Matrix Effects in a Fluid Catalytic Cracking Catalyst Particle: Influence on Structure, Acidity, and Accessibility

Abstract Matrix effects in a fluid catalytic cracking (FCC) catalyst have been studied in terms of structure, accessibility, and acidity. An extensive characterization study into the structural and acidic properties of a FCC catalyst, its individual components (i.e., zeolite H‐Y, binder (boehmite/silica) and kaolin clay), and two model FCC catalyst samples containing only two components (i.e., zeolite‐binder and binder‐clay) was performed at relevant conditions. This allowed the drawing of conclusions about the role of each individual component, describing their mutual physicochemical interactions, establishing structure‐acidity relationships, and determining matrix effects in FCC catalyst materials. This has been made possible by using a wide variety of characterization techniques, including temperature‐programmed desorption of ammonia, infrared spectroscopy in combination with CO as probe molecule, transmission electron microscopy, X‐ray diffraction, Ar physisorption, and advanced nuclear magnetic resonance. By doing so it was, for example, revealed that a freshly prepared spray‐dried FCC catalyst appears as a physical mixture of its individual components, but under typical riser reactor conditions, the interaction between zeolite H‐Y and binder material is significant and mobile aluminum migrates and inserts from the binder into the defects of the zeolite framework, thereby creating additional Brønsted acid sites and restoring the framework structure.


Introduction
Zeolite-based catalysts are employed on al arges cale in many differenti ndustrial processes, such as crude oil refining (e.g., catalytic cracking, isomerizationa nd aromatization reactions) and methanol-to-hydrocarbons (MTH). [1][2][3] For optimal functionality,z eolitesa re often heterogeneously dispersed within a matrix and shaped into catalyst bodies. Typical matrix components include clays, such as kaolinite and attapulgite,a nd amorphous alumina or silica. These matrix components offer severali mportant advantages,s uch as heat and attrition resistance, but also mechanical and chemical stabilitya nd proper accessibility to the acids ites. Moreover,s uch shaped catalyst bodies are easier to handle and recover from the reactort han zeolite powders. [4][5][6] The influence of matrix components, however,c an reach further than merely the physical advantages described above. Only recently,t he academic literature has started to shift its focus from zeolite powderst of ull catalyst bodies with all the intrinsic complexity to investigate the influence of matrix components on the catalytic performance of the catalyst. This importance has, for example, been highlighted in ap erspective article by Hargreaves and Munnoch. [5] Matrix materials can have both beneficial or detrimental effects on the catalytic performance parameters, such as decreased or increased catalystl ifetime, acidity, pore accessibility,a nd product selectivity. [5][6][7][8][9][10] Fluid catalytic cracking (FCC)i sa ni mportantc atalytic technologyi nacrude oil refinery that employs such ac omplex and multi-component zeolite-based catalystm aterial. [11,12] In the FCC process, hot catalyst is mixed with vacuum gaso il (VGO) or heavy gas oil( HGO) and transferred to the riser reactor.I nt he order of seconds, the oil is cracked into smaller fragments at temperatures of approximately 550 8C. Due to the rapid formation of gaseous products,t he mixture rises to the top of the reactor,w here the products are stripped from the catalyst. The spent catalystp articles then travel further to the regenerator,w here the catalyst is regenerated by burning the coke from the catalyst. It is then ready for reuse in the catalytic oil cracking. [12] Crystalline microporous aluminosilicates of the faujasite (FAU) framework type (i.e.,astabilized version of zeolite Y) are the main responsible species for the cracking activitya nd selectivity in the FCC process. These zeolite domains are mixed with ab inder (typicallyas ilica/aluminap hase) and af iller (typically ac lay mineral) and, subsequently,s pray-dried to form spherical liquid-like behaving catalyst bodies of 50-150 mm. The dilution of the active zeolitep hasei sh ighly necessary to prevent excessive gaseous product formation in the riser reactor, [6,11,13] but the matrix components also play an important role in the catalytic performance of the shaped catalyst bodies. [14] The overall reactivity of the FCC catalystr elies on the presence, strength and accessibility of acids ites. [11,15] The substitution of SiO 4 tetrahedra with AlO 4 creates an egativelyc harged zeolite framework. Counter ions in the form of protons can balance this negative charge, resulting in the formation of Brønsted acid sites in zeolitem aterials. The zeolite framework gives rise to various channels and cages of moleculard imensions, providing shape-selectivep ores for the selectivep roduction of the desired products,i ncluding gasoline or lower olefins, such as propylene. [6,16] Lewis acid sites in the matrix, originating from aluminum species, are capable of pre-cracking the long-chain oil molecules, prior to entering the zeolite micropores.
Previousw ork from our research group on structure-acidity relations in the FCC catalyst has mainly dealt with individual FCC catalyst particles and the visualization of Brønsted acid sites by chemicals taininga nd subsequentc onfocal fluorescence microscopy. [17][18][19][20] Staining probe molecule reactions (i.e., acid catalyzed oligomerization reactions of styrene, thiophene, and furfuryl alcohol derivatives yielding fluorescent reaction products)h aver evealed the influence of different binderso r clays on the amount and strength of different Brønsted acid sites in the catalyst body. [7,9,17,18,[20][21][22][23][24] Lewis acid sites are, however,m ore difficult to localize and visualize and have not yet received considerable attention. The nature of these sites is, therefore, not as understood as for Brønsted acid sites. By definition, Lewisa cidity is an electron deficiency,o riginating from coordinatively unsaturated sites (CUS). Under ambient conditions, such sites often interact with water molecules, thereby saturating the sites and losing their Lewis acidic properties. [25] Under reaction conditions, the temperature is higher (550 8C) and the catalysti sd ehydrated, making the Lewis acid sites availablef or catalysis. It is, therefore, highly necessary to study acidic properties at similartemperatures to FCC reaction conditions to ensuret he relevance and reliability of the study.
In this work, we have investigatedw hether the acidity in an FCC catalyst originates from the inherent acidic properties of the individual catalyst components (e.g.,z eolite material), or whether synergistic or detrimental interactions occur between the different components within the catalystb ody,t hereby either creatingo rd eletinga cids ites. Such interactions are referred to as matrix effects. These resultso nm atrix effects then lead to draw conclusions about the structural naturea nd location of all acid sites within the FCC catalyst. The FCC catalyst under study contains af aujasite zeolite (i.e.,z eolite H-Y), a kaolin clay and as ilica/boehmite binder.A ne xtensive comparison between the structural, physical, anda cidic properties of the single components versus the fully shaped FCC catalyst is presented. Te mperature-programmed desorption (TPD) of NH 3 and FTIR spectroscopy of adsorbed CO provide information on the acidic properties of the samples under study and reveal the synergistic effect of component interactionw ithin the FCC catalyst. Next, structure-acidity relations are established employing multiple quantum-magic angle spinning-nuclear magnetic resonance (MQ-MAS-NMR) spectroscopy,t ransmission electron microscopy (TEM), X-ray diffraction (XRD), and Ar physisorption.
