A Ziegler-type spherical cap model reveals early stage ethylene polymerization growth versus catalyst fragmentation relationships

Polyolefin catalysts are characterized by their hierarchically complex nature, which complicates studies on the interplay between the catalyst and formed polymer phases. Here, the missing link in the morphology gap between planar model systems and industrially relevant spherical catalyst particles is introduced through the use of a spherical cap Ziegler-type catalyst model system for the polymerization of ethylene. More specifically, a moisture-stable LaOCl framework with enhanced imaging contrast has been designed to support the TiCl4 pre-active site, which could mimic the behaviour of the highly hygroscopic and industrially used MgCl2 framework. As a function of polymerization time, the fragmentation behaviour of the LaOCl framework changed from a mixture of the shrinking core (i.e., peeling off small polyethylene fragments at the surface) and continuous bisection (i.e., internal cleavage of the framework) into dominantly a continuous bisection model, which is linked to the evolution of the estimated polyethylene volume and the fraction of crystalline polyethylene formed. The combination of the spherical cap model system and the used advanced micro-spectroscopy toolbox, opens the route for high-throughput screening of catalyst functions with industrially relevant morphologies on the nano-scale.


X-ray Photo-electron Spectroscopy 28 29
The La 3d5/2, La 4p and Cl 2p X-ray photo-electron spectroscopy (XPS) data of the three reference 30 materials, namely La2O3, LaOCl and LaCl3, are given in Figure S1. Li et al., have shown in an in-situ 31 study on the energy calibration of Lanthanum compounds with XPS that the component splitting (ΔE) 32 of the La 3d5/2 signal, which in principle can be used for chemical diagnostics, is highly prone to the 33 measurement conditions and surface chemistry [1]. They found, for instance, that the ΔE value of an 34 as-prepared La2O3 material changed from 3.6 eV to 4.3 eV after vacuum treatment at 800 °C, which 116   Figure S3. Raman micro-spectroscopy to determine the bulk chemical phase of the LaOCl spherical cap 150 model system. In blue and black, the Raman spectra are given for respectively the LaOCl powder reference that 151 was also used for XPS studies and a clean Si(100) wafer substrate. In red, the Raman spectrum is given for the 152 LaOCl spherical cap model system, where the stars denote peaks belonging to the LaOCl chemical phase based 153 on the blue spectrum and the octothorp denotes peaks belonging to the Si(100) wafer substrate.

154
In Figure S4, the Raman spectra are given for the LaOCl spherical caps after specified (0,1,2,5,10,20 155 and 60) min of ethylene polymerization, as well as the Si (100)  183 but with TEAL after 60 min of ethylene polymerization and a HDPE reference film. In all cases, normalization was 184 performed using the Si (100) background 521 cm -1 peak as an internal standard.

185
The strength of Raman micro-spectroscopy is the ability to differentiate and map chemical phases 186 over the region of interest. In this case, it allows us to visualize the distribution of polyethylene on the 187 LaOCl spherical caps at different polymerization times. In Figure S5, the Raman micro-spectroscopy 188 maps of the 2700-3100 cm -1 are given, which visualize the asymmetric and symmetric -CH2-stretching 189 modes of polyethylene. Unlike the normalized spectra in Figure S4, using the Si(100) 512 cm -1 190 substrate peak, these maps aren't normalized since this internal standard vibration fell outside of the 191 imaged region. The inset in each Raman micro-spectroscopy map is the correlated optical microscopy 192 image that shows the contours of the LaOCl spherical caps imaged. As shown in Figure S5, the     film. In all cases, normalization was performed using the Si (100) background 521 cm -1 peak as an internal standard.

245
In Figure S7 the Raman maps are given of the 2, 10 and 60 min ethylene polymerized reference

246
MgO/MgCl2 caps. The 2700-3100 cm -1 region is portrayed that visualizes the asymmetric and symmetric 247 -CH2-stretching modes of polyethylene. Whereas only a weak signal for polyethylene is observed for 248 the 2 min polymerized sample, which is in agreement with the spectrum given in Figure S6

