Cd-driven surface reconstruction and photodynamics in gold nanoclusters

Abstract

With atomically precise gold nanoclusters acting as a starting unit, substituting one or more gold atoms of the nanocluster with other metals has become an effective strategy to create metal synergy for improving catalytic performances and other properties. However, so far detailed insight into how to design the gold-based nanoclusters to optimize the synergy is still lacking, as atomic-level exchange between the surface-gold (or core-gold) and the incoming heteroatoms is quite challenging without changing other parts. Here we report a Cd-driven reconstruction of Au₄₄(DMBT)₂₈ (DMBT = 3,5-dimethylbenzenethiol), in which four Au₂(DMBT)₃ staples are precisely replaced by two Au₅Cd₂(DMBT)₁₂ staples to form Au₃₈Cd₄(DMBT)₃₀ with the face-centered cubic inner core retained. With the dual modifications of the surface and electronic structure, the Au₃₈Cd₄(DMBT)₃₀ nanocluster exhibits distinct excitonic behaviors and superior photocatalytic performances compared to the parent Au₄₄(DMBT)₂₈ nanocluster.

---

Introduction

Metal synergy is of paramount importance as the rationale to modulate the intrinsic properties of metal nanoparticles.^1,2 However, the precise synergistic interaction in an intermetallic nanoparticle has so far been elusive, due to the challenges in determining the atomic-level arrangement of the metal heteroatoms in the nanoparticle. Atomically precise metal nanoparticles (often called nanoclusters) lead to unprecedented opportunities in signalling clear directions to exploit the cooperativity between the two metal elements within a single nanocluster.³ Thiolate-protected gold nanoclusters, Au_n(SR)_m, where n is the number of gold atoms and m is the number of thiolate ligands, SR, have gained momentum over the past few years as an exciting area and have opened up new horizons in precise tailoring of the composition and structure to control the physicochemical properties.^4–12 The Au_n(SR)_m nanoclusters are typically configured with an inner gold core (or kernel) and various surface motifs, in which the motifs containing both gold and thiolate resemble staples. Both the gold core and the surface motifs can contribute to the physicochemical properties such as the optical and electronic properties, as well as catalysis.^13–18 It has been recognized that substituting one or more gold atoms in either the core or the motifs with other metals can tune the overall performances of the parent nanoclusters.^19–26 Therefore, it has become possible to access the previously inaccessible metal synergy in the bimetallic nanoclusters with atomic-precision.

Among the gold-based bimetal nanoclusters, cadmium-containing bimetal clusters provide synergistic strategies to adjust the electronic structures and further modulate the physicochemical properties in the clusters, since Cd has one more valence electron than Au.^21,26,27 Cd introduction usually causes surface reconstruction of gold nanoclusters. For example, Au₁₉Cd₂(SR)₁₆ was obtained through the substitution of two neighboring surface Au atoms with one Cd with the cuboctahedral Au₁₃ unchanged.²⁶ Au₁₉Cd₃(SR)₁₈ was formed by retaining the icosahedral Au₁₃ core but only changing the surface of Au₂₅(SR)₁₈.²⁷ However, the surface reconstruction strategy remains challenging and no examples of bimetal clusters formed without breaking the face-centered cubic (fcc) core of the parent gold clusters have been documented, which might thus impede gaining a higher understanding of how to tailor the surface structure of gold-based nanoclusters and accordingly optimize their synergy.

Herein, we report our success in synthesizing a Au₃₈Cd₄(DMBT)₃₀ (DMBT = 3,5-dimethylbenzenethiol) nanocluster that is obtained from the surface reconstruction of Au₄₄(DMBT)₂₈ induced by Cd²⁺. The fcc Au₂₆ kernel originating from Au₄₄(DMBT)₂₈, which is assembled from Au₄ tetrahedra, is retained in the Au₃₈Cd₄(DMBT)₃₀. The Au₃₈Cd₄(DMBT)₃₀ nanocluster exhibits distinctly different excited-state dynamics from Au₄₄(DMBT)₂₈. More interestingly, the Au₃₈Cd₄(DMBT)₃₀ nanocluster as a photocatalyst shows better visible light-driven catalytic activity than the parent Au₄₄(DMBT)₂₈ catalyst.