The investigation into matrix effects in FCC catalysts not only requires maximum exploitation of the characterization toolbox at hand, but also aw ell-defined set of samples, comprising the individual catalystc omponents, the fully mounted spray-dried FCC catalyst, but also spray-dried samples containing only two out of three components (i.e.,b inder-zeolite and binder-clay combinations)t od ifferentiate more clearly between the variousi nteractions within the FCC catalystm aterial. Characterization of the acidic properties of the individual components allows for modeling the acidic properties of the spraydried samples. The comparison between model and experiment clearly visualizes synergistic anda ntagonistic effects as a result of component interaction. Importantly,t his work consid-ers all types of acid sites, pays equal attention to all FCC catalyst components,a nd, moreover,t akes into account the reaction conditions of theF CC process. As ar esult,d etailed structure-acidity relationships for the entire FCC catalyst material are established, to the best of our knowledge,f or the first time.

Temperature-programmeddesorption of NH 3
The individual TPD profiles for the six samples under study are showni nF igure S1 in the Supporting Information, while Ta ble 1s ummarizes the quantified amount of acid sites (expresseda sm molg À1 )p er sample from the NH 3 -TPD analysis. This amount includes the whole range of Brønsted and Lewis acid sites with different strengths. The zeolite component is the most acidic sample with % 1.34 mmolg À1 acid sites in comparison to the binder ( % 0.38 mmol g À1 )a nd clay ( % 0.03 mmol g À1 ), as was also evidencedf rom the NH 3 -TPD profile intensities in Figure 1. It is interesting to look at the acidic properties of the spray-dried samples. Upon the interaction between thez eolite and binder component, additional acid sites are created.W ith ap hysical 1:1mixture of the two components, % 0.86 mmolg À1 of acid sites weree xpected. Instead,t he ZeBi sample exhibits an acidity concentrationo f % 0.92 mmol g À1 .T his is considered the matrix effecto nacidic properties. [7] Such synergistic effectsa lso take place upon the formation of the FCC sample ( % 0.66 mmol g À1 in comparison with the expected % 0.58 mmol g À1 ). The interaction between Table 1. Overview of the accumulated integrateda mount of NH 3 desorbed from the differents amples under study as studied with temperature programmed desorption (TPD). binder and clay,o nt he other hand, resultsi na na lmost negligible diminishing of acid sites.
To understand which type of acid sites are affected, created, or removed upon the interaction between different components, we have modeled the NH 3 -TPD profiles of the spray-dried samples based on ap ure physical 1:1weight ratio mixture of the single component andp lottedt hem in the same graph with the experimentally obtained NH 3 -TPD profiles from FigureS1t oo bserve matrix effects.T he results are shown in Figure1.I ne ach graph, the experimentally obtained NH 3 -TPD profile is indicated in black, the model in red, and the difference of these two in blue. The difference plot represents the synergistic or antagonisticm atrix effect. The pie chart in the top right corner of each graph illustrates which sample was modeled and how the modelw as built. As such, the matrix effects, as observed for the fully mounted FCC catalyst in Figure 1a,a re now assorted in Figure 1b-e, where the influence of clay and binder on the acidic properties is demonstrated in detail.
The difference between the experimental and modelled profile for the FCC catalyst relies on the increase of relative weak Brønsted acid sites ( % 360 8C) [26] and the decrease of acid sites at % 225 8Ca nd % 495 8C. [7,26,27] The former is ascribed to weaker Lewis acid sites and the latter corresponds to NH 3 desorbing from strong Lewis acid sites. [27] The increase in Brønsted acid sites is mainly caused by the interaction betweenz eolite ands ilica/boehmite binder (Figure 1d)a nd, to al esser extent,through the interaction with the kaolin clay ( Figure 1b). Whitinge tal. previously reportedt he influence of ap ure alumina binder on the acidic properties of zeolite ZSM-5, proposing the migration of aluminumi nto the zeolite framework, creating additional Brønsted acid sites. [7a, 9a] In this work, prior to the treatment with probe molecules NH 3 or CO, the samples were dried at 550 8Cf or 1h.I nc orrespondence with the work from Whitinga nd co-workers, we propose that it is possible that, upon this in situ calcination, while the boehmite in the binder undergoes phase transformation, [28] am obile alumina speciesisf ormed that is able to migrate into the zeolite framework, thereby creating additional Brønsted acid sites.
The decrease in strong Lewis acids ites ( % 495 8C) is caused by an interaction between the zeolite and binder material (Figure 1d,e), while the clayd oes not seem to affect the strong Lewis acidic properties in the FCC catalyst ( Figures 1b,c). The matrix effect on the presence of weak Lewisa cid sites ( % 200 8C) in the FCC catalystisnegligible as anet result of different interactions. When two different components interact, it results in either as mall increaseo rd ecrease of weak Lewis acid sites. Within the full FCC catalyst, however,t hese effects have practically cancelled each other out. Coordinatively unsaturated( aluminum) sites (CUS) are considered strong Lewis acid sites. [29][30][31][32] Upon aqueous mixing with other catalyst components,t hese sites are the most reactive and are likely to reorganize with other reactive CUS in the mixture. As such, strong Lewis acid sites on the edge of the zeolite domains can interactw ith the CUS from the binder.A saresult, these Al centers saturate their coordination sphere, losing their strong Lewis acidic properties. In other words, the loss of Lewis acid sites is directly related to the interaction between matrix and zeolite within the FCC body.
Overall, the matrix effects are most pronounced upon the interaction between the zeolite andb inder components.T he net matrix effect as observed for the fully mounted FCC catalyst ( Figure 1a)s trongly resembles the binder effect within the ZeBi sample (Figure 1d). The influence of the clay component on the acidic properties of the FCC catalysti sl ess pronounced, but not negligible. As mall amount of Brønsteda cid sites is createdd ue to the interaction between clay and zeolite/binder ( Figure 1b)a nd the number of Lewis acid sites decreasesd ue to the interaction between clay and binder ( Figure 1c). The loss of Lewis acid sites can also be explained by the decreased materials accessibility.

