Influence of Pore Structure and Metal‐Node Geometry on the Polymerization of Ethylene over Cr‐Based Metal–Organic Frameworks

Abstract Metal–organic frameworks (MOFs) have received increasing interest as solid single‐site catalysts, owing to their tunable pore architecture and metal node geometry. The ability to exploit these modulators makes them prominent candidates for producing polyethylene (PE) materials with narrow dispersity index (Ð) values. Here a study is presented in which the ethylene polymerization properties, with Et2AlCl as activator, of three renowned Cr‐based MOFs, MIL‐101(Cr)‐NDC (NDC=2,6‐dicarboxynapthalene), MIL‐53(Cr) and HKUST‐1(Cr), are systematically investigated. Ethylene polymerization reactions revealed varying catalytic activities, with MIL‐101(Cr)‐NDC and MIL‐53(Cr) being significantly more active than HKUST‐1(Cr). Analysis of the PE products revealed large Ð values, demonstrating that polymerization occurs over a multitude of active Cr centers rather than a singular type of Cr site. Spectroscopic experiments, in the form of powder X‐ray diffraction (pXRD), UV/Vis‐NIR diffuse reflectance spectroscopy (DRS) and CO probe molecule Fourier transform infrared (FTIR) spectroscopy corroborated these findings, indicating that indeed for each MOF unique active sites are generated, however without alteration of the original oxidation state. Furthermore, the pXRD experiments indicated that one major prerequisite for catalytic activity was the degree of MOF activation by the Et2AlCl co‐catalyst, with the more active materials portraying a larger degree of activation.


Introduction
The production processes of plastics,o fw hichp olyethylene has the largest market volume share, stands as one of the most mature, sustainable,a nd efficient technologiesr elying on fossil and, more recently,r enewable feedstocks. Despite increaseda ttentiont ot he environmental issuest hat plastics cause, the broad range of applicationse nsures that the productiono fP Ew ill continue to growi nt he coming years. [1] These factors continue to drive research towards improved production of PE, as well as towards finding new types of PE.
Metal-organic frameworks( MOFs) are defined by ah igh degreeo fo rdering in terms of pore size and pore structure, as well as the possibility to exploit aw ide variety of metal-sites, tailoringt heir activity by rational design. [12][13][14][15][16][17] This has been elegantly exemplified by the linear correlation between catalytic CO 2 photoreduction, as am odel reaction, and the electronic structure of the linker,s howing that functional groups on the organic linkerd irectly affect the reactivity. [15][16][17][18][19][20] This method is comparable to conventional strategies for heterogenous catalyst design,s uch as using dopants (promoters) or tuning the support, [3,[21][22][23][24][25][26] but with the added value of as traightforward characterization and understanding of the material. The ability to change the pore structure,m etal-node and electronic structure with relative ease, make MOFs increasingly prominent candidates forh eterogenized single-site catalysis. [27][28][29] Until now,t hey have been used as such in numerous reactions, for example, others alkene oligomerization reactions [30][31][32][33][34][35][36][37][38][39] as well as polymerization reactions. [30,37,[40][41][42][43][44][45][46][47][48] Dincǎ,R oman-Leshkova nd co-workers investigatedt he ethylene polymerizationp roperties of MFU-41 (MFU = metal-organic framework Ulm-University) MOFs. [41,45] The gas-phase eth-ylene polymerization properties of Cr 3 + ion-exchanged MFU-4l MOF material were investigated, revealing ap olymerization MOF that produced PE with ad ispersity index ()o fa bout 1.4, the being defined as the M w /M n ,w ith the M w representing the weight averagedm olecular weighto ft he polymer and the M n representingt he number averaged molecular weight of the polymer.T his indicates that active sites are similar in nature, and is an excellent example of aC r-based MOF acting as ah eterogenized single-site ethylene polymerization catalyst. However,i nt he case of MOFs, forming true single-site catalysts depends strongly on the materials tructure and its evolution upon reaction with co-catalyst species; and it merits detailed studies on ap er case basis. Previous work from our group compared the polymerization properties of MIL-101(Cr) vs. MIL-100(Cr) and revealed that the type of linker is of paramounti mportance for the ability,o ri nability,o ft he MOF to fragment. Here, the ability of the MOF (MIL-101(Cr)) to fragment was correlated to the ability to polymerize ethylene, whereas the inability to fragment (MIL-100(Cr)) was related to predominantly a-oligomerization and negligible polymerization activities. [35,43] Despite all these advantages and already numerous applications, one should be cautious in using Cr-based catalysts due to their toxicity as well as related environmental hazards. Fortunately,t his is of no real concern in olefin polymerization, since state of the art catalysts portray such high activities that noncorrosive andn ontoxic catalyst residuesc an be safely left in the PE product material in sub-ppb levels. [49,50] In view of an increasing interestt owards the application of MOFs in ethylene polymerization, we opted to investigate several well renowned Cr-based MOFs in this reaction, their structures demonstratedi nF igure 1, specifically:M IL-101(Cr)-NDC (NDC = 2,6-dicarboxynapthalene), [17,[51][52][53][54] MIL-53(Cr) [55][56][57] and HKUST-1(Cr). [58] In order to establish structure-activity relationships, this selectiono fM OFs comprises av ariety of pore sizes, 3D-structures and metal-node geometries.W ei nvestigated the fate of these materials when reacting with Et 2 AlCl to activate the MOF towards ethylenep olymerization in the slurry-phase. Twod ifferent hydrocarbon solvents, heptane and toluene, were studied as reaction media. The molecular architecture and morphology of the resulting polymers were, respectively investigated with gel permeation chromatography (GPC), differentials canning calorimetry (DSC), and with scanning electron microscopy (SEM). Moreover,w es ought to understand the underlying mechanisms of how Et 2 AlCl activates the MOF in terms of crystallinity,surfacearea, Cr oxidation state and surface site accessibility,b ym eans of ab road array of tools including X-ray Diffraction (XRD), UV/Vis-NIR diffuse reflectance spectroscopy (DRS),C Op robe molecule Fourier transform infrared (FTIR),N 2 physisorption, diffuse reflectance infrared spectroscopy (DRIFTS) and scanning electron microscopy (SEM).

