Synthesis of a Hydrogenated Zigzag Carbon Nanobelt

Abstract

Open AccessCCS ChemistryRESEARCH ARTICLE1 Jan 2021Synthesis of a Hydrogenated Zigzag Carbon Nanobelt Han Chen, Shaojun Gui, Yiqun Zhang, Zhifeng Liu and Qian Miao Han Chen Department of Chemistry, The Chinese University of Hong Kong, Sha Tin, New Territories, Hong Kong, , Shaojun Gui Department of Chemistry, The Chinese University of Hong Kong, Sha Tin, New Territories, Hong Kong, , Yiqun Zhang Department of Chemistry, The Chinese University of Hong Kong, Sha Tin, New Territories, Hong Kong, , Zhifeng Liu Department of Chemistry, The Chinese University of Hong Kong, Sha Tin, New Territories, Hong Kong, and Qian Miao *Corresponding author: E-mail Address: [email protected] Department of Chemistry, The Chinese University of Hong Kong, Sha Tin, New Territories, Hong Kong, https://doi.org/10.31635/ccschem.020.202000189 SectionsSupplemental MaterialAboutAbstractPDF ToolsAdd to favoritesDownload CitationsTrack Citations ShareFacebookTwitterLinked InEmail Zigzag carbon nanobelts are a long-sought-after yet unrealized target for organic synthesis. Herein, we report a study toward the synthesis of tetrabenzo[10]cyclacene, an undocumented zigzag carbon nanobelt. The synthetic precursor of tetrabenzo[10]cyclacene is its box-shaped tetraepoxy derivative, which is synthesized through iterative Diels–Alder reactions utilizing a "C"-shaped building block. Attempted aromatization of this tetraepoxy nanobox toward the formation of a fully conjugated nanobelt results in an octahydro derivative of tetrabenzo[10]cyclacene. The structures of the tetraepoxy nanobox and octahydrotetrabenzo[10]cyclacene were both unambiguously identified with single-crystal X-ray crystallography. Download figure Download PowerPoint Introduction Zigzag carbon nanobelts are a long-sought-after for organic synthesis, yet unrealized target. The shortest zigzag carbon nanobelts are [n]cyclacenes (Figure 1a), which were first proposed in 1954 as hypothetical molecules for theoretical study.1 Consisting of a loop of fully fused benzene rings, carbon nanobelts represent sidewall fragments of single-walled carbon nanotubes (SWCNTs). As a result, carbon nanobelts can be classified into armchair, zigzag, and chiral nanobelts according to the chirality of the corresponding SWCNTs, which are described with a chiral index (n,m).2 Chirality-specific synthesis of SWCNTs remains the top challenge in the science of carbon nanotubes and a bottleneck limiting their applications.3 To meet this challenge, carbon nanobelts and other hoop-shaped molecular nanocarbons4 (e.g. cycloparaphenylenes5–7 and related carbon nanorings8–12) can, in principle, serve as templates or precursors for the growth of SWCNTs.13,14 This concept has been proved by Itami's study on selective growth of armchair carbon nanotubes initiated by a well-defined carbon nanoring15 and Fasel's study on chirality-specific growth of (6,6) armchair SWCNTs on platinum surfaces from an end-cap fragment seed.16 Under a broader definition, carbon nanobelts may also have nonhexagonal rings embedded in a loop of fully fused benzene rings.17 The first pentagon-embedded carbon nanobelt was proposed by Schlüter,18 and the first octagon-embedded carbon nanobelt was synthesized by Gleiter and co-workers.19,20 The first armchair and chiral carbon nanobelts were synthesized recently by the Itami21,22 and Miao groups,23 respectively. In contrast, zigzag carbon nanobelts remain unrealized, although great efforts have been made to synthesize them.24–26Attempts to synthesize [n]cyclacenes have been reported by groups led by Stoddart,27,28 Cory,29,30 Schlüter,31,32 Peña,33 and Wang,34,35, who all successfully constructed hoop-shaped molecular frameworks but failed to achieve fully conjugated structures. A