FTIR spectroscopyo fCOa dsorption
The acidic properties of the FCC catalyst, its individual components, and the two-component spray-dried samples werea lso studied with FTIR spectroscopy making use of CO as ap robe molecule. Figures S2 aa nd S2 bi nt he SupportingI nformation summarize the FTIR spectra of the six samples before (black) and after CO adsorption (red) in the OH and CO vibrational region,r espectively.I nt he CO vibrational region (2250-2050 cm À1 ), the blue-shift of the original CO stretching vibration (i.e.,2 143 cm À1 in its liquid state) upon interaction with an acid site, can be taken as am easureo ft he acidic strengtha nd nature. By lookinga tt he corresponding consumption of the bands in the OH vibrational region (3800-3500 cm À1 ), as tructure-acidity correlation can be established. [33][34][35][36] For am ore detailed discussion of the individuals pectra,w er efer to the Supporting Information. As triking observation from these spectra, however,i st hat the OH vibrational bandsa t% 3695 and % 3745 cm À1 ,a ssigned to aluminol and silanol groups,r espectively,s how as ignificant blue-shift upon the interaction of CO with Lewis acid sites. When CO donates electrond ensity into the empty orbital of the Lewis acidic aluminum center,t his transfer is possibly stabilized by transferring some of the electron density into the neighboring OÀHb ond. This strengthens the OÀHb ond, causing ab lue-shift in the FTIR spectrum.T he indirect indications of Lewis acidity via the observed perturbation of hydroxyl groups has been demonstrated before by Busca and co-workers. [37] This observation can provide important information on the structuralo rigin of Lewis acid sites in the FCC catalyst, as it indicates that Lewis acidic Al sites in the FCC catalyst are in the near proximity of hydroxyl groups. In particular, Figure 3d emonstrates that the Lewis acid sites near silanol groups ( % 3742cm À1 )a re inherent to the zeolite and binderd omains, whereas strong Lewis acid sites near aluminol groups can originate from the clay (3695 and3 675 cm À1 )o r the zeolite (3670 cm À1 ).
Matrix effects,a sd etermined through the comparison between experimental and modelN H 3 -TPD profiles in Figure 1, were also established with CO FTIR spectroscopy.S ince the FCC catalyst is composed of zeolite,b inder,a nd clay in a 1:1:1weight ratio, the CO FTIR spectrum of the FCC catalyst can be modeledv ia al inear combination of the weight-corrected CO FTIR spectra of the zeolite,b inder,a nd clay,s imilar to the models used for the NH 3 -TPD profiles. The resulto ft his linear combination model is included in Figure2(red), together with the experimentally obtained spectrum of the FCC catalyst (black). The differenceb etween the model and the experiment is considered to be the matrix effect and is depicted in blue.
The matrix effects,a so bserved for the fully mounted FCC catalysti nF igure 2a,a re assorted in Figures 2b-e, where the effects of clay and binder on the acidic properties are demon-strated in detail. It is important to note here, that small differences between the modela nd the experimentals pectrumc an also derive from minor inconsistencies in the model,t hus to the experimentally assumed equal sample density,n eglecting small differences in for example, IR transmission.  with increasingi ntensity at % 2175a nd % 2165cm À1 . [37][38][39] This increasew as also observed in the NH 3 -TPD results in Figure 1. The band at % 2165 cm À1 ,c orresponding to weak Brønsted acid sites increases more strongly in comparison to the band at % 2175 cm À1 ,a ssigned to strongB rønsted acid sites. This implies that both types of Brønsted acid sites are formed upon the interaction between bindera nd zeolite, but the weak ones prevail. This also explains the fact that the average Brønsted acid site weakens in strength, as was observed in the NH 3 -TPD results in Figure 1.
In accordance with the NH 3 -TPD results, strong Lewis acid sites (characterized by an absorption at % 2230 cm À1 ) [30,40] are suppressed upon the interaction between the zeolite and the binder material, but remain present in the BiC sample (Figure 2c). This suggestst hat strong Lewis acid sites in the binder and zeolite are located on the edgeo ft heir respective domains,w hile they are protected inside the layered clay domains.E xcept for ac leard ilution of the acidic properties and the accessibility (decrease of the peak at % 2139 cm À1 ), the spectralf eatures for the BiC sample simply seem to be the averageo ft he individual two components. Interestingly,t he strong Lewisa cid sites (at % 2230 cm À1 )i nt he binder are preserved upon interaction with the clay component, whereas these had disappeared upon the interaction between binder and zeolite.T his either implies that the clay prevents the full removalo fs trong Lewis acid sites in the FCC catalyst as a result of zeolite-binder interactions or it meanst hat the strong Lewis acids ites in the FCC catalysta ll originate from the clay.

Matrix effects on accessibility properties
The synergistic or antagonistice ffectso nt he acidic properties of the FCC catalyst upon individualc omponent interactions can only be probedw hen the sites are. Also, it is important to note that acid sites are only relevant when reactants can reach these acid sites during the FCC process within the riser reactor. To investigate the accessibility of the six samples under study and assess the extento fm atrix effects, Ar physisorption was employed. The corresponding isotherms are depicted in Figure S3 in the Supporting Information and the quantified surface areas and pore volumes are summarized in Ta ble2.F or the spray-dried samples ZeBi, BiC, and FCC, the expected values for the BET surfacearea, pore volume, microporesurface area, and micropore volume are also indicated.T hese values are calculated based on the 1:1physical mixture of the components withouta ny additional matrix effects. Interestingly, Ta ble 2demonstrates that the bindingo fd ifferent components together in as pray-dried particle, can affect the accessibility of the resulting samples.
The zeolite component has the highest surface area with a value of % 504 m 2 g À1 ,f ollowedb yt he binder with av alue of % 194 m 2 g À1 .K aolin is av ery dense materialw ith as mall surface area ( % 21 m 2 g À1 )a nd an egligible pore volume. The binderc ontains mesopores,w hile the zeolite contains both micropores and mesopores. Upon the mixing of components, spray-drying, and subsequentc alcination, some interesting matrixe ffects are observed. The ZeBi sample contains al arger surfacea rea than expected. This is mainly accompanied with a larger micropore volume.T his meanst hat the silica/boehmite bindera nd zeolite component interacttogether during synthesis conditions and create additional micropores,r esulting in a larger surface area.A no pposite effect occurs upon the mixing of the binder and clay component.
The resulting BET surfacea rea is lower than expected ( % 84 m 2 g À1 in comparison to the expected value of % 108 m 2 g À1 ). Also, the total pore volumed ecreases.W ea scribe this to the dense clay component filling and blocking the larger pores and channels. This is in accordance with the observedd ecrease in physisorbed CO, as evidencedb yt he 2139 cm À1 FTIR peak in Figure 2c.I nt he fully shaped FCC material, we observe similarp atterns as observed for ZeBi. This indicatest hat the clay does not interferew ith the zeolite-binder interactions. The increase in surface area due to zeolite-binder interactions in combination with ad ecrease in surface area due to binder-clay interactions, therefore, gives an et result of as urface area increase.

