Heteroatom-Doped Nanographenes with Structural Precision

Conspectus Nanographenes, which are defined as nanoscale (1–100 nm) graphene cutouts, include quasi-one-dimensional graphene nanoribbons (GNRs) and quasi-zero-dimensional graphene quantum dots (GQDs). Polycyclic aromatic hydrocarbons (PAHs) larger than 1 nm can be viewed as GQDs with atomically precise molecular structures and can thus be termed nanographene molecules. As a result of quantum confinement, nanographenes are promising for next-generation semiconductor applications with finite band gaps, a significant advantage compared with gapless two-dimensional graphene. Similar to the atomic doping strategy in inorganic semiconductors, incorporation of heteroatoms into nanographenes is a viable way to tune their optical, electronic, catalytic, and magnetic properties. Such properties are highly dependent not only on the molecular size and edge structure but also on the heteroatom type, doping position, and concentration. Therefore, reliable synthetic methods are required to precisely control these structural features. In this regard, bottom-up organic synthesis provides an indispensable way to achieve structurally well-defined heteroatom-doped nanographenes. Polycyclic heteroaromatic compounds have attracted great attention of organic chemists for decades. Research in this direction has been further promoted by modern interest in supramolecular chemistry and organic electronics. The rise of graphene in the 21st century has endowed large polycyclic heteroaromatic compounds with a new role as model systems for heteroatom-doped graphene. Heteroatom-doped nanographene molecules are in their own right promising materials for photonic, optoelectronic, and spintronic applications because of the extended π conjugation. Despite the significant advances in polycyclic heteroaromatic compounds, heteroatom-doped nanographene molecules with sizes of over 1 nm and their relevant GNRs are still scarce. In this Account, we describe the synthesis and properties of large heteroatom-doped nanographenes, mainly summarizing relevant advances in our group in the past decade. We first present several examples of heteroatom doping based on the prototypical nanographene molecule, i.e., hexa-peri-hexabenzocoronene (HBC), including nitrogen-doped HBC analogues by formal replacement of benzene with other heterocycles (e.g., aromatic pyrimidine and pyrrole and antiaromatic pyrazine) and sulfur-doped nanographene molecules via thiophene annulation. We then introduce heteroatom-doped zigzag edges and a variety of zigzag-edged nanographene molecules incorporating nitrogen, boron, and oxygen atoms. We finally summarize heteroatom-doped GNRs based on the success in the molecular cases. We hope that this Account will further stimulate the synthesis and applications of heteroatom-doped nanographenes with a combined effort from different disciplines.


