<?xml-model href='http://www.tei-c.org/release/xml/tei/custom/schema/relaxng/tei_all.rng' schematypens='http://relaxng.org/ns/structure/1.0'?><TEI xmlns="http://www.tei-c.org/ns/1.0">
	<teiHeader>
		<fileDesc>
			<titleStmt><title level='a'>Spatially resolved reaction environments in mechanochemical upcycling of polymers</title></titleStmt>
			<publicationStmt>
				<publisher>Elsevier</publisher>
				<date>10/01/2025</date>
			</publicationStmt>
			<sourceDesc>
				<bibl> 
					<idno type="par_id">10651085</idno>
					<idno type="doi">10.1016/j.chempr.2025.102754</idno>
					<title level='j'>Chem</title>
<idno>2451-9294</idno>
<biblScope unit="volume"></biblScope>
<biblScope unit="issue"></biblScope>					

					<author>Kinga Gołąbek</author><author>Yuchen Chang</author><author>Lauren R Mellinger</author><author>Mariana V Rodrigues</author><author>Cauê de Souza Coutinho Nogueira</author><author>Fabio B Passos</author><author>Yutao Xing</author><author>Aline Ribeiro Passos</author><author>Mohammed H Saffarini</author><author>Austin B Isner</author><author>David S Sholl</author><author>Carsten Sievers</author>
				</bibl>
			</sourceDesc>
		</fileDesc>
		<profileDesc>
			<abstract><ab><![CDATA[Not Available]]></ab></abstract>
		</profileDesc>
	</teiHeader>
	<text><body xmlns="http://www.tei-c.org/ns/1.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xlink="http://www.w3.org/1999/xlink">
<div xmlns="http://www.tei-c.org/ns/1.0"><head>INTRODUCTION</head><p>Plastics have significantly contributed to modern standards of living, but their accumulation in landfills and the environment has become a globally significant environmental concern, <ref type="bibr">1</ref> which could be addressed with a circular economy. Poly (ethylene terephthalate) (PET) is the world's most widely used polyester, with tens of millions of tons produced per year. <ref type="bibr">2</ref> Chemical recycling routes can depolymerize PET into its monomers via hydrolysis, alcoholysis, or glycolysis, <ref type="bibr">[3]</ref><ref type="bibr">[4]</ref><ref type="bibr">[5]</ref> but the limited solubility of PET in liquid reaction media, the need for homogeneous catalysts, and separations issues make improved processes desirable. Recently, a mechanochemical approach <ref type="bibr">[6]</ref><ref type="bibr">[7]</ref><ref type="bibr">[8]</ref> has been demonstrated in which mechanical collisions in ball mills at ambient conditions are used to depolymerize solid PET using solid reactants like NaOH. <ref type="bibr">[9]</ref><ref type="bibr">[10]</ref><ref type="bibr">[11]</ref> Complete conversion of PET to its constituent monomers in minutes has been reported with this approach. <ref type="bibr">11</ref> In addition, the mechanochemistry of polymers has been studied for a long time, <ref type="bibr">[12]</ref><ref type="bibr">[13]</ref><ref type="bibr">[14]</ref> mostly motivated by the desire to understand wear and tear.</p><p>THE BIGGER PICTURE Mechanochemical processes offer a promising approach to integrating solid feedstocks, such as waste plastics and biomass, into chemical plants. The complexity of reaction environments in ball mills has hindered predictive reactor design. Although recent studies highlight the potential of mechanochemistry, they focus on lab-scale reactions without addressing scalability. This work bridges this gap by exploring polymer upcycling through a combined approach: simulations with a material point method model and spatially resolved spectro-microscopic characterization of a single collision. This provides a fundamental understanding of mechanochemical reaction kinetics and enables bottom-up modeling of these processes across different scales.</p><p>An array of driving forces has been invoked in describing mechanochemistry, including solid-solid mixing, <ref type="bibr">15,</ref><ref type="bibr">16</ref> distortion of chemical bonds, <ref type="bibr">17</ref> formation of activated phases, <ref type="bibr">[18]</ref><ref type="bibr">[19]</ref><ref type="bibr">[20]</ref> triboelectric charge separation, <ref type="bibr">21,</ref><ref type="bibr">22</ref> and thermal hot spots. <ref type="bibr">23,</ref><ref type="bibr">24</ref> These potential phenomena are ultimately driven by the transfer of kinetic energy to the solid reactants during collisions with the balls in a mill. The complex and transient nature of these collisions, even when studied as individual events across a variety of systems, <ref type="bibr">[25]</ref><ref type="bibr">[26]</ref><ref type="bibr">[27]</ref> make them extremely challenging to probe, <ref type="bibr">28</ref> thereby limiting optimization of mechanochemical processing of PET and similar polymers.</p><p>In this paper, we show that carefully controlled single-impact collisions with PET films can be combined with multiple spatially resolved analytical measurements and multiscale numerical simulations to provide mechanistic insight into the phenomena associated with the biphasic reaction of NaOH with PET. This combination of techniques provides, for the first time, a comprehensive view of the reaction microenvironments occurring during ball milling of PET, opening avenues for rigorous kinetic descriptions of mechanochemical reactions. In combination with discrete element method (DEM) models, <ref type="bibr">29</ref> which can provide a strength distribution of collision events, these results will enable </p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>MPM modeling of impact pressure and specific kinetic energy distributions within impact craters</head><p>A three-dimensional projection of the crater's deformation colored by the equivalent total strain, and a two-dimensional comparison of the crater's region (A). Profile of smoothed average pressure (GPa) at the moment of maximum compression (B), and profile of the temporal averaged specific kinetic energy (J/kg) (C), along the diameter of the crater resulting from the impacted uncoated PET film by a steel ball at the three impact velocities of 4.6, 4.1, and 3.5 m . s -1 .</p><p>bottom-up simulations of a wide range of mechanochemical processes.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>RESULTS AND DISCUSSION</head><p>Energy transfer during the impact of a milling ball Context for our experiments was developed using simulations with a material point method (MPM) model. <ref type="bibr">30,</ref><ref type="bibr">31</ref> A 101.6-&#956;m-thick PET film was simulated to match the experiments described below. Our MPM simulations solve the coupled thermoelasticity equations over a fully discretized volume representation of the contacting solid ball and PET film. The velocity of the impacting ball was chosen based on the motion of a steel ball inside a 25 mL shaker mill vessel operated at 30.0, 27.5, and 25.0 Hz, respectively, <ref type="bibr">23</ref> and will be denoted in terms of the impact velocity (i.e., 4.6, 4.1, and 3.5 m . s -1 ). At an impact velocity of 4.6 m . s -1 , the MPM simulations predict the formation of a persistent crater with a diameter of 3.0 mm (Figure <ref type="figure">1A</ref>). Impacts with lower velocities of 4.1 and 3.5 m . s -1 produced smaller craters with diameters of 2.4 and 2.1 mm, respectively (Figure <ref type="figure">1B</ref>). For the collision of two elastic bodies, the MPM simulations predict a dynamic mechanical response in the impacted region, where compression/tension as well as shear stress waves radially propagate away from the crater center along the surface and within the interior of the film. The interaction of the stress waves creates localized zones of varying pressure within the crater (zones of high and low pressure or pressure fluctuations) (Figure <ref type="figure">1B</ref>).