Jie
Zou‡
abc,
Yi
Pan‡
b,
Di
Zhang
abc,
Jie
Zhang
ab,
Weilong
Chen
abc,
Hongyan
Sun
ab,
Kai-Chung
Lau
*b and
Jingdong
Luo
*abc
aShenzhen Research Institute, City University of Hong Kong, Shenzhen 518057, China
bDepartment of Chemistry, City University of Hong Kong, Kowloon, Hong Kong SAR, China. E-mail: kaichung@cityu.edu.hk; jingdluo@cityu.edu.hk
cHong Kong Institute for Clean Energy, City University of Hong Kong, Kowloon, Hong Kong SAR, China
First published on 9th April 2024
We report an unexpected direct transfer hydrogenation (TH) from 8-alkoxyjulolidine as a strong electron donor to tricyanofuran-based hemicyanines and tandem condensation for the synthesis of push–pull chromophores. Density functional theory calculations identify the α-hydride transfer of julolidine derivatives, instead of nucleophilic addition, as the key initial step of reaction cascades, followed by efficient β-deprotonation to generate a nucleophilic enamine of isojuloline that condenses in situ with a second equivalent of hemicyanines as electrophiles. The scheme represents an exceptional example of TH reactions from 8-alkoxyjulolidine to highly polarizable hemicyanines, notably under simple reaction conditions without using potent catalysts of metal complexes or Lewis acids. The resulting products of isojuloline-based merocyanines exhibit strong near-infrared (NIR) absorption and large first hyperpolarizabilities. The study suggests that direct TH can be exploited to construct new NIR merocyanines by following rational, regioselective condensation pathways for photonic applications.
Only very few methods have been reported to enable the TH reactions from tertiary amines, such as by reacting diethyl azodicarboxylate (DEAD) with tertiary amines through nucleophilic addition to form zwitterionic intermediates (Fig. 1A),4 or by selecting elegant noble/transition metal complexes (Fig. 1B).5 Furthermore, beyond a better synthetic capability in constructing molecular complexity is the lack of examples using TH-enabled reactions in the synthesis of amine-containing functional molecules and molecular materials. Most tertiary amine products were obtained through addition reactions to the sp2 carbon of iminium or enamine intermediates, keeping the sp3 hybridization for the carbon centers unchanged after the transformation (Fig. 1B).4–7 Therefore, it is highly desirable to functionalize the tertiary anilines as the electron donor of π-conjugated molecules through efficient TH reactions of polarizable olefins, which would mark a significant step toward the exploration of new molecular structures for potential applications in nonlinear optics,8–10 fluorescent bioimaging,11,12 and phototheranostics.13,14
In the meantime there is growing interest in concise synthesis of push–pull polymethines containing the tricyanofuran (TCF) acceptors for photonic applications (Fig. 2A–C).8–10,15 Many researchers are particularly interested in the study of modular reaction sequences, consisting of consecutive multiple-step, one-pot reactions, for efficient synthesis of key intermediates and push–pull chromophores. In our quest of simple and highly efficient synthetic protocols for new TCF-based push–pull structures, the present work was initiated in studying the feasibility of direct condensation between the julolidine donors and TCF-based hemicyanine precursors, which are considered as among the most widely used conjugated units with structural diversity and synthetic scalability.8–10,15–18 Following the condensation with methylene bases, we expected that the C-9 position of the substituted phenyl ring would be the predominant nucleophilic site for julolidines to react with π-conjugated TCF hemicyanines and form type-I products under normal conditions (Path I, Fig. 2D). Interestingly, we obtained an unexpected major product (type-II product) of a fully π-conjugated push–pull structure containing an isojuloline donor, π-conjugated bridge, and TCF acceptor (Path II, Fig. 2D), in which the isojuloline group is the dehydrogenated product from the julolidine substrate.
