This is the submitted version of the article: Panighel M., Quiroga S., Brandimarte P., Moreno C., Garcia-Lekue A., Vilas-Varela M., Rey D., Sauthier G., Ceballos G., Peña D., Mugarza A.. Stabilizing Edge Fluorination in Graphene Nanoribbons. ACS nano, (2020). 14. : 11120 - . 10.1021/acsnano.0c01837. Available at: https://dx.doi.org/10.1021/acsnano.0c01837 ACS Nano This document is confidential and is proprietary to the American Chemical Society and its authors. Do not copy or disclose without written permission. If you have received this item in error, notify the sender and delete all copies. Stabilizing edge-fluorination in graphene nanoribbons Journal: ACS Nano Manuscript ID nn-2020-018376 Manuscript Type: Article Date Submitted by the 02-Mar-2020 Author: Complete List of Authors: Panighel, Mirco; Institut Catala de Nanociencia i Nanotecnologia Quiroga, Sabela; CIQUS Brandimarte, Pedro; Donostia International Physics Center Moreno, Cesar; Institut Català de Nanociència i Nanotecnologia, Garcia-Lekue, Aran; Donostia International Physics Center, Vilas-Varela, Manuel; Universidade de Santiago de Compostela, Organic Chemistry Rey, Dulce; CIQUS Sauthier, Guillaume; Catalan Institute of Nanoscience and Nanotechnology Ceballos, Gustavo; Institut Catala de Nanotecnologia, Peña, Diego; Universidade de Santiago de Compostela, CIQUS Mugarza, Aitor; Institut Catalá de Nanotecnologia, ACS Paragon Plus Environment Page 1 of 28 ACS Nano 1 2 3 4 5 6 7 8 Stabilizing edge-fluorination in graphene 9 10 11 12 nanoribbons 13 14 15 Mirco Panighel,∗,†,⊥ Sabela Quiroga,‡ Pedro Brandimarte,¶ Cesar Moreno,† 16 17 18 Aran Garcia-Lekue,¶,§ Manuel Vilas-Varela,‡ Dulce Rey,‡ Guillaume Sauthier,† 19 20 Gustavo Ceballos,† Diego Peña,∗,‡ and Aitor Mugarza∗,†,k 21 22 23 †Catalan Institute of Nanoscience and Nanotechnology (ICN2), CSIC and The Barcelona 24 25 Institute of Science and Technology, Campus UAB, Bellaterra, 08193 Barcelona, Spain. 26 27 ‡Centro de Investigación en Química Biolóxica e Materiais Moleculares (CiQUS) and 28 29 Departamento de Química Orgánica, Universidade de Santiago de Compostela. Santiago de 30 31 Compostela 15782, Spain. 32 33 ¶Donostia International Physics Center, Paseo M. de Lardizabal 4, 20018 San Sebastian, 34 35 Spain. 36 37 §Ikerbasque, Basque Foundation for Science, 48013 Bilbao, Spain. 38 39 kICREA Institució Catalana de Recerca i Estudis Avançats, Lluis Companys 23, 08010 40 41 Barcelona, Spain. 42 43 ⊥Current address: CNR-IOM Laboratorio TASC, SS 14 km 163.5 - Basovizza, I-34149 44 45 Trieste, Italy. 46 47 48 E-mail: panighel@iom.cnr.it; diego.pena@usc.es; aitor.mugarza@icn2.cat 49 50 51 Abstract 52 53 54 The on-surface synthesis of edge-functionalized graphene nanoribbons (GNRs) is 55 56 challenged by the stability of the functional groups throughout the thermal reaction 57 58 59 1 60 ACS Paragon Plus Environment ACS Nano Page 2 of 28 1 2 3 steps of the synthetic pathway. Edge-fluorination is a particularly critical case, where 4 5 the interaction with the catalytic substrate and intermediate products can induce the 6 7 complete cleavage of the otherwise strong C-F bonds before the formation of the GNR. 8 9 Here, we demonstrate how a rational design of the precursor can stabilize the functional 10 11 group, enabling for the first time the synthesis of edge-fluorinated GNRs. The survival 12 13 of the functionalization is demonstrated by tracking the structural and chemical trans- 14 15 formations occurring at each reaction step with complementary X-ray photoelectron 16 spectroscopy and scanning tunneling microscopy measurements. In contrast to pre- 17 18 vious attempts, we find that the C-F bond survives the cyclodehydrogenation of the 19 20 intermediate polymers, leaving a thermal window where GNRs withhold more than 21 22 80% of the fluorine atoms. We attribute this enhanced stability of the C-F bond to the 23 24 particular structure of our precursor, which prevents the cleavage of the C-F bond by 25 26 avoiding interaction with the residual hydrogen originated in the cyclodehydrogenation. 27 28 This structural protection of the linking bond could be implemented in the synthesis 29 30 of other sp2 -functionalized GNRs. 31 32 33 34 Abbreviations 35 36 37 CB, CDH, DFT, DOS, GNR, PDOS, RT, STM, VB, XPS 38 39 40 41 42 Keywords 43 44 45 Graphene nanoribbons, on-surface synthesis, edge-functionalization, fluorination, scanning 46 47 tunneling microscopy, density functional theory, self-assembly 48 49 50 51 Introduction 52 53 54 The functionalization of graphene nanoribbons (GNRs) is an effective method for tailoring 55 56 their structural, chemical and electronic properties. 1–9 Edge functionalization by replacing 57 58 59 2 60 ACS Paragon Plus Environment Page 3 of 28 ACS Nano 1 2 3 hydrogen with different substituents is particularly interesting since it offers a wide flexibil- 4 5 ity in the choice of functional groups that can, in addition to band tuning, 1 bring chemical 6 7 selectivity for sensing. 8 Their peripheral position can also be employed to tune inter-ribbon 8 9 and substrate interactions and drive their organization, one of the most important challenges 10 11 in bottom-up synthetic methods. 7,10,11 The atomic precision offered by bottom-up synthetic 12 13 methods in the positioning of each component makes this approach the natural candidate for 14 15 a controlled functionalization of the ribbons. The thermally activated sequential on-surface 16 17 synthesis has shown to be particularly effective for the realization of atomically precise 18 19 GNRs. 12–14 However, such multi-step methods also face a major challenge, namely the sta- 20 21 bility of the functional groups that have to overcome the whole reaction pathway that leads 22 23 to the GNR synthesis. This is critical in the functionalization of edges, where the linking 24 25 groups are not as strongly bound to the aromatic backbone as hetero-atoms in substitutional 26 27 doping. In the absence of stabilizing intra- or inter-ribbon reactions, 3,6,7 the thermal energy 28 29 supplied during the synthesis can surpass the linking bond energy, leading to a partial 4 or a 30 31 total 15 dissociation of the functional groups. Dissociation can occur even in carefully chosen, 32 33 strongly bonded groups, since the intermediate products formed during the reaction pathway 34 35 