283
A selection of mass fragments is shown in Figure S9. Based on the ion generated SE images, the 284 polyethylene phase can be clearly distinguished from the LaOCl spherical caps. It should be 285 mentioned here that the ion counts for the LaOClfragment is typically orders of magnitude higher than 286 for polyethylene fragments such as C9H13 -. This may be related to charging effects and the low 287 ionisation probability of polyolefins, making them particularly challenging to be measured with ToF-288 SIMS. Additionally, the sample geometry (spherical caps with rough surfaces after polymerization 289 instead of planar systems) could also cause adverse effects here. The C9H13mass fragment was 290 chosen here to represent the polyethylene phase as based on the work by Kern et al. [5]. This 291 polyethylene mass fragment is also found for the pristine sample, but with considerably lower ion 292 counts or intensity and is thus more likely to be from adventitious carbon due to air exposure of the 293 samples. Nevertheless, as is mostly visible for the 2 and 10 min maps, the C9H13mass charge 294 fragment shows strong correlation with the polyethylene phase visible in the SE images.
The integrated intensity for this fragment, found in the different mass spectra, is plotted in Figure S10.

296
As the polymerization time increases, more of this fragment is found (note that all measurement points 297 have the same sputter time) until after 10 min a saturation effect starts to appear. This is in line with 298 the AFM analysis performed in Figures S11-13       The total volume of all features measured, which represents the LaOCl spherical cap and polyethylene 416 phases, minus the estimated volume of the sole LaOCl spherical cap as described above, results in 417 the estimated volume of polyethylene (per micrograph). This volume was plotted in Figure S13 for all 418 micrographs recorded as shown in Figure S11, and for all full micrographs recorded with PiFM as 419 shown in Figure S15 (vide infra), to visualize the increase of polyethylene over time ignoring the initial 420 pore volume of the catalyst which might be filled up at the early stages.

484
In Figure

504
The extracted point spectra from the three markers, shown in the 1461+1471 cm -1 PiFM map in Figure   505 S15 for the 5 min ethylene polymerized sample, are given in Figure S16.

520
The spectra shown in Figure 3c were obtained after averaging all spectra obtained on 9 different 521 patterns per polymerization time. The spectra were pre-processed by applying a Whittaker baseline 522 correction and normalization using the PLS Toolbox of Eigenvector. All individual spectra were then 523 subjected to a multivariate curve resolution (MCR) analysis with non-negativity constraints, in which 4 524 spectral components were generated, as shown in Figure S17.  Table S2 for 551 both the MgO/MgCl2 and LaOCl system (also shown in the main manuscript Figure 3c).

552
Noteworthy, we see a similar saturation of the fraction of crystalline components for both systems.

553
However, the saturation for the Mg system seems to happen at lower ratios. This might be related to the 554 difficulty experienced while measuring the MgO/MgCl2 system: the caps were very irregular due to 555 hydration under ambient conditions leading to frequent presence of PiFM tip artefacts that can influence 556 the quality of the spectra obtained. This becomes clear in Table S2 with the larger error for the 557 MgO/MgCl2 system, especially at earlier polymerization times where less polyethylene is formed and 558 more MgO/MgCl2 is exposed. It should be noted that these percentages represent a ratio of MCR  In Figure S20 an overview is given of the SEM results on the reference MgO/MgCl2 caps for a pristine, 614 2, 10 and 60 min ethylene polymerized samples. From left to right, full single caps are first shown in a 615 top-down fashion followed by a zoom-in and then repeated as cross-sectional images on the exact 616 same caps with again a full size image and a zoom-in. The pristine MgO/MgCl2 caps, which have been 617 exposed to moisture due to the absence of a transfer chamber for the SEM, shows a rough appearing 618 surface with some large exposed cracks and seems to bleed onto the Si (100)

697
The SEM images and EDX maps, shown in Figure

725
The SEM and EDX images, shown in Figure S24, belong to the cross-sectional regions of the pristine,

780
In Figure S26, the cross-sectional SEM-EDX images are shown for the pristine, 2 and 60 min ethylene 781 polymerized reference MgO/MgCl2 caps. The pristine sample shows that the extent of MgO chlorination 782 seems to be uniform from the external surface towards the interface with the Si(100) substrate and 783 therefore TiCl4 pre-active sites could potentially speaking be chemisorbed also uniformly throughout the 784 MgO/MgCl2 volume (on exposed, unsaturated lattices that is) all though again the Ti weight loading 785 seems to be below the detection limit of these EDX experiments.