A comparison study of MnO₂ and Mn₂O₃ as zinc-ion battery cathodes: an experimental and computational investigation

Abstract

The high specific capacity, low cost and environmental friendliness make manganese dioxide materials promising cathode materials for zinc-ion batteries (ZIBs). In order to understand the difference between the electrochemical behavior of manganese dioxide materials with different valence states, i.e., Mn(III) and Mn(IV), we investigated and compared the electrochemical properties of pure MnO₂ and Mn₂O₃ as ZIB cathodes via a combined experimental and computational approach. The MnO₂ electrode showed a higher discharging capacity (270.4 mA h g⁻¹ at 0.1 A g⁻¹) and a superior rate performance (125.7 mA h g⁻¹ at 3 A g⁻¹) than the Mn₂O₃ electrode (188.2 mA h g⁻¹ at 0.1 A g⁻¹ and 87 mA h g⁻¹ at 3 A g⁻¹, respectively). The superior performance of the MnO₂ electrode was ascribed to its higher specific surface area, higher electronic conductivity and lower diffusion barrier of Zn²⁺ compared to the Mn₂O₃ electrode. This study provides a detailed picture of the diversity of manganese dioxide electrodes as ZIB cathodes.

---

1. Introduction

In recent years, zinc-ion batteries (ZIBs) have received significant attention because Zn is naturally abundant, and ZIB devices, which are based on an aqueous electrolyte, are safer and more environmentally friendly compared to lithium-ion batteries.^1–4 Manganese dioxide materials are considered as promising candidates for cathode materials for ZIBs.^5–14 This is due to their high specific capacity, low cost, and environmental friendliness.³ The studies available on manganese dioxide cathodes for ZIBs mainly focus on MnO₂,^15–18 Mn₂O₃,¹⁹ Mn₃O₄ (ref. 9 and 20) or mixed-valence manganese dioxide containing multiple phases.^3,21 For instance, Fu et al. synthesized porous MnO_x (Mn(IV) and Mn(II)) nanorods with N-doped carbon by the metal–organic framework method as a new cathode for ZIBs.²¹ These ZIBs delivered a relatively high specific capacity of 385 mA h g⁻¹ after 120 cycles. The composite cathodes showed excellent rate performance and good cycle retention.

In our previous study, we utilized MnO_x/polypyrrole (PPy) composites containing Mn(III) and Mn(IV) phases as Zn intercalation host cathodes in aqueous ZIBs.³ The electrochemical properties of the MnO_x/PPy electrode were fairly sensitive to the amount of manganese dioxides phases, i.e., Mn(III) and Mn(IV). This inspired us to explore the difference between the electrochemical performance of manganese dioxides with Mn(III) phase and Mn(IV) phase as well as the contribution of the different phases to the electrochemical properties of the mixed-valence MnO_x electrode. However, the mixed-valence MnO_x electrode is composited with other additives (i.e., PPy), and possible contributions from multiple factors make it difficult to identify the role of each phase. In addition, a few factors (e.g., the diffusion barrier of Zn²⁺) that cannot be directly confirmed by experimental means are often ignored, leading to an incomplete analysis. It has been reported that density functional theory (DFT) can be used to simulate the diffusion barrier of Zn²⁺ in cathode materials. Thus, in order to understand and reveal the underlying mechanisms, it is necessary to adopt the related computational methods, and combine the theoretical predictions with the experimental results.

In this study, we investigated and compared the electrochemical properties of pure MnO₂ and Mn₂O₃ as ZIB cathodes via a combined experimental and computational study. While manganese dioxides containing Mn(III) and Mn(IV) phases in mixed-valence MnO_x are often more complex than those presented in this study, an understanding of the basic behavior of the component materials can be helpful in interpreting the behavior of related systems. This study, in combination with our previous work on MnO_x/PPy composites containing Mn(III) and Mn(IV) phases, provides a detailed picture of the diversity of manganese dioxide electrodes.

2. Experimental

2.1 Preparation of MnO₂ and Mn₂O₃ cathode

The α-MnO₂ powder was prepared by a hydrothermal method.²² 0.1264 g of potassium permanganate, 0.0428 g of (NH₄)₂SO₄, and 40 mL of distilled water were added to a Teflon-lined reactor and mixed, and then, treated hydrothermally at 140 °C for 24 h. Then, the resulting powder was filtered by an aqueous filter paper (JINTENG, with a pore size of 200 nm), and washed using distilled water, and dried at 70 °C for 4 h. The as-obtained α-MnO₂ powder was then annealed at 300 °C for 1 h.