can destabilize the bond between the carbon backbone and the functional group, and induce 36 37 a premature cleavage. This was recently shown in an attempt to synthesize edge-fluorinated 38 39 GNRs, where it was found that the activation barrier of the otherwise strong C-F bond was 40 41 reduced by one order of magnitude, mainly by the formation of an additional C(sp3 )-H bond 42 43 with residual H migrating from nearest neighbor C atoms during the cyclodehydrogenation 44 45 (CDH) step. 15 The understanding of this C-F cleaving mechanism can be used for the ra- 46 47 tional design of fluorinated precursors that avoid the formation of undesired, destabilizing 48 49 transient structures, as explicitly stated by the authors. Here, we demonstrate how a suit- 50 51 able design of the molecular precursor can lead to the successful synthesis of edge-fluorinated 52 53 GNRs. These are synthesized on a Au(111) single crystal in a step-wise thermal process, 54 55 initially via the surface-assisted Ullmann coupling of the fluorinated monomers into poly- 56 57 58 59 3 60 ACS Paragon Plus Environment ACS Nano Page 4 of 28 1 2 3 meric chains and then through the subsequent CDH leading to the formation of the graphene 4 5 nanoribbons. 16 The functionalized precursors have been specifically synthesized to overcome 6 7 these thermally activated reaction steps. The key element in the design is the separation 8 9 between the fluorinated carbon and those involved in the CDH reaction, which minimizes 10 11 the defluorination rate by introducing new paths for residual H release onto the surface. 12 13 Regarding the electronic properties of the functionalized GNRs, our ab-initio calculations 14 15 reveal a significant n-type doping effect of fluorination, shifting frontier bands downwards 16 17 by about 200 meV and leaving the size of the gap unchanged. The doping effect, however, 18 19 can be fully screened out by the interaction with the metallic substrate, as demonstrated by 20 21 both ab-initio and scanning tunneling spectroscopy results. 22 23 24 25 26 Results and discussion 27 28 29 First, we designed a GNR precursor core which could allow us the easy introduction of differ- 30 31 ent functional groups in peripheral positions, with the goal to minimizing the possible group 32 33 cleavage during the CDH step. With this idea in mind, we selected the benzo[b]triphenylene 34 35 core which is presented in compounds 1a,b (Figure 1a), including two Br atoms to induce 36 37 the on-surface Ullmann coupling to form polymers 2a,b and two phenyl groups to facilitate 38 39 the CDH step to obtain GNRs 3a,b. These GNRs resemble the previously reported chevron 40 41 GNR, 16 although 3a,b are narrower and characterized by the combination of cove and triple- 42 43 bay regions along the edges. Besides the unsubstituted GNR 3a (R = H) we planned to 44 45 obtain the fluorinated GNR 3b (R = F). Remarkably, these GNRs were designed with the 46 47 functional groups attached to remote positions with respect to the carbon atoms involved in 48 49 the CDH step, in order to minimize the potential weakening and subsequent cleavage of these 50 51 groups. Precursors 1a,b were easily obtained by solution chemistry, in three steps from com- 52 53 mercially available reagents (Figure 1b). First, bromination of 9,10-phenanthroquinone (4) 54 55 followed by double Knoevenagel condensation with 1,3-diphenylacetone afforded dienone 5 in 56 57 58 59 4 60 ACS Paragon Plus Environment Page 5 of 28 ACS Nano 1 2 3 good yield. Then, Diels-Alder reaction of 5 with arynes 6a,b which were in-situ generated by 4 5 fluoride-induced decomposition of triflates 7a,b led to the formation of benzo[b]triphenylenes 6 7 1a,b after CO extrusion of the corresponding adducts. 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 Figure 1: (a) On-surface synthesis of GNRs 3a,b (b) In-solution synthesis of 34 benzo[b]triphenylene derivatives 1a,b used as GNR precursors. 35 36 37 38 With compounds 1a,b in hands, we launched the on-surface studies to attempt the prepa- 39 40 ration of GNRs 3a,b on Au(111). The formation of the expected GNRs and the stability of 41 42 the functional groups throughout the on-surface synthetic protocol have been investigated by 43 44 complementary scanning tunneling microscopy (STM) and X-ray photoelectron spectroscopy 45 (XPS) measurements. Exploiting the chemical sensitivity of XPS, the reactions resulting in 46 47 the synthesis of the GNRs have been systematically characterized, as summarized in Figure 48 49 2. After depositing compounds 1a,b at room temperature (RT) and annealing at 200 °C, 50 51 these GNR precursors undergo a C-Br homolytic bond cleavage, leading to the formation of 52 53 aryl radicals and the subsequent coupling into polymeric chains 2a,b via the Ullmann cou- 54 55 pling. Figure 2a shows the XPS spectra of the Br 3d core levels for precursor 1b (R = F). 56 57 58 59 5 60 ACS Paragon Plus Environment ACS Nano Page 6 of 28 1 2 3 4 5 6 7 (a) Br 3d RT (d) 8 Br[C] 200 °C Intensity (a.u.) 380 °C F F 9 F F Br[Au] 10 Br 11 F F F F F F 12 Br 13 F F 14 74 72 70 68 66 F F 15 Binding energy (eV) Ullmann CDH 16 (b) C 1s C[C2H] Br 3d Br[C] Br[Au] 17 C[C3] 1.0 18 19 C[C2F] 0.5 20 Intensity (a.u.) 21 200 °C 0.0 22 Normalized XPS peak area 23 1.0 C 1s 24 25 C[C2H] 26 27 28 0.5 C[C3] 29 30 288 286 284 282 31 Binding energy (eV) C[C2F] 32 (c) F 1s RT 33 310 °C 0.0 F 1s Intensity (a.u.) 34 335 °C C[C2F] 380 °C 1.0 35 405 °C F[C] 36 0.5 37 38 39 0.0 690 688 686 684 40 0 100 200 300 400 Binding energy (eV) Temperature (°C) 41 42 Figure 2: (a)-(c) XPS core level spectra of Br 3d (a), C 1s (b), and F 1s (c) after annealing 43 44 precursor 1b at different temperatures. The C 1s spectra fit (continuous lines) is done with 45 multiple peaks to account C[C3 ], C[C2 H] and C[C2 F] contributions. (d) Thermal evolution of 46 the XPS atomic ratios, normalized against gold, of the different peaks in (a)-(c). The vertical 47 lines indicate the Ullmann and CDH reaction steps of the GNR synthesis, as indicated by 48 the respective changes in Br[C]/Br[Au] and C[C3 ]/C[C2 H] ratios. In the C 1s series, the 49 50 open circles at the left and right limits represent the stoichiometric relation of the different 51 components for the precursors and the nanoribbons respectively. A normalized C[C2 F] curve 52 has been added to the F[C] data of F 1s. 53 54 55 56 57 58 59 6 60 ACS Paragon Plus Environment Page 7 of 28 ACS Nano 1 2 3 The binding energy of the spin-orbit split doublet measured for precursors deposited at RT, 4 5 69.9/70.9 eV, corresponds to the C-bonded Br. The doublet downshifts by 2.1 eV upon 6 7 annealing at 200 °C, which can be assigned to C-Br cleavage and consequent formation of 8 9 Au-bonded Br adsorbates. 