Carbon‐Carbon Coupling on Inert Surfaces by Deposition of En Route Generated Aryl Radicals

Abstract To facilitate C−C coupling in on‐surface synthesis on inert surfaces, we devised a radical deposition source (RDS) for the direct deposition of aryl radicals onto arbitrary substrates. Its core piece is a heated reactive drift tube through which halogenated precursors are deposited and en route converted into radicals. For the proof of concept we study 4,4′′‐diiodo‐p‐terphenyl (DITP) precursors on iodine‐passivated metal surfaces. Deposition with the RDS at room temperature results in highly regular structures comprised of mostly monomeric (terphenyl) or dimeric (sexiphenyl) biradicals. Mild heating activates progressive C−C coupling into more extended molecular wires. These structures are distinctly different from the self‐assemblies observed upon conventional deposition of intact DITP. Direct deposition of radicals renders substrate reactivity unnecessary, thereby paving the road for synthesis on application‐relevant inert surfaces.


Introduction
On-surface synthesis is am aturing approach for the realization of extended covalent organic nanostructures that spark tremendous interest in fundamental science and await apromising future in applications. [1] In terms of electronic pconjugation and overall stability,reactions that afford carboncarbon (C À C) coupling are of primary importance.I nt he established synthetic routes precursor molecules are deposited onto metal surfaces with accompanying or subsequent heating to thermally activate coupling. [2] On-surface CÀC coupling was implemented by various reactions inspired from their solution analogues as Sonogashira, [3] Glaser-Hay, [4] and vastly explored Ullmann-type couplings [5] as well as cyclodehydrogenation. [6] These reactions proceed exclusively on metal surfaces,w hose chemically active contributions are required either for kinetic or thermodynamic reasons.M etal surfaces can lower reaction barriers and enhance reaction energies by stabilizing reaction intermediates,by-products or final products. [2] Alternatively,C À Cc oupling can be nonthermally activated by light. [7] Despite encouraging first results,t he generality and utility of alternative activation schemes remains elusive.
Although metal surfaces are beneficial for characterization with electron-based techniques such as scanning tunneling microscopy (STM) or X-ray photoelectron spectroscopy (XPS), an impasse is reached when it comes to the fabrication of devices that exploit the unique electronic properties. Ultimately,t his requires semiconducting or insulating substrates.Post-synthetic decoupling by physically separating the nanostructures through an inert spacer layer represents av iable means for diminishing the influence of the metal surface.P roven strategies encompass intercalation, [8] chemical transformation of the upper metal layers into an inert compound (e.g.s ilicides or oxides) [9] or lateral manipulation on top of ap osteriori deposited insulating thin films (mostly alkali halides). [10] While decoupling facilitates characterisation of the nanostructuresinherent structures and electronic properties,i ta ppears to be of limited use for actual device fabrication. Therefore,t ransferring the nanostructures from the growth to the target substrate remains the prevalent method. [11] Yet, commonly applied template stripping is not only tedious, [12] but is also carried out wet-chemically,bearing high risks of compromising purity and integrity. Accordingly,s ynthesis by direct C À Cc oupling on inert surfaces represents an important milestone.O nly few successful examples of either inter-o ri ntramolecular CÀC coupling are reported on various non-metallic surfaces as TiO 2 , [13] calcite [14] and alkali halides. [15] Even though these examples represent major achievements,t heir specific requirements and conditions cannot alleviate the need for afacile and more general approach.
An intriguing concept is the direct deposition of activated species that can readily undergo CÀCc ouplings on arbitrary surfaces.T his idea was implemented by Gleason and coworkers as "initiated Chemical Va pour Deposition-iCVD". Thereby up to micron thick polymer films can be grown with the aid of directly deposited radicals that either polymerize themselves by addition reactions or initiate apolymerization in pre-adsorbed monomers. [16] Foro n-surface synthesis,t he temporal and spatial separation of activation and coupling removes the need for chemically active substrates,w hich could then be chosen for optimized coupling.T he obvious activated species for C À Ccoupling are radicals.The utility of halogen-substituents as leaving groups for thermal cleavage prior to deposition was already alluded in the seminal work by Grill et al., [5a] where they suggested the direct deposition of radicals for crucible temperatures above 590 K. Such an incrucible activation, however, does not allow sustainable deposition of radicals due to progressive polymerization of the evaporand. Am ore sophisticated scheme based on the separation of precursor sublimation, activation and deposition was proposed by Sakaguchi. [17] They used aq uartz tube reactor in at wo-zone furnace,w here af lux of brominated precursors is generated by sublimation. Debromination is supposed to take place in the high temperature zone at the 625 Kh ot reactor walls.T he supposedly created radicals are deposited downstream onto the target substrate in al ower temperature zone.Y et, in this work, conditions were applied where chemical activation is still feasible on the target substrate.T he Au(111) surface was held at at emperature of 525 K, that is,h igh enough for debromination of intact precursors. [18] Albeit this study proposes an intriguing concept, it could not provide unequivocal evidence for precursor activation prior to deposition.