challenge in synthesizing [n]cyclacenes is their instability,36 which arises from the fact that the closed-shell structure of [n]cyclacene does not contain a Clar's aromatic sextet.37 To design more stable zigzag carbon nanobelts, [n]cyclacenes can, in principle, be stabilized by annulation of benzene rings as demonstrated by [n]benzo[n]cyclacenes (Figure 1a) proposed by Itami in 201638 containing 2 n Clar aromatic sextets but have not yet been synthesized. Recently, our group also used this strategy to design N-doped zigzag carbon nanobelts.39 Figure 1 | (a) Documented zigzag carbon nanobelts; (b) structures of tetrabenzo[10]cyclacene (1a/b), tetraepoxy deriviatve 2, and octahydrotetrabenzo[10]cyclacene 3. Download figure Download PowerPoint Herein, we report a new design of zigzag carbon nanobelt, tetrabenzo[10]cyclacene, and studies toward the synthesis of it. As shown in Figure 1b, tetrabenzo[10]cyclacene ( 1a) and its t-butylated derivative ( 1b) have four benzene rings fused to [10]cyclacene. As a result, 1a/ b have six Clar's aromatic sextets, which are highlighted in light blue in Figure 1b, and are therefore expected to be more stable than [10]cyclacene. The synthetic precursor toward tetrabenzo[10]cyclacene 1b is 2, a box-shaped tetraepoxy derivative as shown in Figure 1b. It has been found that reductive deoxygenation of 2 results in 3, a hydrogenated carbon nanobelt as shown in Figure 1b. Further details describing the synthesis and structural analysis of 2 and 3 are shown below. Experimental Methods Experimental methods for synthesis and characterization are detailed in the Supporting Information. Results and Discussion Synthesis Scheme 1 shows the synthesis of 2 and 3 starting from bisfuran ( 4), which was synthesized following the reported procedures with some modification.40 Treatment of bistriflate ( 5)41 with 1.2 equivalents of CsF resulted in the corresponding benzyne in situ, which reacted with 0.38 equivalents of 4 to afford the Diels–Alder adduct 6 as a mixture of isomers. According to the relative positions of two oxygen bridges, 6 exists as a pair of stereoisomers, syn- 6 and anti- 6, and both show two regioisomers with different positions of the trimethylsilyl and triflate groups. Syn- 6, which was isolated with a yield of 36%, is shaped like the letter C and can provide the curvature and rigidity necessary for macrocyclization in the next step. Treatment of syn- 6 with an excessive amount of CsF resulted in the corresponding dibenzyne in situ, which reacted with 1.5 equivalents of 4 to give 2 as a white solid at a yield of 17%. To convert 2 to the fully conjugated nanobelt, 1b (Figure 1), several conditions for reductive aromatization were examined. Treatment of 2 with TiCl4/LiAlH4, TiCl4/Zn, or i-PrMgBr42 resulted in partial removal of the four bridging oxygen atoms as found from the mass spectra of the crude products, while treatment of 2 with SnCl2/HCl or Et3SiH/CF3CO2H43 did not give deoxygenated products. In contrast, reaction of 2 with an excessive amount of NaI/trimethylsilyl iodide (TMSI) resulted in 3, the octahydro derivative of 1b, as the sole product at an excellent yield of 91%. When 3 was heated with 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ), Pd on activated charcoal, or CuO44 for dehydrogenation, no reaction occurred. Based on the aforementioned results, we hypothesize that 1b is not stable enough under conditions of reductive aromatization. As a result, it was converted into 3 although the hydrogenation mechanism remains unclear. Scheme 1 | Synthesis of 2 and 3. Download figure Download PowerPoint Crystal structures Single crystals of 2 suitable for X-ray crystallography were obtained from its solution in chloroform and heptane by slow solvent evaporation.a As shown in Figure 2a, each unit cell contains