17–19 The completion of the C-Br cleavage is in agreement with 10 11 the Ullmann polymerization observed by STM at the same temperature (see Figure 3). At 12 13 higher temperature Br adatoms desorb from the surface, as indicated by the absence of any 14 15 Br 3d peak in the spectrum obtained after annealing at 380 °C. This is clearly visible also 16 17 in Figure 2d which reports the thermal evolution of the XPS atomic ratios of the different 18 19 peaks normalized against gold. The CDH reaction that convert polymers 2 into GNRs 3 20 21 can be tracked by following the temperature evolution of the C 1s core level components. 22 23 As representative examples, Figure 2b shows the C 1s core level XPS spectra after annealing 24 25 1b at 200 °C and 335 °C, which illustrate the evolution of C bond ratios that correspond to 26 27 the CDH step. The main C peak is best fitted with two components that can be assigned to 28 29 C[C3 ] (284.3 eV) and C[C2 H] (284.0 eV) bonds respectively. The minor C-Br component is 30 31 only present at RT according to the Br 3d peak evolution, and therefore is not considered in 32 33 this analysis. The additional peak present at higher binding energies accounts for the C[C2 F] 34 35 (286.5 eV) bond component. As we increase the annealing temperature, a clear inversion of 36 37 the C[C3 ] and C[C2 H] components is visible around 300 °C. This is evident in Figure 2d which 38 39 reports the thermal evolution of the XPS peak area related to the different components. By 40 41 comparing to the stoichiometric references of the polymeric and GNR phases indicated at the 42 43 two sides of the graph by open circles, this transition of the C[C3 ]/C[C2 H] ratio corresponds 44 45 to the cyclodehydrogenation that leads to the formation of GNRs at this temperature, as 46 47 also confirmed by the STM measurements (see Figure 3). The relevant features regarding the 48 49 stability of the functional group are the C[C2 F] component of C 1s and the correlated F 1s 50 51 peaks. The evolution of the latter is reported in Figure 2c for the whole set of temperatures 52 53 investigated. Interestingly, the C[C2 F] bond survives well above the CDH threshold. As can 54 55 be seen in detail in Figure 2d, the intensity of the C[C2 F] components of C 1s and the F[C] 56 57 58 59 7 60 ACS Paragon Plus Environment ACS Nano Page 8 of 28 1 2 3 of F 1s remains unaltered during the formation of GNRs at 310 °C, and still withhold 80% 4 5 of the intensity at 335 °C when the CDH is complete. Their gradual decrease down to about 6 7 ~25% at the maximum studied temperature of 405 °C is accompanied by an increase of the 8 9 C[C2 H] component above the stoichiometric ratio of 3b, suggesting that cleaved C radicals 10 11 are passivated by residual H. 12 13 14 Polymer GNR 15 (a) (c) 16 17 18 19 20 21 H 22 23 24 25 26 27 28 (b) (d) 29 30 31 32 33 F 34 35 36 37 38 39 40 41 Figure 3: STM images of the polymeric chains and GNRs obtained with the hydrogenated 42 (1a) and fluorinated (1b) precursors. (a) Polymer 2a after annealing at 225 °C (It = 85 pA, 43 VS = 1.36 V. Inset: It = 130 pA, VS = 1.75 V). (b) Polymer 2b after annealing at 225 °C 44 (It = 46 pA, VS = 2.52 V. Inset: It = 60 pA, VS = 1.28 V). (c) GNR 3a after annealing 45 at 400 °C (It = 75 pA, VS = 1.36 V. Inset: It = 250 pA, VS = -0.1 V). (d) GNR 3b after 46 47 annealing at 335 °C (It = 80 pA, VS = 1.30 V. Inset: It = 210 pA, VS = 1.23 V). The 48 surface stacking domains of the Au(111) herringbone reconstruction are marked with red 49 (face-centered cubic, fcc) and cyan (hexagonal closed packed, hcp) dashed lines. Images size 50 is 150x150 nm2 . All inset scale bars correspond to 2 nm. 51 52 53 54 We next compare the XPS spectroscopic analysis with the structural information provided 55 56 by STM at each reaction step. Figure 3 shows representative STM topographic images for 57 58 59 8 60 ACS Paragon Plus Environment Page 9 of 28 ACS Nano 1 2 3 the polymeric chains and GNRs synthesized with precursors 1a (R = H) and 1b (R = F). 4 5 In both cases, annealing at 200 °C leads to the Ullmann polymerization of precursors into 6 7 chains with a zig-zag backbone of an apparent height of 0.2 nm, and protrusion pairs with 8 9 a center-to-center distance of 0.8 nm. The comparison of the polymeric unit structure with 10 11 the precursor monomer confirms that the double protrusions correspond to the out-of-plane 12 13 phenyl rings and that functional groups stay on the protruding edge of the zig-zag backbone 14 15 (see insets in Figure 3a,b). Further annealing (at 335 °C for F-GNR and 400 °C for H- 16 17 GNR)leads to the formation of planar chevron-type structures (insets in Figure 3c,d), as 18 19 expected from the incorporation of the side phenyl rings into the backbone after the CDH 20 21 reaction. The overall apparent height of 0.15 nm of this structure is in good agreement with 22 23 previously reported GNRs. 16,19 Despite their similar intra-molecular structural evolution, 24 25 the substituents of each precursor leads to inter-molecular interactions that dictate distinct 26 27 assembly patterns. This is most evident at the intermediate polymeric phase (Figure 3a). In 28 29 the case of 2a the interaction with the herringbone reconstructed Au(111) surface guides the 30 31 polymer assembly. The single chains fill the face-centered cubic (fcc) tracks, terminating at 32 33 the kinks of the herringbone zigzag, and hence giving rise to a superlattice of chains with a 34 35 narrow length distribution that is determined by the zigzag period. In contrast, F-mediated 36 37 inter-chain interactions for 2b are strong enough to bunch them in parallel arrays that 38 39 appear off-registry with the herringbone template, avoiding any size restriction imposed by 40 41 it. The differences are minimized after the aromatization of the polymers into GNRs (Figure 42 43 3b). In this step, GNR 3a simply replicates the polymer superlattice, whereas the stronger 44 45 interaction of the planar structures with the substrate causes a debunching of 2b arrays into 46 47 single 3b GNRs that now try to follow the herringbone reconstruction. The overgrown GNRs, 48 49 however, do not fit in the zigzag units of the reconstruction and consequently the quality of 50 51 the superlattice is significantly reduced. The stability of 3b, the chevron-type fluorinated 52 53 GNR synthesized in this work, is in contrast to previous attempts of synthesizing edge- 54 55 fluorinated armchair GNRs using different anthracene derivatives as precursors. 