</p><p>During impact, the stress waves follow a nonlinear evolution due to the changing contact geometry and the elastic response of the film. Figures 1B and 1C depict the spatial average of pressure (at the moment of maximum compression) and temporal average of specific kinetic energy (from the moment of impact to the moment of maximum compression), respectively, as a function of distance from the crater's center. Remarkably, the pressure waves exhibit a substructure that deviates from the Hertzian contact theory with a spatial frequency in the order of hundreds of &#956;m. The dissimilarity of the elastic moduli of the steel impactor and PET film is hypothesized to be at least partly responsible for causing imperfect contact across the growing crater, in turn causing the PET surface to experience non-uniform compressive loads, as seen in Figure <ref type="figure">1B</ref>. In Figure <ref type="figure">1C</ref>, the temporal average of the specific kinetic energy of the film had a higher central peak with higher collision energies, but the spatial structure of the crater led to non-trivial overall distributions of this energy. These MPM results highlight the importance of the contact mechanics in the local partitioning of the impact energy across the crater. The size and shape of the mechanical pressure and shear stress variation (and thus the local elastic strain energy) produce a rich energy gradient landscape, which is shown below to have significant implications for the chemical outcomes of collisions.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Morphology of single-impact craters in PET</head><p>Single-impact experiments were performed on unsupported PET films with and without NaOH coating by dropping a 25-mm steel ball from a well-defined height so that controlled impact velocities were reached. After the collision, the ball bounced to the side (Video S1) so that each spot was only impacted once. Optical micrographs showed that impacts at 4.6 m . s -1 resulted in persistent, nearly circular impact craters on PET films with (Figures <ref type="figure">2A</ref> and <ref type="figure">2C</ref>) and without (Figures <ref type="figure">2B</ref> and <ref type="figure">2D</ref>) NaOH coating. For the uncoated film (PET <ref type="bibr">4.6</ref> ), the crater diameter and depth were 3.30 mm and 175 &#956;m, respectively, whereas for the alkali-coated film (PET NaOH 4.6</p><p>), the diameter and depth were 11%-13% smaller. These crater depths exceeded the original thickness of the PET film (101.6 &#956;m), indicating that the impact resulted in significant plastic deformation (stretching and warping). Visually, the texture of PET inside the craters differs from the original film and shows significant variation across the crater's radial direction. Scanning electron microscopy (SEM) images in Figures <ref type="figure">2C</ref> and <ref type="figure">2D</ref> illustrate the formation of different regions within the craters.</p><p>During the collision, these different regions experienced different forces as the ball initially impacted the center of the crater and transferred some of its kinetic energy, whereas the outer ring of the crater was impacted slightly later by a slower ball and absorbed a lower dose of kinetic energy. The diameters of the entire crater and the central area with visibly stronger transformation were within 5% of the predictions from the MPM model. The observed morphology of these single-impact craters illustrates the complex microenvironments created in ball milling of polymers. The outer ring of the crater for both films remained relatively smooth, without drastic morphological changes. The inner ring and center of the craters was a region of 0.5-1 mm in diameter characterized by cracks and substantial surface roughness, pointing to physical/structural changes in the film due to plastic deformation as well as potential changes in chemical composition. Strikingly, the region with a diameter of &#8764;200 &#956;m in the center of the crater on PET NaOH 4.6 was significantly elevated from the surrounding film (Figure <ref type="figure">2C</ref>, top left). This observation suggests that the PET in this zone temporarily adhered to the milling ball during impact and was lifted up as the ball rebounded. This viscous behavior implies transformation of significant kinetic energy of impact into thermal energy, locally raising the temperature above the melting point of PET (260 o C) 32</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Figure 3. Structural and chemical transformations observed inside the impacted region</head><p>Representation of crystalline lamella and amorphous region of PET (A). Spatially resolved SAXS intensity profiles of PET impacted and uncoated (B). Representation of the phase transition from crystalline to amorphous phase, measured outside of the crater (gray) as well as along the crater diameter of impacted, uncoated PET film (C). Focused ion beam (FIB) cross-section transmission electron microscopy (TEM) image of the outside of the crater collected for PET <ref type="bibr">3.5</ref> (D). Spatially resolved SAXS intensity profiles of NaOH coated, impacted, and then washed PET films (E). Representation of the phase transition from crystalline to amorphous phase, measured outside of the crater (gray) as well as along the crater diameter of impacted, coated PET films (F). FIB crosssection TEM image of, and inside of, the crater collected for PET <ref type="bibr">3.5</ref> (G). Spatially resolved conversion of PET, presented as the drop in the intensity of the band at 1292 cm -1 along the crater diameter of impacted, coated (H) and uncoated (I) films.</p><p>and creating a region where reactive mixing between the PET and NaOH phases is possible.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Spatially resolved amorphization</head><p>Because the reaction of PET with NaOH requires the formation of reactive interfaces, it is essential to characterize structural changes of the polymer. PET is a semi-crystalline polymer consisting of crystalline (ordered) and amorphous (disordered, flexible, and loose) domains with different molecular conformations (Figure <ref type="figure">3A</ref>). The extent to which mechanochemical impacts affect the alignment of polymer chains was mapped spatially with smallangle X-ray scattering (SAXS), whereas changes in short-range molecular conformations (i.e., trans or gauche orientation of adjacent C=O bonds in the ester linkages <ref type="bibr">33</ref> ) were mapped with Raman micro-spectroscopy with a resolution of 65 &#956;m. Although the sample volume of each spectrum (approx. 6 &#956;m 3 ) allows for a highly spatially resolved analysis of gradients of transformations, each spectrum still represents the average of an ensemble of many polymer molecules.</p><p>In SAXS profiles of PET, a peak located at scattering vector (q) around 0.5 nm -1 is associated with lamellar crystalline domains. <ref type="bibr">[34]</ref><ref type="bibr">[35]</ref><ref type="bibr">[36]</ref> The peak position is defined by the average distance between the centers of neighboring crystalline domains known as the long period (L M ; Figure <ref type="figure">3A</ref>). L M decreased from 13.1 nm in the unimpacted film to 12.2 nm in the outer ring of the crater in uncoated PET, indicating a compression or removal of the amorphous domains between the crystalline ones (Figure <ref type="figure">3B</ref>). Compression of crystalline domains is unlikely due to the rigidity of PET crystallites. However, a 12% loss in peak intensity suggests a reduction in crystallinity as the alignment of some of the lamellar features was disrupted. In the center of the crater, the crystallinity was reduced by as much as 75%, indicating that the compact crystalline domain became disordered, which likely made them more accessible to reactants like NaOH. The amorphous domains in the outer ring of alkali-coated PET were compressed to a more limited extent, with L M decreasing to only 12.7 nm while the abundance of crystalline domains was reduced by at least 17% (Figure <ref type="figure">3E</ref>). In the center of the crater, crystalline domains disappeared almost completely for the coated film.