Composite Resin Formulation and VP-3DP
In the present study, we selected a mixture of bisphenol A-glycidyl methacrylate (bis-GMA) and triethylene glycol dimethacrylate (TEGDMA) with a fixed weight ratio (50 wt%/50 wt%) with diphenyl(2,4,6-trimethylbenzoyl)phosphine oxide (TPO) as an initiator (Fig. 1a), which is a well-known monomer mixture for dental composites[15], as a resin for adding fillers. The dependence of the mixing ratio on the mechanical properties after curing has recently been revisited[16]. According to this report, changing the ratio of bis-GMA to TEGDMA in the mixture from 2:8 to 8:2 results in a minor shift in the mechanical properties, and in the case of the elastic modulus, only a change less than ca. 0.5 GPa can be observed, while the rate of polymerization is much greater when the ratio of bis-GMA exceeds 50 wt%[16]. As such, we fixed the weight ratio at 50 wt%/50 wt% because of its lower viscosity, easier handleability, and reasonably fast rate of polymerization. To test 3D printability, we created CAD data of cuboid specimens with dimensions of 33 mm × 5 mm × 1 mm, and the 3D printability of the formulated resin (BT) was confirmed before proceeding to VP-3DP of the composite resins.
Next, a series of photoresins were formulated upon mixing BT and mica flakes with different sizes, bulk densities, and aspect ratios (Supplementary Table 1). The sol–gel boundaries at room temperature were evaluated for mixtures of these mica flakes and BT by varying the weight fraction of mica flakes by 5 wt% and the maximum loading while maintaining flowability for A-11, A-21S, AB-25S, J-31 M, Y-1800, Y-2300, and Y-3000 was determined to be 30, 30, 30, 40, 30, 35, and 35 wt%, respectively (Fig. 2a). As mixtures containing the mica flakes at these boundaries were highly viscous, we used the loading of 25 wt% for further 3DP investigations. Interestingly, at a loading of 25 wt%, which is equivalent to that mixed with typical cosmetic eyeshadows, the formulated resins reflect and scatter light like such cosmetic products due to their angle-dependent optical effects (Fig. 2b)[17]. This visual feature enhances the design and functionality of the final products. The formulated composite resins containing a series of mica flakes (A-11, A-21S, AB-25S, J-31M, Y-1800, Y-2300, and Y-3000) at a loading of 25 wt% were then subjected to VP-3DP with the CAD model with three different layer addition (printing) directions (Fig. 2c). Importantly, simple 3DP operation without any special treatment readily afforded the desired CPMCs with highly precise geometries (Fig. 2d), demonstrating the feasibility of the fabricated mixtures for ready-to-use photoresins for 3D printers.
Analysis of anisotropy based on SEM observation
Having developed composite resins and confirmed their excellent printability, we next analyzed the anisotropy of the printed CPMCs containing a series of mica flakes (CPMCA−11, CPMCA−21S, CPMCAB−25S, CPMCJ−31M, CPMCY−1800, CPMCY−2300, and CPMCY−3000) by scanning electron microscopy (SEM). SEM provides valuable insights into the filler distribution and orientation within composites. As the thickness (t) of each layer during VP-3DP was set to 25 µm (Fig. 1b), the rotational barrier and alignment of the mica flakes in the presence of compressional force in the confined space would be different depending on their sizes. Thus, mica flakes with a size smaller than 25 µm would be more isotropically distributed and behave more like isotropic materials. On the other hand, if there is a rotational barrier to the micas in the confined space (Fig. 1b), the micas will be aligned, and the resulting printed objects become anisotropic materials. To reveal this, SEM images were taken from the direction perpendicular to the printing direction (xy-plane) and from the direction in which the stacked layers were observed from the side (yz-plane) (Fig. 3a). Indeed, the distributions of A-11, where the average size (3 µm) is sufficiently smaller than 25 µm, in the xy- (Fig. 3b) and yz- (Fig. 3c) planes observed for CPMCA−11 were quite similar. Although slightly more faces of the flakes can be observed in the xy-plane (Fig. 3b), no significant differences were observed that could clearly indicate the stacking of A-11 along the layer addition direction. In sharp contrast, the in-plane aligned morphology of J-31M, where the average size (35 µm) is sufficiently larger than t, was observed in the SEM image of the xy-plane (Fig. 3d) with CPMCJ−31M. Interestingly, the SEM image of the yz-plane sample showed a longitudinal streaked morphology (Fig. 3e). At first glance, the pattern appears to be a side view of the stacked layers during VP-3DP, but the spacing between the stacked mica flakes in the z-direction is apparently narrower than that of each layer (25 µm). Thus, the compressional stress during VP-3DP (Fig. 1b) could be sufficient to stack several layers of mica flakes in the confined space within t, and it is also worth noting that as mica flakes are stacked, it becomes more difficult for them to rotate within the layer (Fig. 1b). If this mechanism is true, it should in principle be possible to print anisotropic materials even with fillers reasonably smaller than 25 µm as long as there is