Multiscale Analysis of Origin and Emergence of Chirality
Figure 1 illustrates our framework to perform a multiscale analysis of the emergence of chirality. We designed a “same connectivity, different chirality” system to unravel the hierarchical emergence of chirality (Fig. 1a): i) from central chirality in small monomeric molecules: ii) to backbone chirality in single polymer chains; iii) and supramolecular order in their assemblies. We selected as model system chiral sulfur fluoride exchange (SuFEx) polymers, a novel class of environmentally friendly and sustainable polymers produced via click chemistry.35,36
To investigate how chirality manifests from the central chirality of molecules to the backbone and supramolecular chirality of polymers, we implemented a multimodal and multiscale approach (Fig. 1b-e). We leveraged bulk Chiral High-Performance Liquid Chromatography (Chiral-HPLC), CD and IR spectroscopies (Fig. 1b-c) to evaluate and compare the average chemical-structural properties of racemic and chiral monomers and polymers. AFM and AFM-IR empowered us to gain a single-molecule understanding of polymer morphology and chemical properties. Phase-controlled AFM nano-imaging allowed visualizing directly backbone helical chirality at the single-chain level, and unravelling heterogeneity among individual polymer chains and their supramolecular self-assemblies (Fig. 1d).37,38 For the first time, the development of AFM-IR enabled to correlate the morphological information with the corresponding molecular and chemical-structural properties of single-chains assemblies (Fig. 1e).
The Chiral Building Blocks, their Synthesis and Characterisation
Our framework with same connectivity but different chirality (Fig. 1a) included monomers, linkers, repeating units, and polymers (Fig. 2a-c, SI Fig. 1).18,35
We synthetised two small molecules serving as building blocks of the polymers: a chiral molecule and a linker (Fig. 2a, SI Fig. 1, SI Methods 1–3). The chiral di(sulfonimidoyl fluoride) molecule (di-SF) had two chiral centres having R (red) or S (blue) configuration (Fig. 2a). Molecules with mixed configurations were termed rac-monomers. Enantiopure chiral di-SFs molecules were termed c-monomers. The linker is a symmetric bis(phenyl ether) (di-phenol) molecule, which is non-chiral and acted as linker for the polymerization (Fig. 2a, link-monomer). The monomers and link-monomer were repetitively bonded via the SuFEx reaction to form the polymers (Fig. 2b, SI Fig. 1, SI Methods 4–5). Rac-monomers may lead to non-helical polymers termed here rac-polymer, while the c-monomers may lead to polymers with helical backbone chirality, termed here c-polymers. Gel Permeation Chromatography (GPC) determined a molecular weight (MW) and polydispersity (PDI) of (SI Fig. 2): c-polymer, Mn~209 kDa with Ð = 1.79; rac-polymer, Mn~220 kDa with Ð = 1.42.
To mimic the smallest c- or rac-repeating unit of the polymers, we synthesised a molecule consisting of a c- or rac-monomer linked to a di-phenol link-monomer with an additional single methyl cap (Fig. 2c, SI Fig. 1, SI Methods 6–7). These repeating units allowed analysing the differences in chirality between monomers and the resulting polymers, focusing on structural differences rather than variations in the composition and presence of specific functional groups.
We finally synthetised both the RR and SS enantiomers of c-monomers, c-repeating units, and c-polymers; to compare their properties with the racemic species.
Bulk Identification of Chirality
Our first aim was to prove the presence/absence of chirality in our building blocks. We applied bulk chiral HPLC (SI Fig. 3), CD (Fig. 2, SI Fig. 4) and Attenuated Total Reflection-Fourier Transform Infrared (ATR-FTIR) spectroscopy (Figs. 2–3, SI Fig. 5–7, SI Note 4).