Ethylenep olymerization reactions
Ta ble 1s hows the activity in ethylene polymerizationo ft he different MOFs in both heptane and toluene as diluents. It is well known that the selected solventc an detrimentallya ffect ethylene polymerization/oligomerization properties, either due to coordination to the actives ite or decomposition of the active site. [59,60] All three MOF structuress howedh igherp roductivity in heptane, likelydue to the absence of potentialCr 3 + -p ring interactions which mayb ep resent when toluene is used. [59][60][61] The catalytic activities in Ta ble 1c onfirmt hat selection of the properM OF precursor is of paramount importance and the topology plays ad ominant role in polymerization activity.M IL-53(Cr) is the most active MOF,w ith an activity of 4.01 kg PE mol Cr À1 h À1 bar À1 ,f ollowed by MIL-101(Cr)-NDCw ith an activity of 1.24 kg PE mol Cr À1 h À1 bar À1 ,a nd HKUST-1(Cr), exhibiting ap oor activity of 0.22 kg PE mol Cr À1 h À1 bar À1 .T able 1a lso  clearly demonstrates the importance of selecting the proper reactionm edium, sinceM IL-53(Cr) loses the majority of its activity upon switchingt ot oluene. Similarly,H KUST-1(Cr) lost about 50 %o fi ts original activity and only MIL-101(Cr)-NDC appeared to be relatively unaffected by the change in reaction medium, retaining about 75 %o fi ts original activity. As already mentioned in the introduction:T he current state of the art polyolefin catalysts portrays such activities that the catalystr esidues can be harmlessly left in the PE material. The MOF residuep ercentages in the final PE products vary from 24 wt %f or HKUST-1(Cr) to 0.4 wt %f or MIL-53(Cr) in the here discussed polymerization reactions, which is not (yet) ideal. However,w ewere limited in terms of catalysty ields by the reaction vessel size. By simply increasingt he reactor volumea nd increasing the reactiont ime, it is possible to increase the PE yield and lower relatedM OF contribution to the final product composition.
As shown in our previous efforts, Cr may partly leach from the MOFs into the solution. [43] Hence, to investigatet he contribution of such speciest op olymer formation,t he reactions were also performed after filtration of the MOFs after reaction with the co-catalyst, andl abeled "filtrate reactions". We found that for all the MOF topologies under study,r esiduala ctivity was obtainedf rom the liquid,i ndeed confirming that ac ertain fraction of polymer mayb ep roduced by species in solution instead of Cr sites on/in the MOF lattice. In the case of MIL-101(Cr)-NDC, an activity of 0.08 kg PE mol Cr À1 h À1 bar À1 is obtained, corresponding to 6.6 %o fi ts original activity,w hile the filtrates of MIL-53(Cr) and HKUST-(Cr) demonstrated higher activities with values of, respectively 29.7 %a nd 40.5 %.
Inductively coupled plasma-atomic emission spectroscopy (ICP-OES) was used to quantify the amount of Cr leachedf rom the framework. These measurements revealed that for MIL-101(Cr)-NDC, MIL-53(Cr) and HKUST-1(Cr) respectively 1.02 %, 4.96 %a nd 0.71 %o ft he originalC rc ontent leached into solution. However,t hese Cr species were responsible for as ignificant fraction of the polymer produced (6.5, 29.7 and4 0.5 %, respectively). This shows that although the MOF materials are relativelys table (< 5% Cr leached) towards disintegration by the co-catalyst, the homogeneous compounds produced by the pre-treatment are very active in ethylene polymerization.