a sufficient aspect ratio. As all mica flakes except A-11 used in this study commonly have high aspect ratios (Table S1), we continued SEM observations of the CPMCs to determine the limits of how small the size of mica flakes can be for printing anisotropic materials. Within the tested range, AB-25S, A-21S, and Y-3000 have average sizes of 24, 23, and 23 µm, respectively (Table S1), which are comparable to t. From the comparison of SEM images of CPMCAB−25S (Figure S1a, b), CPMCA−21S (Figure S1c, d), and CPMCY−3000 (Figure S1e, f) taken to observe the distribution of mica particles in the xy-plane and yz-plane, it becomes evident that all of these mica flakes are aligned so that their plate faces are parallel to the xy-plane. Moreover, despite the average sizes of Y-2300X and Y-1800 mica being 19 µm and 10 µm, respectively (Table S1), which are apparently smaller than 25 µm, they aligned in similar manners (Figure S2). Therefore, with a t of 25 µm, there should be a threshold between 5–10 µm, which is the barrier for restricting the rotation of mica flakes in the present confined space. This led us to further consider the narrowing effect of t down to 10 µm, which is the minimum layer thickness that can be set with the 3D printer used in this study. However, the distributions of A-11 in the xy- (Figure S3a) and yz- (Figure S3b) planes were quite similar, indicating that mica flakes with a size of approximately 1/3 of t can rotate in a confined space within a layer to be cured. Considering that Y-1800 was well aligned despite being less than half the t, a size of approximately above 1/3 of t would be a rough indication for inducing anisotropy.
Evaluation of the mechanical anisotropy based on flexural tests
Achieving the simple, affordable, and easily manipulable fabrication of highly precise and intricate anisotropic materials upon AVP-3DP without requiring external field-generating modules could be a catalyst for the dissemination of anisotropic materials. With the anisotropy observed for our CPMCs, realizing this intent may now be possible. Flexural stress–strain curves of test specimens derived from CPMCs were measured for three different directions, i.e., the layer (L), transverse (T), and perpendicular (P) directions, to evaluate direction-dependent flexural strengths (σs: σL, σT, and σP) and flexural moduli (Es: EL, ET, and EP) (Fig. 4a). These directions were determined based on the difference in the relationship between the layer addition (printing) and indentation directions. For the control experiment, first, the σs and Es of BT were evaluated based on its flexural stress–strain curve (Supplementary Fig. 2), and the average σs (σL = 150 MPa, σT = 141 MPa, and σP = 130 MPa) (Fig. 4b, BT) and Es (EL = 2.90 GPa, ET = 3.22 GPa, and EP = 4.59 GPa) (Fig. 4c, BT) were comparable to those reported in the previous literature[16]. Perhaps as a consequence of the layer addition operation, which inevitably has a directional effect, slight anisotropy in the mechanical properties could also occur for samples without mica flakes. This effect corresponds to those reported in the previous literature[18]. However, in the present case, the observed differences are not significant, as suggested by the range of error bars (Fig. 4b, c). With mica flakes, the hard and brittle properties derived from inorganic materials are imparted to BT, and it is evident that the σs (Fig. 4b) are reduced and the Es (Fig. 4c) are increased for all CPMCs in comparison with those of BT. For CPMCA−11, essentially small differences between the three directions were observed for σs (Fig. 4b, A-11) and Es (Fig. 4c, A-11). This could be reasonably explained by the fact that the distributions of A-11 in the xy- and yz-planes were quite similar, and little anisotropy was confirmed in the comparison of the SEM images (Fig. 3a, b). In sharp contrast, except for those of CPMCA−11, σL and σT are significantly greater than σP and, moreover, in the case of CPMCY−3000, the σL/σP ratio reaches 4.23. In addition, the EP tends to be smallest within Es (Fig. 4c). Interestingly, CPMCY−3000 again showed outstanding directional dependence among all the tested CPMCs. The average Es values were 11.3 GPa (ET), 5.24 GPa (EL), and 5.64 GPa (EP), respectively, and both the ET/EP and ET/EP ratios exceeded 2. Given that the tensile modulus and Vickers hardness of composites reinforced with platelet Al2O3, fabricated based on magnetic-field-assisted VP-3DP in the previous literature, in the strong-axis direction were approximately 900 MPa and 7.5, respectively, and those in the weak-axis direction were approximately 700 MPa and 6, respectively[11a], the present differentiation effect on the mechanical properties would be prominent.
Moreover, the exceptionally high average E of 13.6 GPa, which is comparable to that of metal lead (E = 13–15 GPa), achieved in the T direction with CPMCY−2300X signifies the effectiveness of our approach in fabricating highly precise and intricate 3D-printed objects with excellent mechanical properties. It is also worth mentioning that the observed differentiation or decreasing effect in the mechanical properties are primarily concentrated in the P direction, as intended for the AVP-3DP design (Fig. 1b).