Structures of ( S/R -HP1A)PbBr 3 and (S/R-HP1A)PbI3. Both bromide and iodide-based compounds crystallize in the non-centrosymmetric polar space group P21 and are isostructural (Supplementary Table 1). Asymmetric units contain two cations and two PbX3 units (Supplementary Fig. 1). The 1D network is based on double-chains of edge-sharing PbX6 octahedra running along the crystallographic [100] direction. The polar axis is thus perpendicular to the propagation of the chains. Both independent octahedra are strongly distorted due to two different Pb–X bond distances (Supplementary Table 2). The shortest distance can be found between the metal center and the only terminal halide ion, while the longest distance point towards the center of the double-chain. Numerous N-H···X intermolecular interactions can be observed between cations and the terminal halides (Supplementary Fig. 2). On the other hand, the equatorial Pb–X bond distances have similar values (from 2.965(8) to 3.109(7) Å in (S/R-HP1A)PbBr3 and from 3.142(3) to 3.294(2) Å in (S/R-HP1A)PbI3). At first, the two pairs of edge-sharing octahedra from both enantiomeric compounds appear to share the same distortions and thus to be superimposable. However, by carefully looking in the equatorial plane, one can observe that the distribution of the bond distances values is clearly inverted between both enantiomeric compounds (Fig. 2). These mirror-image octahedra clearly highlight the chiral nature of the inorganic network.
These compounds share many similarities such as the crystal structure, symmetry, nature of the metal and of the cation, distortion within the inorganic network, and only differ from each other by the nature of the halide. Therefore, they represent interesting materials to disclose the role of halide composition in the chiroptical properties. Thin films of (S/R-HP1A)PbBr3 and (S/R-HP1A)PbI3 were prepared using the spin-coating method starting from DMF solution of pure crystalline samples, using the typical procedure previously reported (see Supplementary Information for experimental procedure). X-ray powder diffraction (XRPD) confirmed the phase and purity for both of the crystalline powders and in the thin films (Supplementary Figs. 3, 4 and 5).
Chiroptical activity in ( S/R -HP1A)PbBr 3 . CD measurements in transmission mode were first performed by orienting the deposited film towards the light source (i.e. “front” position). The film was rotating along the optical axis and the CD response was measured for eight different azimuth angles θ, with a step of 45° between each measurement (Supplementary Fig. 6). Indeed, as previously described for organic molecules and coordination complexes, the small artefacts resulting from the third and fourth terms in Eq. (1) could be corrected by averaging the experimental values obtained at different θ angles. The same procedure was applied for the two enantiomeric compounds, and the averaged spectra are presented in Fig. 3a. However, this first series of measurements gave no mirror-image spectra between (S-HP1A)PbBr3 and (R-HP1A)PbBr3. Although the signals appear at similar wavelengths associated to the excitonic peak of the inorganic 1D network, with a sign inversion between the two enantiomeric compounds, the relative peak intensities were very different from each other. Since they were prepared starting from the same concentration of the precursor solution, thicknesses are expected to be very similar between both films. This is further confirmed by the relative absorbance, the same for both compounds (around 1.1). Using the formula gabs = CD / 32980 x Abs, it resulted in apparent dissymmetry factors of + 5.10− 3 and − 1.10− 2 for (S-HP1A)PbBr3 and (R-HP1A)PbBr3, respectively. Very different results were obtained by flipping the sample and measuring the CD spectra in the “back” position (thin film oriented towards the detector), particularly in the case of (R-HP1A)PbBr3 for which the spectrum was almost inverted. These results confirm an important contribution of aLDLB effects in these crystalline films. Using the equations
$${CD}_{iso}= \frac{1}{2}({CD}_{front}+{CD}_{back})$$
3
$$aLDLB= \frac{1}{2}({CD}_{front}-{CD}_{back})$$
4
it was possible to isolate the contribution of both terms (Fig. 3b). As expected, the CDiso display mirror-image spectra between the two enantiomers. This finally allows to obtain accurate gabs factors of + 1.10− 3 and − 1.10− 3 for (S-HP1A)PbBr3 and (R-HP1A)PbBr3, respectively. On the other hand, the aLDLB effects present different intensities between (S-HP1A)PbBr3 and (R-HP1A)PbBr3, almost four times more intense in the latter. This stronger aLDLB explains the sign inversion in (R-HP1A)PbBr3 between CDfront and CDback, which was not observed in the compound based on the S enantiomer.
