Additive manufacturing platforms such as 3D printing reduce the manufacturing development time and cost of individually tailored objects with intricate shapes, through the layer-by-layer assembly of printable inks5,6. Recent advancements in printing techniques have expanded the possibilities for creating complex bioinspired hierarchical structures in synthetic polymers, including rotational and Bouligand structures7-10. Despite the proven utility, 3D printing methods, fundamentally operate deterministically, requiring careful pre-design, coding, and active control throughout the manufacturing process to produce desired architectures with limited resolution. Non-deterministic, autonomous manufacturing of geometrically intricate, free-form objects endowed with architected microstructures and submicron-scale properties remains an unmet scientific challenge. Recent advances by Rylski, et al. demonstrated the effectiveness of a light-sensitive dual catalyst system in fabricating polymeric multimaterials with microscopic precision4. While this manufacturing approach has the ability to spatially control the resulting properties of the polymer, it is still constrained by its reliance on a pre-defined configuration, with limitations on sample dimensions imposed by the depth of light penetration.
Drawing inspiration from biological systems where structural complexity develops through dissipative reaction-diffusion processes, this paper explores a transformative synthetic manufacturing strategy aimed at harnessing the principles underpinning morphogenic growth, unlocking new avenues for advanced materials design and fabrication. Synthetic coupled reaction-transport processes offer a versatile yet relatively underexplored method to manipulate the spatial attributes of synthetic materials11. While controllable multiscale patterns have been achieved through reaction-mass transport processes12-17, these accomplishments have been limited to solutions, gels, or thin membranes. Here, we introduce an innovative manufacturing approach based on frontal ring-opening metathesis polymerization (FROMP) that draws parallels with morphogenic growth and development, enabling the formation of architected microstructures within polymeric materials.
Frontal polymerization (FP) is a thermally driven reaction-transport process in which a polymerization wave propagates through a monomer, transforming it into a polymer18-22. While stable front propagation in a uniform, planar configuration allows for rapid polymeric material production, disruptions in the system, such as heat loss or variations in the initial composition, can destabilize this planar mode, resulting in pulsating, spinning, and aperiodic front propagation12. Nonplanar spin modes have been studied for acrylate-based chemistries11,23-27 and are characterized by the presence of one or more localized high-temperature regions, i.e., hot spots, which exhibit a helical or zig-zag movement pattern dependent on the shape and the size of the reactor geometry28,29. This mode of propagation also introduces non-uniformities in the resulting polymer product30,31. We hypothesized that by strategically harnessing nonplanar front propagation, coupled with the temperature-responsive crystallization of poly(cyclooctadiene) (pCOD)32,33, unique polymer microstructures could be formed through a spontaneous dissipative manufacturing process controlled by the thermochemistry, reaction kinetics, and boundary conditions of the system (Fig. 1a).
Thermal FROMP of our initial formulation of 1,5-cyclooctadiene (COD) with a Grubbs’ second-generation catalyst (Ru-1) as the initiator and a phosphite inhibitor (Fig. 1b) displays uniform front propagation in the X2 direction in a closed mold (Fig. 1c and Supplementary Video 1). Nonplanar front dynamics are triggered by extremely small perturbations in the reaction kinetics34. Increasing the concentration of the phosphite inhibitor (from 1.0 to 2.0 equiv. to initiator), while keeping all other conditions constant, resulted in the formation of fronts with highly regular spin modes that traverse back and forth in the X1 direction until complete conversion of monomer to polymer (Fig. 1e and Supplementary Video 2). This slight increase in inhibitor concentration results in a lower initiation rate for the spin mode formulation, indicated by a higher exothermic onset temperature observed in the dynamic cure profiles (Extended Data Fig. 1). The spin mode formulation also exhibits a marginally lower heat of reaction (Hr), resulting in less heat diffusion to the monomer resin ahead of the front and greater heat loss to the surrounding environment (Extended Data Table 1).
This small difference in the resin formulation and associated front dynamics drives remarkable differences in time-dependent crystallization and resulting structure of the final polymer product (Figs. 1d, f). Consistent with prior reports in literature32,33, relatively homogenous crystallization was observed in pCOD 24 h after FP (Fig. 1d). In great contrast, regularly spaced crystallization patterns emerged in spin mode samples consistent with the direction of the front propagation (Fig. 1f and Supplementary Video 3). The evolution of the polymer structure was characterized by nuclear magnetic resonance (NMR), Raman spectroscopy, and differential scanning calorimetry (DSC) (Extended Data Fig. 2 and Supplementary Figs. S1-7). The trans content and the heat of fusion increased with time in both uniform and spin mode pCOD specimens. This evolution is attributed to the ruthenium initiator remaining active on the chain ends after FP has completed, allowing for chain transfer events to occur post polymerization35-37.