Results and discussion

X-ray crystallography analysis shows that the parent Au₄₄(DMBT)₂₈ nanocluster is composed of an Au₂₆ kernel, six Au₂(SR)₃ and two Au(SR)₂ staples (Fig. 1a, c and Table S1†). The formula of Au₄₄(DMBT)₂₈ is further confirmed by electrospray ionization mass spectroscopy (ESI-MS, Fig. S1a†). The structural framework of Au₄₄(DMBT)₂₈ is identical to that of the reported Au₄₄(TBBT)₂₈ (TBBT = 4-tert-butylbenzenethiol) (Fig. S2†),²⁸ both of which can be assembled into the layered structures (Fig. S3–S5†). Notably, a significant difference is observed in the layer's interior, where all the molecules of Au₄₄(TBBT)₂₈ in the layer (marked with the same color, Fig. S3†) are packed along the same direction, while Au₄₄(DMBT)₂₈ molecules are arranged in different directions (Fig. S5†). Such a difference may be ascribed to the different steric hindrance between TBBT and DMBT. The UV-vis-NIR spectra of the two Au₄₄(SR)₂₈ nanoclusters show only small deviations. As shown in Fig. S6,† the prominent peak at 380 nm for Au₄₄(TBBT)₂₈ is slightly red-shifted to 388 nm for Au₄₄(DMBT)₂₈, and the broad peaks at 650 and 725 nm become apparent when TBBT is replaced by DMBT.

Fig. 1 Structural analysis of Au₄₄(DMBT)₂₈ and Au₃₈Cd₄(DMBT)₃₀ nanoclusters in the space-filling mode. Total structures of (a) Au₄₄(DMBT)₂₈ and (b) Au₃₈Cd₄(DMBT)₃₀ viewed from the front. Total structures of (c) Au₄₄(DMBT)₂₈ and (d) Au₃₈Cd₄(DMBT)₃₀ viewed from the side. Au₂₆ kernels of (e) Au₄₄(DMBT)₂₈ and (f) Au₃₈Cd₄(DMBT)₃₀ viewed from the front. Au₂₆ kernels of (g) Au₄₄(DMBT)₂₈ and (h) Au₃₈Cd₄(DMBT)₃₀ viewed from the side. Note that green frames show kernel distortions. Various motifs of the two nanoclusters: (i) Au₂(SR)₃; (j) Au(SR)₂; (k) Au₅Cd₂(SR)₁₂; (l) SR; color codes: magenta/green = Au, yellow = S, red = Cd. C and H atoms are omitted for clarity.

With Au₄₄(DMBT)₂₈ as a starting unit, a Cd-doped nanocluster was further synthesized via an ion-exchange strategy. From ESI-MS data (Fig. S1b†), the prominent peak at 6025.43 m/z with a +2 charge is assigned to Au₃₈Cd₄(DMBT)₃₀ (theoretical value: 6025.48 m/z), which is further confirmed by the excellent match between experimental and calculated isotopic patterns (inset of Fig. S1b†). Single crystallography analysis reveals that Au₃₈Cd₄(DMBT)₃₀ contains a 26-Au-atom kernel, two Au₅Cd₂(SR)₁₂ staples, two Au(SR)₂ staples and two bridging SR ligands, as shown in Fig. 1b, d, and Table S2.† Note that the retained kernel of Au₃₈Cd₄(DMBT)₃₀ experiences a slight distortion from “slender” to “stocky” in comparison with that of the parent Au₄₄(DMBT)₂₈ (Fig. 1e–h). Further analysis shows that the Au₂₆ kernel in Au₃₈Cd₄(DMBT)₃₀ can be viewed as the assembly of tetrahedral Au₄ units in a double-helical mode, as well as that in Au₄₄(DMBT)₂₈ (Fig. 2). Furthermore, the two nanoclusters have almost identical distances between neighboring Au₄ units, which is clearly manifested in the similar Au–Au bond lengths according to the different positions of the Au atoms (Fig. S7†). Therefore, Au₃₈Cd₄(DMBT)₃₀ can be viewed as the gentle surface reconstruction without breaking the double-helical Au₂₆ kernel based on the parent Au₄₄(DMBT)₂₈. In addition, Au₃₈Cd₄(DMBT)₃₀ is also patterned along different directions in the layer structure (Fig. S8†).