Matrix effects on structural properties
The acidic and accessibility properties of an FCC catalysta re considerably influenced by matrix effects, as clearly indicated by the results from the NH 3 -TPD and CO FTIR spectroscopy measurements, pointing towards effects on acidity,a sw ell as by the resultsf rom Ar physisorption, pointing towards effects on accessibility.H ere, we will discuss the structural nature of interactions between the different components in aF CC catalyst. [a] Expected value based on the single components. Figure 3s hows aT EM image of am icrotomedF CC catalyst sample in whicht he three different componentsa re highlighted in blue (zeolite), yellow (binder) and red (clay). The recorded TEM images of the three individual FCC catalyst components are also illustrated. The FCC morphology is ac learh eterogeneous mixture in whichd ifferent components can clearly be distinguished. The zeolitec onsists of large crystalline structureso f % 700 nm with visible mesopores. The clay consists of smaller crystallites % 200 nm in different shapes,w hich, when overlaid, are difficult to distinguish as single crystallites. The binder comprises three components with different morphologies, namely the amorphous spherically shaped silica particles mixed with both crystalline and amorphousb oehmite.
To further identify the presence of all components in the FCC catalyst and other two spray-dried samples (i.e., ZeBi and BiC), XRD patterns of all six samples were recorded. The results are summarized in Figure S4 of the Supporting Information. The XRD patterns confirm the presence of crystalline zeolite H-Yi nt he zeolite component, the boehmite in the binder,a nd the kaolinite in the clay component. Furthermore, it can be observed that the method of aqueous mixinga nd subsequent spray-drying does not affect the crystallinity of thesei ndividual FCC components. Indeed, the FCC catalyst reveals an XRD pattern that preserves all relevant XRD peaks from the individual FCC components, suggesting ap hysical mixture. Also, the ZeBi sample and BiC sample seem merely physical mixtures of their respective single components based on the XRD patterns.
Whereass tructuralc haracterization of the FCC catalyst, the ZeBi and BiC sample, and individualc omponents did not demonstrate as ignificant interaction between the components, the study into the acidic properties of these samples revealed clear matrix effects. The main differences between the structural and acidity study concerns the sample pre-treatment. In order to study the acidic natureo fasolid surface, samples require a heat-treatment to remove adsorbed water molecules from the acid sites. At the employed drying temperatures, the samples are subjecttostructural changes.Matrix effects, as demonstrated in the previouss ection, can thus be better rationalized with as tructural study under similar conditions. To that extent, samples were heat-treated at 550 8Cu nder aN 2 flow in af luidized bed reactor for 1hprior to XRD, TEM and MQ-MASN MR measurements to mimic the in situ sample pre-treatment in the acidity studies and establish matrixe ffects in terms of structure-acidity relations.
Effects of heat-treatment on structural properties  Figure S4. The XRD patterns are compared with the FCC as such, that is the FCC catalyst prior to heat-treatment.A ll XRD peaks corresponding to the crystalline zeolite are indicated with at riangle. Alreadya fter 15 min of heat-treatment,a ll XRD peaks corresponding to the binder and clay materials have disappeared and only the zeolite material has preserved its crystallinityuponprolongedh eating of the FCC catalyst.
Figures 4b-d depict the TEM images of the microtomed FCC catalysts before and after heat-treatmenta nd confirm the changing morphology of the FCC catalyst with prolonged  heatingt imes. Whereas different FCC components could still be distinguished in the fresh FCC catalyst, the morphology has become more homogeneous and amorphous upon prolonged heatingt imes. This indicates that not only did the components lose their crystallinity, but they have also interacted with each other and merged in an ew activematrix phase. This is also demonstrated in Figure 5t hat presentst he XRD patterns of the single components before and after heat-treatment. The decrease in crystallinity of the binder (Figure 5b)i s associated with the phase transformationo fb oehmite into galumina,w hich has also been reported in literature. [28,41] The kaolin clay (Figure 5c)i sk nown to transform into amorphous meta-kaolinite upon heatingt o5 00 8C. [42] Zeolite H-Y,o nt he other hand, is capable of preserving its crystallinity upon heating, as indicated in Figure 5a.
The six samples under study were furthera nalyzed with 27 Al MQ-MASNMR spectroscopy before and after heat-treatment to investigate the structural changest hat are accompanied with heat-treatment. Nuclear magnetic resonance (NMR) of quadrupolar nuclei, such as aluminum in zeolites, is ap owerful spectroscopic technique. [43][44][45][46] Due to its ability to provide insights into the local atomice nvironment of aluminum, it is av aluable methodt op robe the structureo ft he materials under study. 27 Al NMR spectroscopyi sp articularly useful in the case of zeolite-type catalysts since it makes it possible to monitort he structural changes that occur in am aterialw hen exposed to differentt reatments. [25,43,[55][56][57][58][47][48][49][50][51][52][53][54] 27 Al multiple quantum (MQ) MAS NMR is especially useful for the study of disordered materials, allowing the acquisition of well-resolved spectra that have an isotropic dimension, free of any anisotropic quadrupolar broadening. [47,[58][59][60][61][62][63] This provides good resolution for the variousA lc oordination environments in the sample, giving insights into the distribution in the NMR parameters and thereby allowing am ore detailed characterization of such materials. Based on the chemicalshift, three regions can be distinguished in 27 Al MQ-MASN MR spectra:0 -20 ppm, 30-50 ppm and 50-80 ppm, which are ascribed to octahedral, penta-coordinated and tetrahedral aluminum coordination environments, respectively,asi so ften assigned in the literature. [47,51,57,58,60,61,64,65] Figure 6d emonstrates the 27 Al 3Q-MAS spectra for the three components before (panel I) and after heat treatment (panel II). Lookinga tt he untreated samples in panel I, the zeolite mainly contains tetrahedral aluminum ( % 63 ppm) in the zeolite Yf ramework, which is the origin of Brønsted acid sites. [6] The binder contains silica and boehmite species, demonstrating mainly octahedrala luminum ( % 13 ppm) with as mall amount of tetrahedral aluminum ( % 58 ppm). Since pure boehmite is known to have merely octahedral aluminum centers, [66] these tetrahedral aluminum speciesm ust be the result from an interaction between boehmite and silica on the interface, creating tetrahedral Al centers in as ilica environment. The clay component is kaolin, which is al ayered mineral,c onnecting tetrahedral silica sheets with octahedral alumina sheets. [67,68]   This is confirmed in the 27 Al MQ-MAS NMR spectra, shown in Figure 6c,w ith one large peak for octahedral aluminum species.