</p><p>The intensities of the C=O stretching bands of ester groups in the Raman spectra showed that the number of trans relative to gauche conformers in the crater of uncoated PET decreased by &#8764;50% but without a gradient across the crater (Figure <ref type="figure">3C</ref>). The conformations of ethylene glycol units are very sensitive to pressure. <ref type="bibr">37</ref> In alkali-coated PET, the loss in trans conformers was less significant (Figure <ref type="figure">3F</ref>), which could be due to the conversion of distorted domains with NaOH to generate monomers (see below). Such surface reactions are fairly efficient on the macroscopic scale. <ref type="bibr">11</ref> Amorphization has also been shown to be beneficial for the rate of dual-enzymatic PET depolymerization. <ref type="bibr">38</ref> The contrast between the spatial extent of the structural quantities inferred from Raman mapping and SAXS implies a difference in the mechanisms of molecular-level and domain-level transformations. Although the rotation of ethylene glycol moieties appears to occur readily when the impacted film is stretched, significant destruction of the lamellar structures in the crystalline domains requires the higher energy dose received in the center.</p><p>To further probe the structural reorganization of the lamellar structure, focused ion beam transmission electron micrographs (FIB-TEM) at the center and outside of the crater were collected.</p><p>The PET <ref type="bibr">4.6</ref> samples suffered from extensive structural and textural deformations, which limited the reproducibility of FIB-TEM experiments, so samples impacted at lower energies were investigated. The cross-section of the unimpacted PET film was featureless (Figure <ref type="figure">3D</ref>), consistent with well-aligned polymer chains. In contrast, TEM micrographs from the impacted region of PET <ref type="bibr">3.5</ref> (Figure <ref type="figure">3G</ref>) contained oval features that averaged 40 nm in diameter (Figure <ref type="figure">S14</ref>). A plausible explanation for these features is the formation of coiled oligomeric PET fragments by homolytic bond cleavage (see next section; Figure <ref type="figure">S21</ref>). As these domains exhibit lower alignment compared with the original PET film, they may have increased accessibility of PET for NaOH.</p><p>These features were mostly clustered near the surface but were observed with diminishing abundance to a depth of about 2 &#956;m. These observations suggest that the mechanical impact induced chain cleavage events forming PET fragments (nanoparticles) that tend to coil. This formation of nanoparticles is consistent with the observed reduction of intramolecular crystallinity by 30% and confirms that impact physically disrupts and mechanically cracks PET to create more phase boundaries for NaOH to attack the polymer.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Spatially resolved extent of reaction</head><p>The spatially resolved extent of alkali depolymerization of PET (Scheme S1) within the impact crater was analyzed by Raman mapping. As ester bonds are cleaved in the reaction, <ref type="bibr">39</ref> two diagnostic Raman bands at 1116 and 1292 cm -1 (Figure <ref type="figure">S10</ref>) assigned to ester groups <ref type="bibr">40</ref> were selected to determine the conversion at each point (Figures <ref type="figure">3H</ref>, <ref type="figure">3I</ref>, and <ref type="figure">S16</ref>). For alkali-coated PET, the conversion profile increased nearly symmetrically from &lt;5% at the outer edges to a maximum of 24%-26% near the crater center. In this central region, several new low-intensity Raman bands were observed (Figure <ref type="figure">S15</ref>). These bands are consistent with the presence of disodium terephthalate, <ref type="bibr">41</ref> suggesting the formation of monomeric products.</p><p>Analogous observations for the uncoated PET film indicate that even in the absence of alkali reagent mechanical impact causes chemical degradation in PET to a minor extent. For this film, the intensity of the C(O)-O band decreased by &#8764;4% around the inner area of crater (in Figure <ref type="figure">3I</ref>), equivalent to the rupture of about 1 in 25 ester bonds in the polymer. This morphologically rough region represents &#8764;14% of the crater area. Mechanically induced chain breakage in polymers usually proceeds via homolytic bond cleavage to form chain end radicals. Radicals generated by impact or shear forces may be quite persistent <ref type="bibr">42</ref> but can recombine and lead to rearrangements of the polymer structure or disproportionation to form stable end groups.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>PET transformation as a function of impact energy</head><p>Correlations between the localized energy dose of a spot within the crater, obtained from MPM simulations, and the experimentally observed transformations provide useful insight into the stages of mechanochemical transformations (Figures <ref type="figure">4A-4C</ref>). A limited amount of conversion can be achieved, even with a temporal average of specific kinetic energy below 32 J/kg. This is attributed to the conversion of exposed amorphous domains of PET that are easily contacted with NaOH by applying a minimal force. However, the local conversion reached in this region decreased with decreasing impact velocity. This might be explained by pressure waves that propagate the conditions in the center of the crater to the outer parts. As the depolymerization products are mobile, some of them will be removed from the space between lamellar crystals.</p><p>With a temporal average of specific kinetic energy between 32 and 290 J/kg, no additional conversion was observed. This suggests that the additional energy is absorbed mostly by elastic deformation of the PET film. The lack of a gradient in the crystalline/amorphous peak ratio in Raman spectra across the crater indicates that the near-range order of the polymer can be reduced even during rather weak elastic deformation. However, the lack of additional conversion indicates that additional interfaces between the reactants are not formed in this energy range.</p><p>Once the temporal average of specific kinetic energy of about 290 J/kg was exceeded, conversion in the affected areas increased significantly. This is attributed to the transition between elastic and plastic deformation at the yield point of the PET film. Beyond this point, new interfaces between the reactants readily form at cracks, ruptures, etc.</p><p>The influence of amorphization can be illustrated by correlating it with the temporal average of specific kinetic energy and local conversion (Figures <ref type="figure">4D-4E</ref>). A small dose of energy significantly reduces the alignment of crystalline domains (Figure <ref type="figure">4D</ref>). However, the PET domains appear to remain Article ll compact, with limited access to the ester bond until a higher energy dose causes nearly complete amorphization (Figure <ref type="figure">4E</ref>).</p><p>Once the reactants are in contact, PET is readily depolymerized. <ref type="bibr">11</ref> A similar conclusion can be drawn by plotting the areaweighted conversion of the whole crater as a function of the kinetic energy of the impinging ball (Figure <ref type="figure">4F</ref>). Extrapolating this trend line to zero conversion indicates that a minimum of total kinetic energy of 0.33 J (corresponding to a velocity of 3.2 m . s -1 ) appears to be required for depolymerization to occur. The loss of near-range order did not follow a monotonic trend for the coated film, which could be due to the tradeoff between amorphization and preferential conversion of more disordered domains.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Conclusions</head><p>Our results show that the collisions driving mechanochemical depolymerization of PET create reaction environments with spatial gradients and that impacts above a threshold energy are required for any reaction to occur. Although the reaction readily occurs at interfaces between the reactants, it is necessary to compress PET beyond its yield point to form new inter-faces to a significant extent. Detailed insight of this kind into the productivity of individual collisions will form the foundation for multi-particle DEM simulations of realistic ball mills that will enable mechanochemical reactor design based on fundamental kinetics.