Fig. 2 Structural anatomy of (a) Au₄₄(DMBT)₂₈ and (b) Au₃₈Cd₄(DMBT)₃₀ with the fcc Au₂₆ kernels assembled from Au₄ building blocks. Color codes: blue/magenta/green = Au, yellow = S, red = Cd. C and H atoms are omitted for clarity.

To gain an in-depth insight into the Cd-induced surface reconstruction mechanism, density functional theory (DFT) calculations were performed. Starting from the Au₄₄(SR)₂₈ cluster, as presented in Fig. S9,† four Au₂(SR)₃⁻ protecting motifs of Au₄₄(SR)₂₈ are substituted by six Cd(SR)₂, resulting in the formation of an unstable intermediate, Au₃₆Cd₆(SR)₂₈⁴⁺, due to the very high substitution energy (24.80 eV, step 1). After this, a surface isomerization of Au₃₆Cd₆(SR)₂₈⁴⁺ is considered via the reorganization of the motifs (step 2). In this step, the Cd atom in Cd(SR)₂ binds with a neighboring S atom in the SR[Au(SR)]₂⁻ motif, which is accompanied by the breaking of a Au–S bond and the formation of a naked Au atom. The S atom in Cd(SR)₂ binds with this naked Au atom to form a new Au–S bond. Therefore, the structure in which the three-coordinated μ₃-Cd atoms bind with two SR⁻ and one Au(SR)₂⁻ can be obtained during the structural isomerization of Au₃₆Cd₆(SR)₂₈⁴⁺. In addition, the Au₃₆Cd₆(SR)₂₈⁴⁺ becomes more stable after isomerization via lowering the energy by 3.37 eV. Then, two S atoms of the SR[Au(SR)]⁻ motif further bind with two μ₃-Cd atoms to form two four-coordinated μ₄-Cd atoms, resulting in the formation of a more stable intermediate structure, Au₃₈Cd₆(SR)₃₂²⁺, with a formation energy of −18.64 eV (step 3). In step 4, the SR⁻ motif binds with the Cd atom of Cd(SR)₂ to form Cd(SR)₃⁻, in which the Cd(SR)₂ is quickly separated from Cd(SR)₃⁻ leaving a bridging SR⁻ motif on the surface of the Au core. Finally, the stable Au₃₈Cd₄(SR)₃₀ is formed with a formation energy of −12.50 eV. The proposed conversion process from Au₄₄(SR)₂₈ to Au₃₈Cd₄(SR)₃₀ includes two key steps: (i) the substitution of SR[Au(SR)]₂⁻ by Cd(SR)₂ and (ii) the structural isomerization of surface ligands.

To investigate the electronic structure changes induced by Cd-atom surface modification, the optical adsorption spectra of the Au₄₄(DMBT)₂₈ and Au₃₈Cd₄(DMBT)₃₀ nanoclusters were measured. The absorption peaks of Au₃₈Cd₄(DMBT)₃₀ are mainly centered at 400, 465, 550 and 678 nm (Fig. 3b), which differ from those observed in the parent nanocluster (387, 452, 635 and 725 nm; Fig. 3a). These optical features can be well reproduced by theoretical calculations (Fig. 3a, b and S10†). The Kohn–Sham (KS) molecular orbital (MO) energy levels and atomic orbital components in each KS MO of Au₄₄(SR)₂₈ and Au₃₈Cd₄(SR)₃₀ suggest that the absorption peaks mainly involve the Au(sp) → Au(sp) transitions (Fig. 3c and d). In particular, for Au₄₄(SR)₂₈, the first absorption peak centered at 734 nm originates from the highest occupied molecular orbital → the lowest unoccupied molecular orbital (HOMO → LUMO) transition, while for Au₃₈Cd₄(SR)₃₀, the first absorption peak centered at 695 nm originates from the HOMO → LUMO, HOMO → LUMO+1, HOMO → LUMO+4, HOMO−1 → LUMO, HOMO−1 → LUMO+1, and HOMO−1 → LUMO+5 transitions. The more complex orbital transitions in Au₃₈Cd₄(SR)₃₀ than in Au₄₄(SR)₂₈ can be attributed to the dopant Cd. This behaviour can also be observed for other absorption peaks.