The α-Mn₂O₃ powder (CAS No. 1317-34-6) was purchased from Xiya Chemical Industry Company. Super P (conductive additive), with the size of ∼40 nm, was purchased from TIMCAL from Switzerland.

To prepare the MnO₂ cathode, 70 wt% of MnO₂ powder, 20 wt% of Super P, and 10 wt% of polyvinylidene fluoride (PVDF) was dispersed in N-methyl-pyrrolidone (NMP), and stirred for 4 h. Then, the mixed slurry was coated onto a carbon cloth and dried at 70 °C for 6 h. The Mn₂O₃ cathode was prepared by the same process. The mass loading of the active mass (MnO₂ or Mn₂O₃) was ∼2 mg cm⁻², and the geometrical electrode area was ∼1.54 cm².

2.2 Material characterization

Scanning electron microscopy (SEM, ZEISS SUPRA, Carl Zeiss) was performed to observe the surface morphology of the materials. X-ray powder diffraction (XRD) patterns were recorded using a Bruker X-ray diffractometer (D8 Advance). Nitrogen adsorption/desorption isotherms and pore-size distributions were collected with an accelerated surface-area and porosimetry system (ASAP 2020 HD88).

2.3 Electrochemical characterization

A full battery was assembled from a cathode (MnO₂ or Mn₂O₃), an anode (Zn foil) and a separator (NKK separator) in a solution of 2 M ZnSO₄ + 0.1 M MnSO₄. The cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) were conducted on an electrochemical workstation (CHI 660e, Shanghai CH Instrument Co., Ltd.) using the full batteries. The CV spectra were recorded within a potential range of 1.0 to 1.9 V vs. Zn/Zn²⁺ at a scan rate at 0.5 mV s⁻¹. The EIS spectra were recorded in a frequency range from 0.01 to 10⁵ Hz after charging to ∼1.9 V vs. Zn²⁺/Zn. The galvanostatic charge/discharge tests were conducted in a potential range from 1.0 to 1.9 V vs. Zn/Zn²⁺ using a Land test system (CT2001A, Wuhan Land Electronic Co., Ltd.).

2.4 Computational method and models

All calculations were carried out using the projector augmented wave method in the framework of the DFT,²³ as implemented in the Cambridge Serial Total Energy Package (CASTEP).²⁴ The generalized gradient approximation (GGA) and Perdew–Burke–Ernzerhof (PBE) exchange functional²³ were used. The plane-wave energy cutoff was set to 500 eV, and the Monkhorst–Pack method²⁵ was employed for the Brillouin zone sampling. The Monkhorst–Pack method with 3 × 3 × 3 and 3 × 3 × 5 k-point meshes were employed for the Brillouin zone sampling of the α-Mn₂O₃ unit cell and α-MnO₂ 1 × 1 × 2 supercell. The convergence criteria of energy and force calculations were set to 10⁻⁵ eV per atom and 0.01 eV Å⁻¹, respectively. The Zn-inserted α-MnO₂ structure was built as a 1 × 1 × 2 supercell with a Zn atom at the hollow site.²⁶ For Mn₂O₃, two different types of adjacent interstitial sites were considered for Zn insertion, namely the tetrahedral and octahedral sites. The linear synchronous transition and quadratic synchronous transition methods²⁷ implemented in the CASTEP were used to calculate the energy landscape and activation energy barrier of the Zn diffusion in α-MnO₂ and Mn₂O₃.

3. Results and discussion

The crystal structures and morphologies of the MnO₂ and Mn₂O₃ powders were measured by XRD and SEM. The XRD characterization, as shown in Fig. 1(a) and (b), shows that the Bragg peaks can be indexed to the crystalline phases of α-MnO₂ (JCPDS: 44-0141) in Fig. 1(a) and α-Mn₂O₃ (JCPDS: 41-1442) in Fig. 1(b). The morphology of the α-MnO₂ powder, as shown in Fig. 1(c), can be described as nanosheets with lengths and width in the micrometer scale and thickness in less than tens of nanometers. On the other hand, the morphology of α-Mn₂O₃, as shown in the SEM image in Fig. 1(d), consisted of nanoparticles of size in the range of hundreds of nanometers.