Chiral-HPLC separated and identified different enantiomers in the monomers and repeating units (SI Fig. 3, SI Note 1). For both monomers and repeating units, we had: i) racemic (rac-) with mixed configurations (RR, RS/SR, SS); ii) enantiomer pure chiral (c-) species. Separation of chiral polymers was hampered in chiral-HPLC because of peak broadening due to polymers heterogeneity.39
We then used CD to identify chirality in all c- versus rac-species (Fig. 2d-f, SI Fig. 4, SI Note 2).40 Racemic molecules and polymers showed a flat spectrum, proving the absence of chiral asymmetry; similar behaviour was observed for isolated RS/SR monomers. All c-species (RR, SS) showed a CD spectrum with two peaks (238 nm, 290 nm), proving chiral asymmetry. However, CD does not provide direct information on the configuration of chiral centres, nor it differentiates between central chirality and new sources of chirality, such as backbone chirality emerging during polymer formation.41
To further identify chirality at the molecular level, we used ATR-FTIR spectroscopy to compare the c- versus rac-species (Figs. 2g-i, SI Figs. 5–7, SI Note 3). As symmetrical c-enantiomers (RR, SS) have the same chemical properties, we focused on the RR species. We acquired ATR-FTIR spectra in the mid-IR range (3200 − 1000 cm− 1), and used chemometrics and Principal Component Analysis (PCA) to evaluate without bias the differences between the species in the regions of 1800 − 1200 cm− 1 (Fig. 2g-i, SI Note 4), and 3200 − 2700 cm− 1 (SI Fig. 5). The spectra and PCA score plots showed minor differences between rac- vs. χ-monomers (Fig. 2g-I, SI Fig. 6, SI Note 3). Instead, we observed statistically significant differences along the first principal component (PC1) for the repeating units (61%) and polymers (77%). Within 1800 − 1200 cm− 1, these differences were associated with the: carbonyl region (C = O, 1780 − 1620 cm− 1, ~ 50%); carbon-hydrogen region (C-H, 1550 − 1330 cm− 1, ~ 20%); and sulphonyl region (S = O, 1300 − 1200 cm− 1, ~ 30%). Within the 3200 − 2700 cm− 1 (SI Fig. 6), the differences were associated to the methyl region (CH3, 2925 − 2800 cm− 1, ~ 60%). Overall, the FTIR analysis suggested that specific molecular bonds could be associated with chirality, with the largest differences observable in polymers.
Bulk Structural Differentiation of Forms of Chirality
We leveraged the molecular information contained in the IR spectra to unravel structural differences between c- versus rac-species (Fig. 3). PCA analysis (Fig. 2) focused our analysis on the most significant spectral regions: i) CH3 region, 3020 − 2800 cm− 1 (Fig. 3a); ii) the C = O region, 1780 − 1620 cm− 1 (Fig. 3b). Our system with same connectivity but different chirality allowed then associating IR spectral differences purely to different structural features.
CH3 groups are abundant and often located directly or closely to chiral centres or the backbone of polymers.42,43 Within 3020 − 2800 cm− 1 the spectra showed significant differences between rac- vs. c- species (Fig. 3a,c), which for monomers and repeating units could only be ascribed to central chirality.8,42,44 Monomers showed only minor differences (SI Fig. 5–7). The repeating units showed a significantly higher IR absorption of the aromatic CH3as and CH3s groups for the rac- than the c-species (Fig. 3c); relative to the aliphatic CH3as (2970 cm− 1) in the link-monomer. The weaker IR absorption of c-repeating units (RR) may be associated with two less co-planar aromatic CH3 groups, with fixed angle respect to the backbone (SI Note 4). While rac-repeating units had a larger set of conformations (RR, SS, RS/SR) contributing to a higher IR absorption. Thus, central chirality conformation was associated to an increased planarity of the rac-molecules compared to the c-molecules. The polymers showed also differences in the C-H region, which however may also arise from polymerisation and phenomena such as backbone and supramolecular chirality.
The C = O stretching has been widely related to the structural conformation of the backbone of (bio)-polymers.15,45 In the range 1700 − 1620 cm− 1, monomers and repeating units did no show significant differences between rac- vs. c-species, while the polymers showed statistically significant differences (Fig. 3b, SI Fig. 7, SI Note 4). Thus, the spectral differences between rac- vs. c-polymers could be ascribed only to other forms chirality arising from polymerisation and supramolecular assembly. We observed that (Fig. 3b,d): i) the C = O peak shifted to lower wavenumbers from monomers (1663 cm− 1), via the repeating units (1655 cm− 1), to polymers (1648 cm− 1); ii) c-polymers had higher absorption than rac-polymers at 1656 cm− 1; iii) c-polymers exhibited a single C = O peak, while rac-polymers showed the emergence of a second C = O peak at 1723 cm− 1.
We associated the shift to lower wavenumbers to electron delocalization caused by the backbone formation in polymers,46 moreover the increased IR absorption at 1656 cm− 1 for the c-polymers suggested the stabilisation of helices in the backbone47 (Fig. 3d, SI Note 4). The emergence of the second C = O peak for the rac-polymer could be associated with (SI Note 4): i) intramolecular interactions in their RS/SR repeating units, via a p-p resonance between the carbonyl groups and phenyl groups, inducing a planar conformation; ii) enhanced intermolecular p-p stacking of the RS/SR phenyls leading to supramolecular assembly and possibly higher chiral order.
Overall, the IR structural analysis identified general key molecular vibrations useful to differentiate central chirality in small molecules (CH3) and backbone/supramolecular chirality in polymers (C = O). However, bulk IR prevented analysing sample heterogeneity at the single-molecule level to prove: the helicity of c-polymers; and determine whether the splitting of the C = O absorption and the emergent peak at 1723 cm− 1 are related to an intramolecular planar conformation or intermolecular assembly of the rac-polymers.