The above critical literature review and preliminary findings motivated us to conduct this study for new push–pull molecules to understand the mechanisms of tandem reactions and material properties. We have investigated the TH from 8-alkoxyjulolidine as a strong electron donor to π-conjugated TCF-type hemicyanines and tandem condensation for the synthesis of push–pull chromophores, as well as the effect of acceptor substituents on regioselectivity of two distinct pathways (Fig. 2D). Density-functional theory (DFT) calculations and spectroscopic analyses demonstrate that the TH process involves direct transfer of both α and β hydrogens of 8-alkoxyjulolidine to the CC double bond of highly polarizable TCF-type hemicyanines, notably under simple reaction conditions without using potent catalysts of noble metal complexes or forming very reactive intermediates through nucleophilic addition (Fig. 1). The formed isojuloline intermediate can be condensed with TCF acceptor precursors in situ to access fully π-conjugated push–pull structures. The discovery of this new method has the potential to overcome the issues of non-conjugated functionalization of amine donors through TH and narrow substrate scope under harsh reaction additions and synthesize diverse push–pull chromophores that could not otherwise be obtained.
Entry | Solvent | JD1/TCF-A1 ratio | Temp. (°C) | Time (h) | Yield (%) | |
---|---|---|---|---|---|---|
TE-I-1 | TE-II-1 | |||||
1 | Ethanol | 1.1 | 70 | 72 | Trace | 15 |
2 | Ac2O | 1.1 | 80 | 14 | 3 | 27 |
3 | Ac2O | 0.6 | 80 | 14 | 2 | 26 |
4 | Ac2O | 2.1 | 80 | 18 | 3 | 29 |
5 | Ac2O | 3.1 | 80 | 18 | Trace | 16 |
In refluxing ethanol with a catalytic amount of pyridine (Pyr), we obtained only a low 15% yield of type-II push–pull tetraene “abnormal byproduct” TE-II-1 and no “normal” product or only a trace amount of type-I phenyltriene TE-I-1 product was detected (Table 1, entry 1). In acetic anhydride, the yields of isolated product TE-II-1 have been improved to 26–29% by changing the reaction time and equivalence ratio of reactants, while TE-I-1 remained the minor product (Table 1, entries 2–4). Compared to TE-I-1, TE-II-1 showed two more downfield proton NMR signals of π-bonds at the expense of two upfield proton signals of the methylene group located around 3.5 ppm, plus a small amount of a geometrical isomer. The mass data and crystal structure further verified the structure of TE-II-1. A low yield of TE-I-1 indicates JD1 as a π-nucleophile is less reactive than methylene bases, such as indoline, benzoindoline, benzothiazoline, and Michler's base derivatives, in condensation with π-electrophile TCF-A1. This can be explained by the aromaticity and steric hindrance of JD1 that reduces its nucleophilicity, and the unexpected dehydrogenation reactivity.
For the formation of TE-II-1 (“abnormal” product), there are two possible reaction pathways: (i) the nucleophilic addition (NA) of nitrogen lone pair of JD1 to TCF-A1 and (ii) the intermolecular hydride transfer (HT) from JD1 to TCF-A1. Herein the hemicyanine TCF-A1 could act as either a hydride acceptor or an electrophile in the reactions. Also, as reported by Li et al.,4,23 the strong electrophile of DEAD can be used to form the zwitterionic adducts with the tertiary amines to activate the α and β hydrogens for further HT (Fig. 1A), which led to the analysis of the NA pathway in this study. Due to the relatively long π-conjugation bridge of TCF-A1, we have screened multiple carbon centers in both the HT and NA pathways. As both the JD1 and TCF-A1 are of considerable size, to help saving the computational cost of DFT calculations, we have (i) replaced the n-butoxy with a methoxy group in the JD1, the resultant molecule (8-methoxyjulolidine) is named as JD1′; (ii) substituted a t-butyl group in the chlorocyclohexene ring of TCF-A1 with a hydrogen, and the molecule is named TCF-A2 (see Table 2 and Scheme S3, ESI†).
a The isolated product yields were reported after flash column chromatography and recrystallization. Optimized reaction time was included. b For entry 1 and 5, the values in parentheses are the calculated highest activation energy (ΔG‡298, kcal mol−1) in acetic anhydride at the B3LYP-D3(BJ)/ma-TZVP level. The n-butoxy group of JD1 was replaced by the methoxy group to save computational time. See Fig. 4 and the ESI for detailed PES and structures. c Entry 8 uses the unsubstituted julolidine JD0 under the same reaction condition for comparison. |
---|
![]() |
Three carbon centers (marked with an asterisk in Fig. S1, ESI†) along the conjugated double bonds of TCF-A2 have been identified by their electrophilicity, i.e., positive atomic charge, for the preliminary calculations of HT. Our preliminary single-point calculations at the B3LYP-D3(BJ)/def2-TZVP//B3LYP-D3(BJ)/def2-SVP level suggest that the HT pathway would be favored with a modest barrier of 26.5 kcal mol−1 by attacking the ethoxylated ethylene terminal, and 26.3 kcal mol−1 by attacking the central ethylene, both possibly as the initial step (Table S1, ESI†).24–26 The HT at the dicyanovinylene terminal and NA pathway can be ruled out due to much higher barriers of 36.5 kcal mol−1 and 37.0 kcal mol−1 (Table S1, ESI†), respectively, for which considerable steric congestion can be expected.