Fig. 1 XRD patterns of (a) MnO₂ and (b) Mn₂O₃. SEM images of (c) MnO₂ and (d) Mn₂O₃.

The specific surface area and the pore size distribution of the electrode materials play an important role in ion transport, which has been confirmed by previous studies on batteries.⁴ Herein, the microporous structure of the MnO₂ and Mn₂O₃ electrodes was confirmed by N₂ adsorption–desorption isotherms and was found to be a typical type-IV isotherm (Fig. 2). The specific surface area of the MnO₂ electrode (61.9 m² g⁻¹) was much greater than that of the Mn₂O₃ electrode (10.6 m² g⁻¹), resulting in faster ion diffusion between the electrolyte and MnO₂. It is worth noting that the pore diameter distribution of the MnO₂ electrode was mainly located between 1 and 7 nm, and the pore volume was 0.44 cm³ g⁻¹. Comparatively, the pore diameter distribution of the Mn₂O₃ electrode lay mainly in the range of 20–50 nm, and the pore volume was 0.14 cm³ g⁻¹. Large numbers of micropores provided more sites for Zn²⁺ storage, and an appropriate pore size distribution could improve the transport of Zn²⁺, resulting in a high-rate performance.

Fig. 2 (a) Nitrogen adsorption/desorption isotherms of the MnO₂ and Mn₂O₃ electrodes. (b) Pore size distributions of the MnO₂ and Mn₂O₃ electrodes.

In order to compare the electrochemical performance of the MnO₂ and Mn₂O₃ electrodes, Zn–MnO₂ and Zn–Mn₂O₃ batteries were assembled with a Zn foil anode in an aqueous electrolyte containing 2 M ZnSO₄ and 0.1 M MnSO₄. Fig. 3(a) shows the cyclic voltammogram of the Zn–MnO₂ and Zn–Mn₂O₃ batteries at a scan rate of 0.5 mV s⁻¹ in the voltage range of 1.0–1.9 V vs. Zn²⁺/Zn. For the Mn₂O₃ cathode, two separate reversible redox peaks, i.e., reduction peaks at 1.20 and 1.32 V, and oxidation peaks at 1.63 and 1.66 V, were observed, corresponding to a two-step reaction. In comparison, the CV curve of the MnO₂ cathode showed that two similar reduction peaks occurred, while the oxidation peaks merged into a broad peak at 1.62 V. It is worth noting that the Mn₂O₃ electrode showed much lower peaks than those of the MnO₂ electrode, indicating that the Mn₂O₃ electrode had a lower capacity than the MnO₂ electrode.

Fig. 3 (a) Cyclic voltammetry curves of Zn–MnO₂ and Zn–Mn₂O₃ batteries. (b) Discharge/charge profiles of Zn–MnO₂ and Zn–Mn₂O₃ batteries at 0.1 A g⁻¹. (c) Rate capabilities of Zn–MnO₂ and Zn–Mn₂O₃ batteries. (d) Long-term cycling performances of the Zn–MnO₂ and Zn–Mn₂O₃ batteries at 3 A g⁻¹.

Fig. 3(b) shows the first, third, and fifth charging and discharging profiles of the MnO₂ and Mn₂O₃ electrodes at a low current density of 0.1 A g⁻¹. In the first cycle, the Mn₂O₃ electrode delivered a low discharge capacity of 61.1 mA h g⁻¹. The capacities at the third and fifth cycles increased to 173.5 and 188.2 mA h g⁻¹, respectively. The increased capacity of the Mn₂O₃ electrode at 0.1 A g⁻¹ can be assigned to the gradual activation of electrodes. In comparison, the MnO₂ electrode showed higher and more stable capacities at the first, third, and fifth cycles. The capacities of the MnO₂ electrode at the first, third, and fifth cycles were 270.6, 270.4, and 271.7 mA h g⁻¹, respectively. Both electrodes presented similar discharge plateaus, which can be attributed to the consequent H⁺ and Zn²⁺ insertion processes.²⁸