two molecules of 2 and four molecules of chloroform. Molecules of 2 assemble into a layer and molecules of chloroform are sandwiched between two adjacent layers of 2. As shown in Figure 2b, the macrocyclic backbone of 2 is shaped like a box, which is 8.0 Å long and 6.1 Å wide. The cavity inside 2 is similar in size to that of tetraepoxy[10]cyclacene reported by Peña and co-workers.33 However, unlike Peña's tetraepoxy[10]cyclacene accommodating the methyl group of an acetonitrile molecule in the crystal, the cavity inside 2 is empty in the crystal mainly because the t-butyl groups in 2 block potential guest molecules to access the cavity. Moreover, when the van der Waals radii of carbon atoms (1.7 Å) are taken into consideration, the cavity inside 2 is only 4.6 Å long and 2.7 Å wide and, thus, cannot accommodate a molecule of chloroform. The pyrene unit in 2 is not as flat as that in 2,7-di(t-butyl)pyrene in its single crystal45 but is slightly bent. The angle between the two yellow benzenoid rings (Figure 2b) is 5.8°, and the angle between the two light blue benzenoid rings (Figure 2b) is 3.2°. The C(sp2)–C(sp3)–C(sp2) bond angles at the bridgehead carbons in the oxanorbornene moieties are 104.2° and 104.4°, which are very similar to that of the oxanorbornene moiety in nonmacrocyclic structures.46 Figure 2 | Crystal structure of 2 • (CHCl3)2: (a) molecular packing (all molecules are shown with space-filling models); (b) side view of 2 (the t-butyl groups are removed for clarity, carbon atoms and oxygen atoms are shown with ellipsoids set at 50% probability). Download figure Download PowerPoint The molecular structure of 3 was characterized with nuclear magnetic resonance (NMR) and high-resolution mass spectrometry (HRMS) as detailed in the Supporting Information and unambiguously determined by X-ray diffraction from single crystals that were obtained from its solution in dichloromethane by slow evaporation of solvent.a It was found that each unit cell of the crystal contains two molecules of 3 and eight molecules of dichloromethane. As shown in Figure 3a, one molecule of dichloromethane occupies the space between two t-butyl groups on each side of 3, while the cavity inside 3 is empty. Moreover, two molecules of dichloromethane are sandwiched between two molecules of 3. As shown in Figure 3b, the macrocyclic backbone of 3 is shaped like an American football, which is 11.0 Å long and 4.6 Å wide as measured between the corresponding carbon atoms. When the van der Waals radii of carbon atoms (1.7 Å) are taken into consideration, the cavity inside 3 is 7.6 Å long and only 1.2 Å wide and is too small to accommodate a molecule of dichloromethane. The bond angles of the sp3 carbon atoms (shown in red in Figure 3b) between the red bonds are in the range of 105.2°–106.7°, which is slightly smaller than the standard bond angle of sp3 carbon. The dibenzoanthracene moiety of 3, highlighted in yellow in Figure 3c, is bent. The angles are 24.5° between rings A and B, 13.6° between rings A and C, 17.7° between rings C and D, and 10.4° between rings D and E. Figure 3 | Crystal structure of 3 • (CH2Cl2)4: (a) molecular packing as viewed along the b axis of the unit cell; (b) side view of 3 (sp3 carbon atoms are highlighted in red); (c) top view of 3. (Carbon atoms in 3 are shown as ellipsoids set at 50% probability; molecules of CH2Cl2 are shown as space-filling models; in panels b and c, t-butyl substituting groups were removed for clarity.) Download figure Download PowerPoint Electronic structures Both 2 and 3 have π-systems separated by sp3 carbons and constrained in a loop. To investigate their electronic structures, 2 and 3 were studied with both computational and experimental methods. To calculate the frontier molecular orbitals of 2 and 3 with reduced computational