Polymeranalysis
The resultso ft he GPC analysis on the PE materials are shown in Ta Figure 2B.
The first observation are the broad MW curvesi nF igure 2, which coincidentally are also typicalf or high-density polyethylene materials produced by (ill-defined) Cr-based Phillips-type catalysts. [3,61] For instance, both MIL-101(Cr) and MIL-53(Cr) demonstrate av ariety of peaks in their respectiveG PC curves. On the other hand, the MWD curve for HKUST-1(Cr) is abit narrower,b ut still:m ultiple peaks are identifiable. These three curves gave rise to high values of 28.2, 16.9 and9 .8 for MIL-101(Cr)-NDC, MIL-53(Cr) and HKUST-1(Cr), respectively.W hile these resultsd or eflect an arrowing MWD, they are still far from values usually observedf or single-site catalysts, ultimately indicating the existence of am ultitude of active sites on the MOFs, as wasalso the case for MIL-101(Cr). [43]   The GPC curves, and related values, are slightly different for the reactions performed with the filtrates.T he traces of the PE obtained when using the Cr speciesf iltered from MIL-101(Cr)-NDC and HKUST-1(Cr) are significantly skewed around log MW % 6, suggesting that the number of active sites participating in polymer formationi ss maller.T his may be related to Cr speciesi ns olution, being Cr atoms with similar ligand coordination andnosteric restrictions.Furthermore, and in contrast to the two other topologies, the value obtained for the reaction performed with the filtrate of MIL-53(Cr) is very similar to that for the reactionp erformed with MIL-53(Cr) itself. Again, this findings hows that the number of Cr species generated from MIL-53(Cr) is larger than in the other two cases.
Additionally,t he differences in polymer architecturea re further manifested in the different crystallinities, shown in Table 1. The resulting PE materials demonstrate varying crystallinities: the lowest being that of the PE produced withH KUST-1(Cr) (39.3 %) and the highest being that of MIL-53(Cr)( 52.4 %). Interestingly,t he PEs produced in the filtrate reactions all demonstrate higherc rystallinitiest han their heterogeneous counterparts. Perhaps the fact that the average increased M n explains this, since this value infers an overall relative increase of averagec hain length, which can be associated to the higher crystallinity of the materials. Take note that the calculated crystallinityo ft he PE/HKUST-1(Cr) materiali su nderestimated due to the significant contribution of the HKUST-1(Cr) MOF residue. The likely PE crystallinity is higher, and correcting for the MOF residue gives av alue of 51.7 %.
Conclusively,i ti se videntf rom Table 1a nd Figure2 that rational selectiono ft he MOF is av alid strategy for exposing and/or attaining desirable PE properties.

Polyethylene morphology
Naturally, the PE morphologyi sh ighly important for assessing post-reaction processability as well as preventing reactor fouling, often relatedt oe xpensive reactor downtimes. [62] From an industrial point of view," good" is considered spherical with narrow particle size distributions,w hich is directly related to high bulk density,c ontrolled porosity,c ontrolled internal compositionand high process flowability. [63,64] Figure 3s hows the obtainedP Em aterials obtained from the ethylene polymerization reactions with the MOFs, and clearly demonstrates that one can affect the final PE morphologyb y selectingt he proper MOF polymerization template, at least under the here described reaction conditions. First, Figure 3A showst hat the PE produced by MIL-101(Cr)-NDC consists of spherical particles, where the inset shows subcrystallites being tied together by PE molecules. This indicates that the MOF acts as as elf-sacrificial template and disintegrates due to the increasing stress generated by the growing polyethylene. Despite this fragmentation, it seemst hat the originally octahedral MOF morphologye nforces as pherical morphology on the PE.S econdly,t he obtained morphology from the reactions with MIL-53(Cr)i sb est described as af ibrous worm-like PE material. [65] Interestingly,C hanzy et al. attributed this mechanism of PE growth to active sites being in very close proximity,h ereby hampering the PE growth in lateral directions while this is not the case perpendicular to the plane of active sites. [65] Thus, this indicates that factors such as active site spacinga re also an highly important parameter to considerwhen using MOFs in ethylene polymerization.
Despite the low activity of HKUST-1(Cr), it mainly produced spherical PE beads, as shown in Figure3E, with some PE fibers as well. The dual morphology is likely explained by the fact that as ignificant percentage of the catalytic activity originates from homogeneous Cr sites that produce the fibrousP Em aterial. Interestingly,t he amount of fibrous material is relatively small and spherical particles are the predominantp roduct. This indicates that even for the homogeneously polymerizing components,t he MOF can act as the preferential growth template.
Figures 3B,Da nd Fa re at estament to the importance of the MOF's role as structuringa gent.I na ll instances, if this structuring agent is removed from the equation, av ariety of structures (e.g. platelets, spheroids and fibers) is obtained. Irregularities on the reactor wall and/ors tirrer now predominantly act as crystallization points for the (relativelys mall) waxes formed over the homogeneous Cr sites, after which PE continues growing. [62] If anything, theseS EM images highlight the importance of selecting the appropriate MOF as ethylene polymerization platform for attainingd esirable PE morphologies. Additionally,a nd possibly more importantly,t he simultaneous polymerization over heterogeneous Cr sites and homogeneous Cr sites is not detrimental for the final morphology,i nferringt hat even for the dissolveda ctive sites, and PE materials, the MOF crystallites act as crystallization point.

