performed ( Figure 3).
The photochemical PSM affects the pore system of the MOF only to am inor degree, with only slightly decreased values for the BET area and pore volume( see Supporting Information, Section1.2).
With the proposed reaction according to path Bi n Scheme 2, where hydroxylation occurs at the phenylene moieties and not at the keto carbon atom, the hybridisation at the central Ca tom remains sp 2 and, correspondingly,t he general geometry of the linker does not change significantly.T hus, the general structureo ft he framework is not affected by the PSM reaction, whiche xplainst hat only minor changes occur in the powderX RD pattern. Also, the addition of small OH groups does not significantly affect the pore system,s ot hat also the physisorption properties after the reaction with water are similar.
The PSM of Zr-bzpdc-MOF with water leads to ap ost-synthetic modification of the linker molecules while preserving its crystallinity and its porosity.I ti sa pparently as uitable and very simple method to introduce hydroxy groupst ot he framework which should greatly increaset he hydrophilicity with regard to the parent Zr-bzpdc-MOF.

Photochemical reactions in methanol and ethanol
The photochemical potentialo fb enzophenoneg roups to bind to molecules containing CÀHb onds has been well knownf or many decades. [23,34,[36][37][38] The Zr-bzpdc-MOF possesses high chemicala nd good thermal stability, basic prerequisites for successful post-synthetic reactions. According to its crystal structure, all benzophenonem oieties present throughout the framework are accessible via the pore system;i na ddition, the rhombic-shaped crystals present benzophenoneg roups on their predominant basal surfaces. [24] The Zr-bzpdc-MOF exhibits moderate permanent porosity consisting of corrugated channels with am inimum free pore diameter of 6.5 (visualized in the Supporting Information, Section 1.3, Figure S13).
In initial studies, we had shown the general possibility to perform PSM reactions on Zr-bzpdc-MOF crystals. As reactants we had used polyethylene glycol and decane (and could in this way change the wetting behaviour of the MOF crystals) [24] or EDOT (and could in this way prepare electrically conducting composites). [29] However,int hese reactions only the benzophenone groups on the outer surfaceh ad reacted. Clearly, the reactants mentioned cannotenter the pore system and therefore could not approachthe inner keto groups.
Therefore, we studied in depth the corresponding reactions with smaller CÀH-bond containing molecules,n amely methanol and ethanol, again comparing the reactivity of the Zrbzpdc-MOF and the free acid H 2 bzpdc. For this purpose, the materials were dispersed in the corresponding alcohol andirradiated under argon atmosphere for 120 hours. H 2 bzpdc itself is only slightly soluble in methanol and ethanol. However,d uring irradiation the free acid dissolved completely in methanol but still not in ethanol. Methanola nd ethanol were removed under reduced pressure at room temperature. Afterwards, the samples were characterised via powder X-ray diffraction and NMR spectroscopy ( Figure 4).
Considering the reaction of the free acid in methanol, the 13 CNMR spectrum of the product (Figure 4t op) shows that the signalo ft he keto carbon atom signal at about 195 ppm is missing, indicating as uccessful andc omplete reaction of methanolw ithH 2 bzpdc. Ad etailed inspection of the 1 HNMR spectra (see SupportingI nformation, Section 1.4, FigureS14) does not allow to assign the signals present to one specific product.
On the contrary,t he irradiation of the free acid in ethanol does not lead to ar eactiona sa ll the 13 CNMR signals of the originalm olecule are still present (see Supporting Information, Section1 .4, Figure S15). This can be explained by the insolubility of H 2 bzpdc in ethanols ot hat there is no access to the keto The Zr-bzpdc-MOF shows as uccessful and complete reaction after irradiation with both alcohols. The 13 CNMR spectra in Figure 4p roof ap ractically complete conversion of the keto groups of the MOF after irradiation by the absence of the keto carbon signal in the corresponding 13 CNMR spectra at about 195 ppm (Figure 4t op). This points to ac onversion according to path A. The correspondingr eactions are shown in Scheme3.N ote that the reaction with ethanol can lead to two different products depending on which of the two carbon atoms of the ethanol molecule becomes attached to the former keto carbon atom. In addition, the signal of the carbon atoms of the benzene rings in alpha positiont ot he keto group experience an upfield shift from 140 ppm to about 150 ppm;t his is as trong hint of the reduction of the keto group to an alcohol duringt he reaction with methanol or ethanol and of ac oncomitantc hange of the hybridisation of the keto carbon atom from sp 2 to sp 3 .A lso, the signals for the aliphatic carbon atoms of methanola nd ethanol, respectively, can be observed in the corresponding 13 CNMR spectra at 70-80 ppm (see SupportingI nformation, Section 1.4, Figure S16 -19). Notwithstanding, as in the case of the reactionp roduct of the free acid with methanol, the signals presentc annot be assignedt oas pecificp roduct.