Single-molecule Imaging of Chirality with Angstrom Resolution
To unravel at the single-molecule level the origin of the emergence of chirality from small molecules to polymers, we further leveraged single-molecule 3D imaging. We investigated the presence of backbone helical chirality and the supramolecular assembly state of the rac- and c-polymers with angstrom resolution imaging via phase-controlled AFM (Fig. 4, SI Fig. 8–11).37
We choose atomically-flat Highly Oriented Pyrolytic Graphite (HOPG) to preserve polymeric structure and mimic the hydrophobicity the diamond ATR-FTIR crystal (SI Fig. 8, Methods). AFM morphology maps showed significant differences in shape and size of c- versus rac-polymers (Fig. 4a, SI Fig. 9). A single-molecule statistical analysis of the volume of the polymers was conducted to estimate their molecular weight and polydispersity (SI Note 5). The c-polymers had AFM-measured molecular weight of 231.8 ± 138.9 kDa and PDI ~ 1.36; while the rac-polymers had higher molecular weight of 1219 ± 1003 kDa and PDI ~ 1.68. The comparison of single-molecule AFM and bulk GPC data (Fig. 4b, SI Fig. 2) demonstrated that c-polymers maintained a single-chain conformation, while the rac-polymers self-assembled on the surface.
Ultra-high resolution AFM 3D imaging was further performed to unravel the topological structure of single rac- and c-polymer chains. The AFM maps and the height cross-sectional profile of c-polymer chains showed a helical structure (Fig. 4c, SI Fig. 10). The rac-polymers did not show any periodic structure at the single-chain level (Fig. 4d). To quantitatively assess the presence of a helical structure in χ- vs. rac-polymers, we calculated the Fast Fourier Transform (FFT) of the cross-sectional profiles of the single-chains against any possible residual periodicity on the HOPG surface (SI Figs. 10–11). The FFT power spectra confirmed that χ-polymers had helical periodicity of ~ 40 nm, while rac-polymers single-chains had no backbone periodicity.
Although rac-polymers did not have helical conformation, the AFM analysis showed that rac-polymers further self-assembled into ordered and periodic structures (Fig. 4a,e). To unravel the supramolecular order of these rac-assemblies, we performed a single-molecule analysis of their shape as a function of their increasing size (Fig. 4e). The smallest and most abundant rac-assemblies showed: a constant height of ~ 0.4 nm (SI Fig. 10), which is the typical height of an organic chain on a surface; a transversal periodicity increasing in multiples of ~ 15 nm (Fig. 4f), corresponding to the convoluted width of a single chain. This analysis indicated that the rac-polymers initially self-assemble via lateral interactions of single-chains, which further assembly into larger supramolecular species with increasing height (Fig. 5).
Overall, the AFM single-molecule analysis was in excellent agreement with the IR structural results. AFM demonstrated that the c-polymers have a helical backbone structure, in agreement with the C = O shift at lower wavenumber and increased IR absorption at 1656 cm− 1, which is a typical signature of helices in (bio)-polymers. It then showed that rac-polymers undergo self-assembly via lateral interactions; in agreement with FTIR conclusion that emergence of a new second C = O peak at 1723 cm− 1 could be associated to intramolecular p-p resonances in the RS/SR polymeric-units with a co-planar conformation and/or their supramolecular assembly via intermolecular p-p stacking. However, structural analysis of rac-polymers by AFM nano-imaging and IR structural analysis separately could not unravel whether the emerging C = O at 1723 cm− 1 had an intra- or inter-molecular origin. To unravel the origin of this structural signature, we further applied AFM-IR nano-chemical analysis.
Achieving Single Polymer Chain Nano-Chemical Analysis
Chemical-structural analysis of single polymer chains and assemblies paves the way for understanding the molecular origin of the emergence of supramolecular order in the rac-polymers. However, to date, AFM-IR single-molecule sensitivity has been limited to molecules > 400 kDa; and only if placed at a metallic nanogap between a gold surface and AFM probe. Indeed, the use of non-metallic substrates like HOPG would cause a further drop in AFM-IR sensitivity larger than 10-folds.