Actives itef ormation
The GPC resultsi ndicatedt he participation of av ariety of active Cr sites in ethylene polymerization:b oth homogeneous and heterogeneous. Therefore, we opted to exploit an array of spectroscopic techniques such as XRD, CO probe molecule FTIR and UV/Vis-NIR DRS experimentsi no rder to investigate the MOF activation stage. Additionally,i nvestigating the morphologyo ft he pristine and activated MOFs required the use of SEM as an imaging technique.
First, XRD serves as an excellent tool to investigate the effect of the co-catalyst on the MOF crystallinity,f or which the results are shown in Figure 4. The X-ray diffractograms of the three pristineM OFs perfectly resemble those from the literature, although noteworthy:T he XRD pattern of MIL-53(Cr) resembles that of the solvated MOF. [51,55,58] Now,t he effect of Et 2 AlCl on the crystallinity of each MOF was different and will be discussed hereafter.I fc onsumption of the MOF is to occur, this would coincide with al oss and/or broadening of the diffraction signals.
For both MIL-101(Cr)-NDC and MIL-53(Cr) this indeed occurs, as demonstrated in Figures 4A and B, respectively.A ctivation of MIL-101(Cr) is related to ad isappearance of the signals below diffraction angles of 108.I tc an be argued that the diffraction signals at 10 can still be identified, albeit less intense, inferring that at least some of the crystallinity is retained. In   Figure 4B,a ctivation of MIL-53(Cr)i sr elated to the disappearance of someo ft he diffraction peaks (most clearly the ones at 2q = 13, 15, 23 and 258), indicating that the crystallinity of MIL-53(Cr) is severelyd isrupted. It is worth stating that MIL-53(Cr) is known to undergo structuralt ransitions and exhibitsasocalled breathing effect,h owever it is highly unlikely that such events occur here, since only some of the typical XRD reflections for large pore (LP) structures is visible in this case. [55,66,67] Interestingly,t he crystallinity of HKUST-1(Cr) is almost unaffected by the activation procedure with 100 mol. equiv.E t 2 AlCl, since the original diffraction pattern is largely retained. The decrease in intensity of diffraction peaks above 2q = 158 does infer some minor disruption of the crystallinity,h owever the severity is far from those found for MIL-101(Cr)-NDC and MIL-53(Cr).T his observation might provide part of the explanation for the low activityo fH KUST-1(Cr), since apparently this MOF is activated to the least extent.
Also, the X-ray diffractograms are at estamentt ot he polymerization of ethylene over the MOFs, since instead of MOF re-lated reflections, only HDPE X-ray diffraction peaks are observed,s pecifically at 238.
Furthermore, UV/Vis-NIR DRS served as at ool to investigate the effect of the co-catalyst on the oxidation state and coordination geometry of the Cr active sites. The spectra of the materials before and after reaction with the organo-aluminum cocatalysts are shown in Figure 5. It is worth stating that the startingo xidation states of MIL-101(Cr)-NDC and MIL-53(Cr) are considered to be Cr 3 + ,d emonstrated by the bands around 16 000 and 22 000 cm À1 .T hat of HKUST-1(Cr) is expected to be Cr 2 + on basis of the original manuscript, which wasa lso confirmedb yt he orange color of the material, consistent with the originalr eport. If oxidation of HKUST-1(Cr) is to occur,acolor change from orange to green would be observed, which was excluded. There are small variations from one MOF to another, due to the different electronic structures of the metal centers. The Charge Transfer (CT) bands above 30 000 cm À1 have been omitted for all MOFs, due to their extreme intensity that satu-rates the detector and therefore renders interpretation impossible.
Activation with the co-catalyst does affect the spectra to some extent.I ne ach case, reacting the MOF materials with the co-catalyst resulted in ad arkening of the powder material, which in the UV/Vis-NIR DRS spectra is directly related to an overall increase of intensity of these spectra.
With respect to the oxidation state, none of the UV/Vis-NIR DRS spectra give strong indicationsf or the formation of any oxidation state besidesC r 3 + ,u nlike the case of previously reported MIL-101(Cr). [43] This is surprising given the identical Cr trimer forM IL-101(Cr) and MIL-101(Cr)-NDC,w ith the pore size being the only difference between these two materials. It is worth stating that the 16 000 cm À1 band broadens significantly for MIL-101(Cr)-NDC after activation with the co-catalyst, with two possible chemical explanations. First, activation with the organo-aluminum complexw as found to generatealarge variety of active sites, naturally, this translates into av ariety of UV/ Vis-NIR absorptions and is related to the observed heterogeneous broadening.S econdly,E t 2 AlCl in factr educes some of the Cr 3 + sites from the pristine MOFs to Cr 2 + ,w hich are known to absorb in the 8000-12 000 cm À1 region. In this case, the exact location is usually determined by the degree of coordinative saturation, and the fact that broadening is observed around 10 500 cm À1 infers that if reductioni so ccurring it is likely that Cr 2 + Oh is formed (i.e.,c oordinatively more saturated Cr 2 + ). With respectt ot he fitted bands, these only serve to emphasize broadening of the band at 16 000 cm À1 band, with its location and FWHM being identical in Figures 5A and B. [25,[68][69][70][71] The UV/Vis-NIR DRS spectra after activation for both MIL-53(Cr) ( Figure 5D)a nd HKUST-1(Cr)( Figure 5F)o nly demonstrate increased intensities of the Cr bands, Cr 3 + for MIL-53(Cr) and Cr 2 + for HKUST-1(Cr). However,t he width and location of these bands remains identical, indicating that activation with the co-catalyst does not particularly affect the oxidation state of the MOFs.
Next, CO probe molecule FTIR is an excellent tool to probe the Cr site accessibility, while simultaneously providing some information on the relatedo xidation state. [72] Similarly as in the previouss ection, CO probem olecule experiments were performed on the pristine MOF materials as well as MOF materials that were activated with 100 mol. equiv.E t 2 AlCl, the results of these experiments are shown in Figure 6. Figure 6A and 6B respectively illustrate the resultso btained for MIL-101(Cr) before and after activation with 100 mol. eq. Et 2 AlCl. It is worth stating that the band at 2137cm À1 can exclusively be attributed to physisorbed CO on the MOF material. The second band,a t2 158 cm À1 can be attributed to CO coordinated to H 2 Om oieties still reminiscentf rom the hydrothermal synthesis. Additionally,t wo bands with low intensity at 2198 and 2088 cm À1 are present in the spectra, which based on the literature can be attributed to minor amountso f, respectively Lewis acidic (LA) Cr 3 + sites and extra-framework Cr (oxidation state 2 + or 3 +). With the UV/Vis-NIR DRS spectra in mind, extra-framework Cr 3 + is more likely since no hard evidence for Cr 2 + in the pristine material was found. Activation with the co-catalyst, as seen in Figure 6B,r esultsi nasignifi-cant decrease of the 2158cm À1 band, likely due to scavenging of the coordinating H 2 Om oieties by Et 2 AlCl. Second, the band at 2088 cm À1 is still similarly intense as before activation,s uggesting that only little additional extra-frameworkC ri sf ormed, with both 2 + and 3 + oxidations tates now being viable. However,t he significant increaseinC r 3 + LA species, testified by the 2198 cm À1 band, does infer that the heterogeneous broadening in Figure 5B is predominantly caused by the heterogenization of the Cr 3 + sites rather than the formation of Cr 2 + species. Last, an ew band emerged at 2265 cm À1 ,w hich is possibly attributed to CO coordinated to Al 3 + from the co-catalyst material which is embedded/reacted with the MOF materiala nd could not be washed away.
MIL-53(Cr) is defined by the ability to only physisorb CO, with its coordinatives aturationb eing well demonstrated by the lack of any additional bands, with exception of the band at 2088 cm À1 attributed to minor amounts of extra-framework Cr, likely oxidation state 3 + ,a sw ell as the very small band at 2175 cm À1 attributed to LA Cr 3 + .I ti sw orth mentioning that in this case MIL-53(Cr)-np (np = narrow-porous) is likelyt he phase of MIL-53(Cr) under investigation. Performing the CO probe molecule FTIR experiments on the activated MIL-53(Cr)r esults produces as trongly changed spectrum.F irst, it appears that the ability to physisorb CO has significantly decreased Now, the relative intensityo ft he LA Cr 3 + bands and extra-framework Cr bands is larger.H owever,o ne should take notet hat the absolute intensities of theseb ands remainv ery small. Additionally, one band at 2025cm À1 can now be discerned, which infers the possibility of few-atom Cr clusters.
Last, the CO probe molecule FTIR spectrum for HKUST-1(Cr), as shown in Figure 6E,i sd efined by only one major signal attributed to the physisorption of CO. This indicates the absence of coordinatively unsaturated sites (CUS) formed after evacuating/drying during the synthesis, suggestingt hat either DMF molecules from the synthesis or OH molecules from the solvent exchanger emainc oordinated to the Cr sites, even under the conditions used in this study.A part from the band at 2137 cm À1 ,o nly minor bands are observeda t2 088 and 2240 cm À1 ,t he former is again attributed to extra-framework Cr 2 + .P erforming the same activation procedure with Et 2 AlCl on HKUST-1(Cr) resulted in ab and at 2240 cm À1 ,a long with bandsa t2 117a nd 2088 cm À1 .D ue to the unlikeliness of any other oxidation state than 2 + ,t he latter signal is again attributed to extra-framework Cr 2 + .I nterestingly,t he 2117 cm À1 band falls in the range where normally few-atom Cr clusters (eithero xidation state 2 + or 3 +)a re observed, which considering the structure of the metal-node is not an unlikely explanation. [72] In any case, the fact that the intensity of this band is low is at estamentt ot he stability of this MOF,i ndicating that only as mall number of the originall inker-Cr bonds is broken.
In summary,i ti sc lear that the XRD experiments indicate that the more activeM OFs are modified to al arger extent. The CO FTIR experimentsf urther corroborate the generation of a variety of active sites, unique to each MOF,w hich retain the oxidation state of the pristineM OF.