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>METHODS</head><p>Detailed methods can be found in the supplemental information.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>RESOURCE AVAILABILITY</head></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Lead contact</head><p>Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Carsten Sievers (carsten.sievers@chbe.gatech. edu).</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Materials availability</head><p>This study did not generate new unique materials.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Data and code availability</head><p>Experimental data and modeling files are available upon request to the authors.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Figure 4. Relationship between the impact energy and transformation of PET</head><p>Correlation between local conversion and temporal average of specific kinetic energy for all three impact velocities (A)-(C). Correlation between local amorphization and temporal average of specific kinetic energy for an impact velocity of 4.6 m . s -1 (D). Correlation between local conversion and local amorphization for an impact velocity of 4.6 m . s -1 (E), the color scheme corresponds to conversions in Figures <ref type="figure">S18-S20</ref>. Correlation between integrated, areaweighted conversion and total kinetic energy (F).</p><p>Please cite this article in press as: Go&#322;a &#731;bek et al., Spatially resolved reaction environments in mechanochemical upcycling of polymers, Chem (2025), <ref type="url">https://doi.org/10.1016/j.chempr.2025.102754</ref> </p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Article</head></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Materials and Methods</head></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Materials</head><p>Poly(ethylene terephthalate) (PET) film measuring 0.004 inch/0.1016 mm total thickness was supplied by Grafix Inc., USA. Sodium hydroxide (&gt;99%), methanol (&gt;99.9%) and dichloromethane (anhydrous, &#8805;99.8%) were purchased from Sigma-Aldrich Inc., USA. 316 and 440C grade stainless steel balls used in impact experiments were purchased from McMaster-Carr, USA.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Sample preparation</head><p>The PET films were coated with a layer of sodium hydroxide (Fig. <ref type="figure">S1</ref>.B). To coat the PET, a piece of film cut to approximately 5 cm by 5 cm was scotch-taped at the corners to a flat piece of acrylic. A square profile o-ring with inner diameter 1.5 cm was placed over the center of the film to expose a specific area for application of coating, then another piece of acrylic (the 'mask') with a circular window of the same dimension as the inner diameter of the o-ring was placed over the o-ring with the exposed areas aligned. The acrylic pieces were bolted together so that the o-ring forms a liquid-tight seal with the PET film. This contraption constituted a 'masked' film sample that was ready to be coated. In the coating of a typical sample, 100 &#956;L of a 50 mg/mL suspension of sodium hydroxide in methanol and 400 &#956;L of dichloromethane (DCM) were dispensed over the exposed circular surface of a masked sample and placed in the fumehood until the solvents had evaporated, leaving a solid alkali coating. The list of samples under study and their preparation details are summarized in the Table <ref type="table">S1</ref>.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Methods</head></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Single drop experiments</head><p>Single impact experiments (Movie S1) with uncoated and NaOH-coated PET films were performed using a custom-built setup (termed the 'ball-drop setup'; Fig. <ref type="figure">S1A</ref>) consisting of an elevated platform of roughly 15 cm by 15 cm made out of a piece of cut acrylic framed tightly by T-slot beams. A DC-powered electromagnet was positioned pointing downwards over the center of the acrylic platform by a pair of heightadjustable T-slot beams (min. to max. height 1.07 to 0.63 m). The acrylic piece also had four clamps located near the corners to tightly secure the sample during impact (Fig. <ref type="figure">S1C</ref>). To impact a coated film sample, the sample was scotch-taped at two opposite ends to a tungsten carbide plate (7 cm x 5 cm x 3 mm), which was then secured at the corners to the acrylic platform of the ball-drop setup via the clamps, with the coated region centered directly below the electromagnet. The electromagnet was turned on, and a 316 or 440C grade stainless steel ball 25 mm in diameter (mball = 63.147 g) was attached to it. Then, the electromagnet was turned off to allow the ball to fall straight down from the desired height, impacting a coated region of the film sample at the end of its trajectory (Fig. <ref type="figure">S1B</ref>). The ball velocities at impact were within the range of collision velocities calculated for mechanochemical depolymerization of poly(ethylene terephthalate) using vibratory ball milling reactors presented in previous studies.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Material Point Method model</head><p>Modeling the single impact collision was performed using material point method (MPM). While some papers on impacts of model polymer-films exists, they all utilize different approaches to study the phenomenon. For example, Giuntoli et al. <ref type="bibr">1</ref> employed coarse-grained molecular dynamics (MD) to study the ballistic impact of thin polymer films. Such MD simulations inherently capture polymer chain motions and nanoscale effects. Zhou et al. <ref type="bibr">2</ref> similarly combined experiments and MD to explore thickness-dependent penetration energy, but they were limited to very thin films and very short times. In contrast, our MPM model will treat the PET film as a continuum, which allows simulation of the 100 &#181;m film and its collision with the ball at experimental scales.</p><p>The continuum assumption, however, has its own limitations. In our model, the polymer is treated as a homogeneous viscoelastic solid, so molecular-scale phenomena (e.g., chain-level plasticity, viscoelastic relaxation, or hot-spot formation) are not explicitly resolved. We neglected effects such as strain-rate dependent behavior, thermal softening, or damage evolution in polymer chains and how they break on a molecular level. Another limitation is numerical artifacts like "cell-crossing instability" that can occur when material points move between grid cells, causing sudden redistribution of mass and stiffness. This can introduce spurious noise in the stress-strain fields unless special interpolation is applied. In practice, we used a sufficiently fine grid and small-time step to minimize these effects, but residual oscillations may remain. In addition, the explicit time integration in MPM requires very small-time increments (defined by the Courant condition), which limits the simulated duration and/or increases computational cost. Finally, our simulation assumed a frictionless ball-film contact and did not model the ball's material deformation; any energy loss in the impact beyond elastic rebound was not captured. To put all this together, the MPM model simplified the real physics, but it did not model chemical reaction kinetics. Therefore, some detailed phenomena from a real mechanochemical collision were outside its scope.</p><p>Despite these limitations, the MPM offers significant advantages for this problem. It is specifically designed to handle large-strain dynamics with evolving contact and free surfaces. Indeed, MPM is suited to modeling large deformation problems where conventional mesh-based methods suffer. In our case, this means we can simulate the ball colliding with and deforming the thin film without mesh entanglement or distortion. The hybrid Lagrangian-Eulerian formulation automatically handles deformation to accommodate any amount of bending, stretching, or even tearing of the film under impact. Compared to traditional finiteelement methods, MPM avoids mesh folding when the film wrinkles or moves. Compared to particle methods (DEM or pure molecular models), MPM efficiently represents the continuum film at the micrometerto-centimeter scale. Thus, MPM captures the macroscopic stress waves and inertia-driven deformation of the film in a physically realistic way. While future work could incorporate more sophisticated polymer physics or adaptive schemes, the current MPM approach provides a practical, physics-informed approximation. It balances computational feasibility with the ability to reproduce the key features of a single-collision impact on a film.