What is the rationale of the different aLDLB between the two enantiomers? In chiral molecules, while CDiso has a pure excitonic origin, aLDLB effects in turn are related to local anisotropy and the non-alignment of LD and LB transition dipoles. In the case of crystalline thin films particularly, as for metal-halide semiconductors, epitaxial growth can be strongly dependent on sample preparation. Therefore, at this step, we hypothesized the possible impact of the spin-coating procedure, precisely the direction of the rotation during deposition (clockwise vs counter-clockwise). Indeed, as any chiral molecule or assembly can be associated to a helical potential, this rotation direction is expected to influence the supramolecular assembly of two enantiomeric compounds to a varying extent. For many years, similar mechano-chiral effects have been demonstrated in (metal-)organic compounds. For example, J. M. Ribó et al. have shown the influence of rotatory evaporation of solutions for the chiral induction in porphyrin-based J-aggregates. In our typical spin-coated thin film fabrication procedure, the direction of rotation was configured clockwise. Therefore, a new series of thin films of (S/R-HP1A)PbBr3 was prepared, using the same precursor solutions and procedure but changing the rotation direction of the spin-coating (i.e. counter-clockwise). CD measurements for both front and back positions of the thin film strongly differ from the measurements performed with the clockwise preparation (Fig. 3c). The same equations were used for the calculation of CDiso and aLDLB (Fig. 3d). The obtained CDiso spectra mostly present the same shape and values as in the clockwise configuration, with a slight difference at lower energy (around 330 nm). This was expected due to the same chemical nature of the thin films. On the other hand, aLDLB was strongly affected by the clockwise/counter-clockwise parameter. It is rather difficult to quantitatively analyze the effect of the rotation on the intensity of these macroscopic effects, although it appeared from a large set of measurements that aLDLB tends to increase (in absolute value) for clockwise rotation compared to counter-clockwise in (S-HP1A)PbBr3 and tends to decrease in (R-HP1A)PbBr3. XRPD measurements were performed for each compound and spin-coating configuration (Supplementary Fig. 7). The strong similarity between diffractograms, with almost no difference in peak relative intensities, shows that both chemical composition and crystal orientation inside the thin films remained unchanged with sample preparation. The absolute absorbance, which was the same for all compounds (Supplementary Fig. 8), also supports the thickness integrity among the series. On the other hand, optical microscopy images showed that the compounds presenting the strongest aLDLB signals have stronger roughness with the presence of larger crystal domains (Supplementary Fig. 9). Therefore, these results suggest that the synergy between the nature of the enantiomer and spin-coating rotation direction can influence the presence and size of crystal domains, and that larger the macroscopic anisotropy, stronger could be the aLDLB effects. Finally, starting the rotation before the deposition of the precursor solution could strongly decrease the strength of aLDLB in both films (Supplementary Fig. 10). These results reveal a real impact of the spin-coating parameters on the supramolecular assembly of chiral molecular materials.