Fig. 3(c) shows the comparison of the rate performance of the MnO₂ and Mn₂O₃ electrodes. Again, the MnO₂ electrode exhibited a better rate capability than the Mn₂O₃ electrode. For instance, the MnO₂ electrode could deliver a discharge capacity of 270.4 mA h g⁻¹ at a low current density of 0.1 A g⁻¹. When the discharge current density increased to higher values, such as 1.8 and 3.0 A g⁻¹, the discharge capacities were maintained at ∼144.6 and ∼125.7 mA h g⁻¹, respectively. In comparison, the Mn₂O₃ electrode exhibited a much lower discharging capacity, i.e., ∼102.0 mA h g⁻¹ at 1.8 A g⁻¹ and ∼87 mA h g⁻¹ at 3.0 A g⁻¹. The cycling performances of the MnO₂ and Mn₂O₃ electrodes are presented in Fig. 3(d). The MnO₂ and Mn₂O₃ electrodes showed around 96% and 93% capacity retention after 600 cycles, respectively, indicating their excellent cycling performance.

To obtain a better understanding of the difference between the MnO₂ and Mn₂O₃ electrodes, electrochemical impedance spectroscopy (EIS) measurements were carried out after charging to ∼1.9 V vs. Zn²⁺/Zn. The Nyquist plots are shown in Fig. 4. An appropriate equivalent circuit model (inset in Fig. 4) was established to fit the Nyquist curves. The electrical parameters in this model, namely the ohmic series resistance (R_s), the surface film resistance (R_f), and the charge transfer resistance (R_ct), were calculated. The values of R_s and R_ct of the MnO₂ electrode were much smaller than those of the Mn₂O₃ electrode, resulting in a much better rate performance for the MnO₂ electrode (Fig. 3(c) and Table 1).

Fig. 4 Nyquist plots of the MnO₂ and Mn₂O₃ electrodes after charging to ∼1.9 V vs. Zn²⁺/Zn.

Table 1 EIS fitting results of the MnO₂ and Mn₂O₃ electrodes

	Rs (Ω)	Rf (Ω)	Rct (Ω)
MnO2	3.2	129.5	35.1
Mn2O3	10.3	210.2	72.9

---

The diffusion barrier of Zn²⁺ in the cathode is also another important factor affecting the electrochemical performance. Since it is challenging to experimentally measure the diffusion barrier of Zn²⁺ in mixed manganese oxides, DFT was used to simulate the diffusion barrier of Zn²⁺ in MnO₂ and Mn₂O₃. The corresponding diffusion pathways are shown in Fig. 5(a) and (b), and the calculated energy profile is displayed in Fig. 5(c) and (d). As shown in Fig. 5(a), the Zn²⁺-inserted into the α-MnO₂ structure was built as a 1 × 1 × 2 supercell with the Zn atom at the hollow site.²⁶ The calculated energy barrier for the migration of Zn²⁺ in the α-MnO₂ structure was 0.497 eV (Fig. 5(c)). For Mn₂O₃, two different kinds of adjacent interstitial sites were considered for the Zn²⁺ insertion, namely the tetrahedral and octahedral sites. The calculated migration energy barrier for Zn²⁺ in the α-Mn₂O₃ structure was 1.989 eV (Fig. 5(d)). The migration energy barrier for Zn²⁺ in α-MnO₂ was found to be much lower than that in α-Mn₂O₃, indicating that Zn²⁺ can migrate more easily in α-MnO₂. The lower-energy barriers endowed Zn–MnO₂ batteries with faster electrochemical kinetics than Zn–Mn₂O₃ batteries.

Fig. 5 (a) Zn diffusion pathway in MnO₂. (b) Zn diffusion pathway in Mn₂O₃. (c) The energy landscapes of Zn diffusion in MnO₂. (d) The energy landscapes of Zn diffusion in Mn₂O₃.

In order to obtain the Zn²⁺ diffusion coefficient of the MnO₂ and Mn₂O₃ electrodes, the galvanostatic intermittent titration technique (GITT) was conducted. The GITT was measured at a 0.03 A g⁻¹ pulse current for 20 min (τ) and relaxed for 120 min (Fig. 6). The diffusion coefficient of Zn²⁺ (D_Zn²⁺, cm² s⁻¹) was calculated as:²⁹

where τ is the pulse duration of constant current, and L corresponds to the Zn²⁺ diffusion length, which is equal to the thickness of the electrode. ΔE_s is the voltage change of the termination voltage of two adjacent relaxation steps; ΔE_τ is the potential change during the constant current pulse after eliminating the IR-drop.