cost, simplified model molecules ( 2' and 3') with methyl groups replacing the t-butyl groups, were optimized at the B3LYP level of density functional theory (DFT) with the 6-31 G(d,p) basis set, and their molecular orbitals were then calculated with the 6-31++G(d,p) basis set. As shown in Figure 4a, the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) of 2' were delocalized on both pyrene and benzene moieties with major distribution on the pyrene moieties. The DFT calculations with the same basis set indicate that 2,7-dimethylpyrene (−5.57 eV) has a lower HOMO energy level than 2' (−5.38 eV) and a higher LUMO level (−1.79 eV) than 2' (−2.22 eV).39 As a result, the calculated HOMO–LUMO gap of 2' is smaller than that of 2,7-dimethylpyrene by 0.62 eV. The LUMO of 3' (Figure 4a) exhibits through-space conjugation as a result of the short distance between the two π-systems. In agreement with the calculated HOMO–LUMO gaps of 2' and 2,7-dimethylpyrene, the UV–vis absorption spectrum of 2 (Figure 4b) is red-shifted relative to that of 2,7-di(t-butyl)pyrene (TBP) by 0.52 eV as measured from the absorption edge. In agreement with the fact that 3 has a larger HOMO–LUMO gap (3.63 eV) than 2, the absorption spectrum of 3 is blue-shifted relative to that of 2 as shown in Figure 4b. Upon UV light excitement, the colorless solutions of 2 and 3 both exhibit blue fluorescence with quantum yields of 16% and 8%, respectively. The cyclic voltammograms of 2 and 3 in CH2Cl2 both exhibit two quasireversible oxidation waves in the testing window. The first oxidation waves of 2 and 3 have half-wave potentials of 0.68 and 0.59 V versus ferrocenium/ferrocene, respectively. The oxidation potential of 2 is smaller than that of TBP (0.84 V) by 0.16 V, which agrees with the difference between the calculated HOMO energy levels of 2' and 2,7-dimethylpyrene. On the basis of these oxidation potentials, the HOMO energy levels of 2 and 3 are estimated as −5.78 and −5.69 eV,47,b respectively, which roughly agree with the DFT calculated values. Figure 4 | (a) Frontier molecular orbitals of 2′ and 3', which are calculated at the B3LYP level of DFT with 6-311++G(d,p)//6-31 G(d,p) basis sets; (b) UV–vis absorption spectra of 2, 3 (both 1 × 10−5 mol/L in CH2Cl2) and TBP (2×10−5 mol/L in CH2Cl2) with the molecular structure of TBP. Download figure Download PowerPoint Strains To better understand the stability of zigzag carbon nanobelts 1a/b, we calculated the strain energies of a series of zigzag carbon nanobelts ( A n in Figure 5) at the B3LYP level of DFT with the 6-31 G(d) basis set using Itami's method,48 which affords the strain energy of each carbon nanobelt as a function of n−1 (n is the number of repeat unit in a carbon nanobelt) by linear regression analysis of the enthalpy of carbon nanobelt per repeat unit as a function of n−2. Therefore, the strain energy of 1a is 124.4 kcal/mol, which is slightly larger than that of the armchair carbon nanobelt synthesized by Itami and co-workers (119.5 kcal/mol)21 and much larger than that of the chiral carbon nanobelt synthesized by our group (28.2 kcal/mol).23 The large strain energy of 1a is in agreement with the hypothesis that 1b is not stable enough under conditions of reductive aromatization and the fact that 1b cannot be obtained from 3 by dehydrogenation. Figure 5 | Structures of a series of zigzag carbon nanobelts with calculated strain energies. Download figure Download PowerPoint Conclusion In summary, nanobox 2 was designed as a synthetic precursor toward tetrabenzo[10]cyclacene ( 1b), an undocumented zigzag carbon nanobelt, and successfully synthesized through iterative Diels–Alder reactions utilizing the "C"-shaped building block (syn- 6). Attempted aromatization of this tetraepoxy nanobox toward the formation of 1b resulted