Powder X-ray diffraction patterns of the reactedM OFs are showni nF igure 4( bottom). These indicate as tructural change of the framework. This is in contrast to the PXRD patterns obtained from samples which had been reacted with water (see above)and with larger molecules like decane and polyethylene glycol [24] as well as with 3,4-ethylenedioxythiophene. [29] In these cases, no differences in the PXRD pattern were observed with regard to the parent Zr-bzpdc-MOF which is taken as evidence that only surface-standing keto groups had participated in the photochemical reaction. The changes in the PXRD patterns of the samples which were reacted with methanolorethanol can be rationalised when it is assumed that the photochemicalr eaction has taken place throughout the MOF crystals. In the course of such reactions, (at least the major part of) the keto carbon atoms changet heir hybridisation from sp 2 to sp 3 ,l eading to af ramework with the same linkages and topology as the parent framework, but with different dimensions due to the change in linker geometry.H owever,d ue to the different possibilities for the reaction, where addition of the hydroxyalkyl group can occura td ifferent sites of the benzophenone moiety and-in the case of ethanol-at differentc arbon atoms of the alcohol, thesen ovel MOFs will not have au niform structure which could be described by as ingleu nit cell (see below). These results lead to the hypothesis that the molecule size is am ajor factor in determining whether the post-synthetic addition reactiont akes place only at the outer surface of or throughout the crystals. Therefore, we carried out photochemical PSM reactions with primary linear alcohols of chain length C 1 to C 8 (methanol to octanol), including methanoland ethanol which were studied here in depth.A long with the size of the molecules, their polarity and hydrophilicity also change. Therefore, in order to also test the possible influence of polarityo f the reactant, reactions with linear alkanes (butane to octane) were included in thes tudy.

Photochemical reactionsw ith alcoholsand alkanes of different chain lengths
To test whether the size and/ort he polarityo fr eactant molecules is am ajor factor in determining the extent of the postsynthetic addition (surface only/pervasive), we carriedo ut photochemical PSM reactions of the Zr-bzpdc-MOF with linear alcohols of chain length C 1 to C 8 (methanolt oo ctanol) as well as with linear alkanes (butane to octane). In the first part, we focus on the reactions with primary alcohols, thereby including the results described for the PSM reactions with methanol and ethanol in the preceding section.
To investigate whether the PSM reaction with ac ertaina lcohol was successful, the samples were characterised via powder X-ray diffraction and NMR spectra of the acid-digested reaction products.P XRD patterns are shown in Figure 5( bottom) and selected ranges of the obtained 13 CNMR spectra are displayed in Figure 5( top) (full NMR spectra are providedi nt he Supporting Information, Section 1.4, Figure S14-S28). The intensity of the signal of the keto carbon atom is most informative.
After the reaction carriedo ut in 1-propanol, no keto carbon signal is observed anymore;i nstead, signals for aliphatic carbon atomsa ta round 70-80 ppm are found. This result is comparable to the situation with methanol and ethanol. On the contrary,i nt he case of the reaction products obtained with 1-pentanola nd also 1-octanol, the keto carbon signal is still present,i ndicating that the keto groups have not reacted or have done so only to av ery small extent. After the irradiation in 1-butanol, the keto carbon signal in the 13 CNMR spectrum of the reactionp roduct is very small, indicatingt hat the reactionw ith this alcoholw as also nearly complete. In fact, when the Zr-bzpdc-MOF is stirred for 24 hours in butanol before irradiation, the following photochemical reactioni s complete, as judged from the 13 CNMR spectrum of the acid-digested product (see Supporting Information Figure S24) where no keto carbon signal can be observed anymore. Therefore, it appearst hat in the reaction with butanol the diffusiono ft he reactantm olecule through the narrowa nd corrugated channels of the host structure is an additional factor influencing the extento ft he reaction. Thus, the PSM with 1-butanol appears to be ab orderline case:I rradiation in smaller alcohols lead to af ull reactiono ft he keto groupsw hereasl arger alcohol molecules probably only react at the keto groups located at the outer surface of the MOF crystals.