</p><p>PicassoMPM is a new material point method application with an emphasis on high-fidelity thermomechanical simulations, developed as part of the ExaAM sub-initiative of the Exascale Computing Project (ECP) by the US Department of Energy (DOE). The core libraries of PicassoMPM's software stack, in increasing levels of abstraction, are Kokkos for performance portability across hardware architectures, the co-designed Cabana library for general particle algorithms and communication, and Picasso for Particlein-Cell (PIC) and MPM motifs. This software design facilitates the development of solvers across various applications, such as rapid prototyping of different particle methods or MPM variants for a different application domain. This design also enables simulation across HPC architectures, including GPU architectures. The code uses an explicit time integration protocol, and it is equipped with three PIC approaches to update particle's velocity, being Polynomial PIC (PolyPIC), Affine PIC (APIC), and Fluid Implicit PIC (FLIP). This work uses the FLIP approach with FLIP to PIC factor of 0.99.</p><p>Particle field data for each simulation was output for every 1 &#181;s, including mass, volume, position, pressure, velocity, stress tensor, strain tensor, and equivalent total strain. The data visualization and postprocessing presented in the next sections (unless mentioned otherwise) were performed using Paraview, several python libraries (pandas, matplotlib, scipy, and numpy), and Microsoft Excel.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Small angle X-ray scattering (SAXS)</head><p>The characteristic lengths of the crystalline and amorphous domains within the PET film were investigated using the small angle X-ray scattering (SAXS) analysis. The experiments were performed in the CATERET&#202; beamline at the Sirius Synchrotron Laboratory (LNLS/CNPEM). A wavelength &#955; = 0.1378 nm and a sample-to-detector distance of 1.7 m were used on the experiments, resulting in a q-range of 0.05-3.05 nm <ref type="bibr">-1</ref> , where q = 4&#960;sin&#952;/&#955;, and &#955; is the X-ray wavelength. A beam of 40 x 40 &#181;m was focused on the sample position. The scattering pattern were collected with a PiMega 540D detector (PITEC). An image of the experimental setup is shown in Figure <ref type="figure">S2A</ref>. The SAXS patterns were obtained measuring a line scan across the impacted regions with a 0.5 mm step. The PET films are showed in Figure <ref type="figure">S2B</ref>, uncoated (top) and coated (bottom) with NaOH layer. Before the measurements, the samples were washed with deionized water. The long period (LM) was calculated using Bragg's equation, where qmax represents the position of the maximum intensity in the SAXS profile.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Focused ion beam (FIB) microscopy</head><p>A 2-4 nm layer of chromium (Cr) was deposited onto the surface to make it conductive. Chromium was chosen because the sample does not contain Cr, and it is easier to form an ultra-thin, homogeneous Cr film compared to gold (Au) or platinum (Pt). After surface examination using scanning electron microscopy (SEM, TESCAN AMBER), three specimens for transmission electron microscopy (TEM) were prepared using a focused ion beam (FIB, TESCAN AMBER). These specimens were taken from the center, near the border, and outside the impact mark of the sample. A standard lamella lift-off process was employed, but with a much lower polishing current of gallium (Ga) ions than the lowest standard settings to minimize potential sample damage. Following successful sample preparation, the TEM lamellas were analyzed using field-emission high-resolution electron microscopy (HRTEM, JEOL JEM 2100F).</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Raman spectro-microscopy</head><p>The physicochemical composition of PET films in uncoated or coated, impacted or unimpacted regions was analyzed using a Renishaw inVia&#8482; confocal Raman microscope equipped with a temperature controlled CCD camera. Considering the size of the impacted regions (the craters were larger than just the objective's field of view), for the creation of crater images, the image montaging option was used at 5x magnification of the objective. Raman spectra were acquired using a 532 nm argon ion laser at 5x magnification of the objective. The laser power at the sample was set to 0.5 % of the total power (50 mW) with 10 seconds of exposure time. Along the crater, the 20 acquisition per spectrum were acquired and each spectrum was collected every 65 &#956;m. The theoretical diffraction spatial resolution and depth of the Raman probe were ca. 5 &#956;m and 2 &#956;m, respectively. Baseline correction and normalization relative to the band at 1615 cm -1 assigned to the C-C stretching vibrations in benzene were applied to each reported spectra using the Windows&#174;-based Raman Environment (WiRE) software.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Supplemental Notes Material Point Method model</head><p>Geometry The simulation's geometrical setup is constructed to follow the experimental setup, in which a 25 mm diameter steel ball is dropped from three different heights of 1.07 m, 0.85 m, and 0.63 m on a 100 &#181;m PET thin film resting on a steel substrate. However, to avoid simulating the free fall of the ball from the actual height, leading to significant save of computational time and resources, the steel ball is given an initial pre-determined velocity, and is placed very close to the PET film. The initial velocity is calculated using &#65533;2&#119892;&#119892;&#8462;, where &#8462; is the height that the ball traveled from the actual drop height to the position at which it is initially placed. The steel ball is placed such that its bottom surface is at 200 &#181;m away from the PET surface as shown in Figure <ref type="figure">S3</ref>. The initial velocity assigned to the ball starting from that position is 4.582 m/s for the drop height of 1.07 m, 4.084 m/s for the drop height of 0.85 m, and 3.515 m/s for the drop height of 0.63 m.</p><p>The simulation domain is discretized into a three-dimensional mesh of single type of equal sized cubic cells across all three directions. Each cell has a side length of 50 &#181;m, and there are 600 cells along X and Y directions, and 530 cells along Z direction. The domain boundaries are set to be rigid walls with no flow through all six faces of the domain. This means that the particle will rebound back into the simulation box upon contact with the simulation domain boundaries. The simulation domain center coordinates are (0, 0, 13.25), and extends 15 mm in opposite directions (negative and positive) along X and Y (total length of 30 mm along each axis), and goes from 0 to 26.5 mm along Z.</p><p>Each component shown in Figure <ref type="figure">S3</ref> is discretized into particles such that each cell in the simulation domain mesh contains 1 particle (1 particle-per-cell) that is placed in the center of the cell. The steel ball has diameter of 25 mm, the PET film is 100 &#181;m thick and has a square area of 20 mm by 20 mm, and the substrate is 1 mm thick and has a square area of 25 mm by 25 mm. The purpose of the substrate is to simulate the actual experiment as close as possible, and to keep the PET thin film away from the simulation domain boundary to avoid any potential boundary effect on the film deformation. This discretization resulted in total number of particles of 70.7 million particles across all three components.