Fig. 3 Experimental and calculated UV-vis-NIR spectra of (a) Au₄₄(DMBT)₂₈ and (b) Au₃₈Cd₄(DMBT)₃₀ nanoclusters and molecular orbital (MO) energy level diagrams for (c) Au₄₄(SR)₂₈ and (d) Au₃₈Cd₄(SR)₃₀. The convolution factor of the convoluted spectra is 0.1 eV.

Moreover, femtosecond and nanosecond carrier dynamics of the two nanoclusters were measured via time-resolved transient absorption (TA) spectroscopy to decipher their potential energy-related applications. The femtosecond-resolved TA spectra of the Au₄₄(DMBT)₂₈ and Au₃₈Cd₄(DMBT)₃₀ nanoclusters are provided in Fig. 4a and b. Similarly, both Au₄₄(DMBT)₂₈ and Au₃₈Cd₄(DMBT)₃₀ nanoclusters showed broad excited state absorption (ESA) signals overlapped with ground state bleaching (GSB) peaks near 675 nm. We selectively extracted the TA spectra at different delay times, combined with the dynamic traces probed at 515 and 675 nm to study the transient evolution and the relaxation dynamics (Fig. 4c–f). A 0.6 ps process at the early stage, which is attributed to the ultrafast internal conversion from higher excited states to lower excited states,²⁹ was observed in the two nanoclusters (Fig. S11 and Table S3†). It is worth noting that the major divergence between the two nanocluster systems emerged after a delay of 2 ps. For Au₄₄(DMBT)₂₈, the TA spectra remained nearly unchanged after 2 ps (Fig. 4c), which is consistent with the flat decay kinetic traces shown in Fig. 4e. A 19 ps process obtained by exponential fitting was ascribed to the structural relaxation caused by conformational changes after pumping.^29–31 For Au₃₈Cd₄(DMBT)₃₀, interestingly, an obvious spectral transformation was observed and the lifetime of this component was determined to be 57 ps (Table S3†), which differs from the 19 ps structural relaxation observed in Au₄₄(DMBT)₂₈ and might be related to the charge transfer states between the ligand and the metal core of Au₃₈Cd₄(DMBT)₃₀,^32–35 which can be manifested by the overall Hirshfeld charge of the Au₂₆ core in Au₃₈Cd₄(SH)₃₀ (0.46) and in Au₄₄(SH)₂₈ (0.56). Of note, deduced from nanosecond-resolved TA analysis, as shown in Fig. S12,† the Au₃₈Cd₄(DMBT)₃₀ nanocluster exhibits a faster carrier recombination process with a lifetime of 253 ns than the Au₄₄(DMBT)₂₈ nanocluster (389 ns lifetime), readily supporting the putative synergy in the Au₃₈Cd₄(DMBT)₃₀ nanocluster being invoked in underpinning the excited-state dynamics.

Fig. 4 Femtosecond-resolved TA spectra of (a) Au₄₄(DMBT)₂₈ and (b) Au₃₈Cd₄(DMBT)₃₀ pumped at 400 nm. Time evolution of femtosecond-resolved TA spectra of (c) Au₄₄(DMBT)₂₈ and (d) Au₃₈Cd₄(DMBT)₃₀. Kinetic traces at selected wavelengths of (e) Au₄₄(DMBT)₂₈ and (f) Au₃₈Cd₄(DMBT)₃₀. The data are plotted in a scale normalized to the amplitude of the signal probed at 515 nm at a delay of 2 ps. The gray dots in (e) and (f) are the original data, while the corresponding multi-exponential fits are plotted as colored lines.

The distinguishable electronic and optical properties of the two nanoclusters would apparently impact their catalytic properties. Thus, visible light-driven degradation of methyl orange was selected to explore the photocatalysis of the two nanoclusters. From Fig. 5a and b, within 50 min, methyl orange can be completely degraded on the Au₃₈Cd₄(DMBT)₃₀ catalyst under visible light illumination, while on the Au₄₄(DMBT)₂₈ catalyst it was completed in 70 min. The plots of methyl orange degradation on the catalysts versus reaction time further indicate the better photocatalytic performance of the Au₃₈Cd₄(DMBT)₃₀ catalyst (Fig. 5c). Electrochemical impedance spectroscopy was performed to investigate the interfacial transfer of electrons. In Fig. 5d, the semicircular diameter of Au₃₈Cd₄(DMBT)₃₀ was smaller than that of Au₄₄(DMBT)₂₈, which implies faster electron-transfer in the Au₃₈Cd₄(DMBT)₃₀ system. The photocatalysis difference in the two cluster catalysts is suggested to arise from their different equilibria established between the carrier recombination and the electron transfer influenced by metal synergy.