15,20 In these 56 57 58 59 9 60 ACS Paragon Plus Environment ACS Nano Page 10 of 28 1 2 3 cases polymers were fully defluorinated before turning into GNRs. With the C-F bond being 4 5 one of the strongest C-heteroatom single bond, the premature cleavage was unexpected and 6 7 attributed to a combination of surface and byproduct assisted bond weakening. According to 8 9 the density functional theory (DFT) calculations, 15 the interaction with the underlying Au 10 11 atoms already reduces the activation energy of the C(sp2 )-F bond cleavage by 100 kJ/mol. 12 13 A more significant weakening is induced by the sp3 bonding of a residual H generated in the 14 15 nearest neighbor C atom during the CDH, which brings the activation energy further down 16 17 by 400 kJ/mol. Altogether, the two effects reduce the C-F cleavage activation energy close 18 19 to an order of magnitude, from ~540 kJ/mol to ~40 kJ/mol. 20 21 We studied the C-F bond stability of 3b by performing gas-phase DFT calculations (see 22 23 Figure 4), leading to similar conclusions as for the armchair fluorinated GNRs of Ref. 15. 24 25 The C(sp3 )-H bond formation leads to a similar reduction of ~400 kJ/mol of the C-F bond 26 27 cleavage activation energy, from 554 kJ/mol in C(sp2 )-F to 152 kJ/mol in H-C(sp3 )-F, in- 28 29 dicating that the local environment is the most important factor determining the stability 30 31 of these bonds. We also find that for H-C(sp3 )-F, defluorination is favorable over dehydro- 32 33 genation by ~75 kJ/mol. The main conclusion of the above analysis is that the C-F bond 34 35 in the H-C(sp3 )-F group is not more stable in the chevron GNRs of the present work than 36 37 in the armchair GNRs described in Ref. 15. The difference between the two cases lies in 38 39 the probability of obtaining the transient H-C(sp3 )-F configuration. Since migration bar- 40 41 riers are expected to be similar for different ribbon topologies, the formation rate of the 42 43 H-C(sp3 )-F group will mainly depend on the specific sites where residual H are originated 44 45 during the CDH reaction with respect to the fluorinated C position. For the armchair GNRs 46 47 described in Ref. 15, H is released from the vicinal C atom, and a low migration energy 48 49 barrier of ~80 kJ/mol to the fluorinated C atom results in a high formation rate of the tran- 50 51 sient H-C(sp3 )-F configuration (k~107 s-1 at 400 °C). In contrast, the particular design of 52 53 our precursor relocates the closest formation of residual H significantly further away, at the 54 55 5th nearest neighbor (see Figure 4b and path 1 in Figure 4c). This introduces new low acti- 56 57 58 59 10 60 ACS Paragon Plus Environment Page 11 of 28 ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 Figure 4: Stability analysis of the C-F bond. (a) Calculated dissociation energies of the 44 45 C(sp2 )-F bond (left) and of C-H and C-F bonds within the H-C(sp3 )-F group during the 46 formation of GNR 3b. (b) The three different families of C(sp2 ) atoms depending on their 47 local environment in the GNR sub-unit. (c) Possible migration paths of the closest H atom 48 towards the edge C, marked with the color code of the families in b. 49 50 51 52 53 54 55 56 57 58 59 11 60 ACS Paragon Plus Environment ACS Nano Page 12 of 28 1 2 3 vation barrier paths for the release of the residual H into the surface via the hydrogenated C 4 5 edge atoms, effectively preventing the formation of the H-C(sp3 )-F groups and therefore the 6 7 corresponding defluorination. In our analysis of H release paths, we only consider residual H 8 9 generation closest to the C-F bond, noting that the others only differ by additional migration 10 11 steps within the GNR backbone. This originates from the cleavage of the C-H pair marked 12 13 in orange circles in the polymer of Figure 4b. Each pair produces one C(sp3 )-H after the C-C 14 15 coupling, since the C-H pointing downwards is directly released to the surface. 15,21 In total, 16 17 we find 5 possible paths from this initial position of the residual H to the nearest C edge 18 19 atoms, which are sketched in Figure 4c. The color of the segments in each path indicates 20 21 hopping to three different carbon types, C[C3 ], C[C2 H] and C[C2 F], corresponding to fully 22 23 carbon-bonded, hydrogenated and fluorinated edge atoms respectively (see Figure 4b). Path 24 25 1 is the one inducing defluorination by the mechanism proposed in Ref. 15, whereas paths 26 27 2-5 reach hydrogenated edge C atoms that can release the residual H to the surface by an 28 29 equivalent dehydrogenation process. 21 A common property shared by both fluorinated and 30 31 hydrogenated edges is that, in the presence of a C(sp3 )-H bond, they are both characterized 32 33 by an activation energy for C-F and C-H bond cleavage that is significantly smaller than the 34 35 migration barrier of the residual H to the vicinal C. 15,21 The probability ratio between C-F 36 37 and C-H cleavage will then exclusively depend on the probability of arrival of the residual H 38 39 to the corresponding edge C atom. This is effectively reduced by the topology of 3b in two 40 41 ways: providing multiple additional paths for H release (paths 2-5) and, more importantly, 42 43 interposing one of these paths (path 2) to the defluorination one (path 1). Altogether, the 44 45 precursor design effectively prevents defluorination during the synthesis, as demonstrated 46 47 directly by the temperature dependent XPS measurements. 