Fig. 6 Scheme for a single titration step of the GITT curves of (a) MnO₂ and (b) Mn₂O₃ electrodes.

As calculated, the Zn²⁺ diffusion coefficient of the MnO₂ electrode was 3.05 × 10⁻⁸ cm² s⁻¹, which was higher than that of the Mn₂O₃ electrode (3.42 × 10⁻⁹ cm² s⁻¹). The results prove the lower Zn²⁺ diffusion barrier for MnO₂ than that of Mn₂O₃, which is consistent with the computational study.

Based on the above-mentioned results, the MnO₂ electrode showed superior electrochemical properties, i.e., higher specific capacity and better rate performance than the Mn₂O₃ electrode. The reaction mechanism for MnO₂ has been reported previously that the MnO₂ cathode experiences a consequent H⁺ and Zn²⁺ insertion/extraction process during the discharging/charging.²⁸ In order to reveal the structural reaction mechanism of the Mn₂O₃ electrode, ex situ XRD was carried out in numerous discharging/charging states of the Mn₂O₃ electrode, as shown in Fig. S1 (ESI†). The ex situ results indicated that the energy storage mechanism of the Mn₂O₃ electrode also involved H⁺ and Zn²⁺ insertion/extraction. The superior electrochemical properties of the MnO₂ electrode were ascribed to the following reasons: (i) the α-MnO₂ electrode had a higher electronic conductivity, as evidenced in the decreased charge transfer resistance (Fig. 4), and (ii) the MnO₂ electrode possessed a higher specific surface area (61.9 m² g⁻¹ for the MnO₂ electrode vs. 10.6 m² g⁻¹ for the Mn₂O₃ electrode), which increased the contact area between the electrode and electrolyte, therefore facilitating the Zn²⁺ insertion process. Moreover, the lower-energy barriers endowed the Zn–MnO₂ batteries with faster electrochemical kinetics. All of these merits contributed to the superior electrochemical performance of the Zn–MnO₂ batteries. It is worth noting that the results also provided an explanation for the lower specific capacity of MnO_x/PPy composites, which contained a higher Mn(III) phase/Mn(IV) phase ratio, as reported in our previous study.³

4. Conclusion

We investigated and compared the electrochemical properties of pure α-MnO₂ and α-Mn₂O₃ as ZIB cathodes via a combined experimental and computational study. We demonstrated that the α-MnO₂ electrode possessed a higher specific surface area, higher electronic conductivity, and lower diffusion barrier for Zn²⁺ than that observed in the case of the α-Mn₂O₃ electrode. As a result, the discharging capacity at 0.1 A g⁻¹ was higher for the α-MnO₂ electrode (270.4 mA h g⁻¹) than for the α-Mn₂O₃ electrode (188.2 mA h g⁻¹). In addition, α-MnO₂ showed a better rate performance (125.7 mA h g⁻¹ at 3 A g⁻¹) than α-Mn₂O₃ (87 mA h g⁻¹ at 3 A g⁻¹). This study provides an insight into the electrochemical mechanisms of manganese dioxide-based systems, which can be used to interpret behavior seen in Mn(III)- and Mn(IV)-based phases.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was financially supported by the Shenzhen Science and Technology Program (Grant No. GJHZ20190820091203667, and No. KQTD20190929172522248) and the Program for Guangdong Introducing Innovative and Enterpreneurial Teams (2019ZT08Z656). Technical support from Tsinghua Shenzhen International Graduate School Materials and Devices Testing Center is gratefully acknowledged.