Estimating collapse
From the X-ray diffractograms and CO-FTIR spectra shown above,c ollapse of the framework may have occurred, affecting the catalytic properties of the MOFs by limiting accessibility of ethylene or chain growth. We used N 2 physisorption at À196 8Ct oe valuate the porosity of the MOFs before and after activation with the co-catalyst.
The respective isothermso ft he pristine MIL-101(Cr)-NDC, MIL-53(Cr) and HKUST-1(Cr) are shown in Figures S12, S14 and S16 with BET surface areaso f1 419, 1531 and 1033 m 2 g À1 .I n any case, Type Ii sotherms are observed for all the MOFS, confirming their porosity.I ti sl ikely that collapse of the MOF is related to ad isappearance of the material's porosity as well as a potentialc hange in the observed type of isotherm. While Figures S13, S15 and S17d oc onfirmt hat part of the surfacea rea is lost, apparent BETsurface areas of 635, 124 and 468 m 2 g À1 , indicatet hat still as ignificant part of the porosity is retained. The largestd rop is observed in case of MIL-53(Cr), suggesting that in this case porosity might not be correlated to catalytic activity.I ti sw orth stating that drying of the pristine MOFs was performedu nder dynamic high-vacuum, while this is supposedly as afe possibility fort hese MOFs, it is important to consider that vacuum might have detrimental effects on the surface area, explaining the slightly lower value compared to those observed in the literaturei nc ase of MIL-101(Cr)-NDC. [73,74] Second, the morphology of all the MOFs,a ss hown in Figure 7, has been extensivelyd escribed in the literature and matches the crystalso bserved in our case. For MIL-101(Cr)-NDC that means nanosizedo ctahedral crystalsc lustered into larger structures, as shown in Figure 7A. [75,76] For MIL-53(Cr) a crystalline materiali so btained, which consists of exclusively needles (sized 5-10 mm). It is worth statingt hat MIL-53(Cr) is the thermodynamic product of MIL-101(Cr). [57,[77][78][79] In case of HKUST-1(Cr), au niform powder consisting of the octahedral crystalsi nF igure 7E is obtained. Again,t his is perfectlyi nl ine with what is reported in the literature on the morphologyo f this MOF. [80][81][82] It is worth stating that our research does not provide basis to comment on any potential MOF morphology/ activity relations, due to the varying metal-nodes of the MOFs which in that case should have been constant.H owever,t his does warrant further researchw here the metal-node structure is kept identical and the particle size/shape is systematically varied.
The morphologies of the activated MOFs are shown in Figures 7B,D ,a nd F, in which no obvious changes are visible, confirmedb yt he preserved shapes and sizes of the MOF crystals. It is worth stating that no Al 2 O 3 is observedi nt he brightfield micrograph as ap ossible by-product from the activation procedure.
Diffuse reflectance infrared spectroscopy (DRIFTS) wasu sed to investigate whether the characteristic MOF fingerprint bands changed during activation with the co-catalyst.
In the case of MIL-101(Cr)-NDC ( Figure 8A)t he obtained DRIFTss pectrumi si nl ine with reported spectra. Worth noting, our preparation of MIL-101(Cr)-NDC yielded ap roduct without free 2,6-dicarboxynapthalene, which may potentially block pores as well Cr 3 + CUS sites, testified by the absence of a v(C= O) band at 1700cm À1 .S econdly,a ctivation of the MOF,a s showni nF igure 8B,d id not decrease the S/N ratio nor did it affect the MOF fingerprint. It is worth stating that newly emerged bands at 2870 and 2950 cm À1 can be assigned to alkylation of the MOF by the Et 2 AlCl co-catalyst, as they correspond to thes tretching vibrations of the ethyl groups.
MIL-53(Cr) (showni nF igure 8C)i sc haracterized by isolated Cr 3 + Oh metal centers, linkedi ni nfinite chains 1D by terephthalic acid with m 2 -OH groups bridging the individual chains. These HKUST-1(Cr) behaves very similar,a ss hown in Figures 8E  and F. The DRIFTS spectrum of pristine HKUST-1(Cr) testifies to the successfuls ynthesis where no free trimesic acid is observed which might potentially block the pores and/or Cr 3 + sites. Interestingly,t he 1640-1650 cm À1 range shows two signals, of whicho ne likely belongs to boundD MF originating from the MOF synthesis. Activation with the co-catalyst resulted in the disappearance of this signal, relatedt oa bstraction of DMF,while alkylation occurred simultaneously.
Despitet he significant concern of MOF collapse after activation with the Et 2 AlCl co-catalyst, the above results show that the chemical bonds comprising the MOF seem to be unaffected. This points to the fact that, in terms of morphologya nd bonding, the MOF remains intact. While the loss of crystallinity is relatedtothe observed loss in porosity,iti shighly important to state that activation with Et 2 AlCl does not result in total collapse of the MOF.T his suggestst hat the resultingm aterials are partially disordered and porousC r 3 + carboxylates with av ariety of Cr 3 + embedded (and potentially extra-framework Cr 3 + and Cr 2 + )a lkylated sites for MIL-101(Cr)-NDC and MIL-53(Cr), whereas this involves Cr 2 + sites for HKUST-1(Cr). From our study,i ts eems that more generated defects, that is, MIL-101(Cr)-NDC and MIL-53(Cr), is relatedt oh ighere thylene polymerization activity.T he catalytic activity was related to large values, confirming the presence of av ariety of active Cr sites. However,f urthere xperimentsa re necessary to disentangle the contribution of each of the solid and/orhomogeneous Cr compounds to the MWD.I ti sa lso clear that al arger percentage of PE produced over homogeneous Cr complexes leads to more fibrousHDPE, probablyd ue to uncontrolledg rowth.