48 49 Some effects of fluorination are imprinted in the electronic properties, despite the appar- 50 51 ent low density of edge fluorination exhibited by GNR 3b. DFT calculations of the gas-phase 52 53 structure reveal a significant downshift of valence and conduction bands of ~200 meV (see 54 55 Figure 5a), which can be seen as an electrostatic n-gating induced by the electron accu- 56 57 58 59 12 60 ACS Paragon Plus Environment Page 13 of 28 ACS Nano 1 2 3 (a) (b) (c) 14x10-3 4 -3.0 C (x0.3) 5 -3.5 12 H H-GNR 6 203 meV 10 PDOS (states/eV) 7 -4.0 H-GNR dI/dV (a.u.) F-GNR E-EV (eV) 8 8 C (x0.3) F-GNR -4.5 9 6 F H 10 -5.0 213 meV 4 11 Au surface Au surface 12 -5.5 2 13 0 14 X -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 15 Wave vector E-EF (eV) E-EF (eV) 16 17 Figure 5: Electronic structure of the GNRs 3a,b. (a) Band structures of free-standing 3a 18 (H-GNR, blue) and 3b (F-GNR, green) for the structures relaxed on Au(111). Energies 19 given with respect to the vacuum energy. (b) PDOS on the different chemical species from 20 3a (blue curves, top) and from 3b (green curves, middle), both calculated on a Au(111) 21 substrate and with a Lorentzian broadening of 35 meV. The PDOS on the surface Au atoms 22 23 far from the GNR is shown in shaded gray at the bottom (see Methods). (c) Representative 24 constant-height dI/dV spectra taken on 3a (after CDH at 400 °C) and 3b (after CDH at 25 335 °C) in the position of the functional group. Spectrum taken on clean Au is reported in 26 shaded gray as reference. 27 28 29 mulation at the fluorine atoms. As opposed to fully edge-halogenated armchair 1 and cove 2 30 31 GNRs, the size of the band gap is barely affected by the fluorination (1.48/1.49 eV for 3a/3b 32 33 respectively), probably due to the lower nominal density of edge-fluorination and different 34 35 topology of our chevron GNRs. Indeed, a similar quasi-rigid downshift have been also ob- 36 37 served in N-doped chevron GNRs. 22,23 The band shift, however, is reduced to negligible 38 39 values (~10 meV) when GNRs interact with the Au substrate, as can be noted from the pro- 40 41 jected density of states (PDOS) plotted in Figure 5b. We attribute this band realignment to 42 43 interfacial charge redistribution, similar to that leading to the Fermi level pinning observed 44 45 in armchair GNRs. 24 An additional effect of the surface is the reduction of the band gap 46 47 from ~1.5 eV to 1.3 eV for free-standing and on-Au GNRs respectively, which is attributed 48 49 to hybridization with Au states. The absence of any sizeable band gap difference and shift is 50 51 reproduced in comparative spectroscopic measurements performed in 3a and 3b GNRs on 52 53 the Au substrate (see Figure 5c). Here we find the onset of the valence band (VB) fixed at 54 55 -0.91 eV for both GNRs, whereas the onset of the conduction band (CB) can be identified 56 57 58 59 13 60 ACS Paragon Plus Environment ACS Nano Page 14 of 28 1 2 3 at 1.05 eV for 3a and 0.95 eV for 3b. The resulting gaps of 1.96 eV and 1.86 eV are slightly 4 5 larger than the 1.3 eV obtained by DFT, which is accounted by the well-known gap under- 6 7 estimation of this method. 25 According to the PDOS analysis of Figure 5b, the bands with 8 9 significant fluorine spectral contribution produces PDOS maxima at around 1 eV below the 10 11 VB, which referred to the experimental spectra would corresponds to a bias voltage of -2 eV. 12 13 Unfortunately, the 3b GNRs were not stable at this large bias voltage, probably due to tip- 14 15 induced C-F bond cleavage, hindering the detection of any direct spectroscopic fingerprint 16 17 of the fluorine atoms. 18 19 20 21 Conclusion 22 23 24 We have demonstrated the on-surface synthesis of edge-fluorinated GNRs. Our combined 25 26 thermal evolution study of XPS and STM measurements provides chemical and structural 27 28 evidences of the formation of chevron type GNRs and the survival of fluorine functionaliza- 29 30 tion throughout the whole synthetic path. As opposed to the case of armchair type GNRs 31 32 based on fluorinated anthracene derivatives (DBTA), where the polymers are fully deflu- 33 34 orinated before the formation of GNRs, we achieve 100% of fluorination at the threshold 35 36 of GNR formation. In a thermal window of 40 °C above that threshold we only observe a 37 38 slight decrease to 80%. According to first-principle calculations, the electronegative fluorine 39 40 atoms tend to shift the valence and conduction bands of free-standing GNRs to lower bind- 41 42 ing energies by about 200 meV without affecting the size of the gap. However, this shift is 43 44 canceled out in the experimental dI/dV spectra due to the pinning effect of the underlying 45 46 metallic substrate. The stability of the fluorine substituents is attributed to the particular 47 48 design of the precursor, in which the F atoms are introduced in remote positions with respect 49 50 to the cyclodehydrogenation regions. By providing alternative H release paths, our design 51 52 effectively hampers the migration of residual H originated during cyclodehydrogenation to 53 54 the fluorinated C edge atom, and the consequent formation of the H-C(sp3 )-F bond that 55 56 facilitates the C-F bond cleavage. These alternative H release pathways are independent of 57 58 59 14 60 ACS Paragon Plus Environment Page 15 of 28 ACS Nano 1 2 3 the functional group, and hence we expect that our approach should enable the synthesis 4 5 of other edge-functionalized GNRs. The structural origin of the stabilization highlights the 6 7 relevance of a rational design of precursors not only for achieving a particular topology of 8 9 the final product but also for defining reaction pathways that avoid destabilizing transient 10 11 structures. 12 13 14 15 16 Methods 17 18 19 STM and XPS measurements were performed in two separated ultra-high vacuum (UHV) 20 21 systems with a base pressure below 5x10-10 mbar. Clean Au(111) samples were prepared by 22 23 cycles of Ar ion sputtering at 1 keV and annealing at 470 °C for 20 minutes. Surface clean- 24 25 liness has been checked by low energy electron diffraction (LEED) and either STM or XPS. 26 27 Molecular precursors have been deposited from a resistively heated crucible on the Au(111) 28 29 substrate kept at RT. For the sub-monolayer coverage studied, 1a and 1b were sublimated 30 31 at 280 °C and 300 °C respectively. The sample was annealead at a rate of about 1 K/s and 32 33 kept 15 minutes at the indicated temperatures to promote the polymerization and cyclode- 34 35 hydrogenation reactions. Sample temperatures were measured with an infrared pyrometer 36 37 in the XPS setup and a thermocouple directly spot-welded on the sample holder in the STM 38 39 setup. The pyrometer was previously calibrated with the sample holder thermocouple of the 40 41 STM setup in order to have reliable temperature comparison between the two measuring 42 43 methods. STM measurements have been carried out with a Createc LT-STM operated at 44 45 samples temperatures of 5 K, except for the image of the inset of Figure 3c, which was ac- 46 47 quired at 77 K . STM topographic images have been acquired in constant-current mode and 48 49 analyzed and processed with Gwyddion software. 