The smaller chemical shift for the Al centers in the binder ( % 58 ppm) in comparison to the zeolite Al centers ( % 63 ppm) indicates ah igherl evel of electron shielding from the Al nucleus, due to as lightly larger coordinationn umber on average. [47] Furthermore, the tetrahedral peak of the binderm aterial is more elongated alongt he "chemical shift (CS) axis", indicating ah igher level of disorderi nt he binderm aterial than in the zeolite material with ar esulting wider distribution in isotropic chemical shifts. [60] This indicates that the tetrahedral Al speciesi nthe zeolitea nd binder materials have different conformations.T he zeolite peak, on the other hand, demonstrates al arger distortion along the F2 axis caused by an asymmetric charge distribution, that originates from the corresponding Brønsteda cidic proton. [47] The octahedral Al peak in the binder also demonstrates as malle longation along the CS axis, but a bigger quadrupolari nteraction can immediatelyb er ecognized from the increased line-broadening. [61] This indicates similaro c-tahedralA lc onformations with reduced symmetry of environment, resulting in the increaseo ft he electric field gradient and, therefore, of the quadrupolar product. [65] Panel II in Figure 6p resents the 27 Al MQ-MASN MR spectra after heat-treatment.The heat-treatmento ft he samples results in the loss of physisorbed water and chemisorbed water from the Brønsted and Lewis acidic sites. This is witnessed with a strong increase in the average quadrupole coupling constant and in the irregular shape of all peaks in the MQ-MAS NMR spectra of the heat-treated samples. Indeed, dehydration of the sample is accompanied by ad ecrease in the symmetry aroundt he Al atoms and,t hus, by an increaseo ft he NMR linewidths. [65] Moreover,w er ecognize higherc hemical shifts for all NMR peaks. This indicates that, overall, the Al centers experience ad ecreased electron density in the first coordination sphere. It is proposed that upon dehydration, the oxygen atoms aroundA lb ecome more polarized. This leads to increasedl evels of electron density deshielding of the Al atoms, causing the increasing chemical shift.
In the zeolite material, we notice an increment of signal corresponding to octahedral and penta-coordinatedA la nd an elongation of the tetrahedral Al peak along the quadrupole induced shift (QIS) axis. This suggestst he partial dealumination of the zeolite framework and consequent transformation of framework Al to extra-framework Al species and demonstrates the reactivity of these tetrahedral Al centers at elevated temperatures. [58] The tetrahedral Al sites are isolated in at etrahedral silica framework and are the origin of Brønsted acid sites. These sites are surrounded with water at room temperature. During heat-treatment, these sites are dehydrated, inducing much more strain in the network because of the electrostatic forces of dehydrated acidic protonst hat tend to affect the local environment. The resulting asymmetric environmento f the Al atoms, therefore, leads to much larger electric field gradients.
The deshielding effect is minimum for the tetrahedral Al sites ( % 63 to % 67 ppm) but more evident for the octahedral Al peak, as opposed to the binder,w here the deshielding is stronger for the tetrahedral Al site, with as ignificant downfield shift of more than 10 ppm. This is ascribed to thep hase transformation of boehmite into alumina upon heating. The appearance of the 77 ppm tetrahedral Al peak is, therefore, at the expense of the octahedral boehmite peak at % 13 ppm and correspondst oa na luminap hase. Since the binder demonstrates Brønsted acidic properties, the tetrahedral Al at % 58 ppm is probablys till present but is hidden under the larger % 77 ppm peak. This peak corresponds to tetrahedral Al in as ilica environment.T he Al coordination of the clay is only octahedralb ut upon heat-treatment, the appearanceo ft etrahedral and especially pentahedrals ites is observed. Furthermore, interestingly, the octahedral resonance shifts upfield after heating, indicating the presence of ah igher electron density,p ossibly the resulto f ad ifferent geometry. Figure 6d emonstrated clearly which Al centers in the individual components become reactive upon ah eat-treatment. A reactionb etween these centers could lead to the matrix effects, as observed in Figures 1a nd 2. Figure 7p resents the 27 Al-sheared 3Q-MAS NMR spectra of the three spray-dried samples before and after heat-treatment.The spectra in Panel I show the different Al coordination environments present in the spray-dried samples before heating. Within the FCC catalyst (Figure 7c), mainly two types of Al species are observed. The small peak at % 60 ppm corresponds to tetrahedral Al species and the larger peak at % 12 ppm corresponds to octahedral Al. [57] Closer inspection of this spectrum reveals that both peaks actually comprise two separate peaks. These individual peaks can be traced back to one of the single components, as observedi nF igure 6. The tetrahedral Al peaks in the FCC catalyst originate from the zeolite ( % 63 ppm) and the binder ( % 58 ppm). The octahedral Al peak also consists of two individual contributions. Both the binder ( % 13 ppm) and the clay ( % 11 ppm) contribute to the octahedral Al species in the FCC catalystmaterial.
The ZeBi sample shows as trong octahedral Al contribution at % 13 ppm with similar spectral features and broadening as observed fort he binder component. The tetrahedral region comprises two separate peaks at % 58 and % 63 ppm, assigned to the binder and zeolite material, respectively.T he BiC sample also seems am erely physicalm ix of the binder andc lay component. The Al species mainly adopt an octahedral coordination with two distinct peaks at % 11 and % 13 ppm, assigned to the clay and binder,r espectively.T he small tetrahedral contributiona t% 58 ppm originates from the binder component.I nterestingly,t he tetrahedral Al peak seems slightly lessd istorted for the FCC catalystt han for the ZeBi sample. Thisi ndicates that the addition of clay to the spray-driedc atalystr esults in the stabilization of the tetrahedral Al species with less induced strain in the particle.
Panel II demonstrates the influence of ah eat-treatment on the structural properties of the spray-dried samples. Interestingly,w hereas the spray-dried samples appeared merely physical mixtures of the individualc omponents, based on the MQ-MAS NMR spectra, this is not the case for the samples after heat treatment. This is in correspondencew ith the XRD and TEM resultsb efore and after heat-treatment and confirms that matrixe ffects come into playa te levated temperatures.
For all three samples, line broadening patterns are similar to what waso bserved for the individualb inder component. This suggestst hat the NMRr esonances in the FCC,Z eBi, and BiC samples are dominated by the binder signal. The factt hat the individual binder component also showed the highest signalto-noiser atio supports this hypothesis, since it indicates that most Al centers originate from the binder component. The downfield shift of the tetrahedral Al speciest hat occurs in all three samples, is ascribed to the phase transformationo f boehmite in the binder to alumina upon heating. Upon heating, we observe that the tetrahedral Al peaks in PanelI Io f Figure7 become strongly dispersed parallel to the F2 axis, meaningt hat its line width is mainly caused by second order quadrupoleb roadening, similara sw as observed for the binder component upon heat-treatment. The tetrahedral Al species from the zeolite materialc annot be unambiguously observed in Figure 7a,c, but are probablyh idden under the large peak from the binder.T his is rationalized by the fact that isolated tetrahedral Al sites in az eolite framework are the origin of Brønsted acid sites, which are clearly present in the FCC and ZeBi samples, as indicatedi nF igure S2 of the SupportingI nformation. The observed high quadrupolar coupling for the zeolite in Figure 6a,i sn ot as evident in the ZeBi andF CC sample.