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Simulation Parameters</head><p>The simulations were performed on HPC SUMMIT at Oak Ridge National Laboratory (ORNL) as part of the ORNL's Leadership Computing Facility (OLCF). SUMMIT is an IBM system with peak doubleprecision performance of 200 Petaflops. The HPC contains IBM Power9 CPUs and NIVIDIA Tesla V100 GPUs. Each of the simulations was performed using 420 CPUs and 60 GPUs such that every group of 42 CPUs and 6 GPUs are together from one full node. The number of nodes were determined such that the whole simulation is divided into 1 MPI task per every 7 CPUS and 1 GPU, and that the MPI task has ~1,180,000 particles. The simulation timestep was chosen to be 5.0 * 10 -10 &#119904;&#119904;, and the end time was 100 &#120583;&#120583;&#119904;&#119904;. The computation wall time, that is the time needed on HPC to perform the simulation until completion, required for each simulation is 0.861 days (20.673 hours) for drop height of 1.07 m, 0.836 days (20.064 hours) for drop height of 0.85 m, and 0.686 days (16.456 hours) for drop height of 0.63 m. The output files were dumped at a frequency of 2000 timesteps (every 1 &#120583;&#120583;&#119904;&#119904;, total of 100 files per simulation).</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Constitutive Models</head><p>The constitutive model used to calculate the stress at a material point is vital in characterizing the material response and is detailed in this section. Constitutive models appropriate for solids were chosen based on their ease of implementation and relative simplicity; an elasto-plasticity model for the steel ball and the supporting substrate, and a visco-elastic model for the thin film. Below, bold symbols denote that the variable is a tensor, unless mentioned otherwise.</p><p>The stress tensor for a solid material point requires stress rate tensor, which is defined as:</p><p>where &#120634;&#120634;&#775;(&#119961;&#119961;, &#119905;&#119905;) is the strain rate tensor, &#119915;&#119915; is the fourth-order tangent stiffness tensor, and &#120648;&#120648; &#8711; denotes the Jaumann stress rate, which ensures that an objective stress rate is calculated in a rotating material frame of reference. In addition to using an objective stress tensor, the total strain is calculated incrementally over very small substeps (on the order of 10 -7 &#119904;&#119904;) using an explicit solver. At each substep, the incremental total strain is additively decomposed into separate elastic and plastic components such that &#119941;&#119941;&#120634;&#120634; = &#119941;&#119941;&#120634;&#120634; &#119942;&#119942; + &#119941;&#119941;&#120634;&#120634; &#119953;&#119953; . The total strain increment on particle p at timestep k+1 is computed as follows:</p><p>where &#916;&#120634;&#120634; &#119901;&#119901; &#119896;&#119896;+1 , &#120634;&#120634;&#775;&#119901;&#119901; &#119896;&#119896;+1 are the particle's incremental total strain and strain rate at the current timestep, respectively, &#916;&#119905;&#119905; is the timestep, &#119959;&#119959; &#119946;&#119946; is the grid nodal velocity, &#119961;&#119961; &#119953;&#119953; is the particle's position, and &#119873;&#119873; is the shape function.</p><p>The stress is decomposed into hydrostatic (pressure) and deviatoric components. The hydrostatic component for all constitutive models (elasto-plastic and visco-elastic) is determined as:</p><p>(3) where &#119861;&#119861; is the bulk modulus, &#120648;&#120648; &#8462;&#119901;&#119901; &#119896;&#119896; is the particle's hydrostatic stress at the previous timestep, and &#120648;&#120648; &#8462;&#119901;&#119901; &#119896;&#119896;+1 is the particle's hydrostatic stress at the current timestep. The deviatoric component is determined as:</p><p>)</p><p>where &#120648;&#120648; &#119889;&#119889;&#119901;&#119901; &#119896;&#119896;+1 is the deviatoric stress at the current timestep, &#119866;&#119866; is the shear modulus, &#8486; &#119901;&#119901; &#119896;&#119896;+1 is the vorticity tensor at the current timestep, and &#119959;&#119959; &#119896;&#119896;+1 is the velocity vector at the current timestep. The third and fourth terms on the right-hand side of equation 4 ensures that the calculated stress is objective, which is necessary to account for rotation during large deformation problems. Equations ( <ref type="formula">2</ref>) and ( <ref type="formula">4</ref>) for the stress and strain calculations above are general relations without reference to the chosen specific constitutive model (elasto-plastic or visco-elastic). For the PET thin film visco-elastic model, the deviatoric component of the stress is divided by a timescale factor, which is calculated as: An elasto-plastic constitutive model is used for the impacting ball and supporting substrate; in particular, a simplified Johnson-Cook (JC) model 3 appropriate for metals is adopted to determine material yielding at every timestep. The PicassoMPM code employs a simple von-Mises yield criterion with radial return mapping using an iterative Newton-Raphson scheme. Figure <ref type="figure">S4</ref> illustrates the radial return mapping algorithm to determine the plastic strain in a timestep. The yield surface is expressed as a function of the stress and one internal variable that is the equivalent plastic strain, &#120576;&#120576;&#771;&#119901;&#119901;, such that the yield surface &#119891;&#119891;&#65533;&#120590;&#120590;, &#120576;&#120576;&#771;&#119901;&#119901;&#65533; reads as shown in Figure <ref type="figure">S4</ref> and as follows: (&#119889;&#119889;&#119889;&#119889; &#119924;&#119924;: &#119889;&#119889;&#119889;&#119889; &#119924;&#119924;) = &#119889;&#119889;&#119889;&#119889; (9) where &#119891;&#119891; is the yield surface, &#119869;&#119869; 2 is the second invariant of the stress tensor, &#119872;&#119872; is the flow rule, &#119873;&#119873; is the yield surface normal, &#119889;&#119889; is the plastic multiplier, and &#120590;&#120590; &#119869;&#119869;&#119869;&#119869; is the yield stress determined by the Johnson-Cook model [ADDING A REFERENCE HERE] as a function of the equivalent plastic strain such that:</p><p>3 &#65533;&#120576;&#120576; 11 2 &#119901;&#119901; + &#120576;&#120576; 22 2 &#119901;&#119901; + &#120576;&#120576; 33 2 &#119901;&#119901; + 2 &#120576;&#120576; 12 2 &#119901;&#119901; + 2 &#120576;&#120576; 13 2 &#119901;&#119901; + 2 &#120576;&#120576; 23 2 &#119901;&#119901; &#65533; (11</p><p>where &#120576;&#120576;&#771;&#119901;&#119901; is the equivalent plastic strain and A, B, and n are material constants. When plastic yielding is detected during the solution, the radial return mapping scheme is activated such that the plastic strain is calculated for the first iteration, and then refined using the Newton-Raphson scheme until convergence is achieved. The values of all material properties and parameters, as obtained from the literature, are shown in Table <ref type="table">S2</ref>.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>The Instant of Maximum Compression</head><p>To simplify the approach for temporal data analysis, the instant of maximum compression is used as a reference snapshot in time for determining the final crater size and to perform spatial analysis of the data, since it was observed that the crater reaches its maximum deformation and also experiences the largest pressure variation. The instant of maximum compression is taken to be the moment in time for which the steel ball achieves zero velocity and then starts reversing its direction of motion (rebound). The instants of maximum compression after initial impact differed slightly for each simulation and was found to be 47 &#120583;&#120583;&#119904;&#119904;, 50 &#120583;&#120583;&#119904;&#119904;, and 50 &#120583;&#120583;&#119904;&#119904; for drop velocities of 4.6 m/s, 4.1 m/s, 3.5 m/s, respectively.</p><p>Also, due to limitations on computational resources to achieve full separation between the rebounding ball and the PET film, it was not possible to examine the crater at times after the separation. The constitutive models considered here do not account for dedicated contact/separation mechanisms between ball and film surface as well as damage and fracture, thus, they would not be expected to be accurate to assess the film state after the ball-film separation (i.e., after maximum compression).