References

1. Y. Li, J. Fu, C. Zhong, T. Wu, Z. Chen, W. Hu and J. Lu, Adv. Energy Mater., 2019, 9, 1802605.
2. P. Yu, Y. Zeng, H. Zhang, M. Yu, Y. Tong and X. Lu, Small, 2019, 15, 1804760.
3. Z. Li, Y. Huang, J. Zhang, S. Jin, S. Zhang and H. Zhou, Nanoscale, 2020, 12, 4150–4158.
4. Y. Huang, Z. Li, S. Jin, S. Zhang, H. Wang, P. Hiralal, G. A. J. Amaratunga and H. Zhou, Carbon, 2020, 167, 431–438.
5. K. Wang, X. Zhang, J. Han, X. Zhang, X. Sun, C. Li and Y. Ma, ACS Appl. Mater. Interfaces, 2018, 10, 24573–24582.
6. M. Alfaruqi, S. Islam, V. Mathew, J. Song, S. Kim, D. Tung and J. Kim, Appl. Surf. Sci., 2017, 404, 435–442.
7. C. Wang, M. Wang, Z. He, L. Liu and Y. Huang, ACS Appl. Mater. Interfaces, 2020, 3, 1742–1748.
8. D. Wang, L. Wang, G. Liang, H. Li, Z. Liu, Z. Tang and C. Zhi, ACS Nano, 2019, 13, 10643–10652.
9. C. Zhu, G. Fang, J. Zhou, J. Guo, Z. Wang, C. Wang and S. Liang, J. Mater. Chem. A, 2018, 6, 9677–9683.
10. W. Qiu, L. Yu, Y. Ao, Z. Zhang, G. Li and X. Lu, J. Mater. Chem. A, 2017, 5, 14838–14846.
11. N. Palaniyandy, M. Kebede, K. Raju, K. Ozoemena, L. le Roux, M. Mathe and R. Jayaprakasam, Mater. Chem. Phys., 2019, 230, 258–266.
12. T. Xiong, Y. Zhang, W. S. V. Lee and J. Xue, Adv. Energy Mater., 2020, 10, 2001769.
13. T. Xiong, Z. G. Yu, H. Wu, Y. Du, Q. Xie, J. Chen, Y.-W. Zhang, S. J. Pennycook, W. S. V. Lee and J. Xue, Adv. Energy Mater., 2019, 9, 1803815.
14. W. Shi, W. S. V. Lee and J. Xue, ChemSusChem, 2021, 14, 1–26.
15. H. Pan, Y. Shao, P. Yan, Y. Cheng, K. Han, Z. Nie, C. Wang, J. Yang, X. Li, P. Bhattachary, K. T. Mueller and J. Liu, Nat. Energy, 2016, 1, 16039.
16. N. Palaniyandy, M. Kebede, K. Raju, K. Ozoemena, L. le Roux, M. Mathe and R. Jayaprakasam, Mater. Chem. Phys., 2019, 230, 258–266.
17. B. Wu, G. Zhang, M. Yan, T. Xiong, P. He, L. He and L. Mai, Small, 2018, 14, 1703850.
18. S. Khamsanga, M. T. Nguyen, T. Yonezawa, P. Thamyongkit, R. Pornprasertsuk, P. Pattananuwat, A. Tuantranont, S. Siwamogsatham and S. Kheawhom, Int. J. Mol. Sci., 2020, 21, 4689.
19. B. Jiang, C. Xu, C. Wu, L. Dong, J. Li and F. Kang, Electrochim. Acta, 2017, 229, 422–438.
20. J. Hao, J. Mou, J. Zhang, L. Dong, W. Liu, C. Xu and F. Kang, Electrochim. Acta, 2018, 259, 170–178.
21. Y. Fu, Q. Wei, G. Zhang, X. Wang, J. Zhang, Y. Hu and S. Sun, Adv. Energy Mater., 2018, 8, 1801445.
22. B. Lee, H. R. Lee, H. Kim, K. Y. Chung, B. W. Cho and S. H. Oh, Chem. Commun., 2015, 51, 9265–9268.
23. W. Kohn and L. J. Sham, Phys. Rev., 1965, 140, A1133–A1138.
24. S. J. Clark, M. D. Segall, C. J. Pickard, P. J. Hasnip, M. I. J. Probert, K. Refson and M. C. Payne, Z. Kristallogr. - Cryst. Mater., 2005, 220(5/6), 567–570.
25. H. J. Monkhorst and J. D. Pack, Phys. Rev. B: Solid State, 1976, 13, 5188–5192.
26. B. Lee, C. S. Yoon, H. R. Lee, K. Y. Chung, B. W. Cho and S. H. Oh, Sci. Rep., 2014, 4, 6066.
27. N. Govind, M. Petersen, G. Fitzgerald, D. King-Smith and J. Andzelm, Comput. Mater. Sci., 2003, 28, 250–258.
28. W. Sun, F. Wang and S. Hou, J. Am. Chem. Soc., 2017, 139, 9775–9778.
29. D. Zhang, J. Cao, X. Zhang, N. Insin, S. Wang, J. Han and Y. Huang, Adv. Funct. Mater., 2021, 2009412.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1ra00346a

‡ These authors contributed equally to this work.

---