We further explored the detailed reaction mechanism of normal C-9 condensation (Path I) and HT pathway (Path II) at the B3LYP-D3(BJ)/ma-TZVP level, for improving the accuracy of both geometrical optimizations and energy corrections, and obtained the relevant PES (Fig. 4, TCF-A2-based in black color) and a complete set of intermediates (INTs) and TSs for the reaction (Fig. S2, ESI†).
In Path II, the HT step firstly proceeds, starting from an hydrogen-bonded intermediate (INT-II-A) formed between JD1′ and TCF-A2, viaTS-II-A where it has two relatively short partial C⋯H bonds around 1.39 Å and a slightly bent Cα⋯H⋯C angle of 166.7°, likely imposed by noncovalent π–π interactions between the donor JD1′ and acceptor TCF-A2 (Fig. 4 red line and 3D structure of TS-II-A). The barrier height (ΔG‡298) of TS-II-A is 27.0 kcal mol−1, relative to JD1′ and TCF-A2. In addition, different from most tertiary amines involving HT with an sp3-hybridized nitrogen center, the nitrogen in the JD1′ is sp2-hybridized to permit more efficient conjugation with the aromatic nucleus.27 Thus the lone pair of nitrogen would impose a better stabilization effect on the α-carbocation in TS-II-A through resonance. The HT proceeds over TS-II-A to give a weakly bound complex intermediate INT-II-B, which is resonance stabilized. Next, like the TS-II-A in HT, two short partial C⋯H bonds around 1.45 Å and a Cβ⋯H⋯C angle of 173.4° were found in TS-II-B with a ΔG‡298 value of 28.3 kcal mol−1, which is slightly higher than that of TS-II-A. The subsequent proton transfer (PT, Fig. 4, blue line) completes the TH reaction to afford the JD1′(−2H), as the dehydrogenated product JD1′(−2H) of JD1′, and TCF-A2(+2H) as the hydrogenated product (in terminal ethoxy-substituted ethylene) of TCF-A2. Then, the in situ generated JD1′(−2H) as an enamine π-nucleophile undergoes the condensation reaction with another TCF-A2 to give the type-II product via several mechanistic steps starting from the intermediate (INT-II-C) by following the common addition/elimination pathway. The ΔG‡298 of TS-II-C is clearly lower than the ΔG‡298 of TS-II-A, suggesting that the reaction cascade would be controlled by HT or PT, depending on the substituents on the TCF acceptors, as the rate-determining step. Furthermore, the “normal” condensation (Path I) has been investigated to obtain a more thorough understanding of the reactivity of the substrates for comparison (Fig. 4, black line on the left). We obtained a somewhat higher ΔG‡298 for TS-I-B by 0.7 kcal mol−1 than that of TS-II-B in Path II. Given the typical uncertainty of a few kcal mol−1 in the current DFT level, although the difference in two ΔG‡298 values is too small to call, in fact, our predictions correlate well with the experimental results, from which the reaction equilibrium favors the product of TE-II-2 over the TE-I-2 by a factor of five in the isolated products (Table 2, entry 1).
In the reaction cascade via the central ethylene (denoted as “c”) of TCF-A2 (Fig. S3, ESI†), the respective ΔG‡298 values of HT and PT are 25.5 kcal mol−1 (viaTS-II-A(c)) and 27.9 kcal mol−1 (viaTS-II-B(c)), both values are quantitatively similar to the ΔG‡298's of the HT/PT steps via the terminal ethoxylated ethylene, 27.0 kcal mol−1 (viaTS-II-A) and 28.3 kcal mol−1 (viaTS-II-B). Upon the formation of the dehydrogenated JD1′ and hydrogenated TCF-A2(+2H,c), the generated JD1′(−2H) together with another TCF-A2 would undergo the condensation reaction as usual and this reaction cascade is expected to contribute to the formation of type-II product.