We unambiguously demonstrate the generation of radicals from an iodinated precursor for subsequent deposition and CÀCc oupling on inert surfaces.A st arget substrates we have chosen Ag(111) and Au(111) passivated with chemisorbed iodine monolayers,b ecause iodine passivation is facile, self-limiting to one monolayer and renders the surface inert with respect to thermally activated dehalogenation. [8,19] Iodine forms hexagonal p 3 p 3R AE 308 8 superstructures on both Au(111) and Ag(111) with similar lattice parameters of a = b = 0.50 nm. [20] More importantly,a dsorbed aryl radicals form covalent-like bonds with the chemisorbed iodine atoms, [21] which may be helpful to stabilize smaller radical species for STM characterization. We used the 4,4''-diiodo-pterphenyl (DITP) precursor as relatively simple and wellstudied model compound. [22] Moreover,t he targeted parapoly-phenylene (PPP) wires received significant interest, [13a, 23] not at least for the on-surface synthesis of graphene nanoribbons by lateral fusion. [22c] Thee nr oute generation of radicals was realized by deposition through ar eactive drift tube.T he resulting nanostructures were characterized by STM either directly as deposited or after subsequent mild heating. As an important control, results are compared to experiments with conventional deposition of intact DITP.

Results and Discussion
Te rphenyl biradicals were generated and deposited with adedicated radical deposition source (RDS,see SI section 2). Its core piece is ah eated drift tube comprised of gold-plated stainless steel through which the precursors are indirectly deposited. While on gold surfaces deiodination readily occurs at room temperature, [18,24] as ufficiently high drift tube temperature of % 500 Kw as required to facilitate passage of en route generated radicals by sequences of adsorptiondesorption processes.
STM images acquired after deposition with the RDS onto iodinated metal surfaces held at room temperature are presented in Figure 1. Theo verview image unveils ah ighly regular arrangement of rod-shaped entities,o rganized in unevenly spaced lamellas.The (2.5 AE 0.1) nm long rods clearly exceed the dimensions of DITP precursors (1.58 nm from iodine to iodine,c f. SI, Figure S2). But their length is consistent with CÀCb onded terphenyl dimers,t hat is, sexiphenyl, as demonstrated by the overlay,a nd further corroborated by the internal STM contrast showing six protrusions,c orresponding to the phenyl rings ( Figure 1b). Moreover,w eo ccasionally observed longer entities that are consistent with CÀCb onded terphenyl trimers ((3.7 AE 0.1) nm) or tetramers ((5.0 AE 0.1) nm) (Figure 1c). Thechemical nature of the termini, however, remains nonspecific.Y et, we can already exclude remaining iodine substituents,simply because there is not enough space to accommodate them for the narrowest observed lamella spacings.
Accordingly,w ep ropose that the rods correspond to biradicals that are stabilized by covalent-like bonds formed at the termini with chemisorbed iodine atoms in the monolayer. [21] Them odel in Figure 1d suggests an excellent geometric match. Thereby strain is minimized, resulting in relatively strong adsorption of the comparatively small sexiphenyl biradicals that facilitates sufficient stability for room temperature STM imaging.D espite the covalent bonding with the iodine monolayer,t he adsorbed biradicals still remain active for progressive CÀCc oupling.T he STM image in Figure 2a was acquired after subsequent mild heating to 375 K. It shows aremaining domain of sexiphenyl biradicals,b ut also corroborates the formation of more extended linear oligomers.T hese are organized in domains with uniform alignment along the three equivalent < 10 > directions of the iodine monolayer.A gain the oligomer lengths are quantized in terphenyl units as illustrated by the overlays (Figure 2b). Statistical analysis of the length distribution of asample heated at 375 Kreveals % 50 %dimers and < 5% of oligomers longer than five terphenyl units (Figure 2e). Higher heating at 425 Kprofoundly alters the length distribution. Only as mall fraction of dimers remains,w hile most of the oligomers (> 80 %) are comprised of five up to 24 terphenyl units (Figure 2e). Apparently,h eating provides additional thermal energy to overcome barriers that prevent further coupling at room temperature.T hese barriers arise from the necessity of breaking radical-iodine bonds to facilitate diffusion and CÀCb ond formation. Even though longer wires are sufficiently stable to be imaged at room temperature,w ed id not resolve individual phenyl rings,i n accord with previous studies of short PPP chains on copper surfaces. [25] This is consistent with ad ominating stabilization on the iodine passivated surfaces by covalent anchoring at the termini, which is comparatively strong for sexiphenyl, but too short-ranged for longer wires.