Conclusions
In this work, we have investigated the performance of MIL-101(Cr)-NDC, MIL-53(Cr) and HKUST-1(Cr) in ethylene polymerization. First, we have found that selecting the appropriate MOF as polymerization platform is of paramount importance for attaining desirable levels of activity with MIL-53(Cr) being the most active, followed by MIL-101(Cr)-NDC and eventually HKUST-1(Cr). Although the three MOFs demonstrated activity, we found that some of the activity originated from leached Cr complexes.
Secondly,t he poly-ethylenes (PEs) produced over the MOFs demonstrated large values as wella sv aryingc rystallinities, indicating that ethylene polymerization occurs over al arge variety of actives ites. Furthermore, selection of the appropriate MOF was critical for templating the final PE morphology,w ith only MIL-101(Cr)-NDC producing favorable PE spheres and HKUST(1)-Cr demonstrating potential by predominantly producing spheres as well, however with some fibers.
Thirdly,o ur spectroscopic investigations indeed confirmed that activation of the MOFs results in the generation of av ariety of active sites, while retainingt he oxidation state of the pristine MOF.F urthermore, we found that the MOFs which were the most modified were also the most active, indicating that propera ctivation of the MOFs is ap rerequisite for ethylene polymerization.
Lastly,i tw as critical to exclude total collapse of the MOF, which on basis of the GPC and spectroscopicr esultsw as a likely event.A ctivation of the pristine MOFs indeed resulted in decreased BET surfacea reas (SA), with porousm aterials (BET SA > 400 m 2 g À1 )s till being the predominant products. Furthermore, activation of the MOF neither affectedt he MOF morphologyo rt he MOF DRIFTS fingerprint, indicating that activation does not result in cleavage of all the bonds constituting the MOF.
In summary,w eh ave explored three renownedC r-based MOFs as ethylene polymerization platforms. In this work, we have shown that selecting the appropriate MOF is criticalf or the activity,PEproperties as well as PE morphologies. More importantly,w eh ave also demonstrated that the active MOF cannotb ec onsidered as ingle-site heterogenized ethylene polymerization catalyst. We believe that these findings can be helpful for the future development of heterogeneousC rc atalysts as well as Cr-based MOFs and their applicationsi nt he important ethylene polymerization reaction.

