Influence of Metal‐Alkyls on Early‐Stage Ethylene Polymerization over a Cr/SiO2 Phillips Catalyst: A Bulk Characterization and X‐ray Chemical Imaging Study

Abstract The Cr/SiO2 Phillips catalyst has taken a central role in ethylene polymerization since its invention in 1953. The uniqueness of this catalyst is related to its ability to produce broad molecular weight distribution (MWD) PE materials as well as that no co‐catalysts are required to attain activity. Nonetheless, co‐catalysts in the form of metal‐alkyls can be added for scavenging poisons, enhancing catalyst activity, reducing the induction period, and tailoring polymer characteristics. The activation mechanism and related polymerization mechanism remain elusive, despite extensive industrial and academic research. Here, we show that by varying the type and amount of metal‐alkyl co‐catalyst, we can tailor polymer properties around a single Cr/SiO2 Phillips catalyst formulation. Furthermore, we show that these different polymer properties exist in the early stages of polymerization. We have used conventional polymer characterization techniques, such as size exclusion chromatography (SEC) and 13C NMR, for studying the metal‐alkyl co‐catalyst effect on short‐chain branching (SCB), long‐chain branching (LCB) and molecular weight distribution (MWD) at the bulk scale. In addition, scanning transmission X‐ray microscopy (STXM) was used as a synchrotron technique to study the PE formation in the early stages: allowing us to investigate the produced type of early‐stage PE within one particle cross‐section with high energy resolution and nanometer scale spatial resolution.

. Used energy profile for the scanning transmission x-ray microscopy measurements along the C K-edge. Table S2. Used energy profile for the scanning transmission x-ray microscopy measurements along the Cr L2,3-edge. Table S3. Used energy profile for the scanning transmission x-ray microscopy measurements along the O K-edge. Figure S1 schematically illustrates the principles in scanning transmission x-ray microscopy (STXM), in which incident X-ray beams were focused by a 37nm gold zone plate, passing through the Order Sorting aperture which filtered zero order, unfocused light, and subsequently passing through the sample and reaching the detector. B) The materials under investigation consisted of 100 nm thick coupes of the pre-polymerized catalyst particles in Struers Epofix Epoxy resin deposited on uncoated Cu TEM grids, which were loaded on the STXM holder and used as such. C) Scanning Electron Microscopy (SEM) image of the microtomed coupes on the TEM grid. Furthermore, optical density (OD) image at the oxygen K-edge, 580 eV, reveals two distinct materials: epoxy resin and SiO2 which are identifiable in the clusters after Principal Component Analysis (PCA) in combination with the respective XANES of the two clusters. Figure S3. Normalized Epoxy reference XANES measured at ALS Beamline 11.0.2 and used, as such, for subtraction from the combination XANES by LCA. The spectra are offset for clarity. Figure S2. A) Non-normalized XANES of the clusters at the Oxygen Kedge. Demonstrating that the Cr L2,3 edge lies on the continuum of the O K-edge. B) Magnification of the Cr L2,3 edge in the 570 -620 eV region. The blue XANES corresponds to the background region: that is the region without catalyst material. The red and green lines correspond to catalyst material regions: the averaged spectrum in Figure 5H is extracted from this. Figure S4. Illustration describing all individual steps of the data processing. I) Data analysis started with the raw data obtained from the beamline. It was opened in the aXis2000 [1] software package and it was converted into a .ncb file. Subsequently, the "Jacobsen Stack Analyze" function was used for aligning the images in the stack based on each previous image. The aligned stack was saved as a .ncb file. Hereafter, the "Stack Analyze" function in the aXis2000 software package was used to open the aligned stack, where the regions of interest (ROI) of the void space (background, I0) and sample (rest, I) were selected and the stack was converted to Optical Density (OD). The stack was saved, along with the individual images at each energy point (.tif) and a Bulk XANES spectrum.