The DFT calculations were validated with the experimental data for comparative purposes in a qualitative sense. The relatively high barrier suggests that the reactions would hardly proceed at room temperature. At the elevated temperature in acetic anhydride, the change of JD1/TCF-A1 ratio from 0.6 to 2.1 gave essentially the same yield (∼27%) of TE-II-1 with trace amount of TE-I-1 (Table 1, entries 2–4). This underscores the dual role of hydrophilicity and electrophilicity of TCF-A1 hemicyanine, especially in the rate-determining step of HT. Besides the final products, we were able to collect the byproduct of hydrogenated TCF-A1(+2H) using HPLC and further verify it by HRMS as explicit experimental evidence (Fig. S5, ESI†). It is worth noting that this result was obtained after many trial reactions in order to capture trace amounts of TCF-A1(+2H) as a highly unstable intermediate under the reaction conditions. In controlled experiments at 80 °C, without adding TCF-A1, JD1 stayed unreacted in acetic anhydride with a catalytic amount of pyridine, corroborating the TH from JD1 to TCF-A1, not to the acetic anhydride. Without pyridine, neither TE-II-1 nor TE-I-1 were obtained by simply mixing JD1 and TCF-A1 in acetic anhydride, supporting the key role of pyridine as a base catalyst for deprotonation in condensation. The DFT calculations predicted a little lower ΔG‡298 value for Path II than that of Path I, reflecting a preliminary selectivity of two competing pathways to afford TE-II-1 as the major product over TE-I-1. As a matter of fact, our calculations in this study use a simplified molecular model (JD1′ and TCF-A2) to mimic the actual reactants. It is worth noting that the isolated product yield of 27% for TE-II-1 in this one-pot, two-step cascade reaction indicates a modest yield roughly around 50% in the critical intermolecular TH reaction via sequential HT and PT, which is considered a remarkable result without using potent catalysts of noble metal complexes or Lewis acids.
We used thin layer chromatography (TLC) to carefully monitor the reaction and optimize the yields of products from the final condensation between JD1 and TCF-A1-7 (Tables S2–S6, ESI†). We found that enhanced acceptor strength from TCF1 to TCF5 enables faster reactions, indicating higher hydrophilicity and/or electrophilicity of π-conjugated TCF hemicyanine precursors. The optimized reaction time of the final condensation was shortened to 1 h for TCF5 as the strongest acceptor, considerably reduced from 14 h for TCF1 as the weakest acceptor in the series. More importantly, the change of acceptor strength has substantial impact on the selectivity of reactions. As initially discovered, the TCF1-based reactions gave the type-II products of TE-II-1 and TE-II-2 as the major products in modest yields (22–27%). TCF2 and TCF3 are slightly stronger acceptors than TCF1 by introducing the substituent of phenyl derivatives at the germinal position of the structures. Compared to TCF1-based results, TCF2- and TCF3-based reactions gave significantly higher yields (11%) of TE-I chromophores, while the yields of TE-II chromophores remained essentially identical (around 25%, entries 2 and 3 in Table 2). TCF4 and TCF5 are two of strongest acceptors in the series due to the introduction of fluorinated phenyl groups at the germinal position of the TCF base structure. Their reactions demonstrated a reversed selectivity relative to the results from TCF1-based reactants, affording TE-I chromophores as the major product and TE-II chromophores as the minor one (Entries 5 and 6, Table 2).
We did the same level of DFT calculations for TCF5-based reactions (Fig. 4 and Fig. S2, ESI,† in green color) and compared the results with those of TCF1-based reactions (Fig. 4 and Fig. S2, ESI,† in black color). Incorporating a strong acceptor of 4-fluorophenyl (FP) substituted TCF5 to the system has reduced the ΔG‡298 of FP-TS-I-B (28.0 kcal mol−1) by 1.0 kcal mol−1 (relative to TS-I-B) in the C-9 condensation (Path I). This may be due to the stabilized effect of “additional” halogen bonding between α hydrogens of the JD1′ and the fluorine atom on TCF-A6 (Fig. 4, 3D structure).