</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Crater Size</head><p>The diameter and extents of the crater region were analyzed at the instant of maximum compression, for which the crater reaches its maximum size. Two independent threshold criteria for defining the crater diameter in our MPM simulations were chosen to reduce systemic bias-kinetic energy and equivalent total strain-with both criteria yielding similar trends over the series of simulations. In each analysis, particles were selected within a 1 mm narrow central band extending over the full length of the impact region along both the X and Y coordinate directions. However, for the second criterion, the strip is taken once along X direction, and another time along the Y direction, and the average of the two is taken to be the crater's diameter. Figure <ref type="figure">S5</ref> shows the per-particle kinetic energy versus position along the X axis. The crater diameter extents, denoted by the red arrows in Figure <ref type="figure">S5</ref>, is determined by finding the least upper bound and greatest lower bound of particle positions for which the kinetic energy is found to start showing an increasing-only trend.</p><p>Figure <ref type="figure">S6</ref> shows similar plots of the per-particle equivalent total strain. In here, we define two threshold values for the equivalent total strain, that is 10% and 20%. The two specific values are reported as the bounds within which the breaking strain of PET fibers are usually determined experimentally in the literature. <ref type="bibr">4</ref> The equivalent total strain is plotted against the position for each drop height, and the extents of the crater's diameter is defined once for the 10% threshold, and also for the 20% threshold. That is why, two diameter values are reported in each subplot shown in the figure. This analysis was done once for a strip that extends along the X-axis and a second time for a strip that extends along the Y-axis. The final reported diameter is the average of the X-diameter and the Y-diameter. The extents of all diameters are noted by the red arrows for all drop heights. Figure <ref type="figure">S7</ref> shows a two-dimensional top view of the craters where particles are colored by the equivalent total strain, and the crater size is highlighted by black circles. This view shows that only some portion of the impact region is the actual crater, and those craters diameters match those obtained from the kinetic energy analysis.</p><p>Multiple experiments were performed for all three drop heights, and craters diameters were measured and found to range between the upper and lower bounds (+/-0.1 mm) similar to those found in the MPM simulations by the equivalent total strain analysis. In other words, all measured experimental values for the case with impact velocity of 4.6 m/s are found to range between 2.6 mm and 3.4 mm, matching the MPM simulation values from the equivalent total strain analysis for the upper and lower bounds (at 10% and 20% thresholds, respectively) for both strips along x and y directions. Table <ref type="table">S3</ref> shows the final reported averaged values from experiment and from both of our analyses here. It is found that the craters diameters are within 3% error compared to experiments.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Gaussian Kernel Density Estimate (G-KDE) and Outlier Estimation</head><p>To simplify the comparison of the transient nature of the deforming impacts and to determine if any pressure values should be considered as outliers, probability densities of the instantaneous pressure throughout the crater were computed at the point in time of maximum thin film compression as shown in Figure <ref type="figure">S8</ref>. This analysis used a Gaussian form of the Kernel Density Estimate (KDE). The kernel density estimate (KDE) is a normalized probability histograms that were spatially averaged and smoothed by convolution with a Gaussian function to determine the pressure variation within the crater's region. A data point with high KDE value suggest that this data point value has higher frequency relative to other values found among all data points.</p><p>where &#119891;&#119891; &#770;(&#119909;&#119909;) is the estimated density at pressure point &#119901;&#119901;, &#119897;&#119897; is the number of data points, and &#8462; is the bandwidth (smoothing) parameter that controls the width of the kernel and it is determined as &#8462; =</p><p>&#119886;&#119886; &#937; &#119899;&#119899; &#119887;&#119887; , where &#119886;&#119886; and &#119887;&#119887; are user inputs to control the bandwidth values, and in turn determine the degree of smoothening desired. &#937; is the standard deviation of the data points.</p><p>The KDE contours shown in Figure <ref type="figure">S8</ref> show data for a crater formed after the impact with 4.6 m/s impact velocity. The region with red has higher frequency than those with other colors. The center of the crater has higher density of blue regions than the ring or the outer side of the crater. Looking at the pressure profiles in the adjacent image and the PDF plot, the red in the KDE contour correspond to the pressure range of -0.30 to -0.7 GPa, and the blue regions in the KDE contours correspond to the pressure range of 0.2 to 1.0 GPa in the pressure profiles. This matches the pressure probabilities shown in the PDF plot, where the range -0.3 to -0.7 GPa have higher probability relative to the range of 0.2 to 1.0 GPa.</p><p>The probability distribution functions of the pressure variation within the crater as shown in Figure <ref type="figure">S8</ref>, suggest that the peak pressure (approximately 600 MPa) as well as the distribution tails tend to converge as drop height decreases. Notably, increasing drop heights (increasing collision energies) corresponds to a narrower (spatially uniform) pressure distribution. The distribution becomes wider with decreasing drop height, indicating higher spatial variation of pressure (less pressure localization and more blended spatial distribution) within the crater region and surrounding area. This is seen in the distinct crater print and ring resulting from the highest drop height compared to the results from the other two lower drop heights in the third column of Figure <ref type="figure">S8</ref>. The observed simulation trends provide evidence of increasing localization of maximum impact pressure with increasing collision energy.</p><p>Pressure and Specific Kinetic Energy Profiles Along the Crater's Diameter Due to the discretization inherent in MPM simulations, significant pressure fluctuations and noise are observed within the crater region. This is a typical behavior in all particle-based simulations . We determined from G-KDE analysis that pressure values outside the range of -3 GPa to 2 GPa are outliers , and thus, they were excluded from subsequent analysis. We divided the crater into rings of equal thickness (100 &#181;m) to identify an average and a standard deviation value for each ring. The thickness of each ring is chosen to be 100 &#181;m, that is 2 cells lengths. Figure <ref type="figure">S9</ref> shows plots of pressure (raw data), averaged pressure with curves showing the bounds of the standard deviation, and a Savitzky-Golay (SG) smoothing of the averaged pressure values. The SG smoothing is adopted as implemented in the scipy library <ref type="bibr">5</ref> in python using a window length of 11 and a polynomial order of 5. In the main manuscript, the plot of the SG smoothed pressure is reported and analyzed.</p><p>The specific Kinetic Energy (SK) is calculated for all particles (p) in every ring at every output frame (i) after filtering the particles whose pressure are in the range of values reported in Fig 1 .B in the main text. Then the SKs of any ring over all frames are summed up and divided by the total number of frames (NF) between the frame corresponding to initial impact (IM) and the frame corresponding of maximum compression (MC) as shown by Equation <ref type="formula">13</ref>. The resulting value is the temporal average of specific kinetic energy (TASK) along the diameter of the crater. The results are shown in Fig 1 .C in the main text.