Our understanding of how the thermochemistry, reaction kinetics, boundary conditions, and initial selection of resin formulations was guided by numerical modeling of the front dynamics using the reaction-diffusion partial differential equations as shown in 1 (Extended Data Fig. 3a and Extended Data Table 2)38,39. Two-dimensional finite element simulations conducted using the Multiphysics Object-Oriented Simulation Environment (MOOSE), reveal that the spin mode of front propagation has two distinct thermal regions, a hot region (normalized maximum temperature, θmax ~ 1.2), and a cold region (θmax ~ 0.8), whereas the uniform mode of front propagation has only one distinct thermal region (θmax ~ 1.0). The uniform mode of front propagation was simulated with an initial ambient temperature of 35 °C, which minimizes the thermal diffusion to the immediate surroundings, whereas the spin mode of front propagation was simulated at 25 °C. Figure 2a shows the simulated evolution of temperature and degree of cure (T-α) at the three distinct thermal regions during the polymerization processes: uniform mode (grey curve), spin mode hot region (red curve), and spin mode cold region (blue curve). The contour of reaction rates (dotted lines) for each mode of propagation was derived from the cure kinetic model and fit with the experimental data (Extended Data Fig. 3b). The slope of the line for the uniform mode is approximately equal to the heat of reaction divided by the specific heat capacity of the resin (Hr/Cp,r), implying that the time scale of the reaction is faster than the heat that is dissipated to the surroundings, thus resulting in the uniform mode of front propagation. The spin mode of front propagation necessitates a higher energy input to trigger the reaction due to a lower initial temperature. Meanwhile, the heat loss at the boundaries is more pronounced than the uniform mode, leading to a more heterogeneous spatial temperature distribution. The hot region accumulates more heat before the polymerization is kicked off, and the reaction rate accelerates, reaching approximately 102 s-1. The slope of (dα/dT) for the spin mode hot region is slightly greater than that of the uniform mode as it forms into a polymer more rapidly and then diffuses the heat from the reaction to the cold regions, continuing the polymerization process. The spin mode cold region receives less heat before polymerization is kicked off, resulting in lower reaction rates than the hot region. The discrepancy in reaction rates between the hot and cold regions contributes to the formation of the observed spin modes, which are evident in both experimental observations and simulations.
Systematic experimental and numerical studies reveal that the boundary conditions of the system control the size and spacing (d) of the architected domains40. Consistent with prior reports for acrylates28,29, an increase in specimen width leads to a greater amount of heat generated per unit surface area, accelerated propagation of the spin mode, and a decrease in domain spacing (Figs. 2b, c and Supplementary Fig. S11). Multi-head spin modes were observed in both experiments and simulations for larger specimen geometries (W > 2 cm) (Supplementary Videos 4 and 5). Increasing the ambient temperature from 25 °C to 30 °C results in enhanced reaction kinetics, an accelerated spin mode of front propagation, and a consequent decrease in the predicted and observed spacing between material domains with the Ru-1 initiator (Figs. 2d, e, f). Increasing the temperature to 35 °C results in the loss of the spin mode and return to the uniform mode of front propagation. This result is attributed to the predominance of the thermochemistry and reaction rate over the thermal transport to the system’s surroundings. Conversely, when the ambient temperature is lowered below 15 °C, the reaction rate is significantly suppressed and the heat loss is enhanced, resulting in front quenching. Numerical simulations consistently predict the experimental trends in material domain spacing and provide a powerful tool for predicting the necessary conditions to produce a desired architecture (Supplementary Table S4).
Comparative analysis of small- and wide-angle X-ray scattering of the uniform and architected polymers revealed unexpected differences in the alignment of the lamellae and polymer chains in the X2 scan direction (Fig. 3a and Supplementary Fig. S8) consistent with prior studies of crystallization in non-FROMP systems41-45. The shear flow induced in the monomer by hot and cold spin mode regions during front propagation significantly impacted the growth direction within the semi-crystalline domains. Azimuthal integrations along the X2, out-of-plane direction provide quantitative insights into the orientation of the polymer chains and lamellae at angles of θ = 0° and θ = ± 90° and the intricate relationships between the architecture, alignment, and properties of the polymers (Extended Data Fig. 4 and Supplementary Fig. S9). S/WAXS data suggests that polymer chains align orthogonally to the lamellae orientation.