INTRODUCTION
Since the isolation of single graphene layers by Geim and Novoselov, 1 graphene has opened up tremendous opportunities for materials science because of its extraordinary properties and broad applications. 2−5 In particular, the ultrahigh charge-carrier mobility of graphene is extremely appealing for electronic devices; however, the use of graphene as the active layer in fieldeffect transistors (FETs) is hindered by the lack of an electronic band gap because the devices cannot be switched off. 6 This disadvantage can be overcome by reduction in the dimensionality of graphene, generating quasi-zero-dimensional graphene quantum dots (GQDs) and quasi-one-dimensional graphene nanoribbons (GNRs) (Figure 1). 5,7 Because of quantum confinement, these nanoscale graphenes (nanographenes) are promising for next-generation semiconductor applications with finite band gaps. 8,9 The fabrication of nanographenes is often accomplished through top-down methods such as "cutting" of graphene and carbon nanotubes. 5 The applications of nanographenes, for example, in electronic devices, catalysis, bioimaging and sensing, have also been demonstrated. 8,10 However, the top-down approach cannot accurately control the size and edge structure, which are highly relevant to the physical and chemical properties of nanographenes. Therefore, control of the nanographene structure at the atomic level is required to achieve reliable structure−property correlations and desired functions. 5 This goal can be realized through bottom-up syntheses starting with molecular precursors, and this approach illustrates how organic chemists can contribute to the materials science of nanographenes, providing atomically precise GQDs (e.g., nanographene molecules) and GNRs. 11,12 The capability to precisely control nanographene structures further promotes the creation of emergent materials such as magnetic GNRs and topological insulators by rational molecular design. 13−15 Atomic doping is a common strategy to tune the properties of inorganic materials, for instance, in silicon-based semiconductor industry: doping of silicon with group III or group V elements can achieve p-doped or n-doped semiconductors, respectively. The atomic doping strategy is also essential to nanographene materials through incorporation of heteroatoms into the graphenic carbon network, effectively modulating the electronic, magnetic, and catalytic properties. 16 Again, the bottom-up method is indispensable to achieve atomic precision not only in terms of the size and edge structure but also with regard to the heteroatom type, doping position, and concentration. This accuracy is an indispensable tool for controlling the physical properties and revealing structure−property relationships. 11,12 In addition, structurally well-defined heteroatom-doped nanographenes are intriguing semiconducting materials. The heteroatoms in the nanographene framework provide more opportunities for materials development (i) by modulating the band gap and thus the photophysical properties, (ii) by offering rich electrochemical activities, (iii) by installing coordinating sites for metals, and (iv) by stabilizing charges and spins in the carbon skeleton. 16 Nevertheless, large nanographene molecules and GNRs incorporating various heteroatoms (e.g., B, 17 N, 18,19 O, 20 S, 21 and P 22 ) are still scarce. 16 Our group has been working on the synthesis, characterization, and application of nanographenes during the past decades. In 1995 we reported the synthesis of alkyl-substituted hexa-peri-hexabenzocoronene (HBC) 1b, 23 which has a diameter of over 1 nm and can thus be regarded as a nanographene molecule (Figure 2a). 5,7 Since this seminal work, we have synthesized various HBC derivatives and other larger nanographene molecules by means of a general synthetic strategy, namely, cyclodehydrogenation of polyphenylene precursors (e.g., hexaphenylbenzene 2, Figure 2a) to afford planarized products. 24 The alkyl substituents not only impart solubility and solution processability to the disc-shaped nanographene molecules but also induce the formation of columnar liquid-crystalline mesophases (Figure 2b). 24 Charges can therefore be transported through the columnar π−π stacking channel, which is a special concept for FETs and photovoltaic devices. 25−27 We have further extended the synthetic strategy to achieve atomically precise GNRs through conventional solution chemistry and modern on-surface syntheses. 28 These advances have been reviewed in our previous contributions. 5,7,11,12,24,28 In this Account, we focus on our works toward heteroatom-doped nanographenes, which have yet been comprehensively summarized. In the following, we will discuss the design, synthesis,  properties, and future opportunities of the research on heteroatom-doped nanographene molecules and GNRs.