At this step, the presence of the symmetric LDLB term in the CD response has not yet been investigated. According to the theory of Tempelaar et al., sLDLB is invariant upon sample flipping and should in principle give bisignate profile of the spectrum, although it has a completely different origin than the Cotton effect in CD.12 How to experimentally discriminate between genuine CD and sLDLB? As far as CDiso is concerned, the gabs factor should be independent to the film thickness l. Indeed, CDiso is a single-scattering event (scaling at l) and results in the thickness-independent gabs factor (CD normalized by absorbance). On the other hand, both aLDLB and sLDLB terms scale as l², since they are considered as double-scattering events. Therefore, the dissymmetry factors corresponding to aLDLB and sLDLB should linearly increase with sample thickness. Very recently, thickness-dependent CD studies were performed on thin films and KBr pellets of 1D and 2D chiral metal-halide compounds, however these measurements could only highlight the possible CD signal overload and absorbance saturation region. In order to confirm the presence or absence of sLDLB in (S/R-HP1A)PbBr3, thin films of different thicknesses were prepared and characterized. Three different concentrations of the precursor solutions have been selected (0.2, 0.5 and 1.0 M based on the formula unit). For each concentration, thin films were prepared under both clockwise and counter-clockwise rotation and CD measurements were performed as previously described (Supplementary Fig. 11). According to surface profilometry measurements, these concentrations resulted in thin films of thicknesses of around 20, 60 and 130 nm (± 10 nm and average roughness of about 5 nm) for 0.2, 0.5 and 1.0 M concentration, respectively. Nevertheless, the apparent and genuine gabs factors were plot as function of the precursor solution concentrations for more accuracy (Fig. 4). The apparent dissymmetry factors can be strongly modulated due to the increasing contribution of aLDLB by increasing film thickness. Although a linear increase of the apparent gabs can be observed in some cases (as depicted by dashed lines in Fig. 4), the evolution of this factor doesn’t necessarily follow any function due to competing effect between CDiso and aLDLB (sometimes being opposite in sign at a given wavelength) and because the aLDLB can be affected by sample preparation, such as clockwise/counter-clockwise spin-coating rotation. After averaging CDfront and CDback measurements, the obtained dissymmetry factors are the same for the three concentrations, confirming the genuine gabs values of about + 1.10− 3 and − 1.10− 3 for (S-HP1A)PbBr3 and (R-HP1A)PbBr3, respectively. This constancy across the range of concentration also confirms the absence of thickness-dependent sLDLB in this series of thin films. It can be concluded that only CDiso and aLDLB are present in the bromide-based hybrid compounds. Finally, it appears that the aLDLB effects almost vanish for the lowest concentration of 0.2 M (Supplementary Fig. 11), proving that a pure molecular CD can be obtained under these conditions.
Previous reports on solid-state CD of (metal-)organic materials reasonably described aLDLB as a reproducible effect rather than an experimental artefact, which allows to make use of these antisymmetric effects in practical devices.9 However, it appears from the present measurements on (S/R-HP1A)PbBr3 that aLDLB can be strongly dependent on sample preparation, pointing out the importance of describing all the selected conditions of sample preparation before reporting any dissymmetry factor. It also reveals the possibility to modulate such conditions in order to optimize the CDiso/aLDLB ratio for the target application.
Chiroptical activity in ( S/R -HP1A)PbI 3 . Lead-iodide compounds have been reported to exhibit the strongest dissymmetry factors, usually between three and five times the values obtained with the lead-bromide counterparts. In the present example, thin films of (S/R-HP1A)PbI3 prepared in the same conditions as (S/R-HP1A)PbBr3 exhibit much stronger CD signals (reaching a degree-scale), suggesting very high dissymmetry factors in the iodide-based compounds (Supplementary Fig. 12). With such amplitude of the CD signal, (S/R-HP1A)PbI3 appeared to be good candidates to perform CD measurements on a series of films prepared from a large range of concentrations of the precursor solution, ranging from 0.06 to 1.0 M based on the formula unit. At each concentration, the thin films presented an excitonic peak located at 417 nm (Fig. 5a). The evolution of the absolute absorbance between samples indicates a linear increase of the film thickness with the concentration and further confirms the integrity of the chemical composition. This linear increase was also suggested based on profilometry surface measurements, with a thickness range estimated at between 20 nm (at 0.2 M) and 150 nm (1.0 M). The evolution in the chiroptical response across the series is less straightforward (Fig. 5b). It thus appears important to verify the presence of artefacts, aLDLB and sLDLB effects for each concentration.