Tuning of the resin formulation through the choice of ruthenium initiator and inhibitor concentration enables further manipulation of the macro- and micro-scale domain architectures (Extended Data Table 1). Use of initiators with different alkylidene species (Ru-1, Ru-2, or Ru-3)46 alters the initiation kinetics (Extended Data Fig. 1), and results in surprising changes in the size and spacing of the polymer semi-crystalline domains (Figs. 3b, c, d). Due to differences in the initiation kinetics of the three resin formulations, the relative amount of inhibitor for Ru-2 and Ru-3 must be decreased to obtain successful spin mode propagation (Extended Data Table 1). Alternatively, we can increase the amount of initiator or change the ambient temperature to achieve spin modes in these formulations (Supplementary Videos 6 and 7).
Comparison of the structure of polymers made with the three different initiators reveals reproducible variations in the orientation of the polymer chains with respect to the X2 direction (Fig. 3d). This diversity in chain orientation provides an additional dimension of control for final material properties. Among the three initiators, Ru-3 exhibited the largest anisotropy in the percentage distribution of chains oriented at 0° versus those oriented at 90° (Extended Data Table 3). Intriguingly, even in uniform specimens, a discernible preference for polymer chain orientation was observed, indicating that planar fronts propagating in the X2 direction exert an influence on the alignment of the polymer chains within the material. The spatial distribution and alignment of these chains, alongside the packing of lamellae within the architected materials impact the resulting mechanical properties as compared to the uniform counterparts.
The preferential orientation of polymer semi-crystalline domains has a profound impact on the heat transport and optical properties of polymeric materials47. To gain a deeper comprehension of how the mechanical properties are impacted by the dynamics of the nonplanar front and the orientations of the polymer chains / lamellae, we prepared tensile coupons with domains oriented along the A1 (θ = 0°) and A2 (θ = 90°) directions (Fig. 4a). The local variation in properties along the length of the specimen was characterized by nanoindentation. The presence of FP spin modes led to the formation of the material domains with a two-fold difference in the reduced modulus along the X2 direction 24 h post polymerization. In great contrast, specimens generated through the uniform mode of front propagation did not exhibit any deviations in the resultant material characteristics (Fig. 4b, Extended Fig. 5, Supplementary Fig. S13).
The anisotropy of architected materials fabricated using the Ru-3 initiator leads to enhanced mechanical properties in the A2 loading direction when compared to the uniform counterparts (Figs. 4c, d and Extended Data Table 4). The Ru-3 A2 architected specimen has 18% higher tensile strength and remarkably a 178% higher strain energy density (area under stress-strain curve) than the uniform specimen. The presence of hard and soft material domains in A2 leads to a significant increase in strain-to-failure and toughness, with little change in elastic modulus (~ 0.7 GPa). A1 specimens have even higher strain to failure, but this ductile failure is accompanied by lower elastic modulus (~ 0.3 GPa) and tensile strength (~ 8 MPa). We attribute the distinctive anisotropic characteristics observed in the Ru-3 architected specimens to the greater percent distribution of polymer chains orientated 90° to the loading direction of the A2 specimens (Fig. 3d).
We found that the tensile mechanical behavior of architected polymers is controlled by the selection of the ruthenium initiator, even for the same monomer composition. The changes in polymer orientation and spacing of the material domains, for different resin formulations, significantly alter the resulting mechanical properties of these architected polymers48 (Supplementary Videos 8, 9, 10). The elastic modulus, strain energy density, and tensile strength in both the A1 and A2 directions can be tailored exclusively through selection of the ruthenium initiator in the resin formulation (Figs. 4e, f; Extended Data Fig. 5). The distribution of oriented polymer chains varies across initiator systems (Fig. 3d), contributing to the significant differences in mechanical response reported in Figs. 4e, f. Judicious slection of initial resin formulation guided by modeling and experiment enables unique possibilities to design properties of engineered architected materials.
The deliberate selection of resin formulation and boundary conditions leads to the emergence of nonplanar front dynamics, facilitating the production of molecularly architected polymers. Minor perturbations in the initiator and inhibitor enable the design of microstructure and macroscopic properties. Informed by numerical simulations, careful manipulation of the ambient temperature and specimen geometry allowed us to change the spacing of the material domains. Initiator formulations further influenced the spacing of these material domains, primarily due to variations in the reaction kinetics. Furthermore, we measured variations in the orientation of polymer chains and lamellae that significantly influenced the material properties of the molecularly architected polymers generated by the three initiators. The specimens with enhanced chain orientation in the ± 90° direction had enhanced toughness in comparison to their uniform counterparts. The formation of complex patterned architectures that depend solely on the initial resin formulation and boundary conditions opens a new avenue for the control of polymer properties and function through dissipative processing methods like FP with significant advantages with respect to resolution and performance over more deterministic additive manufacturing platforms.