26 STM dI/dV spectra have been measured 50 51 at 5 K in constant-height mode with the lock-in technique and a bias modulation of 20 mV. 52 53 Due to the difficulty of determining the VB/CB maxima/minima in the broad experimental 54 55 features, both the theoretical and experimental band gaps have been determined by consid- 56 57 58 59 15 60 ACS Paragon Plus Environment ACS Nano Page 16 of 28 1 2 3 ering the onset of the corresponding features. This represents a lower limit in the estimation 4 5 of the band gaps. XPS measurements were carried out at RT with a SPECS Phoibos 150 6 7 hemispherical energy analyzer using a monochromatic Al Kα X-ray source at 1486.6 eV. 8 9 XPS spectra have been referenced to the Fermi level, fitted with a Fermi-Dirac distribution 10 11 at the sample temperature. 12 13 First-principles calculations were performed via Density Functional Theory as imple- 14 15 mented in the SIESTA package. 27,28 Here the valence electrons are described with a basis 16 17 set given by a linear combination of numerical atomic-orbitals, in our case a double-ζ plus 18 19 polarization (DZP) basis set with the orbital radii defined using a 50 meV energy shift. 28 We 20 21 adopted for the top Au atomic layer an extended DZP basis optimized for the description of 22 23 the Au(111) surface 29 as well as a Hydrogen passivation of the slab bottom layer to prevent 24 25 spurious effects due to the proximity of the surface states on the slab representation. 30 More- 26 27 over, for a better description of the carbon states, the basis set was extended to included 3s 28 29 and 3p orbitals. 6 The core electrons were represented by norm-conserving Troullier-Martins 30 31 pseudopotentials. 31 Dispersive interactions were considered via the van der Waals density 32 33 functional by Dion et al 32 with the modified exchange by Klimeš, Bowler and Michaelides. 33 34 35 The fineness of the grid for integrations in real space was defined using a 300 Ry energy cut- 36 37 off. The self-consistency cycle was stopped when the changes of the density matrix elements 38 39 were less than 10-5 and also lower than 10-4 eV for the Hamiltonian matrix elements. The 40 41 smearing of the electronic occupations was defined by an electronic temperature of 300 K 42 43 with a Fermi-Dirac distribution. The structures were allowed to fully relax until the forces 44 45 were lower than 10 meV/Å on top of a 3-layer thick slab within a 17.65x30.57x60 Å3 sim- 46 47 ulation cell (with the two bottom layers kept fix with experimental lattice). We used a 15 48 49 k-point mesh along the GNRs periodic direction (x) and for calculations involving the Au 50 51 substrate 3 k-points were adopted along the other periodic direction (y) as well. The Au 52 53 surface density of states (DOS), shown in Figure 5b, was obtained by the difference between 54 55 the DOS projected on the Au atoms in the surface layer and far from the GNR and the DOS 56 57 58 59 16 60 ACS Paragon Plus Environment Page 17 of 28 ACS Nano 1 2 3 projected on the Au atoms in middle slab layer, which has been shown to provide a good 4 5 estimation using similar parameters. 34 6 7 8 9 10 Acknowledgement 11 12 13 We are indebted to D. Pérez and E. Guitián for fruitful discussions. C.M. was supported 14 15 by the Agency for Management of University and Research grants (AGAUR) of the Cata- 16 17 lan government through the FP7 framework program of the European Commission un- 18 19 der Marie Curie COFUND action 600385 funded by the CERCA Program/Generalitat de 20 21 Catalunya. We acknowledge support from the Spanish Ministry of Economy and Competi- 22 23 tiveness, MINECO (under Contracts No. MAT2016-78293-C6, FIS2017-83780-P and Severo 24 25 Ochoa No. SEV-2017-0706), the European Regional Development Fund (ERDF), the In- 26 27 terreg V-A España-Francia-Andorra program (Contract No. EFA 194/16 TNSI), the EU 28 29 project SPRING (863098), the Xunta de Galicia (Centro singular de investigación de Galicia 30 31 accreditation 2016-2019, ED431G/09). 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 17 60 ACS Paragon Plus Environment ACS Nano Page 18 of 28 1 2 3 4 Supporting Information Available 5 6 7 Molecular synthesis 8 9 10 General aspects 11 12 All reactions were carried out under argon using oven-dried glassware. THF, CH2 Cl2 , and 13 14 CH3 CN were purified by a MBraun SPS-800 Solvent Purification System. Finely powdered 15 16 CsF was dried under vacuum at 100 °C, cooled under argon and stored in a glove-box. TLC 17 18 was performed on Merck silica gel 60 F254; chromatograms were visualized with UV light 19 20 (254 nm and 360 nm). Flash column chromatography was performed on Merck silica gel 21 22 60 (ASTM 230-400 mesh). 1 H and 13 C NMR spectra were recorded at 300 and 75 MHz or 23 24 500 and 126 MHz (Varian Mercury 300, Bruker AMX-500 instruments, respectively). Low- 25 26 resolution electron impact mass spectra were determined at 70 eV on a HP-5988A instrument. 27 28 High-resolution mass spectra (HRMS) were obtained on a Bruker Microtof spectrometer. 29 30 Commercial reagents were purchased from ABCR GmbH or Aldrich Chemical Co. and 31 32 were used without further purification. Molecules 5 and 7a-b were prepared following pub- 33 34 lished procedures (Figure S1)1 . 35 36 Br 37 38 TfO TfO F 39 O 40 TMS TMS F 41 7a 7b 42 Br 5 43 44 Figure 6: Structures of starting materials 5 and 7a-b. 