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Spatially-Resolved Raman spectroscopy</head><p>The Raman spectrum of the clean, unimpacted commercial PET film is displayed in Figure <ref type="figure">S10</ref>, where multiple overlapping bands can be noted. A peak fitting procedure was applied to extract individual peak positions. All characteristic vibrations corresponding to terephthalate as well as ethylene glycol units were identified, and their positions matched those reported in the literature. <ref type="bibr">6</ref> </p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>Conversion</head><p>The conversion of PET resulting from a single collision was calculated as a function of position along the diameter of crater, by tracking the changes in the intensity of ester group stretches at 1292 cm -1 and 1116 cm -1 7 , respectively, defined by the equations presented below. However, firstly several spatially resolved spectra were collected for clean, unimpacted PET (Fig <ref type="figure">S11</ref>) and based on peak fitting analysis, a statistical areas distribution of the band at 1116 cm -1 at 1292 cm -1 was established to compute the mean and standard deviation (SD). The calculated mean values with SD are 0.998 cm -1 &#177; 0.027 for C(O)-C and 3.15 cm -1 &#177; 0.042 for C(O)-O band. These were used as the standard values of peak areas determined for unmodified PET films in the conversion equation. </p><p>To interpolate the conversion distribution known in this procedure as a z-value on a 2-dimensional grid representing the crater area, we found an ellipse intersecting (x,y) from the following equation:</p><p>where the ratio R = r1/r2 is known and taken from the measurements. Then points (x',0) and (0,y') were established where such an ellipse intersects x axis and y axis, respectively:</p><p>Next, zx' and zy' were found, that are the nearest-neighbor interpolations from the known measurements taken on the x axis and y axis, respectively.</p><p>Finally, the z-value was found by weighing zx' and zy' relative to the elliptical distance between (x,y) and (x',0) and between (x,y) and (0,y'), respectively. After plugging the distance calculation and the weighted average calculation, we obtain the following equation:</p><p>for all (x,y) != (0,0). For (x,y) = (0,0) we simplify it to z=(zx'+zy')/2.</p><p>The method described above can be reduced to the nearest-neighbor interpolation on the x axis: z(x,0)=zx', and the y axis z(0,y)=zy'. On the other hand, it places all the points not lying on the axes on a set of concentric ellipses, each with identical R=r1/r2 and centered in (0,0). The method then computes the zvalue corresponding to (x,y) by weighing the two nearest known measurements zx' and zy' inversely proportionally to their elliptical distance from (x,y). Amorphization According to the literature 6,8 , the 1726 cm -1 , 1721 cm -1 and 1733 cm -1 are subcomponents of C=O band in PET Raman spectrum. The peak located at 1726 cm -1 can be assigned to the crystalline domains (see Fig. <ref type="figure">S10</ref>), where in terephthalate units, C=O groups are coplanar with the benzene ring and glycol units are in trans conformation. The peaks at 1721 cm -1 and 1733 cm -1 correspond to the amorphous region, where glycol units are in gauche conformation and terephthalate units are completely disordered. The curve fitting procedure had revealed the presence of all three overlapping bands in the Raman spectra of unimpacted PET film (Fig. <ref type="figure">S10</ref>). To reduce the error margin, the phase transformation of PET was tracked by calculating the ratio between crystalline and amorphous conformers (crys./amorph) according to the equation below.</p><p>crys. amorph.</p><p>= 1726 cm -1 band area (1733 cm -1 + 1721 cm -1 ) band area (20)   Because the intensity of these peaks may vary by location even in fresh, unmodified samples, it was necessary to quantify the variance in intensity. Spectra were taken at several randomly selected spots on a film sample and areas were calculated for each characteristic peak and plotted in Figure <ref type="figure">S12</ref>. The subtle but noticeable differences in the intensity of all aforementioned bands with no clear pattern were observed. The range of ratios for unmodified PET was 0.959-1.033 (Fig. <ref type="figure">S12</ref>). The obtained range of crys./amorph ratios allows to establish the direction of phase transformation. If the crys./amorph ratios calculated for impacted films are below the range of ratios for unmodified PET, it indicates an increase in the amorphous conformation of PET chain, if the crys./amorph ratios increases, it indicates an increase in the crystalline conformation. Fig. S3. Simulation geometrical setup showing the steel ball bottom surface initially placed at 200 &#181;m away from the PET film top surface, which is resting on a 1 mm thick substrate. The setup is visualized using Paraview software. 9,10  Fig. S6.</p><p>2D scatter of the equivalent total strain (as a percentage) against position for all particles found within 1 mm thick strip that extends across the length of the impact region (once along x direction (top row) and another along y direction (bottom row)) for all three velocities. From left to right is the 4.6 m/s impact velocity simulation analysis, 4.1 m/s impact velocity simulation analysis, and 3.5 m/s impact velocity simulation analysis. Fig. <ref type="figure">S7</ref>. 2D scatter plots from MPM simulations in which particles are colored by the value of the equivalent total strain, and the crater regions are defined by the black circles. Top row is showing the crater with the 10% equivalent total strain threshold as a maximum value (lower bound from the equivalent total strain analysis), and the bottom row is showing the crater with the 20% equivalent total strain threshold as a maximum value (upper bound from the equivalent total strain analysis). From left to right is the 4.6 m/s impact velocity simulation analysis, 4.1 m/s impact velocity simulation analysis, and 3.5 m/s impact velocity simulation analysis.      Fig. S14 High-resolution TEM image from inside the crater of PET 3.5 .         Table S3. Craters diameters comparison between simulations and experiment. Impact Velocity (m/s) Diameter From Simulation (mm) Diameter From Experiment (mm) Error (%) 4.6 3.00 3.07 0.62 4.1 2.51 2.43 2.88 3.5 2.30 2.23 3.05 Table S4. Height of the ball drop, total crater area and maximum conversion at given crater area observed for PET films under studies as well as corresponded collision velocity and Energy. The area of maximum conversion was calculated based on range of crater where the max. Con values varies &#177; 1.34%. sample height of the ball drop [m] total crater area [m 2 ] Max. conv. [%] Area of max. conv. [m 2 ] Max. value of amorph.* [%] collision velocity [m&#8226;s -1 ] collision Energy [J] collision velocity* [m&#8226;s -1 ] frequency [Hz]** PETNAOH 4.6 1.07 6.96&#8226;10 -6 27.05 2.68&#8226;10 -6 39.63 4.58 0.662 4.6585 30 PET 4.6 1.07 7.35&#8226;10 -6 4.11 -50.37 4.58 0.662 4.6585 30 PETNAOH 4.1 0.85 5.99&#8226;10 -6 20.42 1.19&#8226;10 -6 19.80 4.08 0.525 4.0832 27.5 PET 4.1 0.85 4.72&#8226;10 -6 6.02 -28.64 4.08 0.525 4.0832 27.5 PETNAOH 3.5 0.63 2.43&#8226;10 -6 19.25 0.53&#8226;10 -6 21.57 3.5 0.387 3.7747 25 PET 3.5 0.63 2.21&#8226;10 -6 5.03 -27.51 3.5 0.387 3.7747 25 * Raman spectro-microscopy data ** collision characteristics established for vibratory ball milling reactors published in Ref. (17)</p></div><note xmlns="http://www.tei-c.org/ns/1.0" place="foot" xml:id="foot_0"><p>Chem 12, 102754, January 15, 2026 Article</p></note>
			<note xmlns="http://www.tei-c.org/ns/1.0" place="foot" xml:id="foot_1"><p>Fig. S16. Representation of the drop in the intensity of the ester band at 1116 cm -1 along the crater diameter of impacted, coated (left) and uncoated (right) PET films.</p></note>
		</body>
		</text>
</TEI>