For example, the thin film of (S-HP1A)PbI3 prepared from a 0.3 M precursor solution (absolute absorbance of around 0.8) shows a small but measurable azimuth angle dependence on both CD and LD values (Supplementary Fig. 13a-b). While the apparent gabs factor was first estimated at + 0.0229, this value was lowered to + 0.0227 after artefacts corrections. It shows that the thin film presents negligible LD. Finally, as confirmed by the superposition of CD measured in both front and back positions (Supplementary Fig. 13c), no measurable aLDLB contribution could be observed, thus confirming (at this step) the calculated gabs for such concentration. We can conclude that both artefacts and aLDLB can here be disregarded. For this reason, such iodide-based thin films don’t behave differently depending on the direction of the spin-coating rotation. Similarly, these corrections have been applied for all concentrations and both compounds, confirming the absence of measurable artefacts and aLDLB in the entire series of thin films. On the other hand, strong modulations both in shape and values of CD spectra could be observed for concentrations of 0.9 and 1.0 M, a clear indication of signal saturation for such high concentrations. In the range 0.06–0.8 M, a non-linear evolution of the maximum CD values was also observed, at the difference of the absolute absorbance (Fig. 5c). Once normalized by the absorption, the values of dissymmetry factor at 417 nm were plot as function of concentration (Fig. 5d). A strong evolution appeared, with an increase of gabs from ± 1.10− 3 to ± 4.10− 2. The strong thickness-dependence suggests that the CD signals are dominated by sLDLB and co-exist with genuine CDiso in (S/R-HP1A)PbI3 thin films. Indeed, the gabs factors for the lowest thicknesses are calculated at + 1.10− 3 and − 1.10− 3 for (S-HP1A)PbI3 and (R-HP1A)PbI3, respectively. These constant values obtained from 0.06 to 0.12 M suggest the pure molecular origin (CDiso) of the CD response. At higher concentration, the gabs factor linearly increases until reaching a plateau at 0.6 M, supporting the presence of sLDLB which increases the dissymmetry factors up to ± 4.10− 2. The results also demonstrate the saturation of the CD signal at the highest thicknesses. Finally, it should be noted that very similar results were obtained by replacing the FTO substrates for glass substrates (Supplementary Fig. 14).
Molecular origins of chiroptical activity. Surprisingly, the same values of genuine dissymmetry factors are obtained for both (S/R-HP1A)PbBr3 and (S/R-HP1A)PbI3 thin films. It is rather contradictory with previous reports that highlighted the larger CD observed in iodide compounds compared to bromide compounds, although it can now be understood that such comparative studies possibly relied on uncomplete characterizations. In addition, such studies were lacking of isostructural bromide and iodide-based compounds. For these reasons, it was challenging to study the influence of chemical and structural properties in the chiroptical activity of metal-halide semiconductors. In this context, two different hypotheses were recently proposed. On the one hand, it was suggested that the higher CD responses observed in iodide networks compared to bromide networks could be explained based on the stronger spin-orbit coupling (SOC) in iodide compounds due to the smaller exciton binding energy. On the other hand, the fine modulation of the structure in nano-confined chiral perovskites suggested that the strength of hydrogen-bonding interactions between chiral cation and inorganic network could be considered as a key parameter for the chiroptical activity.15 These two explanations on the origin of CD in metal-halide networks can be complementary. In the light of the present results, the hypothesis on intermolecular interactions seems to have more influence on the genuine CD, since the hydrogen-bonding contacts are strictly the same in both (S/R-HP1A)PbBr3 and (S/R-HP1A)PbI3. This has also been proposed to be at the origin of chirality transfer strength in a series of 2D chiral halide perovskites. Nevertheless, the first hypothesis on the influence of chirality-induced SOC cannot be disregarded, and could be envisaged to slightly influence the gabs values, even though it was not observed in the present compounds.