This delineationa lso becomes apparent in the powder X-ray diffraction patterns ( Figure 5, bottom). The PXRD patterns of the samples that were irradiated with long-chain alcohols (pentanol to octanol) show no changes in comparison to that measured on the pristineZ r-bzpdc-MOF.I nc ontrast, for those samples which werep ost-synthetically modified with small alcohols (methanolt ob utanol), changes can be observed:t he first reflectionsa re shifted to slightly higher 2V values, the reflection pattern at larger 2V values changes, and in general a minor decrease in crystallinity is indicated by weaker and broader reflectionsint his range.
The shifto ft he two first, most intense reflections( 110a nd 200) to higher 2V values can be rationalised by the changes in the linker geometry. Each linker molecule connects twoI BUs in the structure of the Zr-bzpdc-MOF.I nc ourse of the photo-chemicalP SM reaction, the hybridization of the initial keto carbon atom changes from sp 2 to sp 3 by the photochemical reaction ( Figure 6a). Correspondingly, the bond angles at this carbon atom change from 1208 to 109.58 which in turn should result in as hrinkage of the distance betweent wo IBUs on certain lattice planes, provided that the topology of the framework remains the same and that every keto group (or most of them) is affected (Figure6a). Concomitantly,c hanges occur in the unit cell dimensions (see Supporting Information, Section 1.6, Ta ble S2). This is shown in Figure 6a-c for the product of the reaction with methanol.N ote that this is the only case where ah omogenous structure of the product can be expected (when all The a and especially the b axis shrink (Figure 6b), explaining the shift of the first two reflectionst oh igher 2V values (Figure 5b ottom). The hydroxymethylener esidues introduced at the former keto carbon atom through the PSM require additional space between the layers of the structure, thereby leading to an expansion of the structure along the c axis (Figure 6c). On the contrary,t he bulk structure is not affected when the reactiono ccurs only at the surface, as illustrated in Figure 6d.
The irradiated samples were further investigated by physisorptionm easurements with nitrogen at 77 Ka nd carbon dioxide at 273 K. Ac ompilation of the resulting isotherms of the N 2 @77 Km easurements is shown in Figure 7( as many of the isotherms closely overlap, the individuals orption curves are showni nt he SupportingI nformation, Section1 .2, Figure S11; valuesf or BET area and pore volume derived from the nitrogen sorptioni sothermsa re given in the Supporting Information, Section1.2, Table S1). In comparison to the pristine Zr-bzpdc-MOF with aB ET area of 680 m 2 g À1 ,t he samples modified with long-chain alcohols (1-pentanolt o1 -octanol)o nly show about half of this value (ca. 350 m 2 g À1 ). Similar values for the nitrogen BET area were observed in our former studies after the irradiation in PEG (250 m 2 g À1 ), [24] decane (350 m 2 g À1 ) [24] and EDOT (520-630m 2 g À1 ). [29] The decreased amountso fg as volumea dsorbed by these samples indicateapartial pore blockingo ft he pore systemsd ue to PSM occurring only at the outer surface with long chain alcohols. When the PSM occurs not only at the surface, but throughout the crystals, as is the case with the alcohols methanol up to butanol,t he porosity is very low.F rom the flat N 2 isotherms depicted in Figure 7, BET areas of about 20 m 2 g À1 are determined, that is, nitrogen moleculesc annotenter anymore the former pore system.