Experimental Section
Catalyst preparation The synthesis of the MOFs MIL-101(Cr)-NDC, [51] MIL-53(Cr) [55] and HKUST-1(Cr) [58] was carried out according to previously published procedures. Details on the preparation can be found in the Supporting Information.

Ethylenep olymerization reactions
In at ypical polymerization experiment, an amount equivalent to 0.05 mmol of each MOF was suspended in 15 mL of anhydrous heptane (99.9 %a nhydrous, stored over molecular sieves, Sigma-Aldrich) or toluene (99.9 %a nhydrous, stored over molecular sieves, Sigma-Aldrich) in as tainless-steel Parr reactor,t ogether with 100:1 (Al:Cr) molecular equivalents of Et 2 AlCl (97 %, Sigma-Aldrich) in an N 2 filled glovebox (O 2 < 1.5 ppm and H 2 O < 0.6 ppm). Subsequently,t he reactor was attached to an ethylene gas/vacuum system allowing the evacuation and flushing of the lines before ethylene (Linde AG, 99.9 %) polymerization was performed at a constant pressure of 10 bar and at emperature of 23 8C. The reaction mixture was stirred at 1000 rpm. The reactor was depressurized after 30 min of polymerization and the residual Et 2 AlCl was quenched with 1 m HCl in ethanol. The solid product was copiously washed with 1 m HCl in ethanol, followed by ethanol. The solid material was dried at 70 8Co vernight and 3hunder vacuum. The catalyst activity was based on the weight of the obtained polymer products. The polymerization reactions with the filtrates were performed as follows. 0.05 mmol of each MOF was suspended in 10 mL of anhydrous heptane (99.9 %a nhydrous, stored over molecular sieves, Sigma-Aldrich), stirred for 15 minutes and subsequently collected by filtration, the filter was washed twice with 2mLo ft he heptane and once with 1mLo ft he heptane in order to reach the desired diluent volume of 15 mL. Subsequently,e thylene polymerization was performed as described before.

X-ray diffraction
The X-ray diffraction (XRD) patterns were obtained with aB ruker-AXS D2 Phaser powder X-ray diffractometer in Bragg-Brentano geometry using Co Ka = 1.78897 ,o perated at 30 kV.T he measurements were carried out between 5a nd 30 8,u sing as tep-size of 0.058 and as can speed of 1swith a0.1 mm slit for the source. The activated MOF material was prepared as follows. 100 mg of MOF was distributed over ar equired number of vials so that each vial contained 0.05 mmol MOF.S ubsequently,1 5mLo fh eptane was added and 100 mol. equiv.o fE t 2 AlCl and the mixture was homogenized for 15 minutes before collecting the powder by filtration.
Hereafter it was washed thrice with 5mLp entane. The MOF was carefully dried in air for 5min before being brought outside and carefully exposed to ambient atmosphere. After this, the materials were measured.