Macroscopic effects on chiroptical activity. Despite the similar natural optical activity of both bromide and iodide-based hybrid compounds (with genuine gabs of ± 1.10− 3), it clearly appears that the nature of the halide has a predominant role in the overall chiroptical response. Here, iodide-based compounds show an apparent dissymmetry factor of one order of magnitude higher than the bromide counterparts. As mentioned in the introduction, in the previous studies on chiroptical properties of metal-halide semiconductors, the nature of the halide (Br vs I) influenced the gabs values in a lesser extent. This was due to the fact that in both bromide and iodide-based 1D networks, a non-negligible aLDLB effect was observed and the calculated gabs values were comprised between 10− 4 and 10− 3. To the best of our knowledge, no sLDLB effects were previously observed in chiral metal-halide semiconductors. Here, aLDLB effects could be observed only in (S/R-HP1A)PbBr3, keeping the apparent dissymmetry factors in the range between 10− 3 and 10− 2. On the other hand, the sLDLB effect observed in (S/R-HP1A)PbI3 resulted in an increase of this factor up to 4.10− 2. The reason for different linear contributions in two isostructural compounds may be explained by a different crystal growth of the thin films. (R-HP1A)PbBr3 and (R-HP1A)PbI3 thin films displayed very similar XRPD patterns, with the presence of only (0kl) peaks that confirmed the predominant deposition of the hybrid compound with the propagation direction of the 1D chain parallel to the substrate (Fig. 6a, d). However, a stronger intensity of the peaks associated to the (002) and (040) planes could be observed in (R-HP1A)PbBr3 (highlighted in blue), as well as the presence of (051) and (071) peaks (highlighted in green). These additional peaks indicated a lower preferential orientation of the 1D chains in (R-HP1A)PbBr3 compared to (R-HP1A)PbI3 and could be at the origin of the very different thin film textures (Fig. 6b, e). As mentioned in the chiroptical characterization of (S/R-HP1A)PbBr3, the CD responses were associated to the presence of large crystal domains in the thin films (Fig. 6c), with roughness that could be dependent on sample preparation. For (S/R-HP1A)PbI3 however, optical microscopy images of the thin films under unpolarized and polarized light confirm the absence of such large crystal domains and reveal an homogeneous texture of the films (Fig. 6f) with a small and gradual increase of the roughness by increasing the film thickness (Supplementary Fig. 15). In conclusion, this huge difference of CD signal observed between thin films based on isostructural compounds and prepared under the same conditions couldn’t be explained based on the sole effect of SOC, but rather arises from different crystal growth and therefore different macroscopic effects inside the thin films, here leading in the case of bromide-based compounds to aLDLB effects due to large crystal domains, and to sLDLB in iodide-based compounds due to the twist angle between stacks of molecules, according to Tempelaar’s model.12 The discussion can be extended to 2D metal-halide networks. As mentioned in the introduction, 1D networks have shown dissymmetry factors of more than one order of magnitude higher than 2D based compounds. For example, the 2D chiral compounds of formula (S/R-HP1A)PbBr4 displayed a gabs of ± 10− 4,16 one order of magnitude lower that the 1D networks obtained starting from the same chiral cation. Such results were understood based on the increase of exciton broadening (due to lower quantum confinement) when increasing the dimensionality of the system.13 However, on the basis of the new results obtained with (S/R-HP1A)PbBr3 and (S/R-HP1A)PbI3, it is strongly envisaged that not only the SOC could influence the CD strength. In layered-based metal-halide networks, the strong preferential orientation on the substrate and optimal overlap between 2D layers naturally prevents crystal domains or twist angle between stacks, and therefore prevents the generation of (antisymmetric and symmetric) LDLB effects. On the other hand, it was observed for many years that 1D metal-halide compounds can show non-homogeneous deposition on surfaces, as observed in (S/R-HP1A)PbBr3. Furthermore, the results obtained with (S/R-HP1A)PbI3 shed light on the strong effect due to the twist angle and non-homogeneous stacking of the molecules on the surface. Therefore, it is reasonable to conclude that 1D metal-halide networks are more interesting candidates than 2D compounds for the generation and modulation of CD strength by controlling aLDLB and sLDLB effects.