In af urther series of experiments,s amples were immediately cooled to % 85 Ka fter room temperature deposition with the RDS.S TM images acquired at low temperatures similarly show ah ighly regular arrangement of rods with domain sizes up to 100 nm. Thes elf-assembly is periodic, hence ap rimitive unit cell can be assigned with lattice parameters a = (2.1 AE 0.1) nm, b = (1.0 AE 0.1) nm, g = 968 8 AE 38 8 that agree well with ac ommensurate (3 5; 20 )s uperstructure with respect to the iodine lattice.N oteworthy,a lso for this preparation some dimers were occasionally observed (see SI, Figure S3). Yet, the vast majority of rods are only (1.2 AE 0.1) nm long,t hat is,s ignificantly shorter as compared to those in Figure 1. Ther ods also appear with appreciable internal STM contrast, featuring two smaller bright protrusions at its termini and al ess bright larger feature at the center. Them easured (0.42 AE 0.05) nm spacing between the protrusions agrees well with the phenyl-phenyl distance of 0.44 nm in terphenyl. These observations are consistent with identifying the shorter rods as terphenyl biradicals as further illustrated by the overlay in Figure 3b.T he model in Figure 3c suggests aless favorable size match between terphenyl biradical and iodine monolayer.Consequently,surface bonding of the smaller biradical strains the terphenyl backbone, resulting in al ower stability as compared to adsorption of sexiphenyl biradicals.T his also explains why terphenyl biradicals are only observed when samples are instantaneously quenched.
Theexperiments with immediate sample cooling confirm the deposition of terphenyl biradicals.T oa lso corroborate generation of the deposited radicals by the RDS,c ontrol experiments were carried out with conventional deposition of intact DITP onto iodinated metal surfaces.
STM images of at ypically observed highly ordered structure are shown in Figure 4, but other polymorphs were similarly observed (see SI, section 5). Most importantly,a ll structures are supramolecular assemblies of intact DITP. Heating the DITP self-assemblies on I-Ag(111) to 425 K resulted in almost full desorption of DITP,l eaving ap ristine surface behind. On the one hand, this confirms the inertness of I-Ag(111) for dehalogenation. On the other hand, it provides further evidence that dehalogenation is indeed achieved by the RDS.W eo ccasionally observed indications of linear oligomers at step-edges,s uggesting some deiodination at sparse reactive sites (see SI, Figure S6). Also incompletely iodine passivated surfaces exhibit some remaining reactivity (see SI, section 7)

Conclusion
In summary,w ed emonstrate the feasibility of direct deposition of radicals generated from the iodinated DITP precursor en route in areactive drift tube.Its hot gold surface catalyzes the dissociation of iodine-substituents from DITP precursors and conducts the generated biradicals.S TM images acquired after deposition onto iodinated metal surfaces reveal regular arrangements of terphenyl biradicals that spontaneously dimerize into sexiphenyl biradicals at room temperature.Based on this observation, we propose an important role of the registry between biradical species and iodine monolayer for their surface stability.C ontrol experiments with conventional DITP deposition demonstrate that I-Ag(111) surfaces are highly inert with respect to dehalogenation, and also corroborate generation of radicals by the RDS for downstream deposition. Thea dsorbed species maintain their radical character as confirmed by their ability to further couple into extended linear structures upon mild heating. Accordingly,iodine passivated metal surfaces proved ideal for initial RDS studies as they offer aunique combination of inertness for dehalogenation and stickiness for the adsorption and stabilization of radical species.A mong thousands of coupled terphenyl units only af ew structures were observed that may be non-linear junctions (see SI, Figure S8). Given that those could still arise from isomer impurities in the precursor,this observation suggests that the RDS does not induce 1,2-rearrangements,w here the radical site migrates to the adjacent carbon atom. This is important good news for using the RDS in the reticular synthesis of covalent nanostructures,b ecause it implies that the sites for C À Cc oupling are fully predetermined by the precursors halogen-substitution pattern. However,apossible favourable role of the covalent bonding between radicals and iodinated surface remains to be explored.
Successful implementation of the RDS is our base camp for further exploring on-surface synthesis with radicals. Exciting future experiment could tackle more abundant brominated or more complex precursors for targeting twodimensional polymers.M oreover,m ore fragile precursors could by tested, as for instance molecules with sulfurcontaining heterocycles that are key components in molecular electronics, [28] but unfortunately prone to on-surface decomposition. [29] An obvious extension would be the use of alternative surfaces,where the RDS facilitates thus far elusive studies of deposition, diffusion and coupling of radicals at arbitrary temperatures.