Fig. 5 Photocatalytic degradation of methyl orange over (a) Au₄₄(DMBT)₂₈/TiO₂ and (b) Au₃₈Cd₄(DMBT)₃₀/TiO₂ under visible light illumination. (c) Visible light-driven degradation of methyl orange by Au₃₈Cd₄(DMBT)₃₀/TiO₂ and Au₄₄(DMBT)₂₈/TiO₂ catalysts. Reaction conditions: 40 mg catalyst, 20 mL H₂O, 0.1 mL (1 g L⁻¹) methyl orange. (d) Electrochemical impedance spectra of the Au₃₈Cd₄(DMBT)₃₀ and Au₄₄(DMBT)₂₈ nanoclusters.

Conclusions

In summary, we have developed a Cd-driven surface reconstruction strategy for synthesizing a new Au₃₈Cd₄(DMBT)₃₀ bimetallic nanocluster with the fcc Au₂₆ core retained from the parent Au₄₄(DMBT)₂₈ nanocluster. The two nanoclusters that exhibit elegant patterns of Au₄ tetrahedra show distinct differences in the electronic structures, optical properties, and photocatalytic performances. Beyond the Cd-mediated surface reconstruction case, we anticipate that this heteroatom-doping mechanism will find applications in using gold and other metals in a series of challenging gold-based nanocluster formations and tuning of their intrinsic properties.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

We acknowledge the financial support from the National Natural Science Foundation of China (21773109, 91845104 and 11974195), China Postdoctoral Science Foundation (2019M650106, 2020T130286), and Programs for high-level entrepreneurial and innovative talents introduction of Jiangsu Province (individual and group program).