Chiroptical activity measurements on metal-halide networks. The present results highlight the importance of measuring the chiroptical properties on thin films of various thicknesses and in various orientations for a same enantiomer. Indeed, it is crucial to analyze all possible macroscopic effects in order to perform a comparative study and conclude about the molecular origins of CD in metal-halide networks. Considering the new hypotheses, some compounds may therefore be re-investigated. For example, one of the first CD study of 1D lead-iodide networks reported a gabs factor of ± 0.02 (Supplementary Fig. 16).7 Interestingly, the compounds of formula (S/R-α-PEA)PbI3 (α-PEA = α-phenylethylammonium) presented a different type of 1D network than in (S/R-HP1A)PbI3 (single-chain of face-sharing octahedra in (S/R-α-PEA)PbI3, double-chain of edge-sharing octahedra in (S/R-HP1A)PbI3), but both compounds share similar apparent gabs factors at the given absorbance of 1 (± 0.02 in (S/R-HP1A)PbI3 at 0.4 M concentration). In the first place, these results could suggest that in 1D networks, the chiroptical properties would be influenced by the nature of the halide rather than the type of connectivity between metal-halide octahedra. By extension, one could anticipate a possible predominance of sLDLB in the CD responses of (S/R-α-PEA)PbI3, such as it was observed in (S/R-HP1A)PbI3. However, such assumptions will be strongly revised after a careful comparative study. Despite proper preparation conditions of (S/R-α-PEA)PbI3 thin films (precursor solution at 0.3 M, discarding any possible saturation of the CD signal) and the highly preferred orientation of the 1D chains as observed by XRPD, crystal domains were observed from optical microscopy images under polarized light. The authors also mentioned the possible presence of LD and LB due to a slight modulation of the CD response at different azimuth angle measurements. Still, the presence or absence of aLDLB couldn’t be confirmed since the measurements were performed for only one position (front or back) of the thin film. On the other hand, thickness-dependence measurements were not performed in order to discriminate between molecular CD and sLDLB. More recently, H. Lu et al. reported the crystal structure and solid-state CD of (R-EBA)PbI3 (EBA = ethylbenzylammonium), which is isostructural to (S/R-α-PEA)PbI3 due to the similarity in the cation structure (Supplementary Fig. 16a).13 Interestingly, optical microscopy images of the film with crossed-linear polarizers revealed a similar aspect than in (S/R-α-PEA)PbI3 thin films with large crystal domains. The chiroptical activity of (R-EBA)PbI3 was analyzed in more details, showing strong aLDLB that could invert the CD response upon sample flipping. After extracting the aLDLB contribution from the apparent CD signal, the calculated gabs factor in (R-EBA)PbI3 was found at around − 7.10− 4 at all thicknesses, demonstrating that the authors could disclose the pure molecular chirality of the material. It is therefore strongly envisaged that the strong CD responses obtained in the isostructural (S/R-α-PEA)PbI3 compounds were mainly altered by the presence of antisymmetric LDLB, at the difference of the predominant sLDLB in thin films of (S/R-HP1A)PbI3. The observation of such different behaviors between two systems with similar chemical composition, which differ only by the organic cation used and the shape of the inorganic chain, further support the importance of hydrogen-bonding interactions between chiral organic cation and metal-halide network. A complete study on a large number of metal-halide networks would be necessary in order to reveal the relative contribution of molecular and macroscopic features on the ratio of CD/aLDLB/sLDLB, and to fully describe the polarization ability of 1D chiral metal-halide semiconductors.
Through all these considerations, a guide for measuring CD of chiral metal-halide semiconductors with total accuracy can now be provided. To be fully validated, the gabs value must be reported for thin films prepared from non-saturated precursor solutions. The absence of artefacts must be controlled by performing measurements at various azimuth angles. More importantly, the presence/absence of aLDLB must be confirmed by flipping the film and comparing both front and back positions measurements. In the case of predominant aLDLB over CDiso, as observed in 1D lead-bromide networks, changing the direction of the rotation during the spin coating process can strongly modulate the CDiso/aLDLB ratio. This effect can be strongly minimized in the case of highly oriented thin films, which can be demonstrated by XRPD and the absence of crystal domains by polarized optical microscopy images. The CDiso or sLDLB nature of the chiroptical response can only be confirmed by the preparation and characterization of various thin films of different thicknesses, i.e., starting from different concentrations of the precursor solutions. Finally, to be fully reproducible, experimental procedures should describe all sample preparation parameters (duration of spin-coating, rpm speed), including the clockwise or counter-clockwise rotation during the spin-coating process.