Revealing structural evolution occurring from photo-initiated polymer network formation

The study addresses how nanoscale morphology evolves during photo-initiated polymerization and how these structural changes correlate with chemical conversion and physical processes such as cross-linking, vitrification, and segmental mobility. Photopolymerization enables spatially and temporally controlled fabrication of materials for diverse applications, yet a comprehensive experimental description of nanoscale structural evolution during curing has been lacking. Prior understanding has largely focused on photochemistry and reaction kinetics, with microscopic-scale physical changes clarified but nanoscale processes remaining hypothetical or simulated. The purpose is to monitor in situ and in real time the coupled chemical and physical transformations during photopolymerization and to determine how initial precursor structure and glass transition behavior influence the development and length scale of nanoscale heterogeneities, ultimately impacting the properties of the cured network.

Previous work has established photopolymerization chemistry and kinetics and described microscopic-scale transformations, but nanoscale descriptions have been limited to hypotheses and simulations. X-ray diffraction has elucidated photopolymerization in specific systems (e.g., single-crystal 2D polymers), and GISAXS has been used to monitor film formation in physically drying dispersed systems. Acrylate-based networks are known to form inhomogeneous structures with density fluctuations due to polymerization-induced phase separation and reaction–diffusion processes creating nano-sized gel domains. Vitrification can arrest conversion, leading to local differences in reactive shrinkage and density. These backgrounds motivate an in situ nanoscale probe to connect initial liquid-state structures to final solid-state morphology in photocurable thermosets.

Two model acrylate resin formulations were studied: LT (low glass transition temperature, Tg = −17.3 °C) polymerizing fully without vitrification, and HT (high glass transition temperature, Tg = 40.5 °C) experiencing vitrification during polymerization. Films were prepared by spin-coating diluted resins onto cleaned silicon wafers (0.2 mL dispensed; 6000 rpm for 60 s) and handled under minimized ambient light. Wafers were stored in ultrapure water prior to use and dried with nitrogen before coating. UV photopolymerization was performed in a nitrogen-purged chamber using a 365 nm UV-LED (FireJet FJ800) at 11.4 mW/cm². Step-wise irradiation was employed: UV exposure segments were interleaved with measurements while the sample remained in situ. Real-time FTIR monitored chemical conversion, normalizing spectra to the cyclic alkene C–C stretching (LT: 1601 ± 1 cm⁻1; HT: 1607 ± 1 cm⁻1) and tracking the decrease of the monosubstituted C=C stretch at 1636 ± 1 cm⁻1. Conversion vs. time was fitted with an exponential model. In situ grazing-incidence small-angle X-ray scattering (GISAXS) and transmission SAXS (T-SAXS) were conducted at the P03/MiNaXS beamline (PETRA III/DESY) with 13 keV X-rays and SDD = 3403 ± 1 mm, using a Pilatus 1M detector. For GISAXS, the incident angle αi = 0.4° (above critical angle) ensured full film penetration; patterns were reduced with DPDAK and integrated along the Yoneda region (αi ≈ 0.1°). Each GISAXS acquisition was 100 ms, with lateral scanning and summing of 10 patterns per step. The in-plane feature size was calculated from the fitted qxy peak positions using d = 2π/qxy. In-plane correlated roughness and correlation length were assessed from resonant diffuse scattering oscillations, with Rc = 2π/Δq (Δq: disappearance of fringes). X-ray reflectivity (XRR) used attenuated beam (0.3 mm Al) with incidence angle sweep 0.085°–1.8°, integrating vertically at 4.0 m detector distance. Multilayer fits (air/resin/SiO2/Si) were performed with PyXRR v0.6 to extract thickness, interfacial roughnesses, and electron density profiles as a function of UV dose. AFM topography (Bruker MultiMode, tapping mode, SCANASYST-AIR-HR tip, k = 0.4 N/m, f0 = 130 kHz, tip radius ~2 nm) assessed surface rms roughness on three 10 × 10 µm² areas; film thickness was confirmed by scratch-depth profiling. For viscoelastic properties, dynamic mechanical analysis (DMA Q800) was performed on thicker films (LT: spin-coat 800 rpm 60 s; HT: 1500 rpm 60 s; leveled 20 min; detached and cut to 5 × 10 mm²). Temperature sweeps at 1 Hz, 3 °C/min were run over −60→50 °C (LT) and −50→120 °C (HT), evaluating tan δ.