Nitrogen-Doped Nanographene Molecules
Following the synthetic strategy of HBCs, it is easy to imagine the possibility of incorporating heteroatoms into the HBC skeleton by replacing the outer phenyl groups with other heteroaromatic rings in the hexaarylbenzene precursor ( Figure  3a). Indeed, Draper et al. synthesized the first N-doped HBC 4 in 2002 through cyclodehydrogenation of the corresponding pyrimidine-substituted precursor 3 (Figure 3b). 29 The incorporation of pyrimidine rings endowed the nanographene molecule with electron-accepting properties compared with the parent HBC. Furthermore, the N atoms in the bay region served as a coordinating site with transition metals (e.g., Pd II and Ru II ). 30 Not surprisingly, such metal complexes displayed profoundly changed absorption and emission features.
To develop electron-donating HBC analogues, in 2007 our group reported the synthesis and properties of hexapyrrolohexaazacoronenes 6 through oxidative cyclodehydrogenation of hexapyrrolylbenzene precursors 5 (Figure 3c). 31 The peripheral benzene rings were formally replaced with six pyrroles, providing a π skeleton isoelectronic to that of HBC. The interior nitrogens played an important role in stabilizing the oxidation states, and thereby, the dication of 6c was generated by chemical oxidation and isolated under ambient conditions. Later on, together with Takase, Nishinaga, and co-workers, we extended the family of pyrrole-fused HBC analogues by incorporating a mixture of phenyl and pyrrolyl groups in the hexaarylbenzene precursors ( Figure 3d). 32 This unique class of molecules has attracted great interest from other groups. For example, Takase, Uno, and coworkers made further modifications by changing the peripheral aryl moieties (compounds 10 and 11) or by extending the benzene core to naphthalene (compound 12) (Figure 3e). 33−35 Stepieńet al. reported disc-shaped naphthalene monoimidefused hexapyrrolohexaazacoronene 13 as well as an interesting    derivative 14 containing five-, six-, and seven-membered rings, obtained from an indole-containing precursor ( Figure 3f). 36,37 As pyrimidine and pyrrole are both 6π aromatic heterocycles, incorporation of these rings into nanographene molecules does not change the aromatic nature because the number of π electrons is unchanged. From a chemistry perspective, we were interested in designing a new class of nanographene molecules involving antiaromatic rings that would lead to low energy gaps between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) as well as excellent redox properties. 38 In 2017, together with Auwarter, Barth, Palma, and co-workers, we conceived an azomethine ylide homocoupling strategy to achieve the synthesis of pyrazineembedded HBC 15, which possessed an 8π central ring through the introduction of a pyrazine core instead of a benzene core (Figure 4a). 39 After deprotonation of dibenzo-9a-azaphenalene salts 16, polycyclic azomethine ylides (PAMYs) 17 were generated in situ with several existing resonance structures, including the zwitterionic and diradical forms (Figure 4d). The PAMYs were highly reactive and underwent dimerization in solution upon heating to afford "dimers" 18. Oxidative dehydrogenation of 18b in a sealed NMR tube under exclusion of air gave N-doped hexabenzoperylene 19, but further cyclodehydrogenation in solution was unsuccessful. Nevertheless, the on-surface method enabled the synthesis of the fully fused N-doped HBC 15 through dimerization and concomitant cyclodehydrogenation on Ag(111) under ultrahigh vacuum (UHV) using dibenzo-9a-azaphenalene salt 16a as the precursor. The chemical structure of 15 was unambiguously characterized by high-resolution scanning tunneling microscopy (STM) and noncontact atomic force microscopy (nc-AFM) with a CO-functionalized tip (Figure 4b,c). Density functional theory (DFT) calculations at the B3LYP/6-311G(d,p) level revealed a significantly narrowed HOMO−LUMO gap of 15 (2.06 eV) compared with that of the pristine HBC 1a (3.57 eV), suggesting that the introduction of antiaromatic rings could be an effective way to tune the HOMO−LUMO gap and develop low-energy-gap nanographene materials.

Sulfur-Doped Nanographene Molecules
Incorporation of thiophene rings into the polyphenylene precursors results in S-doped nanographene molecules after cyclodehydrogenation. In 2007 we synthesized tribenzothiophene-fused HBC 21a as the first example of S-doped nanographene molecules through cyclodehydrogenation of the C 3 -symmetric precursor 20a (Figure 5a). 40 Recently, the

Accounts of Chemical Research
Article tribenzothiophene-fused HBC core of 21a was derivatized by attaching alkoxy and alkyl chains (21b and 21c) to ensure good solubility and to tune their HOMO and LUMO levels. 41 Remarkably, the electron-rich thiophene rings in the alkylsubstituted tribenzothiophene-fused HBC 21c could be oxidized into electron-deficient thiophene-S,S-dioxide moieties, greatly lowering both the HOMO and LUMO levels by about 0.5 eV and representing an appealing feature of thiophene incorporation. In 2011, Draper et al. described another way of introducing thiophene rings in S-doped HBC analogue 24 ( Figure 5b). 42 During the oxidative cyclodehydrogenation, dimerization of 24 was also observed via intermolecular oxidative coupling at the thiophene α-position.
Apart from the above-mentioned method through cyclodehydrogenation of suitable precursors to incorporate thiophene rings, we have also developed a postfunctionalization strategy to fuse thiophenes onto the HBC core based on edgeperchlorinated HBC 26. 43 A thiolation reaction was performed to provide trithiophene-annelated HBCs 27 in 35−50% yield and dithiophene-annelated products 28 in 1.5−3% yield ( Figure  6). 44 Furthermore, Tan, Zheng, and co-workers employed 27a as hole-transporting material (HTM) in perovskite solar cells, providing the best efficiency of 12.8%, which could be improved to 14.0% through further device optimization. 45 Further studies indicated that 27a endowed the devices with enhanced stability compared with the well-established HTM 2,2′,7,7′-tetrakis-[N,N-bis(4-methoxyphenyl)amino]-9,9′-spirobifluorene (Spiro-OMeTAD). In addition, together with Wu, Feng, and coworkers, we used nanographene molecule 27a as a precursor to fabricate ultrathin S-doped graphene films, which were found to be applicable to ultrahigh-rate microsupercapacitors. 46 Intrigued by the beautiful structure of contorted hexa-catahexabenzocoronene (29) (Figure 7a) 47 as well as its partially thiophene-substituted analogues such as dibenzotetrathienocoronenes 30 (Figure 7b) reported by Nuckolls et al., 48 we synthesized two fully thiophene-fused coronenes, that is, hexathienocoronenes (HTCs) 31 and 32a−c (Figure 7c,d). 49 HTCs 31 and 32a−c exhibited almost identical absorption features and a highly electron-rich nature with higher HOMO levels compared with 30. Single-crystal X-ray analysis revealed a columnar π-stacking structure of HTC 32b. A discotic liquidcrystalline phase was observed for HTC 32c with longer dodecyl chains. Together with Chen et al., we employed the backbone of HTC 32a to form a gemini-type amphiphile by attaching hydrophobic and hydrophilic side chains on opposite sides of the aromatic core. 50 The obtained amphiphilic molecule 32d self-assembled into nanofibers in solution. After chemical oxidation with nitrosonium tetrafluoroborate (NOBF 4 ), the nanofibers became electroconductive with a resistivity of 1.1 MΩ, which is lower than the value for the amphiphilic HBC nanotubes (2.5 MΩ). 51