On the other hand, the TCF5-based pathway has a higher ΔG‡298 in the initial HT step, 27.0 kcal mol−1 for TS-II-Avs. 30.3 kcal mol−1 for FP-TS-II-A (Fig. 4). In the FP-TS-II-A, although a halogen bonding effect also exists between two reactants, the steric hindrance, which forces JD1′ to do a rotation displacement, may play an important role and results in a higher ΔG‡298 value (Fig. 4, 3D structure).
Further examination on the atomic charge of TCF-A2 (Fig. S1, ESI†) indicates that the carbon (0.036) on the central ethylene bears a similar positive charge to the carbon (0.053) next to the ethoxy terminal. It appears that the conjugation length to the TCF correlates with a smaller ΔG‡298 of the TS-II-A(c), 25.5 kcal mol−1vs. 27.0 kcal mol−1 of the TS-II-A. The TCF5-based TCF-A6 is expected to display a higher electrophilicity and lower ΔG‡298's in the initial HT step, especially at the much more positively charged carbon (0.397) of the central ethylene. The DFT results show that the condensation reaction following the HT-PT process occurs at the central ethylene of TCF-A2 and TCF-A6 (Fig. S3, ESI†) and is rate-determining due to a higher PES.
Since a change of 1.4 kcal mol−1 in standard Gibbs free energy of formation (ΔG0298) changes the equilibrium constant by approximately a factor of 10, the DFT analysis can well explain the essential features of experimental results, such as reversed selectivity and reactivity change with the change of acceptor strength. The TH reactivity of two CC double bonds of TCF-based hemicyanines and relatively high barriers for a few key steps may explain the limited yields of products. Overall, the remarkable correlation shown between experimental and computational results supports the adequacy of DFT calculations used in this work and provides key insight into the reaction mechanisms for further improvement of yields. Also, the study presents an unprecedented example of catalyst-free TH sequentially from two C(sp3) hydrogens of a strong electron donor to the double bond of highly polarizable hemicyanines assisted by noncovalent π–π interactions, which is fundamentally different from the common catalytic or direct TH reactions via a six-membered cyclic transition state for HT.1,3
The solvatochromic and electrochemical properties of these chromophores were measured to understand the ICT of new push–pull structures (Fig. 5 and Fig. S85, ESI†). We select the compounds with reasonable solubility in the measurements. These chromophores, both type-I and type-II, feature a near-Gaussian-like broad absorption band and positive solvatochromism in nonpolar solvents such as toluene and trichloroethylene, suggesting the neutral ground state (NGS) of chromophores. In polar solvents from DCM to acetonitrile, type-I chromophores showed the same broad absorption bands and inverted solvatochromism for the polar (zwitterionic) ground state (ZGS) of chromophores, while the type-II chromophores displayed a substantial change in absorption band shape with the appearance of long-wavelength shoulders or main peaks at 850–900 nm. For TE-II-5 and TE-II-7 with strong TCF acceptors, larger extinction coefficients of relatively sharp peaks may point to an increased contribution from a cyanine-like resonance structure or ideal polymethine state (IPS) to the electronic structures of chromophores.28 Herein the IPS can also be treated as arising from equal contributions of NGS and ZGS resonance structures.10 The solvatochromism data of this study show that the electronic structures of new push–pull chromophores could be primarily governed by two resonance structures of NGS ↔ ZGS for type-I products, and by three resonance structures of NGS ↔ IPS ↔ ZGS for type-II products, respectively.15–17
Cyclic voltammetry (CV) measurements were conducted in dichloromethane solutions to determine the electrochemical properties of chromophores (Fig. 5F and Fig. S85F, ESI†). From the relative oxidation and reduction potentials of chromophores, the HOMO levels of type-I chromophores were about −4.94 eV, slightly deeper than the −4.90 eV of type-II chromophores, both different from the typical value of −4.85 eV for TCF-based push–pull heptamethines containing F1, F2, and M1 donors.15 The LUMO levels of these chromophores varied from −3.73 eV for the TCF1 acceptor to −3.83 eV for the TCF5 acceptor. The analysis shows that new chromophores from this modular one-pot synthetic method adds a unique feature of fine-tuned energy levels to the molecular systems of push–pull phenyltrienes and heptamethines.