CO probemoleculeF ourier transform infrared spectroscopy
Fourier transform infrared (FTIR) spectroscopy measurements with CO probe molecules were recorded on aP erkinElmer 2000 instrument, in as pecially designed cell fitted with CaF 2 windows. The dried MOF materials were pressed into 5-7 mg wafers inside an N 2 glovebox (O 2 < 1.2 ppm and H 2 O < 0.6 ppm). The cell was sealed and connected to the gas/vacuum system. Subsequently,t he cell was carefully evacuated to 10 À5 bar at 25 8C, after which the sample was cooled to liquid N 2 temperature. Am ixture of 10 % CO/ 90 %H E( v/v) was dosed with small increments while measuring FTIR spectra 1min after each CO dosing to ensure equilibration. Experiments performed on the activated materials were performed as follows:5 -7 mg of the pristine MOF was pressed in a self-supporting pellet, after which it was carefully suspended for 15 min in am ixture containing 15 mL heptane and 100 mol. eq. Et 2 AlCl. Subsequently it was suspended in 15 mL pentane to wash away the excess Et 2 AlCl. Subsequently,t he CO probe molecule experiments were performed as previously described. Extra care was taken to ensure that there were no cracks in the pellet after activation.
UV/Vis-NIR diffuse reflectance spectroscopy UV/Vis-NIR Diffuse Reflectance Spectroscopy (DRS) measurements were performed using aP erkinElmer Lambda950s spectrophotometer with aP raying Mantis DRS accessory.T he measurements were performed in the 40 000-4000 cm À1 region with a6 0msd atapoint scan time and 4nms pectral resolution. For every measurement, the Praying Mantis DRS was loaded with 10-20 mg of MOF inside an N 2 filled glovebox (O 2 < 1.2 ppm and H 2 O < 0.6 ppm). The samples were measured against aT eflon white, measured in the same cell, consisting of 30 mmT eflon beads. The activated MOFs were performed as follows:0 .05 mmol MOF was suspended in 15 mL heptane, to which 100 mol. eq. Et 2 AlCl was added. The mixture was stirred for 10 min, after which the powder was collected by filtration and washed twice with 5mLh eptane. The powder was dried in the atmosphere for an additional 10 min before being measured.
Inductively coupled plasma-atomic emission spectroscopy ICP-OES was used for chemical analysis. 0.05 mmol of the pristine MOF was weighed and suspended in 15 mL heptane to which 100 mol. eq. Et 2 AlCl was added. Subsequently,t he liquid was collected by filtration and the filter was washed twice with 5mLh eptane. Subsequently,t he diluents were removed by slow evaporation. The residue was then dissolved in am inimal amount of aqua regia before being diluted to the same pH as a5%H NO 3 solution. 10 mL of the diluted samples were taken for measurements on a PerkinElmer Optima 8300 and an average of three samples was used. Cr (267.7, 205.6 and 283.6 nm) were measured and referenced to 0, 0.2, 0.4, 0.6, 0.8 and 1.0 mg L À1 were prepared of all the metals.

Scanningelectron microscopy
SEM measurements were carried out with aF EI Helios NanoLab G3 UC (FEI Company) instrument equipped with aS ilicon Drift Detector (SDD) at 10.0 kV acceleration voltage and a0.10 nA current. The samples were dispersed on an aluminum SEM stub with ac arbon sticker and were subsequently coated with aPtlayer.

Differential scanning calorimetry
DSC was performed on aT AI nstruments DSC Q20 with 1-2 mg of the nascent material. Each sample was heated from À40 8Ct o 200 8Ca tar ate of 10 8Cmin À1 after which it was held isothermally at 200 8Ct oe rase thermal history of the PE. Subsequently the cooling cycle was initiated to À40 8Ca tarate of 10 8Cmin À1 followed by an additional heating cycle to 200 8Ca tar ate of 10 8Cmin À1 . The crystallinities of the polyethylene materials were determined assuming DH m 0 = 293 Jg À1 for 100 %c rystalline ultrahigh-molecular-weight polyethylene (UHMWPE).

Gel permeation chromatography-size exclusion chromatography
Gel permeation chromatography-size exclusion chromatography (GPC-SEC) was carried out on aP olymer Laboratories PL-GPC220 instrument in 1,2,4-trichlorobenzene at 160 8C, equipped with aP L BV-400 refractive index detector.T he column set consisted of three Polymer Laboratories 13 mmP Lgel Olexis 300 7.5 mm columns, and the calibration was performed with linear polyethylene (PE) and polypropylene (PP) standards. N 2 physisorption N 2 adsorption measurements for MIL-101(Cr)-NDC and MIL-53(Cr) were measured at 77 Ko naMicromeritics TriStar 3000 instrument. Prior to all measurements, samples were dried at 150 8Cu nder dynamic vacuum. Specific surface areas (SSAs) were calculated using the multipoint BET method (0.05 < p/p 0 < 0.25). N 2 measurements for HKUST-1(Cr) were performed as follows:h igh-resolution lowpressure adsorption measurements were measured on aM icromeretics ASAP2010 gas adsorption analyzer equipped with additional 1mmHg and 10 mmHg pressure transducers. Ar elative pressure range from p/p 0 = 10 À7 to 0.99 has been applied. Before the actual measurements on this apparatus, the samples were degassed for 16 ha t1 50 8C. Specific surface areas (SSAs) were calculated using the multipoint BET method (0.05 < p/p 0 < 0.25).