To allow for the first time the chemical analysis of single polymer chains of molecular weight down to ~ 200 kDa on non-metallic surfaces such as HOPG, we developed acoustical-mechanical suppressed (AMS) AFM-IR with sub-Angstrom noise level (Fig. 5, SI Fig. 12). We leveraged composite foams (Fig. 5a, Methods) to achieve RMS noise of 25 ± 5 pm in tapping-IR and 60 ± 5 pm in contact-IR mode; allowing chemical-imaging of single monoatomic steps of HOPG with ~ 3 Å height (Fig. 5b, SI Fig. 13). To reduce AFM thermal drift and allow single-molecule localisation with high-accuracy, we stabilised room temperature with variations < 0.1°C/h (Figure SI 13).48
We then applied AMS AFM-IR to extract structural information of single polymer chains from nano-resolved IR maps and spectra as in bulk FTIR. We first focused on c-polymers (Fig. 5c-f). Tapping-IR allowed achieving multimodal imaging of morphology and IR absorption of a single polymer chain with a height down to ~ 5 Å (Fig. 5c); the phase locked loop (PLL) tracking of contact resonance frequency assured that chemical contrast was not affected by mechanical effects. Next, we pointed the AFM probe on a single polymer chain in the map (Fig. 5d) to acquire nano-localised IR spectra (Fig. 5e); for chains with a size down to ~ 0.8 nm height and ~ 5 nm width (Fig. 5f), corresponding to a molecular weight of ~ 200 kDa. While preserving better topography, tapping-IR detection was limited to the S = O region of a single c-polymer chain by low signal-to-noise ratio (SNR).
We thus employed contact-IR mode to detect the chemical properties of rac-polymers (Fig. 5g-k). Figure 5g shows a representative rac-polymer assembly. Instead of single wavenumber IR maps (Fig. 5c), we acquired 4D hyperspectral chemical maps of the polymers. Figure 5h shows four 3D-slices of the hyperspectral map, where each pixel represents an IR spectrum. The IR spectra showed the typical S = O absorption of the rac-polymer; while the HOPG surface showed significantly lower and uncorrelated IR absorption, which was subtracted to correct the spectra of the polymers (SI Fig. 13–14). The high SNR allowed acquiring nano-localised spectra of single rac-polymer assemblies in the full 1800 − 1200 cm− 1 fingerprint region (Fig. 5i-j), for assemblies with a size down to ~ 4.5 nm in height and ~ 4 nm in width, corresponding to ~ 2–4 chains (Fig. 5k, SI Fig. 14).
The emerging supramolecular C = O stretching at 1723 cm− 1 was observed in all acquired IR spectra of different rac-assemblies (SI Fig. 14), but with significant variations between different assemblies. We thus proceeded to a statistical analysis of their nano-chemical heterogeneity by AMS AFM-IR.
Unravelling the Origin of the Emergence of Supramolecular Chirality
We leveraged our AFM-IR unprecedented sensitivity to investigate how the chemical signature of the rac-polymers varied as a function of their size and supramolecular state, to unravel if the emerging C = O at 1723 cm− 1 is related to intra- or inter-molecular interactions (Fig. 6).
We acquired AFM-IR spectra from assemblies with varying size, from the smallest one (Fig. 6a, ~ 4.5 nm height; ~4 nm deconvoluted radius) to the largest ones (Fig. 6b, ~ 10 nm height; ~6 nm deconvoluted radius). Although at the limit of AFM-IR sensitivity, smaller assemblies presented lower IR signal than larger ones in accordance with Beer-Lambert Law (Fig. 6c, SI Fig. 14). We then normalized the spectra to their maximum, to reduce variability due to signal intensity and ran a PCA analysis in the full fingerprint region (1750 − 1200 cm− 1). The PCA analysis showed that smaller (3.0–6.4 nm) and larger assemblies (7.5–15 nm) belong to separate spectroscopic clusters (Fig. 6c, 95% confidence). Significant part of the variance related to the conformation of the C = O group (Fig. 6e), suggesting that rac-polymers of different sizes have different structural properties.
We consequently focused our analysis on the C = O groups (Fig. 6f, 1780 − 1600 cm− 1), which were associated by IR to intra- and/or inter-molecular interactions in the rac-polymers. We normalised the spectra to the backbone C = O peak at 1646 cm− 1, to investigate the structural differences related to the emergent C = O peak at 1723 cm− 1, independently of the number of C = O bonds present in the assemblies. The shape of the emergent C = O peak was significantly different for smaller vs. larger assemblies; as proved by PCA analysis of the second derivatives of the spectra (Fig. 6g, 95% confidence). The PC1 loading plot of the second derivatives (Fig. 6h, 81% variance) proved that larger assemblies had increased absorption of the emergent C = O peak, which also shifted its maximum at higher wavenumber (1730 cm− 1), compared to smaller assemblies (1710 cm− 1). Since we normalised the spectra to the backbone C = O peak (1646 cm− 1), these modifications could only be associated to an increasing number of intermolecular (p-p stacking) interactions between the rac-polymer chains in the larger assemblies; rather than the constant number of intramolecular interactions in the single chains.
Overall, AMS AFM-IR proved the ability of studying the heterogeneous chemical-structural properties of single polymeric assemblies, composed of only few chains. The analysis enabled identifying the splitting of the C = O group absorption and the second peak at 1725 cm− 1 as key signature of intermolecular p-p stacking and the emergence of supramolecular order and chirality in the rac-polymers (Fig. 6i).