Whereas the nitrogen physisorption isotherms of the samples treated post-synthetically with methanol, ethanol, propanol and butanol thus indicate the formation of essentially nonporousp roducts, this statement is negated by physisorption Small alcohol molecules (methanol is shown here as example) can diffuse through the whole pore system and the PSM reaction can occurw ith every photoexcited benzophenonegroupl eadingtoc onsiderable structural changes:a )atthe keto carbon atom,t he hybridisationc hanges from sp 2 to sp 3 ;b )the change in bond angles leads to as hrinkageoft he unit cell in the a-b plane;c )the additional space requiredbyt he hydroxymethylene residues leads to an elongationa long the caxis. d) Long-chaina lcohols (octanol is shown here as an illustrative example) cannot penetrate deeply into the framework;only benzophenonemoieties at or neart othe surface react,t he structure and the unit cell within the interior of the crystals is not affected. experiments with carbon dioxide at 273 K. Because of the high temperature and the high relative pressures( p/p 0 % 3 10 À2 ) for this experiment, diffusion is much faster and therefore it is possible to resolve ultra-small micropores (below 0.7 nm). Althought he general trend in adsorbed volume of carbon dioxide is similara sf or nitrogen (Figure 7b ottom), the samples treated post-synthetically with short-chain alcohols stille xhibit am oderate free pore volume which is not measurable with nitrogen at 77 K, possibly due to diffusionh indrance.B ased on these results, the pores of the samples should be below ca. 4 diameter.W hen, with alcohols with more than four carbon atoms the PSM occurs only at or near the surface, al arger accessiblep orosity results. Like in the case of nitrogen sorption, this porosity is considerably smaller than for the pristine Zrbzpdc-MOF.
Ad ependence of the amenability of the Zr-bzpdc-MOF to the photochemical modification on the chain lengths of the alcohols is thus clearly apparent. Although judging from their largestd imension (diameter of the methylene chain of the alcohols:4 .8 ), all alcohol molecules should be able to enter the pore system of the Zr-bzpdc-MOF (minimum free pore diameter: 6.5 ), the corrugated channel structure could strongly impede the diffusion of longer-chain alcohols (C 5 to C 8 )i nto the interioro ft he crystal whereas small-chaina lcohols (C 1 to C 4 )c an diffuse through the whole pore system (although the diffusion of butanol takes considerable time, see above). Another possibly important aspect is that with the chain length of the alcohols, their hydrophobic characteri ncreases, which could lead to ad ecreased diffusioni nto the pore structure of the rather hydrophilic Zr-bzpdc-MOF. [24] Therefore, we furthermore tested the PSM reaction with more hydrophobic alkanes, startingw ith butane and increasing the chain lengthu pt o octane.F or butane, the reaction was carriedo ut in the liquid phase in aclosed ampoule;reactions with even smaller alkanes were precluded due to the high vapour pressures of methane, ethane and propane at room temperature. When we acid-digested the samples to measure NMR spectra, insoluble residues were still present after the treatment. Although we cannote xclude that part of the organic substances liberated by the digestion of the MOF may have become adsorbed on the solid, we have measured NMR spectra of the supernatants. The results obtained are in line with reactions which take place (nearly)e xclusivelya tt he surface (see Supporting Information, Section1.5, Figure S29 and S30). The PXRDs pattern of the samples after the PSM reactiona re compared to that of Zrbzpdc-MOF in Figure8.
In case of the samples post-synthetically modified with alkanes, no changes can be observed in the X-ray diffraction patterns. Even for butane, the reflection positions and intensities are comparable to those of Zr-bzpdc-MOF.T he PXRD results are corroborated by physisorptionm easurements, carried out with nitrogen at 77 Ka nd carbon dioxide at 273 K, shown in Figure 9( as many of the isotherms closely overlap, the individual sorption curves are shown in the Supporting Information, Section1 .2, Figure S12;v alues for BET area and pore volume derived from the nitrogen sorption isotherms are given in the Supporting Information, Section1.2, Ta bleS1).
Nitrogen physisorption isotherms show ac onsiderable decrease in adsorbed volumef or all post-synthetically modified samples. The pore volumes and BET areas are in the range of about 350 m 2 g À1 ,c omparable to the values obtained on samples after reaction with long chain alcohols. As imilarc onclusion can also be drawn from the carbon dioxide isotherms.
The results showt hat also with butane-and in contrast to butanol-the PSM occurs only at the outer surface of the crystals, that is, no complete pervasion of the crystalso ccurs. This is ah int that in addition to molecular size also the polarity of the molecule may be importantf or the penetration of mole-  cules into the framework and, accordingly,f or the extent to which PSM occurs.