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 18 60 ACS Paragon Plus Environment Page 19 of 28 ACS Nano 1 2 3 Br Br 4 5 TfO CsF 6 O + 7 TMS CH3CN/CH2Cl2 8 7a 50 ºC 9 Br Br 1a 10 5 F- Br 11 12 CO 5 13 CO extrusion 14 Diels-Alder 15 7a 16 Br 17 18 Figure 7: Synthesis of compound 1a. 19 20 21 Synthesis of 2,7-dibromo-9,14-diphenylbenzo[f]tetraphene (1a) 22 23 24 Finely powdered anhydrous CsF (170 mg, 1.09 mmol) was added over a solution of dienone 25 26 5 (125 mg, 0.24 mmol) and triflate 7a (110 mg, 0.36 mmol) in CH3 CN/CH2 Cl2 (1:1, 6 27 28 mL). The mixture was heated at 50 °C under argon for 16 h. Then, the solvent was evap- 29 30 orated under reduced pressure and the residue was purified by column chromatography 31 32 (SiO2 ; hexane/CH2 Cl2 4:1) to isolate compound 1a (87 mg, 61%) as a yellow solid. 1 H 33 34 NMR (300 MHz, CDCl3 ) δ: 8.05 (d, J = 8.6 Hz, 2H), 7.95 (dd, J = 6.6, 3.3 Hz, 2H), 7.61 – 35 36 7.50 (m, 12H), 7.48 (d, J = 3.3 Hz, 2H), 7.43 (dd, J = 8.6, 1.9 Hz, 2H) ppm. 13 C NMR (125 37 38 MHz, CDCl3 ) δ: 141.2 (2C), 136.6 (2C), 133.6 (2CH), 133.2 (2C), 132.7 (2C), 132.5 (4CH), 39 40 130.1 (2CH), 129.5 (4CH), 129.2 (2C), 128.3 (2CH), 127.7 (2C), 127.1 (2CH), 126.5 (2CH), 41 42 124.7 (2CH), 120.6 (2C) ppm. MS (EI), m/z (%): 588 (M+, 100), 508 (14), 428 (15), 350 43 44 (20), 212 (16). HRMS (EI) for C34 H20 Br2 , calculated: 585.9932, found: 585.9949. 45 46 47 Synthesis of 2,7-dibromo-11,12-difluoro-9,14-diphenylbenzo[f]tetraphene (1b) 48 49 50 Finely powdered anhydrous CsF (198 mg, 1.31 mmol) was added over a solution of dienone 51 52 5 (150 mg, 0.29 mmol) and triflate 7b (136 mg, 0.41 mmol) in CH3 CN/CH2 Cl2 (1:1, 24 53 1 54 a) For synthesis of dienone 5, see: Vo, T. H.; Shekhirev, M.; Kunkel, D. A.; Morton, M. D.; Berglund, 55 E.; Kong, L.; Wilson, P. M.; Dowben, P. A.; Enders, A.; Sinitskii, A. Nat. Commun. 2014, 5, 3189. b) For 56 preparation triflates 7a-b, see: Peña, D.; Cobas, A.; Pérez, D.; Guitián, E. Synthesis 2002, 1454. 57 58 59 19 60 ACS Paragon Plus Environment ACS Nano Page 20 of 28 1 2 3 Br Br 4 5 TfO F CsF F 6 O + TMS F CH3CN/CH2Cl2 F 7 50 ºC 7b 8 Br Br 9 5 1b 10 F- 11 Br 12 13 5 F CO extrusion CO 14 Diels-Alder F 15 7b 16 Br 17 18 Figure 8: Synthesis of compound 1b. 19 20 21 mL). The mixture was heated at 50 °C under argon for 16 h. Then, the solvent was evapo- 22 23 rated under reduced pressure and the residue was purified by column chromatography (SiO2 ; 24 25 hexane/CH2 Cl2 4:1) to isolate compound 1b (72 mg, 40%) as a pale yellow solid. 1 H NMR 26 27 (353 K, 500 MHz, C2 D2 Cl4 ) δ: 8.02 (d, J = 8.6 Hz, 2H), 7.64 – 7.52 (m, 10H), 7.49 – 7.39 28 29 (m, 6H) ppm. 13 C NMR (353 K, 126 MHz, C2 D2 Cl2 ) δ: 150.43 (dd, J = 253.4, 17.7 Hz, 30 31 2CF), 140.74 (2C), 136.18 (2C), 133.51 (2CH), 132.65 (2C), 132.06 (4CH), 130.49 (2CH), 32 33 130.44 (2C), 130.22 (4CH), 129.87 (2CH), 128.81 (2C), 128.38 (4CH), 124.80 (2C), 120.76 34 35 (2C), 113.19 (dd, J = 13.7, 6.6 Hz, 2CH) ppm. MS (EI), m/z (%): 623 (100), 543 (24). 36 37 HRMS (APCI (M+1)) for C34 H18 Br2 F2 , calculated: 621.9738, found: 621.9730. 38 39 40 1 13 41 H and C NMR spectra 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 20 60 ACS Paragon Plus Environment Page 21 of 28 ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Br 16 17 18 19 20 Br 21 8.1 8.0 7.9 7.8 7.7 7.6 7.5 7.4 1a 22 23 24 25 26 27 28 29 30 10.0 9.5 9.0 8.5 8.0 7.5 7.0 6.5 6.0 5.5 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 -0.5 31 32 33 34 35 36 37 38 39 40 136 134 132 130 128 126 124 122 120 41 42 43 210 200 190 180 170 160 150 140 130 120 110 100 90 80 70 60 50 40 30 20 10 0 -10 44 45 Figure 9: 1 H, DEPT 135 and 13 C NMR spectra of compound 1a. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 21 60 ACS Paragon Plus Environment ACS Nano Page 22 of 28 1 2 3 4 5 6 7 8 9 Br 10 11 F 12 F 13 14 Br 15 8.0 7.9 7.8 7.7 7.6 7.5 7.4 1b 16 17 18 19 20 21 22 23 24 25 26 27 10.0 9.5 9.0 8.5 8.0 7.5 7.0 6.5 6.0 5.5 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 150 145 140 135 130 125 120 115 43 44 45 46 47 48 49 210 200 190 180 170 160 150 140 130 120 110 100 90 80 70 60 50 40 30 20 10 0 -10 50 51 Figure 10: 1 H, DEPT 135 and 13 C NMR spectra of compound 1b. 52 53 54 55 56 57 58 59 22 60 ACS Paragon Plus Environment Page 23 of 28 ACS Nano 1 2 3 4 References 5 6 1. Wagner, P.; Ewels, C. P.; Adjizian, J.-J.; Magaud, L.; Pochet, P.; Roche, S.; Lopez- 7 8 Bezanilla, A.; Ivanovskaya, V. V.; Yaya, A.; Rayson, M. et al. Band Gap Engineering via 9 10 Edge-Functionalization of Graphene Nanoribbons. The Journal of Physical Chemistry C 11 12 2013, 117, 26790–26796. 13 14 15 2. Tan, Y.-Z.; Yang, B.; Parvez, K.; Narita, A.; Osella, S.; Beljonne, D.; Feng, X.; 16 17 Müllen, K. Atomically precise edge chlorination of nanographenes and its application 18 19 in graphene nanoribbons. Nature Communications 2013, 4, 2646. 20 21 22 3. Ruffieux, P.; Wang, S.; Yang, B.; Sánchez-Sánchez, C.; Liu, J.; Dienel, T.; Talirz, L.; 23 24 Shinde, P.; Pignedoli, C. A.; Passerone, D. et al. On-surface synthesis of graphene 25 26 nanoribbons with zigzag edge topology. Nature 2016, 531, 489–492. 27 28 29 4. Carbonell-Sanromà, E.; Hieulle, J.; Vilas-Varela, M.; Brandimarte, P.; Iraola, M.; Bar- 30 31 ragán, A.; Li, J.; Abadia, M.; Corso, M.; Sánchez-Portal, D. et al. Doping of Graphene 32 33 Nanoribbons via Functional Group Edge Modification. ACS Nano 2017, 11, 7355–7361. 34 35 36 5. Slota, M.; Keerthi, A.; Myers, W. K.; Tretyakov, E.; Baumgarten, M.; Ardavan, A.; 37 38 Sadeghi, H.; Lambert, C. J.; Narita, A.; Müllen, K. et al. Magnetic edge states and 39 40 coherent manipulation of graphene nanoribbons. Nature 2018, 557, 691–695. 41 42 6. Moreno, C.; Vilas-Varela, M.; Kretz, B.; Garcia-Lekue, A.; Costache, M. V.; Parad- 43 44 inas, M.; Panighel, M.; Ceballos, G.; Valenzuela, S. O.; Peña, D. et al. Bottom-up 45 46 synthesis of multifunctional nanoporous graphene. Science 2018, 360, 199–203. 47 48 49 7. Shekhirev, M.; Zahl, P.; Sinitskii, A. Phenyl Functionalization of Atomically Pre- 50 51 cise Graphene Nanoribbons for Engineering Inter-ribbon Interactions and Graphene 52 53 Nanopores. ACS Nano 2018, 12, 8662–8669. 54 55 56 57 58 59 23 60 ACS Paragon Plus Environment ACS Nano Page 24 of 28 1 2 3 8. Cho, K. M.; Cho, S.-Y.; Chong, S.; Koh, H.-J.; Kim, D. W.; Kim, J.; Jung, H.-T. 4 5 Edge-Functionalized Graphene Nanoribbon Chemical Sensor: Comparison with Carbon 6 7 Nanotube and Graphene. ACS Applied Materials & Interfaces 2018, 10, 42905–42914. 8 9 10 9. Joshi, D.; Hauser, M.; Veber, G.; Berl, A.; Xu, K.; Fischer, F. R. Super-Resolution 11 12 Imaging of Clickable Graphene Nanoribbons Decorated with Fluorescent Dyes. Journal 13 14 of the American Chemical Society 2018, 140, 9574–9580. 