HETEROATOM DOPING ON THE ZIGZAG EDGES
The edge topologies of nanographene molecules, such as armchair and zigzag structures (Figure 8), are essential to their chemical and physical properties. 12 The majority of bottom-upsynthesized nanographene molecules have armchair edges, as represented by HBC and its derivatives. Although relatively rare, zigzag-edged nanographene molecules are particularly appealing, as they often have low HOMO−LUMO gaps and an openshell ground state with long zigzag peripheries. 52 Despite the appealing characteristics of zigzag-edged nanographene molecules, their syntheses and characterizations are largely hampered by their instability under ambient conditions. On the one hand, replacement of the all-carbon zigzag edges with isostructural heteroatoms provides new access to stable nanographene molecules featuring zigzag architectures. On the other hand, heteroatom-doped zigzag edges are often theoretically considered as active sites in electrochemical catalysts for the oxygen reduction and hydrogen evolution reactions. 53,54 The synthesis of nanographene molecules having such edge structures can thus provide atomically precise models for investigating the electrocatalytic mechanism.
Inspired by the high performance of N-doped graphene-based materials for electrocatalysis, we synthesized dibenzo-9aazaphenalene 35 as a model compound to study the catalytic activity of N-doped zigzag edges. Although we could not demonstrate a reasonable activity of 35 in electrocatalysis, we further achieved the synthesis of the linearly extended "dimer" 38 ( Figure 9), indicating the potential of further extension toward the corresponding GNRs. 55 The PAMY unit in 35 was highly reactive, stemming from its diradical and zwitterionic resonance structures (Figure 4d). Therefore, compounds 35 and 38 were generated only under inert conditions by deprotonation of their charged precursors 34 and 37. In addition, because of the extremely high chemical reactivity of the N-doped zigzag edges, the dibenzo-9a-azaphenalene-based PAMY offered a new entry to internally N-doped nanographene molecules through 1,3-dipolar cycloaddition with dipolarophiles. 56 This chemical reactivity was also independently exploited by Ito, Tokimaru, and Nozaki 57 and used for syntheses of azapentabenzocorannulene derivatives. 58,59 We have also considered the incorporation of multiple heteroatoms to obtain stable heteroatom-doped zigzag edges and deeper insights into the effects of doping on the chemical and physical properties. Together with Zhang, Feng, and coworkers, we conceived nitrogen−boron−nitrogen (NBN)doped zigzag peripheries based on 1,9-diaza-9a-boraphenalene as the next structural motif. 60 Both monomers 40 and the linearly fused "dimer" 45 were synthesized through electrophilic borylation ( Figure 10). The NBN-doped zigzag edges exhibited high chemical stability, allowing for electrophilic bromination of 40b and further modifications through cross-coupling reactions to give 42 (Figure 10a). Furthermore, chemical oxidation of monomers 40 led to σ dimers 43 connected at the positions para to the N atoms. To avoid such a dimerization reaction during oxidation, the active positions were protected by additional phenyl substituents (i.e., compound 42). Clean single-electron oxidation of the NBN-doped zigzag peripheries to afford radical cation 42 •+ was demonstrated by in situ spectroelectrochemis-