• Real-time FTIR: LT achieved nearly 100% double-bond conversion; HT reached ~80% conversion due to vitrification and restricted topological mobility in a more densely cross-linked network. - GISAXS revealed nanoscale heterogeneities present pre-cure that grow during polymerization. In-plane characteristic length scales: LT increased from (47.2 ± 4.3) nm to (189.2 ± 12.4) nm within 1 s of UV exposure and then remained constant; HT increased from (13.4 ± 0.1) nm to (17.2 ± 0.2) nm with growth largely arrested after ~0.5 s. - The in-plane correlation length (smallest replicated structure size from substrate roughness): Rcorr,LT = (52.3 ± 4.1) nm; Rcorr,HT = (28.6 ± 1.2) nm. - Film thickness after 28 s UV: dLT = (143.2 ± 1.5) nm; dHT = (160.1 ± 1.9) nm. - AFM rms surface roughness of fully cured films: σLT = (0.74 ± 0.07) nm; σHT = (0.55 ± 0.05) nm. - XRR interfacial roughness evolution: For HT at the air–resin interface, roughness decreased from (2.3 ± 0.2) nm to (1.5 ± 0.1) nm within 300 ms, remained constant to 1 s, and decreased further to (1.3 ± 0.1) nm at 28 s; both interfaces showed similar trends with lower roughness at the resin–substrate interface. For LT, air–resin roughness was ~ (1.3 ± 0.2) nm during the first second, increasing to (2.1 ± 0.1) nm by 28 s; resin–substrate roughness also increased during curing. - Guinier radii (uncured formulations, SAXS): LT = (1.84 ± 0.10) nm; HT = (1.46 ± 0.08) nm. - The magnitude and kinetics of heterogeneity growth correlate with segmental mobility differences associated with Tg (LT: −17.3 °C; HT: 40.5 °C). LT heterogeneity length scales after curing approach the film thickness, whereas HT heterogeneities remain far below the film thickness and become homogenously distributed through thickness upon vitrification arrest. - Overall heterogeneity length scales formed during curing lie within ~10–200 nm.

The combined in situ GISAXS, XRR, and FTIR measurements establish that nanoscale structural evolution during photopolymerization is strongly governed by the liquid-state precursor structure and glass transition behavior. Early in the reaction, reaction–diffusion processes nucleate gel-like network domains that grow by monomer influx and polymerization; simultaneous cross-linking and vitrification impose local physical arrest. In the LT system (low Tg, higher segmental mobility), domains coarsen rapidly to ~190 nm within 1 s and then stabilize, yielding heterogeneities comparable to film thickness. In the HT system (higher Tg), vitrification restricts conversion (~80%) and segmental mobility, limiting domain growth to ~17 nm and homogenizing features across the film thickness. Interfacial roughness trends seen by XRR support opposite curing behaviors: smoothing in HT consistent with rapid arrest and densification, and increasing roughness in LT as continued network formation and shrinkage proceed. The correlation between Tg, conversion arrest, and heterogeneity size provides a mechanistic link between initial monomer state and final network morphology. These insights indicate that nanoscale morphology—and thus macroscopic properties—can be predictively tuned via precursor selection and control of vitrification during curing.

This work demonstrates a robust in situ, real-time methodology combining GISAXS, XRR, and FTIR to reveal and quantify nanoscale structural evolution during photo-initiated polymer network formation. It shows that initial precursor nanostructure and glass transition behavior dictate the emergence and arrest of nanoscale heterogeneities (10–200 nm), with LT systems developing larger domains and HT systems arresting at much smaller scales due to vitrification. The approach enables predictive tailoring of local morphology and therefore macroscopic properties in photocurable thermosets through informed choice of liquid precursors and curing conditions. Future work could extend these measurements to broader resin chemistries, different film thicknesses and substrates, continuous (non-stepwise) curing protocols, and 3D printing-relevant exposure patterns to further generalize structure–processing–property relationships.

The study focuses on two specific acrylate-based model formulations (LT and HT) and thin films prepared on smooth silicon substrates; substrate-induced correlated roughness and film thickness may influence observed in-plane correlations. Polymerization was conducted in a step-wise manner under specific UV intensity and wavelength, which may not capture all aspects of continuous industrial curing. Generalization to other chemistries, bulk samples, and different processing conditions will require further investigation.