To gain insight into the asymmetric WGM resonant spots shown in Fig. 3, the arrangement of the PFBT main chain on the surface of (S)-TB-MS was studied by means of polarized FL imaging, schematically shown in Fig. 4a. Figure 4b shows a plot of the FL intensity collected at the centre of (S)-TB-MS with θ = 90° versus the rotation angle of the analyser (φA) upon nonpolarized light excitation (λex = 400–440 nm). The plot shows a clear φA dependency: The FL intensity has a maximum and a minimum at φA = 140° and 50°, respectively, at which the direction of the analyser coincides well with the directions parallel and perpendicular to the electronic transition dipole moment (ETDM) in the PFBT main chain (Fig. 4c, d). Polarization mapping, reconstructed from a series of polarized FL images, confirms the counterclockwise spiral alignment of the PFBT main chains on the surface of (S)-TB-MS (Fig. 4e and Supplementary Fig. S16a), which is further corroborated by the polarized FL images collected at θ = 0° (Fig. 4f and Supplementary Fig. S16b, c). These images correspond well with a twisted director field line in an LC droplet with a TB arrangement.32, 33
These results are informative for explaining the azimuthally selective WGMs and their λem-dependent angular movement along the circumference. Because FL radiates perpendicularly to the direction of the ETDM, FL at the surface of (S)-TB-MS is preferentially directed towards Q1 and Q3 and is confined via TIR. Accordingly, the peculiar surface orientation of PFBT results in the right-left asymmetric WGM PL images in Fig. 3f, which is schematically drawn in Fig. 4g. Moreover, due to the distinct anisotropy in the extraordinary and ordinary refractive indices (ne and no, respectively) of the PFBT main chain,34 the swirl molecular orientation makes the morphologically spherical surface an optically anisotropic surface along the latitudinal angle. As a result, the diagonal optical pathways on (S)-TB-MS show a λ-dependent optical path length, as depicted in Fig. 4g.
To verify the λ-dependent angular shift of the WGM, we constructed an analytical model and conducted numerical simulations. In our model, we suppose that the bipolar axis of (S)-TB-MS is aligned along the y-axis, and the WGM orbits a great circle inclined by φ from the x- towards the y-axis (Fig. 5a). Here, the effective permittivity \(⟨\epsilon ⟩\) along the WGM trajectory is given by tracking the orientation of PFBT (p) along the circle. Since the TE modes are expected to be sensitive to the molecular orientation due to the tangential anchoring of the main chain of PFBT on the surface of the microsphere, the φ -dependent shift of the WGM resonance frequency is simulated by substituting \(⟨\epsilon ⟩\) for the resonance condition of the TE mode (Eq. S2 in the Supporting Information). In this context, the electric field component in the travelling direction (\(\widehat{\phi }\)) of the WGM is sufficiently small with respect to that in the perpendicular direction (\(\widehat{\theta }\)). Therefore, \(⟨\epsilon ⟩\) is approximated as
\(⟨\epsilon ⟩\approx \frac{1}{2\pi }{\int }_{0}^{2\pi }{{n}_{e}}^{2}{\text{cos}}^{2}\xi \left(\phi \right)+{{n}_{o}}^{2}{\text{sin}}^{2}\xi \left(\phi \right)d\phi\)
where ε is the angle between p and \(\widehat{\theta }\) (Fig. 5a). In ε, the function f(y) characterizes the swirl pattern determined by the angle between p and the y-axis (see SI for details). Here, we assume \(f\left(y\right)={\varphi }_{e}\sqrt{1-{\left(\frac{2y}{d}\right)}^{2}}\) to dictate a twisted geometry, where φe represents the angle at which the ETDM crosses the equator of (S)-TB-MS (Fig. 5a; see SI for details). According to the detailed analysis of the polarized FL images shown in Fig. 4 and Supplementary Fig. 16, the φe of (S)-TB-MS is expected to be ~ 50°.
To simulate φ-dependent WGM resonance, four parameters, ne, no, d, and φe, should be considered. As reliable values, we set ne, no, and d as follows: ne = 1.80 ± 0.05 and no = 1.55 ± 0.05 from reference 34 and d = 4.4 ± 0.1 µm from the optical micrographs of (S)-TB-MS in Fig. 3c. Then, the φe value that well fits the experimental plots of φ in Fig. 3e is systematically assigned. The simulated curves with φe = 55° best fit the experimental results (Fig. 5b, Supplementary Fig. 17, and Supplementary Table 1), which is comparable to the experimentally obtained value (~ 50°). Furthermore, the analytical model indicates the appearance of WGM spots around the bipolar positions (φ = 70–120°), although it does not fit well at φ ~ 90°, possibly because of the disorder of the polymer arrangement at the topological defect position (spot (iii) in Fig. 3f and Supplementary Fig. 18). These results strongly suggest that the proposed swirl molecular orientation can indeed explain the azimuthally selective WGMs with an λem-dependent shift along the circumference.
As an enantiomeric control, the WGM resonance of (R)-TB-MS was also studied. In contrast to (S)-TB-MS, HSC images of (R)-TB-MS display resonant PL in the second and fourth quadrants (Q2 and Q4, respectively, Supplementary Fig. 19 and Supplementary Video 5). Furthermore, polarized FL images of (R)-TB-MS are mirror images to those of (S)-TB-MS, indicating that the swirl directions of the PFBT main chains on the surfaces of these spheres are opposite (Supplementary Fig. 20). These results corroborate that the origin of the azimuthally anisotropic WGM confinement in the TB microcavity can be attributed to chirality-dependent twisted alignment of the ETDM.