In contrast, for all octahedral peaks in Panel II of Figure 7, the distribution in the electric field gradient dominates, observing as trong broadening along the QIS axis of the two-di-mensional spectrum upon heat-treatment. [61] This indicates the presence of many different Al species with all slightly different quadrupolar coupling constants, which can be ascribed to the formation of an amorphous matrix.Anew pentahedralA lp eak becomes evident after heat-treatment at 550 8C, which can be either ascribed to the dehydration of octahedral Al atoms inherentt ot he clay or originates from reactive tetrahedral Al in the zeolite that has increased its coordination number after interaction with the binder.
Finally,F igure S5 of the Supporting Information shows the FTIR spectra of the six samples under studya fter differentd ehydration conditions and demonstrates that the zeolite and binder are already completely dehydrated when the desorption temperature reaches 550 8C. Clay,o nt he other hand, requiresafull hour of heat-treatment under vacuum to dry, which is ascribed to the dense structure of the kaolin clay.T his meanst hat the surface groupso fz eolite and binder become reactive an hour before the clay is capable of interacting with otherc omponents.A sm entioned before,t he reactivity of the components is ascribed to coordinatively unsaturated surface species.A tr oom temperature, these surfaceg roups are saturated with water molecules. Therefore, matrix effects in terms of structure-acidity are mostly ascribed to the zeolite-binder interaction, whereas clay plays ad ilution and pore-blockingr ole. It is possible that-in an industrial environment-with prolonged times in the FCC reactor,t he clay will become more active and play am ore important role in the catalytic cracking mechanism.

Structure-Acidity Relationships
From the characterization of the six samples under study, structure-acidity relationships can be derived that aid the understanding of matrix effects in FCC catalysts. It is well reported that the origin of Brønsted acid sites are tetrahedral Al centers in at etrahedral SiO 2 framework, the most famouse xample being zeolites. [6, 11]2 7 Al MQ-MASN MR spectroscopy in Figure 6 indicates that theseB rønsted acid sites are characterizedb ya peak at % 60 ppm. The zeolite possessed tetrahedral Al species with an NMR signal at % 63 ppm, while upon dehydration, there is as mall downfield shift to % 67 ppm since the oxygen atoms around Al become more polarized. The binder also contains Brønsteda cid sites, as characterizedb ya nN MR peak at % 58 ppm.
The interaction between binder andz eolite materials resulted in the formation of additional Brønsted acid sites. Figure6a suggestst hat at elevated temperatures, the zeolite framework can partially collapse due to dealumination, leaving defects in the zeolite framework. In the presence of the binder material, on the other hand, we propose that mobile Al species, originating from the boehmite binder,c an insert into these zeolite framework defects, thereby creating additional Brønsted acid sites. This hypothesis on the matrix effects is supported by the Ar physisorption resultss howing an increasei nm icropores upon the interaction between zeolite and binder.
The extent of elongation along the QIS axis upon dehydration seems to be correlated with the strength of the Brønsted Chem. Eur.J.2020, 26,1 1995 -12009 www.chemeurj.org 2020 The Authors. Published by Wiley-VCH GmbH acid sites. [47] The average Brønsteda cidic strengthi sh igh for the zeolite that demonstrates as ignificant elongation along the QIS axis, whereas this elongation is smaller for the ZeBi and FCC materials. These latter samples indeed contain on average weakerB rønsted acid sites. Ap ossible rationalization for this correlation is the fact that as trong dehydrated acidic proton induces much more strain in the network because of the electrostatic forces than aw eak Brønsted acidic proton does, leadingt oah igher quadrupolarc oupling. Further research, however,m ust be performed to confirm this hypothesis.
In addition to the Brønsted acid sites, the zeolite contains a considerable amount of Lewis acid sites. Since the zeolite component mainly consists of tetrahedral Al centers, this indicates that not all tetrahedral Al is Brønsted acidic.I ndeed, part of the tetrahedral Al sites is probed as Lewis acid sites. Thepostulation that strong Lewis acid sites originate from (distorted) tetrahedral Al sites is supported by the observation that also the binder contains strong Lewis acids ites and ac onsiderable amount of tetrahedral Al sites. Also the clay,a fter activation at 550 8Cf or 1h,d emonstrates the presence of tetrahedral Al speciesa nd Lewis acid sites. Moreover,i ti sr eportedi nl iterature that Al centers can expand their coordination number up to 6l igands. [37] Extrapolating this theory,o ctahedral Al centers should not be acidic as their coordination sphere is saturated, penta-coordinated Al centers are weak Lewis acid sites, and tetrahedral Al centers could be strong Lewis acid sites. Lewis acidity in zeolites is often ascribed to extra-framework structures, such as Al(OH) 2 + ,A l(OH) 2 + ,A l(OH) 3 ,a nd AlO(OH),a lthought he exact structures are still debated. [15,69] In these structures, Al is at rivalent cation, whichi sh ighly acidic.T he presenceo fs trong Lewis acid sites is, therefore, also often observed in combinationw ith more extra-framework Al. [15,69] Literature has also often stated that trivalent Al centers in extraframework species re-coordinate to form extra-framework clusters. [15,69] In these clusters,t he Al centers can be four up to sixcoordinated and will be detected as such with methodology such as 27 Al MQ-MAS NMR spectroscopy. [15,56] Upon the entrance of as trong basic probe molecule, the Al centerd isconnects from the clustera nd coordinates to the probe molecule. As such, it is characterizeda sas trong Lewis acid site. It is postulated, in accordance with the hypothesis in this research work, that as trong Lewis acid site is at rivalent Al centert hat disconnects from at etrahedral extra-framework and that a weak Lewisa cid site is at etrahedral or penta-coordinated Al center that disconnects from an octahedral extra-framework Al.