in 3, an octahydro derivative of 1b. The molecular structures of 2 and 3 were identified with single-crystal X-ray crystallography, and their HOMO and LUMO were studied with both computational and experimental methods. The DFT calculations suggest that the larger analogues of 1a, A n (n > 2), are more reachable synthetic targets with lower ring strains. Studies toward the synthesis of A n (n > 2) are currently in progress in our laboratory. Footnotes a CCDC 1975215 and 1975216 contain the supplementary crystallographic data for 2 • (CHCl3)2 and 3 • (CH2Cl2)4, respectively. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre. b The commonly used formal potential of the redox couple of ferrocenium/ferrocene (Fc+/Fc) in the Fermi scale is –5.1 eV, which is calculated on the basis of an approximation neglecting solvent effects using a work function of 4.46 eV for the normal hydrogen electrode (NHE) and an electrochemical potential of 0.64 V for (Fc+/Fc) versus NHE. See Cardona et al.47 Supporting Information Supporting Information is available. Conflict of Interest There is no conflict of interest to report. Acknowledgments The authors wish to acknowledge Ms. Hoi Shan Chan (the Chinese University of Hong Kong) for the single-crystal crystallography. This research was made possible as a result of generous grants from the Research Grants Council of Hong Kong (GRF 14300218) and the Croucher Senior Research Fellowship. References 1. Heilbronner E.Molecular Orbitals in Homologen Reihen mehrkerniger Aromatischer Kohlenwasserstoffe: I. Die Eigenwerte yon LCAO-MO's in Homologen Reihen.Helv. Chim. Acta.1954, 37, 921–935. Google Scholar 2. Saito R.; Fujita M.; Dresselhaus G.; Dresselhaus M. S.Electronic-Structure of Chiral Graphene Tubules.Appl. Phys. Lett.1992, 60, 2204–2206. Google Scholar 3. Liu B.; Wu F.; Gui H.; Zheng M.; Zhou C.Chirality-Controlled Synthesis and Applications of Single-Wall Carbon Nanotubes.ACS Nano.2017, 11, 31–53. Google Scholar 4. Kammermeier S.; Jones P. G.; Herges R.Ring-Expanding Metathesis of Tetradehydro-Anthracene—Synthesis and Structure of a Tubelike, Fully Conjugated Hydrocarbon.Angew. Chem. Int. Ed.1996, 35, 2669–2671. Google Scholar 5. Jasti R.; Bhattacharjee J.; Neaton J. B.; Bertozzi C. R.Synthesis, Characterization, and Theory of [9]-, [12]-, and [18]Cycloparaphenylene: Carbon Nanohoop Structures.J. Am. Chem. Soc.2008, 130, 17646–17647. Google Scholar 6. Takaba H.; Omachi H.; Yamamoto Y.; Bouffard J.; Itami K.Selective Synthesis of [12]Cycloparaphenylene.Angew. Chem. Int. Ed.2009, 48, 6112–6116. Google Scholar 7. Yamago S.; Watanabe Y.; Iwamoto T.Synthesis of [8]Cycloparaphenylene from a Square-Shaped Tetranuclear Platinum Complex.Angew. Chem. Int. Ed.2010, 49, 757–759. Google Scholar 8. Hitosugi S.; Yamasaki T.; Isobe H.Bottom-Up Synthesis and Thread-In-Bead Structures of Finite (N, 0)-Zigzag Single-Wall Carbon Nanotubes.J. Am. Chem. Soc.2012, 134, 12442–12445. Google Scholar 9. Golling F. E.; Quernheim M.; Wagner M.; Nishiuchi T.; Müllen K.Concise Synthesis of 3D Pi-Extended Polyphenylene Cylinders.Angew. Chem. Int. Ed.2014, 53, 1525–1528. Google Scholar 10. Tran-Van A. F.; Huxol E.; Basler J. M.; Neuburger M.; Adjizian J. J.; Ewels C. P.; Wegner H. A.Synthesis Of Substituted [8]Cycloparaphenylenes by [2 + 2 + 2] Cycloaddition.Org. Lett.2014, 16, 1594–1597. Google Scholar 11. Lu D.; Wu H.; Dai Y.; Shi H.; Shao X.; Yang S.; Yang J.; Du P.A Cycloparaphenylene Nanoring with Graphenic Hexabenzocoronene Sidewalls.Chem. Commun.2016, 52, 7164–7167. Google Scholar 12. Huang Z. A.; Chen C.; Yang X. D.; Fan X. B.; Zhou W.; Tung C. H.; Wu L. Z.; Cong H.Synthesis of Oligoparaphenylene-Derived Nanohoops Employing an Anthracene Photodimerization-Cycloreversion Strategy.J. Am. Chem. Soc.2016, 138, 11144–11147. Google Scholar 13. Bunz