The thermal stability of chromophores was evaluated by TGA and the results are shown in Fig. S86 (ESI†). Except for TE-II-5, these chromophores show a high decomposition temperature (Td), with the onset Td greater than 200 °C, promising for potential photonic applications in phototheranostics and optoelectronics.
Chromophore | λ max in nm | n TE/nTM at 1304 nm | n TE/nTM at 1541 nm | r 33, pm V−1 at 1304 nm | Φ | β μ , 1.3c μm (10−30 esu) |
---|---|---|---|---|---|---|
a A loading density of chromophores at 1.30 × 1020 cm−3 and the host polymer poly(styrene-co-methyl methacrylate) were used to formulate the guest–host systems. b The order parameters of poled films were calculated from the poling-induced optical birefringence of nTE and nTM with a Sellmeier fit of refractive indices. c The βμ is the vector part of β tensors along the dipole moment direction of chromophores and was calculated using the rigid oriented gas model (ROGM). d Data of two previously reported heptamethine chromophores, F1 and F2, were included for comparison.15a See Fig. 2 for detailed structures, where X = Cl, R = tert-butyl, and R1, R2 = methyl. | ||||||
TE-I-1 | 678 | 1.5592/1.5733 | 1.5534/1.5581 | 29.1 | 0.08 | 2103 |
TE-II-1 | 754 | 1.5680/1.5853 | 1.5609/1.5787 | 46.3 | 0.10 | 2931 |
TE-I-3 | 698 | 1.5643/1.5797 | 1.5585/1.5757 | 43.7 | 0.09 | 2838 |
TE-II-3 | 767 | 1.5730/1.5954 | 1.5652/1.5840 | 47.6 | 0.14 | 2799 |
TE-I-4 | 712 | 1.5681/1.5853 | 1.5613/1.5758 | 37.9 | 0.10 | 2673 |
TE-II-4 | 804 | 1.5759/1.5958 | 1.5670/1.5878 | 37.1 | 0.11 | 2330 |
TE-I-5 | 740 | 1.5681/1.5904 | 1.5601/1.5784 | 67.1 | 0.11 | 4106 |
TE-II-5 | 854 | 1.5832/1.6093 | 1.5726/1.5953 | 29.9 | 0.13 | 1718 |
TE-I-7 | 752 | 1.5742/1.5989 | 1.5661/1.5832 | 65.1 | 0.11 | 4106 |
F1 | 835 | 1.5840/1.6233 | 1.5702/1.6022 | 23.1 | 0.155 | 1273 |
F2 | 863 | 1.5748/1.6184 | 1.5645/1.5905 | 27.4 | 0.19 | 1332 |
We found that type-I and type-II chromophores exhibited comparable NLO properties in poled polymers. The obtained r33 values of poled guest–host polymers at 1304 nm were relatively large while maintaining rather blue-shifted absorption bands of chromophores. The NLO activities of two types of chromophores differed considerably with the change of acceptor strength. For type-I series, optimal r33 values around 67 pm V−1, corresponding to the largest βμ values from this study, were demonstrated for TE-I-5 and TE-I-7 containing stronger TCF4 and TCF5 acceptors, respectively. For type-II series, however, the best r33 value and corresponding βμ value were achieved for TE-II-1 containing a simple TCF1 acceptor and the values gradually dropped for TE-II-3, TE-II-4, and TE-II-5, despite their red-shifted absorption bands and large order parameters with stronger acceptors. The results corroborate the electronic structures of these chromophores that are governed by the relative contribution of difference resonance forms, as revealed by the solvatochromism study.
At the wavelength of 1.3 μm, the NLO activities of new type-I and type-II chromophores are much higher than those from known push–pull heptamethines with similar optical bandgaps, improving the βμ values by 165% for TE-I-1vs. F1 and 308% for TE-I-7vs. F2.15a Interestingly, the order parameters of these new chromophores were lower than those of F1 and F2, suggesting room for further improvement of the material's performance. On the other hand, the resulting reduced optical birefringence for the poled films would be beneficiary to the polarization-independent operation of poled polymers in optical waveguides.32
Footnotes |
† Electronic supplementary information (ESI) available: Computational details, synthetic schemes, experimental section, spectra, and X-ray crystallography. CCDC 2289992–2289995. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d4nj00820k |
‡ These authors contributed equally. |
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2024 |