Conclusions
The possibility for photochemical post-syntheticm odification of the recently introduced Zr-bzpdc-MOF is as pecial characteristic of this porousc oordination polymer,a si tc an be carried out directly after synthesis and work-up and does not need any further preparations. We have already shown that the photochemical reactivity of the benzophenone linker can be employed to adaptt he surface properties of the MOF crystals [24] and to generate interesting composite materials by direct grafting-from polymerization reactions. [29] Therefore, it is worthwhile to furthero ur understanding of this reaction. We do this here by elucidating the question with which reactants ac omplete PSM extending throughout the crystal occurs and with which molecules the reactioni sr estricted to the surface. In this respect, we have shown that with smalla lcohol molecules (C 1 to C 4 alcohols), the PSM pervades the whole crystals whereas alcohol molecules with longer aliphatic chains (C 5 to C 8 alcohols) react only at the outer surface. Interestingly,i nt he borderline case n-butanol, diffusion hindrance is possibly important.T he polarity of the molecule probablya lso plays a role, as with the C 4 alkane butane, the reaction occurs only at the outer surface.
Retrospectively,i ti so fi nterest to note the differences between the products of the PSM obtained after irradiation in PEG and decane which we had described in our original paper on the Zr-bzpdc-MOF. [24] According to PXRD, the reactions only occurred at the outer surface. After the PSMi nd ecane, nitrogen physisorption measurements revealaB ET area of 350 m 2 g À1 , [24] agreeing very well the values obtained after PSM in C 5 to C 8 alcohols and C 4 to C 8 alkanes. WithP EG, however,a value of only 250 m 2 g À1 is obtained. Thisi ndicates that the very hydrophilic PEG molecules might have enteredt he pore system to some degree before they were restrained from further diffusion by their polymeric nature, and thus substantiates the idea that polarity of the reagent is an important factor in the selection between reactants which can react throughout the pore system and those whichonly react at the surface.
The concept of selectivef unctionalisation of aM OF-either only at the surface or throughout the whole crystals-is of general interest for various applications like gas separation, optics, electronics or for the generation of composite materials. In case of the Zr-bzpdc-MOF,t his can be accomplished photochemically directly after synthesis and work-up. The selectivity of the functionalisation is closely coupled to the crystal structure of this MOF,n amely on the one hand the exposure of reactive benzophenone moieties at the surface, on the other hand the peculiarities of the pore system,n amely rather small apertures and corrugated channels as well as high hydrophilicity.I na nother bzpdc-based MOF,n amely CAU-8 (Al-bzpdc-MOF), photochemical PSM reactions are also possible, but these always occur throughout the pore system. [27,28] In this respect, the synthesis of bzpdc-based MOFs with other metalsa ppears interesting, perhaps offeringf urthers electivities with which the attractive reactivity of the benzophenone-containingl inker can be exploited.

Experimental Section
Materials:Z irconium(IV)-oxychloride octahydrate, formic acid, N,Ndimethylformamide (DMF), n-butane were purchased from Linde Group, pentane, hexane, heptane, octane, methanol, ethanol, 1-butanol, 1-pentanol, 1-hexanol, 1-heptanol, 1-octanol were purchased from Merck Munich and benzophenone-4,4'-dicarboxylic acid from abcr chemicals Karlsruhe. All chemicals were used without further purification. Instrumentation:P owder X-ray diffraction (PXRD) measurements were performed at room temperature using aS TOE STADI-P transmission diffractometer equipped with curved Ge(111)m onochromator with Cu Ka1 radiation (l = 1.540594 )a nd al inear positionsensitive detector.T he samples were fixed between X-ray amorphous foils. Scanning electron microscopy was carried out with a JEOL JSM-6700F (2 keV). The samples were prepared by the use of ethanolic dispersions dried on ag raphite pad under reduced pressure. The resulting images were edited with the software ImageJ 1.49v.L iquid-phase 1 Ha nd 13 CNMR spectra were measured on a Bruker instrument at 400 MHz and were analysed with ACD NMR Processor 12. Sorption isotherms were measured at an Autosorb 1 (Ar) and Autosorb 3( N 2 ,C O 2 )i nstrument from Quantachrome and were evaluated with the software ASiQwin 2.0 (Quantachrome). Measurements were performed with nitrogen at 77 Ka nd with carbon dioxide at 273 K. From nitrogen sorption isotherms, BET areas of the microporous samples were determined with the micropore BET assistant of the accompanying software and total pore volumes were determined at ar elative pressure value of about 0.95. UV irradiation experiments were performed with ah igh power UV LED Spot epiSPOT by Laser 2000 with aw avelength of 365 nm and 176 mW optical power. Synthesis of (Zr 6 O 4 (OH) 6 (HCO 2 ) 2 (bzpdc) 4 ):F or reproducibility and consistency,a ll post-synthetic modification reactions were carried out on Zr-bzpdc-MOF single crystals synthesized in ao ne-batchserves-all approach by ar oute slightly modified from the already published recipe. [24] Solvothermal synthesis of Zr-bzpdc-MOF was carried out in Te flon-sealed screw cap glass vessels using am odulation approach. [25,39] Favourable conditions to obtain single crystal samples without impurities are as follows:3 .22 g( 10 mmol) ZrOCl 2 ·8 H 2 Ow ere dissolved in 200 mL DMF.A fter adding 56.59 mL formic acid (150 molar equivalents (eq) based on amount of Zr 4 + ) and 5.40 g( 20 mmol) H 2 bzpdc the clear solution was transferred in a5 00 mL Teflon-sealed glass vessel and was kept at 120 8Cf or 2weeks in ac irculating air oven. After the reaction, the solution was cooled down to room temperature and the resulting solid was centrifuged and washed once with 100 mL of DMF and two times with 50 mL acetone and then dried under reduced pressure. For further investigations the sample was activated via Soxhlet extraction with acetone for 24 hours in the dark. The white powder was kept under reduced pressure until further usage.
Post-synthetic modification:T he photochemical reactions were performed using aU V-LED with aw avelength of 365 nm and ar adiant flux of about 100 mW cm À2 .AMOF sample of about 100 mg was placed in aq uartz cuvette and ca. 3mLo ft he neat reactant were added. The reaction batch was flushed for 30 minutes with and kept under argon while irradiating the samples under stirring. The irradiation time was set to 120 hours for alkanes and alcohols. For investigations in aqueous media the irradiation times were 72 hours. After irradiation, the samples were extracted for 24 hours with methanol (Soxhlet extractor) and dried under reduced pressure. For the post-synthetic modification with butane, the MOF was sealed with 5mLl iquid butane in aq uartz vial with as tir bar. After irradiation for 120 hours under stirring, unbound molecules were removed under reduced pressure. This sample was used for further investigations without Soxhlet extraction.
Preparation for solution-phase NMR spectroscopy:F or liquidstate NMR experiments, 50 mg of the MOF were dispersed in [D 6 ]DMSO. Under vigorous stirring 20 mLa queous HF (40 %) were added and stirred for 18 hours. After complete dissolution of the Zr-bzpdc-MOF,a ne xcess of CaCl 2 was added to the clear solution.
The supernatant was used for 1 Ha nd 13 CB BNMR investigations. The spectra were analyzed with ACD/NMR Processor Academic Edition v12.01.
Structural modelling:T oi nvestigate structural changes of the MOF by the PSM with methanol, the program suit Materials Studio from Dassault Systme BIOVIA was used. First, the crystal structure of pristine Zr-bzpdc-MOF was subjected to an energy minimisation employing the UFF (Universal Force Field) within the Forcite program. [40][41][42] The minimisation employed the Ewald summation method for electrostatic interactions. For this kind of interaction as well as for the atom-based van der Waals interactions the cut off was set to 15 .I nt he first step of the minimisation procedure, only atom coordinates were relaxed while the cell parameters were retained. In the second step, both the unit cell parameters and the atoms were allowed to relax. In order to find the global minimum, this structure was employed for aQ uenched Dynamics procedure leading to 10 000 energy-minimised structures. The parameters are given in Ta ble 1.
Our model for the sample after PSM with methanol is based on the reaction according to path A. The sp 2 keto carbon atoms of the benzophenone units were transformed to sp 3 Ca toms. Instead of the double-bonded oxygen atoms, hydroxy groups and hydroxymethylene groups were placed at these carbon atoms. By doing so, this carbon atom is connected to the corresponding alkyl residue. This structural model then underwent similar procedures as described above for the pristine Zr-bzpdc-MOF.