U. H.; Menning S.; Martin N.Para-Connected Cyclophenylenes and Hemispherical Polyarenes: Building Blocks for Single-Walled Carbon Nanotubes?Angew. Chem. Int. Ed.2012, 51, 7094–7101. Google Scholar 14. Fort E. H.; Donovan P. M.; Scott L. T.Diels-Alder Reactivity of Polycyclic Aromatic Hydrocarbon Bay Regions: Implications for Metal-Free Growth of Single-Chirality Carbon Nanotubes.J. Am. Chem. Soc.2009, 131, 16006–16007. Google Scholar 15. Omachi H.; Nakayama T.; Takahashi E.; Segawa Y.; Itami K.Initiation of Carbon Nanotube Growth by Well-Defined Carbon Nanorings.Nat. Chem.2013, 5, 572–576. Google Scholar 16. Sanchez-Valencia J. R.; Dienel T.; Gröning O.; Shorubalko I.; Mueller A.; Jansen M.; Amsharov K.; Ruffieux P.; Fasel R.Controlled Synthesis of Single-Chirality Carbon Nanotubes.Nature.2014, 512, 61–64. Google Scholar 17. Chen H.; Miao Q.Synthesis of Tribenzo[a, c, e]cyclooctene Oligomers: Toward Negatively Curved Nanocarbons.ChemPlusChem2019, 84, 627–629. Google Scholar 18. Neudorff W. D.; Lentz D.; Anibarro M.; Schlüter A. D.The Carbon Skeleton of the Belt Region Of Fullerene C84 (D2).Chem. Eur. J.2003, 9, 2745–2757. Google Scholar 19. Esser B.; Rominger F.; Gleiter R.Synthesis of [6.8]3cyclacene: Conjugated Belt and Model for an Unusual Type of Carbon Nanotube.J. Am. Chem. Soc.2008, 130, 6716–6717. Google Scholar 20. Esser B.; Bandyopadhyay A.; Rominger F.; Gleiter R.From Metacyclophanes to Cyclacenes: Synthesis and Properties of [6.8]3cyclacene.Chem. Eur. J.2009, 15, 3368–3379. Google Scholar 21. Povie G.; Segawa Y.; Nishihara T.; Miyauchi Y.; Itami K.Synthesis of A Carbon Nanobelt.Science.2017, 356, 172–175. Google Scholar 22. Povie G.; Segawa Y.; Nishihara T.; Miyauchi Y.; Itami K.Synthesis and Size-Dependent Properties of [12], [16], and [24]Carbon Nanobelts.J. Am. Chem. Soc.2018, 140, 10054–10059. Google Scholar 23. Cheung K. Y.; Gui S.; Deng C.; Liang H.; Xia Z.; Liu Z.; Chi L.; Miao Q.Synthesis of Armchair and Chiral Carbon Nanobelts.Chem.2019, 5, 838–847. Google Scholar M. R.; L. Jasti Implications Toward the Synthesis of Aromatic Google Scholar Wegner H. the to Carbon The Synthesis of an Aromatic Chem. Int. Google Scholar Lu X.; Wu of The Synthesis of a Carbon 2, Google Scholar P. R.; R.; S.; D.; F. H.; J. P.; A. M. Z.; R.; J. F.; 1. Synthesis Am. Chem. Google Scholar P. R.; D.; F. H.; J. P.; F. M.; A. M. Z.; J. F.; 2. of and Structures.J. Am. Chem. Google Scholar M.; C. of a Belt into and 37, Google Scholar M.; C. L.; A. J.; J. of by Synthesis of a to an 37, Google Scholar A.; Schlüter A and the Chem. Int. Google Scholar O.; P.; M.; Schlüter an and a Chem. Google Scholar F.; F.; K.; D.; E.; L.; Peña a to by Chem. Int. Google Scholar Shi S.; the Synthesis of a Hydrocarbon Am. Chem. Google Scholar Shi S.; Chem. Int. Google Scholar Chen Z.; E.; Lu X.; H. F.; Dai S.; P.; K. of and Zigzag 9, Google Scholar Clar Aromatic J. New Google Scholar K.; M.; Segawa Y.; Itami and Reactivity of a Toward the Synthesis of Zigzag Carbon Google Scholar J.; Miao Toward Synthesis of Zigzag Carbon Google Scholar D.; J.; T.; P. of and Google Scholar B.; S.; D.; E.; G.; Peña D.; and of by Google Scholar H.; of by Google Scholar H.; Y.; T.; K.; S.; Z.; Isobe H.Synthesis of and of Google Scholar Liang Z.; W.; S.; C.; Miao of Google Scholar A. C.; J. Crystal Structure of Google Scholar K. G.; F. R.; H. as a for for Google Scholar Cardona C. M.; W.; A. E.; D.; for Frontier of Conjugated for Google Scholar Segawa Y.; A.; H.; Itami on the of Carbon Google Scholar Information Chinese authors wish to acknowledge Ms. Hoi Shan Chan (the Chinese University of Hong Kong) for the single-crystal crystallography. This research was made possible as a result of generous grants from the Research Grants Council of Hong Kong (GRF 14300218) and the Croucher Senior Research Fellowship.