Accounts of Chemical Research
Article try, shedding light on the photophysical perperties of its isoelectronic dibenzophenalenyl radical.
In parallel to the studies on heteroatom-doped nanographene molecules, we also tried to tackle the syntheses of peri-acenes (e.g., peri-tetracene 61,62 and peri-pentacene 63 ), which comprise two acenes laterally fused at all of the peri positions and can be viewed as segments of full zigzag GNRs (Figure 11). It was thus a logical extension to consider the synthesis of heteroatom-doped

Accounts of Chemical Research
Article peri-acenes as the next targets. Our initial attempts to make NBN-doped peri-tetracene turned out to be challenging, and we are still working on different routes to synthesize it, but we were successful in installing oxygen−boron−oxygen (OBO) segments on the zigzag peripheries of peri-acenes. More specifically, we developed a tandem demethylation−borylation method to construct OBO-fused tetrabenzo[a,f,j,o]perylenes 48 by simple heating of precursors 47 with BBr 3 in o-dichlorobenzene ( Figure  12a). 64 Further cyclodehydrogenation of 48 provided the first heteroatom-doped peri-tetracenes 49 in 2016, before the allcarbon peri-tetracenes were achieved in 2018. 65,66 The OBOdoped peri-tetracene exhibited excellent ambient stability, blue fluorescence, and a large HOMO−LUMO gap with a low-lying HOMO level compared with pristine peri-tetracene. The incorporation of OBO units drastically changed the properties of peri-tetracene, and therefore, isoelectronic heteroatom doping would be important to maintain the open-shell character of peri-tetracene while improving its stability for potential applications. It is noteworthy that Hatakeyama et al. independently reported compound 48a obtaind by a different method (Figure 12b), in which a precursor containing two extra iodo groups was used to generate 48a in 55% yield. 67 Our method gives higher yields (>90%) and is simpler since it is based on direct C−H borylation without the need for iodo groups in the precursor. It must be noted that Hatakeyama et al. also independently reported the use of the direct C−H borylation method to synthesize heteroatom-doped benzo[fg]-

Accounts of Chemical Research
Article tetracene structures with OBO, NBN, and SBS substitution on the zigzag peripheries. 68 In our pursuit of longer peri-acene-type nanographene molecules, we synthesized OBO-doped peri-hexacene as the longest peri-acene analogue. 69 We first tried to furnish the OBOfused precursor 54 by our tandem demethylation−borylation method from 1,2,4,5-tetrakis(3-methoxynaphthalen-2-yl)benzene (52) (Figure 13a), but the reaction was not successful, probably because of the high reactivity of naphthalene αpositions during the electrophilic borylation. Therefore, we adopted Hatakeyama's method by introducing iodo groups on the central benzene to synthesize 1,4-diiodo-2,3,5,6-tetrakis(3methoxynaphthalen-2-yl)benzene (53). After lithiation and reaction with BBr 3 , the OBO-fused precursor 54 was obtained in 43% yield (Figure 13a). However, oxidative cyclodehydrogenation of 54 in solution did not provide the planarized product after attempts under different conditions. We thus employed onsurface synthesis in collaboration with Fasel et al. and achieved the OBO-doped perihexacene 55 on Au(111) by heating precursor 54 at 380°C under UHV (Figure 13a). The structure of 55 was unambiguously characterized by STM and nc-AFM (Figure 13b,c). Furthermore, OBO-doped peri-hexacene 55 formed two kinds of 1D superstructures via intermolecular hydrogen bonding and metal coordination with the OBO segments on the zigzag edges (Figure 13b,d), demonstrating a potential edge-doping strategy for engineering self-assembled structures of graphene nanoarchitectures.
The OBO-fused precursor 54 featured a double [5]helicene substructure, which underwent a conformational change upon interactions with the surface. 69 On the other hand, in the crystal structure of double [5]helicene 48a, we observed a pair of (P,P) and (M,M) enantiomers in one unit cell. 64 Hatakeyama et al. separated the enantiomers of such a double [5]helicene through the incorporation of bulky tert-butyl groups (compound 51) to increase the isomerization energy barrier. 67 The interesting properties of double helicenes inspired us to further push the limit of double helicene synthesis and to explore their chiroptical properties. By adjusting the connectivity of the naphthalene rings, we synthesized the first double [7]helicene 57 ( Figure  14a), 70 which is an isomer of the π-extended OBO-fused double [5]helicene 54. Compound 57 was obtained in 7% yield through a tandem demethylation−borylation reaction of 1,2,4,5-tetrakis-(2-methoxynaphthalen-1-yl)benzene (56). The isolated yield is apparently lower than those of compounds 48 (over 90%),

