*3.1. Sequential Growth Based on the Ullmann Coupling/Aromatization Combination*

In all the aforementioned reaction combinations, 2D covalent networks have been obtained by sequential or hierarchical growth of nanostructures. However, a series of combinations has also been specifically developed for the design of 1D macromolecular organic structures. At present, it has to be noticed that these combinations have not been used for the design of 2D structures yet, even if from a technical point of view, these combinations could be easily transposed to 2D structures. In this field, graphene, by its unique properties, has driven a great deal of interest and numerous works have been devoted to design graphene-like structures. To produce regular and extended structures, the Ullmann coupling followed by an aromatization reaction is a promising approach (see Figure 18).

**Figure 18.** (**A**) Reaction mechanism and STM images of the formation of nanoribbons with 10,10 -dibromo-9,9 -bianthracene. (**B**) Reaction mechanism and STM images of the formation of zigzag nanoribbons with 6,11-dibromo-1,2,3,4-tetraphenyltriphenylene. (**C**) STM images of the copolymerization reaction. All experiments have been carried out on Au(111) surfaces. Reprinted by permission from Macmillan Publishers Ltd: Nature from [101], copyright 2010.

In this field, the first on-surface synthesis was reported in 2010 [101]. Graphene nanoribbons (GNRs) were obtained by first depositing 10,10 -dibromo-9,9 -bianthracene on Au(111) surfaces at 200 ◦C, enabling to interlink the biradicals resulting from the dehalogenation reaction, the planarization of the structure and the formation of a linear polymer. At 400 ◦C, a cyclodehydrogenation reaction converted the polymer to a fully conjugated and linear structure (see Figure 18A). Versatility of the approach was demonstrated by the design of several types of nanoribbons, such as chevron-type nanoribbons resulting from the polymerization and aromatization reaction of 6,11-dibromo-1,2,3,4-tetraphenyltriphenylene (see Figure 18B). In this last case, steric hindrance generated by the tetraphenylene groups enforces these bulky substituents to stand on each side of the polymer main axis and a zigzag structure was obtained. The possibility of codepositing two differing precursors (6,11-dibromo-1,2,3,4-tetraphenyltriphenylene and 1,3,5-*tris*(4"-iodo-2 -biphenyl)benzene) was also examined, demonstrating the possibility to design chemically modified graphene-like structures (see Figure 18C). Recently, the different experiments carried out by Cai et al. on 10,10 - dibromo-9,9 -bianthracene and 10,10 -dibromo-9,9 -bianthracene were reproduced using another deposition process for the first step. In this last case, the direct contact transfer (DCT) consisting in using a stamp with the corresponding molecule at its surface was used [102]. Using this strategy, the precursor could be deposited on the surface without taking recourse to sublimation. This finding constitutes a major advancement, considering that the molecular weight often constitutes a drastic limitation. Indeed, increase of the molecular size results in higher sublimation temperatures that can degrade the molecule. Over the years, a wide range of molecular tectons giving access to GNRs have been reported in the literature and these different structures are depicted in the Figure 19 [103–117].

**Figure 19.** Chemical structures of precursors used to elaborate graphene nanoribbons (GNRs) and the corresponding reference in brackets.

The beauty of these approaches relies in the fact that nitrogen-doped GNRs [111,116], ultranarrow GNRs [103,118], chevron-type GNRs [102], or zigzag GNRs [106,117,119] with chemically modified edges could be elaborated with these precursors. Among the most interesting findings and by properly choosing the molecular precursors, GNRs could notably be used as elemental building blocks to trigger the design of more extended structures. This goal was achieved by substituting the precursors with nitrogen atoms [112]. Ordered nanomaterials could be obtained by an atomically precise bottom-up synthesis based on the creation of hydrogen bonds and van der Waals interactions between GNRs (see Figure 20A). Growth of structures in the third dimension could be even obtained by π–π stacking of GNRs sheets. Modification of the electronic structures of GNRs is another long-standing challenge and, in this field, a remarkable example has been reported in 2016 [115].

**Figure 20.** (**Aa**–**f**) Bottom-up synthesis of GNRs containing nitrogen atoms with the possibility to create lateral interactions between GNRs (a: scale bar: 10 nm, b: scale bar: 2 nm, c: scale bar: 10 nm, d: scale bar: 2 nm). Adapted with permission from Vo et al. [112]. Copyright 2015 American Chemical Society. (**B**) Schematic representation of GNRs comprising both carbazole and phenanthridine units. (**Ba**–**c**) High-resolution AFM images of GNRs with the demonstration of the presence of phenanthridine (**Bb**) and carbazole (**Bc**) units in GNRs. Reproduced from [113] with permission from The Royal Society of Chemistry.

By a thermally activated ring expansion/dehydrogenation reaction, the electron-rich carbazole could be converted into the electron-deficient phenanthridine [120]. In this work, authors succeeded to partially convert the carbazole groups into phenanthridine, opening the way towards the fine tuning of the materials bandgaps by the presence of electron-rich and electron-poor groups onto the same structures (see Figure 20B).

Choice of the surface as well as a careful selection of the shape of the molecular tectons (planar or nonplanar) can impact the geometry of the final GNRs and greatly help in designing GNRs with precise structures [121]. These opportunities were demonstrated with an extensive study carried out on two molecular building blocks, i.e., 10,10 -dibromo-9,9 -bianthracene (DBBA). The debate concerning the molecular arrangement of DBBA on Cu(111) is a long-standing issue [105,107,108,122,123] and by using noncontact atomic force microscopy (nc-AFM), the controversy could be definitely solved. In the specific case of DBBA and despites the presence of bromine atoms, the Ullmann reaction proved to be ineffective on Cu(111) substrates to couple the molecular tectons and GNRs resulting from another coupling mode could be detected on the surface. Interestingly, the same outcome than the one obtained with DBBA could be produced with the unsubstituted 9,9 -bianthracene (BA) or 10,10 -dichloro-9,9 -bianthracene on Cu(111) substrates (see Figure 21Aa–Ad).

**Figure 21.** (**A**) GNRs obtained with different bianthracene (BA) precursors. (**B**) Connection of bianthracene precursors at the C2 and C2' positions (scale bar: 1 nm for all noncontact AFM images). Adapted with permission from Schulz et al. [121]. Copyright 2017 American Chemical Society.

Conversely, upon deposition of DBBA on Ag(111) or Au(111), the conventional Ullmann coupling reaction occurred, providing GNRs with armchair edges. By combining STM and nc-AFM at low temperature, the mechanism could be elucidated. Authors demonstrated that upon cleavage of the

carbon–halogen bond, the biradicals of DBBA were stabilized by the strong interactions existing with the Cu(111) surface and/or adatoms, so that a severe reduction of the energy barrier for the carbon–hydrogen bond scission at the C2 and C2' positions could be obtained. Parallel to this, the twisted structures of the different BA derivatives enforced a specific arrangement of the molecular tectons on the surface, favorable to intermolecular interactions at the C2/C2' positions (see Figure 21(Ba,Bb). As a result of this, and irrespective of the substitution pattern of BA, the same GNRs could be obtained with all monomers, the homocoupling being governed by the C–H scission and not by the recombination of radicals.