The strength of Lewis acid sites does not seem to merely depend on the coordination number.F or every sample under study,s trong Lewis acid sites alwaysa ppear to be in the proximity of (non-acidic) hydroxyl groups, as indicated in Figure S2 ci nt he Supporting Information. The acceptance of an electron pair by at etrahedral Al center( strong Lewis acid site) is facilitated by the partial transfer of electron density to nearbyh ydroxyl groups.C onsequently,t he OÀHb ond is strengthened andt he Al center expands its coordination number.W ithout the option to stabilizet he received electron density,h owever,t he Al center is only weakly Lewis acidic. This appearst ob et he case in the ZeBi sample. Although there are many tetrahedral Al sites presenti nt he ZeBi sample, they are either Brønsted acidic ( % 60 ppm) or weakly Lewis acidic ( % 77 ppm). There is no indication of OÀHb ond strengthening in Figure S2 that suggests the presence of strongL ewis acid sites. The proximity of (non-acidic)h ydroxyl groups to Lewis acid sites has been observed before and is built on this academic literature. [37,[70][71][72][73] In particular, Lewis acidic extra-framework structures, such as Al(OH) 2 + ,A l(OH) 2 + ,A l(OH) 3 ,a nd AlO(OH), are all Lewis acidic Al centers in the proximity of hydroxyl groups. [15,69] Figure 2d shows that Brønsted acid sites are createdu pon binder-zeolite interactions at the expense of strong Lewis acid sites. This suggests that tetrahedral Al centers near the edge of the zeolite domain are often strong Lewis acid sites. Upon heat-treatment, these centers are extracted from the zeolite framework and replaced with Al from the binder material. Binder-zeolite interactions lead to the dehydroxylationo ft he nearbyh ydroxyl groups. As such, theser enewedA lc enters are no longers trong Lewis acid sites, but new Brønsted acid sites. The remaining strong Lewis acid sites in the FCC catalyst are inherentt ot he clay.T hese sites are protected from interaction as they are hidden in the poorly accessible clay,w hich is only dehydrated after afull hour of heat-treatment.

Conclusions
This work has evaluated the matrix effects in af luid catalytic cracking (FCC) catalysti nt erms of structure, accessibility,a nd acidity. An extensive characterization study into the structural and acidic properties of the FCC catalyst, its individual components,a nd samples containing only two components (zeolitebindera nd binder-clay) was performed at relevant conditions. This allowed drawing conclusions about the role of each individualc omponent, describing their mutualp hysicochemical interactions, establishing structure-acidity relationships, and determining matrix effectsinF CC catalyst materials.
The most important matrix effects are schematically illustrated in Figure 8. It is important to note that the observed matrix effectsi nt his work rely heavily on the structurala nd acidic properties of the individual FCC catalyst components. This meanst hat, for instance, the use of ad ifferent alumina source in the binder or ad ifferent zeolite material (e.g.,z eolite ZSM-5 replacing zeoliteY )c an lead to different mutual interactions and different acidic properties of the FCC catalyst. The preparation of as pray-dried FCC catalystc onsisting of zeolite H-Y, kaolin clay,a nd as ilica/boehmite binder results in ap ractically physicalm ixture of the individualF CC components.T his was demonstrated with TEM, XRD, and 27 Al MQ-MASN MR analysis. The only observed effect in the spray-dried catalyst was the interaction betweens ilica and boehmite. 27 Al MQ-MASN MR spectroscopy demonstrated ap eak at 58 ppm, assigned to tetrahedral Al in aS iO 2 environment.
Each individual FCC component contributes to ac ertain extentt ot he acidic properties of the fully shaped FCC catalyst particle. The zeolite contains the highest amount of Brønsted and Lewis acid sites. These sites originate from tetrahedral Al centers in the tetrahedral zeolite framework. The silica/boehmite binder mainly contains octahedral Al and as mall amount of tetrahedral Al in as ilica environment, providing minimal Brønsted aciditya nd strongL ewis acid sites. The boehmite in the binder material is converted into alumina at elevated temperatures that introduces weak Lewis acidity in the FCC matrix. The clay material appearsi nert, as it preserves al ot of water inside its small pores at high temperatures that blocks acid sites. It, therefore, mainly fills the large pores and blocks ap art of the accessibilityt ot he acid sites, hereby fulfillingi ts role as ad iluent. After longer heating times, however,t he octahedral Al sites in the clay mineral convert into tetrahedral and pentacoordinated Al, thereby creatingB rønsted and Lewis acid sites.
Upon temperature treatment, however,s ignificant matrix effects come into play.F irst, an interaction between binder and zeolite occurs that relies on highly reactive Lewis acid sites on the binder-zeolite interface. We have observed the disappearance of strongL ewis acid sites and the creationo fn ew Brønsted acid sites as ar esult of this interaction. This was ascribed to mobile Al species inherent to the binder materialt hat are inserted into zeolite defects, most probably at the outer layers of the zeolite particles. Secondly,t he interaction between zeolite and bindera lso results in as ignificant increase of the microporef raction. This supports the hypothesis that mobile Al speciesf rom the binder insert in zeolite defects, restoring the zeolite framework. Thirdly,t he interaction between binder and clay leads to as ignificantr eduction of accessibility, as they merge together forminga na morphous matrix consisting of alumina, silica, and metakaolin. The clay is definitely not inert, as it contains somes trong Lewis acid sites that are preserved after 1hof heat-treatment, but it does not chemically interact with other FCC components. Therefore, the clayd oes not play ar ole in the acidic matrix effects, which are dominated by the zeolite-binder interaction.
We have demonstrated in this work that even as hort exposure of the spray-dried samples to typical riser temperatures already inducess ignificant modificationsi nt heir properties. It must be noted that the state and properties of FCC catalysts, after multiple reaction-regeneration cycles might be considerably different compared to this 1h heat treatment. Sintering, coking,m etal deposition, ageing etc. will certainly induce severe modificationsint he catalyst.
Finally,w ew erea ble to correlate structural characteristics to acidic properties in the samples under study.ABrønsted acid site is characterized by ap eak around 60 ppm in the 27 Al MQ-MAS NMR spectra, corresponding to at etrahedral Al site in a tetrahedral SiO 2 environment. The strengtho ft he Brønsted acid site is signified by the extent of elongation alongt he QIS axis upon dehydration, as this is am easure of the strain in the zeolitef ramework because of the electrostatic forces caused by an acidic proton. AL ewis acid site is an Al center that can still accept electron density,t hereby expanding its coordination number.B ased on the described results, it is postulated that the strength of the Lewis acid site dependso nt wo factors. The first parameter is the coordination number,i ndicated by the downfield shift in 27 Al MQ-MASN MR data. An Al center can coordinate up to six ligands. Therefore, al ower coordination number generally indicates ah igher Lewis acidic strength. The second factor involves the facilitation of electron pair acceptance. If an Al center is in close proximity to ah ydroxyl group, it can partially transfer the accepted electron density to this hydroxyl group upon coordinating to an additional ligand. These hydroxyl groups can be detected with FT-IR spectroscopy.