II)
The previously generated image sequence (.tif) was loaded in TXM Wizard. Noise filtering is performed based on the edge jump and normalization quality of every single pixel XANES to remove pixels that show insufficient signal to noise region. A detailed explanation is given in Y. Liu, F. Meirer et al. [2] Hereafter, PCA and K-means clustering was performed to pool clusters based on spectral similarities. The number of Principal Components was selected based upon inspection of the scree plot, the eigenspectra and eigenimages. Hereafter, K-means clustering was performed using a number of clusters that was initially based on the number of PC kept and refined upon inspection of the obtained XANES of the generated clusters.
III) The clustered image was now generated, along with the C K-edge XANES of the individual clusters. This allowed to localize both pure epoxy components as well as mixed phases in the Field of View. Hereafter, a normalized reference XANES of the pure epoxy resin was loaded and compared to the XANES of the clusters. The pure epoxy resin contains a pre-edge feature at 285 eV that is absent for PE materials. Therefore, it can be used as a quantitative marker for the presence of epoxy in each pixel by inspecting the magnitude of this feature in each normalized single-pixel XANES. This reference for pure epoxy resin was weighted by the magnitude of the feature at 285 eV in the XANES of each pixel and subtracted according to the formula: Where index i indicates the pixel index, Xi,corr the corrected XANES, Xi the uncorrected XANES, R the epoxy reference, and wi the weight for pixel i based on the magnitude of the 285 eV feature recorded for that pixel and scaled between 0 and 1; wi =1 indicates a pure epoxy spectrum based on the magnitude of the feature in the epoxy reference and wi=0 the absence of any contribution from epoxy. The effectiveness of this method is confirmed by the fact that all XANES of pixels containing epoxy have been reduced to their baseline and subsequently removed by the edge jump filter.

IV)
This step generated a new image sequence: the pixels that contained pure epoxy were removed and the epoxy contribution in mixed phase pixels has now been removed. Hereafter, the same procedure for normalization and clustering as described above was applied.
V) The clustered image was generated along with the related C K-edge XANES. The C K-edge XANES showed the successful removal of the 285 eV feature as well as pure PE components.      Differential Scanning Calorimetry (DSC) was performed on a TA Instruments DSC Q20 with 1-2 mg of the nascent material. Each ample was heated from -40 °C to 200 °C at a rate of 10 °C min -1 after which it was briefly held isothermally at 200 °C. Subsequently the cooling cycle was initiated to -40 °C at a rate of 10 °C min -1 followed by an additional heating cycle to 200 °C at a rate of 10 °C min -1 . The crystallinities of the materials, as shown in Table S4, were determined assuming ΔHm 0 = 293 J/g for 100% crystalline polyethylene, with the fraction of the measured ΔHm representing the crystallinity of the nascent early stage materials. The residual catalyst masses were not taken into account, resulting in a significant underestimation of the crystallinity. Due to possible variations in PE yields, we refrained from a discussion on the crystallinity in the main text. Yield variations in 1-1.5 g/g already result in 0-50% differences in the determined crystallinity. Figure S11. Differential Scanning Calorimetry (DSC) profiles for the Early-Stage polyethylene Materials produced with A) 1.5 mole equivalents and B) 5.0 mole equivalents of tri-ethyl borane (TEB). The first heating ramp was from -40 °C to 180 °C with 10 °C/min, after which the temperature was kept constant at 180 °C to erase the thermal history. Subsequently the sample was cooled from 180 °C to -40 °C with a ramp of 10 °C/min and the second heating cycle was started from -40 °C to 180 °C with a 10 °C/min ramp.  Figure S12. Differential Scanning Calorimetry (DSC) profiles for the Early-Stage polyethylene Materials produced with A) 1.5 mole equivalents and B) 5.0 mole equivalents of tri-ethyl aluminum (TEAl). The first heating ramp was from -40 °C to 180 °C with 10 °C/min, after which the temperature was kept constant at 180 °C to erase the thermal history. Subsequently the sample was cooled from 180 °C to -40 °C with a ramp of 10 °C/min and the second heating cycle was started from -40 °C to 180 °C with a 10 °C/min ramp.