Notes and references

1. A. C. Deacy, A. F. R. Kilpatrick, A. Regoutz and C. K. Williams, Nat. Chem., 2020, 12, 372–380.
2. N. Zou, X. Zhou, G. Chen, N. M. Andoy, W. Jung, G. Liu and P. Chen, Nat. Chem., 2018, 10, 607–614.
3. Y. Sun, W. Pei, M. Xie, S. Xu, S. Zhou, J. Zhao, K. Xiao and Y. Zhu, Chem. Sci., 2020, 11, 2240–2247.
4. P. D. Jadzinsky, G. Calero, C. J. Ackerson, D. A. Bushnell and R. D. Kornberg, Science, 2007, 318, 430–433.
5. C. Zeng, Y. Chen, K. Kirschbaum, K. J. Lambright and R. Jin, Science, 2016, 354, 1580–1584.
6. J. Yan, B. K. Teo and N. Zheng, Acc. Chem. Res., 2018, 51, 3084–3093.
7. Y. Negishi, K. Nobusada and T. Tsukuda, J. Am. Chem. Soc., 2005, 127, 5261–5270.
8. Y. Song, F. Fu, J. Zhang, J. Chai, X. Kang, P. Li, S. Li, H. Zhou and M. Zhu, Angew. Chem., Int. Ed., 2015, 127, 8550–8554.
9. Z. Wu, Y. Du, J. Liu, Q. Yao, T. Chen, Y. Cao, H. Zhang and J. Xie, Angew. Chem., Int. Ed., 2019, 58, 8139–8144.
10. A. Baksi, P. Chakraborty, S. Bhat, G. Natarajan and T. Pradeep, Chem. Commun., 2016, 52, 8397–8400.
11. C. Zeng, Y. Chen, C. Liu, K. Nobusada, N. L. Rosi and R. Jin, Sci. Adv., 2015, 1, e1500425.
12. L. Shi, L. Zhu, J. Guo, L. Zhang, Y. Shi, Y. Zhang, K. Hou, Y. Zheng, Y. Zhu, J. Lv, S. Liu and Z. Tang, Angew. Chem., Int. Ed., 2017, 56, 15397–15401.
13. R. Jin, C. Zeng, M. Zhou and Y. Chen, Chem. Rev., 2016, 116, 10346–10413.
14. M. A. Tofanelli, K. Salorinne, T. W. Ni, S. Malola, B. Newell, B. Phillips, H. Häkkinen and C. J. Ackerson, Chem. Sci., 2016, 7, 1882–1890.
15. X. Cai, G. Saranya, K. Shen, M. Chen, R. Si, W. Ding and Y. Zhu, Angew. Chem., Int. Ed., 2019, 58, 9964–9968.
16. K. Kwak, W. Choi, Q. Tang, M. Kim, Y. Lee, D. E. Jiang and D. Lee, Nat. Commun., 2017, 8, 14723.
17. S. Li, H. Chen, X. Liu, H. Liu, J. Ma and Y. Zhu, Chem. Sci., 2020, 11, 8000–8004.
18. X. Wan, J. Wang, Z. Nan and Q. Wang, Sci. Adv., 2017, 3, e1701823.
19. S. Takano, S. Hasegawa, M. Suyama and T. Tsukuda, Acc. Chem. Res., 2018, 51, 3074–3083.
20. Q. Li, T. Luo, M. G. Taylor, S. Wang, X. Zhu, Y. Song, G. Mpourmpakis, N. L. Rosi and R. Jin, Sci. Adv., 2017, 3, e1603193.
21. S. Zhuang, D. Chen, L. Liao, Y. Zhao, N. Xia, W. Zhang, C. Wang, J. Yang and Z. Wu, Angew. Chem., Int. Ed., 2020, 59, 3073–3077.
22. W. Fei, S. Antonello, T. Dainese, A. Dolmella, M. Lahtinen, K. Rissanen, A. Venzo and F. Maran, J. Am. Chem. Soc., 2019, 141, 16033–16045.
23. H. Qian, D. E. Jiang, G. Li, C. Gayathri, A. Das, R. R. Gil and R. Jin, J. Am. Chem. Soc., 2012, 134, 16159–16162.
24. X. Cai, W. Hu, S. Xu, D. Yang, M. Chen, M. Shu, R. Si, W. Ding and Y. Zhu, J. Am. Chem. Soc., 2020, 142, 4141–4153.
25. M. Suyama, S. Takano, T. Nakamura and T. Tsukuda, J. Am. Chem. Soc., 2019, 141, 14048–14051.
26. Q. Li, K. L. Lambright, M. G. Taylor, K. Kirschbaum, T. Luo, J. Zhao, G. Mpourmpakis, S. Mokashi-Punekar, N. L. Rosi and R. Jin, J. Am. Chem. Soc., 2017, 139, 17779–17782.
27. C. Yao, C. Xu, I. Park, M. Zhao, Z. Zhu, J. Li, X. Hai, H. Fang, Y. Zhang, G. Macam, J. Teng, L. Li, Q. Xu, F. Chuang, J. Lu, C. Su, J. Li and J. Lu, Angew. Chem., Int. Ed., 2020, 59, 8270–8276.
28. C. Zeng, Y. Chen, K. Iida, K. Nobusada, K. Kirschbaum, K. J. Lambright and R. Jin, J. Am. Chem. Soc., 2016, 138, 3950–3953.
29. M. Zhou, C. Zeng, M. Y. Sfeir, M. Cotlet, K. Iida, K. Nobusada and R. Jin, J. Phys. Chem. Lett., 2017, 8, 4023–4030.
30. M. Zhou, S. Tian, C. Zeng, M. Y. Sfeir, Z. Wu and R. Jin, J. Phys. Chem. C, 2017, 121, 10686–10693.
31. M. Zhou, T. Higaki, Y. Li, C. Zeng, Q. Li, M. Y. Sfeir and R. Jin, J. Am. Chem. Soc., 2019, 141, 19754–19764.
32. Z. Wu and R. Jin, Nano Lett., 2010, 10, 2568–2573.
33. S. A. Miller, C. A. Fields-Zinna, R. W. Murray and A. M. Moran, J. Phys. Chem. Lett., 2010, 1, 1383–1387.
34. K. L. D. M. Weerawardene and C. M. Aikens, J. Am. Chem. Soc., 2016, 138, 11202–11210.
35. X. Kang, S. Wang, Y. Song, S. Jin, G. Sun, H. Yu and M. Zhu, Angew. Chem., Int. Ed., 2016, 55, 3611–3614.

---

Footnotes

† Electronic supplementary information (ESI) available. CCDC 2022214 and 2022215. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/d0sc05163b

‡ These authors contributed equally to this work.

---