Experimental Section Materials
The spray-dried fluid catalytic cracking catalyst (further denoted as FCC), the corresponding individual components (further denoted as zeolite, binder,a nd clay), and the spray-dried samples containing only two components, namely zeolite and binder (further denoted as ZeBi)a nd binder and clay (further denoted as BiC) were provided by Albemarle Corporation. The FCC catalyst contains the three individual components (i.e.,z eolite H-Y,b inder (silica/boehmite binder) and kaolin clay) in a1 :1:1 weight ratio. The combined spray-dried model samples ZeBi and BiC contain the two individual components in a1:1 weight ratio. The six catalyst samples were used as received for X-ray diffraction (XRD), transmission electron microscopy (TEM) and multiple quantum magic angle spinning nuclear magnetic resonance (MQ-MAS- Figure 8. Survey of the experimentally observed matrix effects in fluid catalytic cracking (FCC) catalysts:The spray-dried FCC catalyst particle consisting of az eolite H-Y,kaolin clay,and as ilica/boehmite binder,ism ainly ap hysical mixture of its individual components.O nly within the binder,the boehmite and silica demonstrate some interaction as evidenced by thep eak at 58 ppm assigned to tetrahedral Al centers, responsible for Brønsteda cidity in the binder.Upon heat-treatmentfor 1h,h owever,significant matrix effects come into play.T he schematic indicates the matrix effects, as revealed by the used characterization toolbox. FTIRs pectroscopy of adsorbed CO on the ZeBisample indicates the removal of strong Lewis acid sites and creation of Brønsteda cid sites. It is proposed to be the result of Al insertion from the binder into the zeolite framework. Transmission electronmicroscopy (TEM) and X-ray diffraction (XRD), in combination with Ar physisorption, revealedt hat the bindera nd clay lose their crystallinity upon heat-treatment, and together form an amorphous matrix withr educed accessibility. This is schematically depicted as ar ed oval (amorphous alumina, instead of crystalline boehmite) that overlays with the partially deconstructed kaolin. Furthermore, advanced nuclear magnetic resonance (NMR) spectroscopy shows the phase transformationo fb oehmite into alumina in the binder material. NMR) measurements. Next, all samples were subjected to ah eattreatment at approximately 550 8Cu nder ac ontinuous N 2 flow of 100 mL min À1 in aq uartz calcination tube. The samples were, subsequently,t ransferred to aN 2 glovebox, to prevent air or moisture exposure after the heat-treatment. Sample preparation for Ar physisorption and MQ-MAS-NMR measurements of the heat-treated samples took place inside the glovebox and samples were transferred under inert atmosphere to the corresponding setups.

Characterization
Fourier transform infrared (FTIR) spectra were recorded in transmission mode on aP erkinElmer 2000 instrument, equipped with a DTGS detector,u sing 32 scans per spectrum and ar esolution of 4cm À1 .S ample preparation took place by pressing approximately 15 mg into as elf-supported wafer that was subsequently placed in aw ell-sealed cell with CaF 2 windows that allows switching between vacuum and the probe molecule gas. Samples were dried at 550 8C( ramp of 10 8Cmin À1 )u nder high dynamic vacuum and kept at that temperature for 1h.C O( 10 %i nH e, Linde Gas Group, purity 99.9 %) was dosed at low temperatures (À188 8C) and at low pressures with spectra being taken after each pulse. CO desorption occurred through vacuum desorption. NH 3 temperature programmed desorption (TPD) experiments were performed on aM icromeritics ASAP2920 apparatus equipped with aT CD detector.T ypically,0 .1 go fs ample was dried in situ in aH e flow with at emperature ramp of 10 8Cmin À1 up to 550 8Ca nd remained at that temperature for 30 min. Subsequently,t he sample was cooled to 100 8C; at this point, NH 3 pulses of 25.17 cm 3 min À1 were applied. After saturation of the acid sites with NH 3 ,t he sample was outgassed for 2h at 100 8Ct oe nsure the removal of physisorbed NH 3 .T he sample was then heated to 550 8Cw ith a ramp of 5 8Cmin À1 to induce desorption of NH 3 .
X-ray diffraction (XRD) patterns were recorded on aB ruker AXS Advance D8 apparatus, equipped with Co Ka radiation, operating at 45 kV and 30 mV.T he XRD patterns were collected between 20-808.T he samples were prepared outside of the diffractometer.I t was assumed that the crystallinity of the heat-treated samples would not change upon exposure to air.
Transmission electron microscopy (TEM) measurements were performed on aF EI Ta losTM F200X instrument. The FCC particles were embedded in an Epofix embedding resin prior to sectioning on a Reichert Jung UltraCut Em icrotome to 70 nm sections that were placed on 200 mesh copper grids with carbon coated Pioloform film. The individual components were directly placed on the grids.
Ar physisorption was performed at À196 8Cu sing aM icromeritics TriStar instrument. The mesopore volumes (the 2-300 nm range) and Barrett-Joyner-Halenda (BJH) pore size distributions of the silica support and solid activators were determined using the adsorption branch of the isotherm with Aerosil 380 as ar eference. Samples were heat-treated as described above and prepared in the N 2 glovebox prior to transfer to the Ar physisorption set-up.
Magic angle spinning nuclear magnetic resonance (MQ MAS NMR) experiments were performed at 11.7 TonaB ruker Avance III spectrometer equipped with a3 .2 mm MAS probe. Spectra were recorded at ambient temperature using 18 kHz MAS. Ar adio frequency (RF) field of 50 kHz was used for the 27 Al p/12 pulse followed by 6.5 ms acquisition. 128 scans were accumulated using an inter-scan delay of 1s.T he 27 Al chemical shift was externally referenced to an aluminum nitrate solution (Al(NO 3 ) 3 (aq)) in milliQ water.T he 1D spectra were processed using al ine-broadening of 100 Hz. Az ero-quantum (ZQ) filtered multiple-quantum magic angle spinning (MQ-MAS) pulse-sequence [59] was used to correlate the 27 Al isotropic chemical shift (F1) with the quadrupolar lineshape (F2), specifically the 3Q-MAS sequence. The RF field for the 3Q excitation pulse was 50 kHz, instead for the soft, selective pulse 3.5 kHz was used. Ar ecycle delay of 1s and acquisition times of 6.5 ms was used for the direct dimension. For the non-heated samples an acquisition time of 1.7 ms was used in the indirect dimension, the MQ-MAS-NMR spectra were recorded using 948 scans and spectral processing was performed using 100 Hz line broadening in both 27 Al dimensions. For the heat-treated samples an acquisition time of 0.9 ms was used in the indirect dimension, the MQ-MAS-NMR data were recorded using 1440 scans and spectral processing was performed using 250 Hz line broadening in both 27 Al dimensions. MQ-MAS-NMR data were Fourier transformed and sheared using the software of Bruker To pspin 3.5. Heat-treated samples were prepared in aN 2 glovebox. After heat-treatment, samples were transferred into 3.2 mm rotors inside aN 2 glovebox, closed with an airtight Te flon cap, and, subsequently,t ransported under N 2 atmosphere to the NMR spectrometer.A 1 H-NMR spectrum was recorded before and after the experiment to ensure the rotor was airtight.