15 16 17 10. Moreno, C.; Paradinas, M.; Vilas-Varela, M.; Panighel, M.; Ceballos, G.; Peña, D.; Mu- 18 19 garza, A. On-surface synthesis of superlattice arrays of ultra-long graphene nanoribbons. 20 21 Chemical Communications 2018, 54, 9402–9405. 22 23 24 11. Keerthi, A.; Radha, B.; Rizzo, D.; Lu, H.; Diez Cabanes, V.; Hou, I. C.-Y.; Beljonne, D.; 25 26 Cornil, J.; Casiraghi, C.; Baumgarten, M. et al. Edge Functionalization of Structurally 27 28 Defined Graphene Nanoribbons for Modulating the Self-Assembled Structures. Journal 29 30 of the American Chemical Society 2017, 139, 16454–16457. 31 32 33 12. Franc, G.; Gourdon, A. Covalent networks through on-surface chemistry in ultra-high 34 35 vacuum: state-of-the-art and recent developments. Physical Chemistry Chemical Physics 36 37 2011, 13, 14283. 38 39 13. On-Surface Synthesis II: Proceedings of the International Workshop On-Surface Synthe- 40 41 sis, San Sebastián, 27-30 June 2016. 2016, 42 43 44 14. Clair, S.; de Oteyza, D. G. Controlling a Chemical Coupling Reaction on a Surface: 45 46 Tools and Strategies for On-Surface Synthesis. Chemical Reviews 2019, 119, 4717–4776. 47 48 49 15. Hayashi, H.; Yamaguchi, J.; Jippo, H.; Hayashi, R.; Aratani, N.; Ohfuchi, M.; Sato, S.; 50 51 Yamada, H. Experimental and Theoretical Investigations of Surface-Assisted Graphene 52 53 Nanoribbon Synthesis Featuring Carbon–Fluorine Bond Cleavage. ACS Nano 2017, 11, 54 55 6204–6210. 56 57 58 59 24 60 ACS Paragon Plus Environment Page 25 of 28 ACS Nano 1 2 3 16. Cai, J.; Ruffieux, P.; Jaafar, R.; Bieri, M.; Braun, T.; Blankenburg, S.; Muoth, M.; 4 5 Seitsonen, A. P.; Saleh, M.; Feng, X. et al. Atomically precise bottom-up fabrication of 6 7 graphene nanoribbons. Nature 2010, 466, 470–473. 8 9 10 17. Basagni, A.; Sedona, F.; Pignedoli, C. A.; Cattelan, M.; Nicolas, L.; Casarin, M.; 11 12 Sambi, M. Molecules–Oligomers–Nanowires–Graphene Nanoribbons: A Bottom-Up 13 14 Stepwise On-Surface Covalent Synthesis Preserving Long-Range Order. Journal of the 15 16 American Chemical Society 2015, 137, 1802–1808. 17 18 19 18. Batra, A.; Cvetko, D.; Kladnik, G.; Adak, O.; Cardoso, C.; Ferretti, A.; Prezzi, D.; 20 21 Molinari, E.; Morgante, A.; Venkataraman, L. Probing the mechanism for graphene 22 23 nanoribbon formation on gold surfaces through X-ray spectroscopy. Chem. Sci. 2014, 24 25 5, 4419–4423. 26 27 28 19. Moreno, C.; Panighel, M.; Vilas-Varela, M.; Sauthier, G.; Tenorio, M.; Ceballos, G.; 29 30 Peña, D.; Mugarza, A. Critical Role of Phenyl Substitution and Catalytic Substrate in 31 32 the Surface-Assisted Polymerization of Dibromobianthracene Derivatives. Chemistry of 33 34 Materials 2019, 31, 331–341. 35 36 37 20. Ohtomo, M.; Jippo, H.; Hayashi, H.; Yamaguchi, J.; Ohfuchi, M.; Yamada, H.; Sato, S. 38 39 Interpolymer Self-Assembly of Bottom-up Graphene Nanoribbons Fabricated from Flu- 40 41 orinated Precursors. ACS Applied Materials & Interfaces 2018, 10, 31623–31630. 42 43 21. Blankenburg, S.; Cai, J.; Ruffieux, P.; Jaafar, R.; Passerone, D.; Feng, X.; Müllen, K.; 44 45 Fasel, R.; Pignedoli, C. A. Intraribbon Heterojunction Formation in Ultranarrow 46 47 Graphene Nanoribbons. ACS Nano 2012, 6, 2020–2025. 48 49 50 22. Bronner, C.; Stremlau, S.; Gille, M.; Brauße, F.; Haase, A.; Hecht, S.; Tegeder, P. 51 52 Aligning the Band Gap of Graphene Nanoribbons by Monomer Doping. Angewandte 53 54 Chemie 2013, 125, 4518–4521. 55 56 57 58 59 25 60 ACS Paragon Plus Environment ACS Nano Page 26 of 28 1 2 3 23. Cai, J.; Pignedoli, C. A.; Talirz, L.; Ruffieux, P.; Söde, H.; Liang, L.; Meunier, V.; 4 5 Berger, R.; Li, R.; Feng, X. et al. Graphene nanoribbon heterojunctions. Nature Nan- 6 7 otechnology 2014, 9, 896–900. 8 9 10 24. Merino-Díez, N.; Garcia-Lekue, A.; Carbonell-Sanromà, E.; Li, J.; Corso, M.; Co- 11 12 lazzo, L.; Sedona, F.; Sánchez-Portal, D.; Pascual, J. I.; de Oteyza, D. G. Width- 13 14 Dependent Band Gap in Armchair Graphene Nanoribbons Reveals Fermi Level Pinning 15 16 on Au(111). ACS Nano 2017, 11, 11661–11668. 17 18 19 25. Yang, L.; Park, C.-H.; Son, Y.-W.; Cohen, M. L.; Louie, S. G. Quasiparticle Energies 20 21 and Band Gaps in Graphene Nanoribbons. Physical Review Letters 2007, 99, 186801. 22 23 24 26. Nečas, D.; Klapetek, P. Gwyddion: an open-source software for SPM data analysis. Open 25 26 Physics 2012, 10 . 27 28 29 27. Artacho, E.; Sánchez-Portal, D.; Ordejón, P.; García, A.; Soler, J. M. Linear-Scaling 30 31 ab-initio Calculations for Large and Complex Systems. physica status solidi (b) 1999, 32 33 215, 809–817. 34 35 28. Soler, J. M.; Artacho, E.; Gale, J. D.; García, A.; Junquera, J.; Ordejón, P.; Sánchez- 36 37 Portal, D. The SIESTA method for ab initio order-N materials simulation. Journal of 38 39 Physics: Condensed Matter 2002, 14, 2745–2779. 40 41 42 29. García-Gil, S.; García, A.; Lorente, N.; Ordejón, P. Optimal strictly localized basis sets 43 44 for noble metal surfaces. Physical Review B 2009, 79, 075441. 45 46 47 30. Gonzalez-Lakunza, N.; Fernández-Torrente, I.; Franke, K. J.; Lorente, N.; Arnau, A.; 48 49 Pascual, J. I. Formation of Dispersive Hybrid Bands at an Organic-Metal Interface. 50 51 Physical Review Letters 2008, 100, 156805. 52 53 54 31. Troullier, N.; Martins, J. L. Efficient pseudopotentials for plane-wave calculations. Phys- 55 56 ical Review B 1991, 43, 1993–2006. 57 58 59 26 60 ACS Paragon Plus Environment Page 27 of 28 ACS Nano 1 2 3 32. Dion, M.; Rydberg, H.; Schröder, E.; Langreth, D. C.; Lundqvist, B. I. Van der Waals 4 5 Density Functional for General Geometries. Physical Review Letters 2004, 92, 599–610. 6 7 8 33. Klimeš, J.; Bowler, D. R.; Michaelides, A. Chemical accuracy for the van der Waals 9 10 density functional. Journal of Physics: Condensed Matter 2010, 22, 022201. 11 12 13 34. Zuzak, R.; Castro-Esteban, J.; Brandimarte, P.; Engelund, M.; Cobas, A.; 14 15 Piątkowski, P.; Kolmer, M.; Pérez, D.; Guitián, E.; Szymonski, M. et al. Building 16 17 a 22-ring nanographene by combining in-solution and on-surface syntheses. Chemical 18 19 Communications 2018, 54, 10256–10259. 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 27 60 ACS Paragon Plus Environment ACS Nano Page 28 of 28 1 2 3 4 Graphical TOC Entry 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 28 60 ACS Paragon Plus Environment