We synthesized a new molecular precursor, 6,11-diiodo-1,4-bis(2-fluorophenyl)-2,3-diphenyltriphenylene (C42H24F2I2), with fluorine substitution placed at internal ortho-positions. The 1H and 19F nuclear magnetic resonance (NMR) spectra are shown in Figs. S1 and S2. This precursor has the potential to form chevron-type GNRs on a nonmetallic substrate suited for device fabrication but here we explore the effects of fluorination on GNR growth on Au(111) surface, as illustrated in Fig. 1, by following a two-step thermally triggered on-surface reaction process.
We find that adsorption of the precursor 1 to the substrate is largely dependent upon the temperature of the substrate during deposition. After deposition with the substrate at room temperature, molecules are only observed along the step edges (Fig. 2a,b). Organic molecules like precursor 1 have been reported to selectively decorate step edges due to enhanced interchain \(\pi\)-\(\pi\) stacking interaction39,40. The precursors 1 that do adsorb form very short oligomers, with individual lobe-like features which we attribute to the lost iodine atoms (Fig. 2b). Keeping the substrate at 100°C during deposition is sufficient to consistently deiodinate the precursors and allows assembly on the Au(111) terraces as observed in Fig. 2c,d. However, the effectiveness of the deiodination seems to be enhanced at higher temperatures such as 180°C (Fig. 2e,f), where much greater coverage and average polymer length are achieved. This behavior is in contrast to previously reported behavior for pristine brominated or iodinated chevron precursors on Au(111)41 (see Fig. S3 as well).
Polymers of 1 are observed to be similar in appearance to previously reported STM images of non-functionalized chevron-type polymers11,12 (see Fig. S4), and our simulated STM images of the pristine and fluorine-functionalized polymers corroborate this similarity – both poly-1 and previously reported chevron polymers appear to adopt a \(\pi\)-interaction-stabilized assembly mode which favors the growth of small 2D islands42. However, a new helical feature is recognized for the polymer intermediates from both experiments and density functional theory (DFT) calculations (Figs. S5 and S6). To discern the relative orientations of the fluorine atoms in the polymers i.e. towards or away from the surface, we have calculated the energies and simulated STM images of these possible configurations (for the latter see Fig. S7). The GNRs arising from the cyclization step demonstrate no structural features to be attributed to lingering fluorine atoms, unlike edge-fluorinated GNRs14, suggesting that the expected HF elimination has occurred by the time cyclodehydrogenation and cyclodehydrofluorination are complete.
In order to better understand the behavior of the precursor 1 upon deposition – particularly if polymers retained the monomers’ fluorine functionalization – we employed X-ray photoelectron spectroscopy (XPS) to characterize a polymer sample grown at 180\(^\circ\)C on an Au(111)/mica thin film. Figure 3 shows the results of these experiments. This sample was transferred from the growth chamber of the STM system, in air, to a separate chamber for the XPS measurements. Measurements were acquired immediately upon introduction to the XPS chamber, 72 hours afterwards, and then upon annealing to 400°C. This was performed to acquire information regarding the short-term air and vacuum stability of the polymer sample as well as whether fluorine atoms survived through the polymerization step.
As can be seen in Fig. 3, the sample presents clear signatures of fluorine in the form of the F 1s peak and the shoulder adjacent to the C 1s peak, attributed to the carbon in C2F, utilized by Panighel et al. as a model for a fluorinated benzene moiety14,43. The I 3d5/2 peak is also observed in the polymer sample at room temperature, consistent with our STM results as well as previously reported XPS results44, where iodine was found to persist in the form of Au-I up to approximately 350°C. Annealing to 400°C induces the expected cyclodehydrogenation step in the fluorinated polymers, resulting in chevron-type GNRs and desorption of iodine-attributed features from the surface. In the XPS spectra, this is marked by the disappearance of both the I 3d5/2 and F 1s peaks. The C2F shoulder in the C 1s peak is likewise lost in this transition, which supports our conclusion that the chevron GNRs observed in STM possessed no lingering fluorine atoms. Additional individual elemental spectra can be found in Fig. S8-S11.
Here we note that the polymer intermediates are observed to exist in helical configurations. This new feature of helicity, a special case of axis chirality, has not been reported in previous bottom-up synthesis of chevron-type GNRs11,15,16,18. Consistent with earlier STM images of non-functionalized chevron-type polymers11,12, each polymer is visualized as a “stripe” shape, featuring a low-lying triphenylene core flanked by periodic bright regions attributed to the peripheral phenyl rings. Upon closer examination, these polymers in fact exhibit distinct helical signatures. As shown in Fig. 4, the line profiles along the top and bottom fringes of a representative polymer chain, where the peripheral phenyl rings are situated, display asymmetric low-high and high-low features, respectively, unequivocally indicating a helical structure. By comparing these experimental line profiles with those generated from our DFT calculations using both P- and M-configurations, we find that the observed polymer intermediates belong to the P-configuration (Fig. S6a,b). In contrast, achiral polymer chains would show two peaks of equal heights in these profiles (Fig. S6g,h). Further investigations are needed to either identify the M-configuration or justify its absence. Additionally, the helical signatures are found more pronounced under lower bias condition and become diminished when a higher bias magnitude is applied (Fig. S5). Evidence from previous studies showed that asymmetric bright spots in STM images appeared at lower bias conditions11,16.
To better understand the effect of the growth temperatures, we now compare the growth of GNRs derived from the samples deposited at different temperatures. Figure 5 shows example STM images of the 180°C (Fig. 5a) and 100°C (Fig. 5c) samples before heating to 400°C for cyclodehydrogenation plus cyclodehydrofluorination. Figure 5b,d show that the greatest average GNR length, greatest individual GNR length, and greatest total GNR coverage are obtained in the 180°C sample, with both average and greatest individual lengths reaching nearly double that observed in the 100°C sample. To shed light on this difference, we carried out DFT energy calculations for fluorine-bearing polymers adsorbed on Au(111) surface. It is found that the fluorine-bearing polymer with F atoms pointing up (Fup) is more stable than the other configuration with F atoms pointing down (Fdn) by 4.6 eV per unit cell of polymer, which is equivalent to ~ 1.2 eV each length section containing only one F atom. Due to the internal ortho-positions of F substitution, the preferred adsorption of Fup over Fdn is opposite to the previously reported trend for F atoms located on the polymer/GNR edges14,33. Despite the more favorable adsorption configuration of Fup over Fdn, the molecular precursors in the Fdn configuration may not flip to Fup configuration once adsorbed. On the other hand, the adsorption of pristine polymers should be weaker than the fluorine-bearing polymers due to the smaller molecular weight and accordingly weaker van der Waals interactions. Consequently, the stronger adsorption and molecular weight of the F-bearing precursors dictate a higher temperature for the precursors needed to overcome the barrier and to diffuse on the Au(111) surface. We conclude from these results that the higher temperature of the surface is more conducive to the diffusion of molecules on contact with the terraces, affording longer and more plentiful polymers and thus GNRs.