Accounts of Chemical Research
Article presumably because of the high strain built up during the ring closure. Single-crystal X-ray diffraction revealed substantial π overlap between the terminal benzene rings (Figure 14b). Compound 57 exhibited excellent chiral stability, which facilitated the optical resolution of the enantiomers by chiral HPLC. The separated (P,P) and (M,M) isomers displayed opposite responses in circular dichroism spectra (Figure 14c). The successful synthesis of the first double [7]helicene with OBO incorporation demonstrated the efficacy of the tandem demethylation−borylation method in the synthesis of even strained architectures, stimulating more efforts toward longer double helicenes and opening up a new avenue to chiral synthetic nanographenes.
To examine the scope of the tandem demethylation− borylation reaction, we tested the direct C−H borylation reactivity toward the chemically inert 2-and 7-positions of pyrene ( Figure 15). 71 We found that the tandem reaction readily occurred under typical conditions, generating a new kind of oxaborin-annelated pyrene-based chromophores (60 and 63) in excellent yields (>95%). The singly annelated compound 60 displayed strong blue fluorescence (λ max = 442 nm, Φ F = 70%), whereas the doubly annelated product 63b resembled perylene diimide (PDI) dyes in terms of bright green fluorescence (λ max = 525 nm, Φ F = 84%) and strong aggregation tendency in solution. The tandem demethylation−borylation reaction thus provided a new method for π extension of pyrene to obtain novel photonic and optoelectronic materials.

TOWARD HETEROATOM-DOPED GRAPHENE NANORIBBONS
The studies of the nanographene molecules have paved the way to heteroatom-doped GNRs by offering model cases. For example, when we established the surface-assisted synthesis of atomically precise GNRs together with Fasel et al. in 2010, we fabricated chevron-type GNR 65 from 6,11-dibromo-1,2,3,4tetraphenyltriphenylene precursor 64 (Figure 16a,b). 72 The chevron-type GNR comprises HBC units that are fused side by side. Therefore, considering the many heteroatom-doped HBC analogues as described previously, one could easily imagine synthesizing heteroatom-doped chevron-type GNRs by modifying the triphenylene-based molecular precursors. Among the various possibilities, N doping is particularly interesting, as it can

Accounts of Chemical Research
Article effectively lower the valence and conduction bands to afford ntype semiconductors. In 2013, Bronner, Hecht, Tegeder, and coworkers synthesized N-doped chevron-type GNRs 67 by replacing one or two phenyl rings with 4-pyridinyl groups in the monomers (Figure 16c). 73 Compared with the pristine chevron-type GNR 65, the introduction of N atoms simultaneously lowered the valence and conduction bands, leaving the band gaps nearly unchanged. In 2014, together with Du, Gao, and co-workers, our group reported N-doped GNR 67b fabricated on Au(111) under UHV and investigated its structure by high-resolution STM. 74 Different from its allcarbon counterpart 65, the N-doped GNR 67b exhibited a sideby-side alignment due to the inter-ribbon N···H interactions (Figure 16d). In the same year, together with Fasel et al., we reported the on-surface synthesis of GNR heterojunction 69 comprising alternating pristine and N-doped GNR segments (Figure 16e−g). 75 The incorporation of N atoms was accomplished by replacing two phenyl groups with 5-pyrimidinyl rings in the triphenylene-based precursor to give monomer 68. The resulting GNR heterojunction 69 displayed a band offset of about 0.5 eV, representing the first GNR-based p− n junction and indicating great potential for future applications in nanoelectronics. N-doping of GNRs in the form of pyridine and pyrimidine rings lowers the energies of the valence and conduction bands, whereas the band gap is nearly unaffected because the lone pair of N atoms is not conjugated with the GNR π system. In this regard, S doping in the form of thiophene incorporation is supposed to involve the lone pair in the π conjugation of GNRs and thus can effectively modulate the electronic structures and band gaps. Recently, together with Du, Gao, and co-workers, our group described the on-surface synthesis of S-doped GNR 71 by employing 6,11-dibromo-1,4-diphenyl-2,3-bis(thien-3-yl)triphenylene (70) as the monomer (Figure 17a), that is, by replacing two phenyl rings with 3-thienyl groups in the typical triphenylene-based monomer. 76 Different from the N-doped chevron-type GNRs, the rotation of the single bonds connecting thienyl rings and the triphenylene core led to different isomers when the precursor was deposited on the Au(111) surface (Figure 17b). After polymerization and cyclodehydrogenation at 450°C, different S-doped GNR segments were obtained (Figure 17c−e), giving rise to a sequence of tunable band gaps (1.28−1.78 eV), as confirmed by experimental STM and scanning tunneling spectroscopy (STS) as well as theoretical calculations. Through molecular engineering of the triphenylene-based monomers, Fischer et al. further achieved heteroatom-doped chevron-type GNRs with trigonal-planar S-, N-, and O-dopant atoms. 77 As we had developed heteroatom-doped zigzag-edged nanographene molecules, we subsequently aimed at synthesizing GNRs with heteroatom-doped zigzag edge topologies. During the past years, although armchair GNRs have been extensively investigated, 79 zigzag GNRs have very rarely been achieved ( Figure 18). 13 Chiral GNRs with a mixture of armchair

Accounts of Chemical Research
Article and zigzag edges have also attracted great attention, but the bottom-up-synthesized structures have been limited only to (3,1)-GNRs (for the definition of the chiral index, see Figure 18) obtained from 9,9′-bianthracene-based monomers, 80 calling for further variations of the chiral edge configurations.

CONCLUSIONS AND OUTLOOK
We have summarized our works on heteroatom-doped nanographene molecules and GNRs during the past decade. On the basis of the prototypical nanographene molecule HBC, we have explored N doping through the incorporation of aromatic pyrrole and antiaromatic pyrazine rings. We have also developed a series of thiophene-annelated HBC derivatives as S-doped nanographene molecules. These results have promoted heteroatom doping as a powerful method for modulating the energy levels, band gaps, aromaticity, electron densities, and magnetic properties of nanographene molecules. Moreover, being aware of the importance of zigzag-edged nanographene molecules, we have explored heteroatom doping on zigzag edge structures, including generation of the PAMY as well as NBN and OBO incorporation. In particular, the OBO-doped compounds are easy to synthesize and are stable under ambient conditions, thereby representing a new class of nanographene molecules, including stable peri-tetracene and peri-hexacene analogues displaying intermolecular self-assembly tendency on the surface and double [7]helicene with high chiral stability as well as new pyrene-based chromophores. All of these achievements in nanographene molecules have paved the way to heteroatom-doped GNRs, such as N-doped and S-doped chevron-type GNRs with tunable band gaps and OBO-doped chiral (4,1)-GNRs with lateral alignment. Nanocarbon science has attracted increasing interest in recent years, giving rise to a variety of carbon nanostructures. The incorporation of heteroatoms into the carbon frameworks provides a number of new opportunities to fine-tune the molecular properties. Certainly this process defines grand challenges in synthesis, thereby requesting the development of new and efficient synthetic methods. Convincing cases have been demonstrated in the PAMY chemistry and the tandem demethylation−borylation reaction, both of which have opened up new avenues to various heteroatom-doped nanographenes. Furthermore, the synthesis of heteroatom-doped GNRs is lagging behind that of nanographene molecules. There is much room to explore new structures such as pyrrole-and pyrazineincorporated GNRs and heteroatom-doped full zigzag GNRs. In any case, close collaborations among chemists, physicists, and engineers are required to realize the applications of GNRs in future nanoelectronic and spintronic devices.

■ ACKNOWLEDGMENTS
We thank all of our distinguished collaborators and group members who enabled the achievements described in this Account. Financial support from the Max Planck Society, the Alexander von Humboldt Foundation, and the China Scholarship Council is gratefully acknowledged.