On‐Surface Ullmann‐Type Coupling: Reaction Intermediates and Organometallic Polymer Growth

Ullmann‐type coupling is the most widely used on‐surface reaction to form rationally designed bottom‐up molecular nanoarchitectures. A commonly observed reaction product in this reaction is an organometallic phase, however little is known about the formation of this phase. The on‐surface polymerization of the prochiral precursor 6,12‐dibromochrysene (DBCh) on Ag(111) is studied. Upon annealing of DBCh on Ag(111), a linear organometallic polymer forms. However, the delicate energy balance involved in the polymerization of DBCh is such that, at room temperature, several reaction intermediates, which eventually lead to the formation of the organometallic polymer, can be observed experimentally. Organometallic monomers, dimers, and trimers are finds, that self‐assemble into distinct networks. The experimental availability of these reaction intermediates provides key insights into the formation of the organometallic polymer. Comparing the chirality of the intermediates and the polymer sheds additional light on the reaction mechanism leading to the formation of the polymer. The main finding is that the organometallic polymer is not formed by a simple coupling of the reaction intermediates, but rather requires the breaking and re‐establishing of the C─Ag bonds. Additionally, a Br‐enhanced growth mode is observed, where the split‐off halogens align the polymers, which results in an increased polymer length.


Introduction
The study of chemical reactions on well-defined surfaces is an excellent way to gain fundamental insight into reaction processes, as the surface science tool kit can provide valuable information to discern the obtained reaction products.Notably, scanning tunneling microscopy (STM) can be used to identify the structural properties of the reaction products with unmatched precision.Specifically, on-surface synthesis is a topic DOI: 10.1002/admi.202300728 of both fundamental and practical interest, as on-surface polymerization can yield products that are difficult to obtain through conventional chemical methods.6][7][8][9][10] Broadly, Ullmann-type coupling occurs in three distinct steps: i) dehalogenation of the precursor resulting in the formation of surface-stabilized radicals, ii) diffusion of the surface-stabilized radicals, and iii) C─C coupling of the radicals.Occasionally, instead of C─C coupling in the third step, an organometallic phase is formed with native substrate adatoms. [8]espite the ubiquity of this reaction, the exact mechanism, and specifically the formation of the organometallic phase is not fully understood.In the past, several works focused on tracking individual molecules with one halogenated site each, which upon Ullmann-type coupling led to the formation of 0D clusters.[21][22][23][24][25][26] We contribute valuable experimental insight by identifying several organometallic intermediate products which eventually lead to the formation of the organometallic (polymeric) phase often observed during on-surface Ullmann-type coupling.Our detailed experimental study can thus help to unravel the reaction mechanism for the formation of on-surface polymers as well as serve as an input for modeling the various reaction steps and obtaining detailed insight into the underlying energetics.
In this study, we focus on the on-surface reaction of the prochiral molecule 6,12-dibromochrysene (DBCh) on Ag(111).This molecule was previously studied on Au(111) and Cu(111). [27,28]n Au(111), DBCh forms a chiral, porous self-assembled network that may be transformed into chevron-like graphene nanoribbons upon annealing through stepwise Ullmann-type coupling and surface-assisted cyclodehydrogenation reactions.
On Cu(111), DBCh readily debrominates at room temperature, leading to the formation of chiral organometallic polymers which cannot be transformed into covalently bonded structures.][33] Simultaneously, the structural properties of Ag(111) are similar to those of Au(111) which allows for probing the influence of the electronic surface properties on the formation of the reaction products.
We could identify upon deposition of DBCh on Ag(111) four distinct organometallic phases whose formation delicately depends on the temperature of the substrate during deposition.Three of the phases are ascribed to self-assembled islands of organometallic monomers, dimers, and trimers, respectively, whereas the fourth phase is attributed to organometallic polymers.Furthermore, the organometallic dimer phase is chiral, while the monomer, trimer, and polymer phases are all achiral.Since all the phases we observed on the sample were organometallic structures and the first three phases disappeared upon annealing at moderate temperatures (>363 K), we interpret the first three phases as reaction intermediates toward the formation of (mainly) heterochiral 1D organometallic polymers.The 1D polymers were observed to be arranged into 2D islands with the help of noncovalent Br⋯H interactions which are mediated by the split-off Br atoms.
Based on a statistical analysis of the polymer length distribution in dependence on the sample preparation, we could identify that the organometallic polymer growth is path-dependent, i.e., postdeposition annealing leads to shorter polymers than deposition of the molecules on a heated substrate.We additionally observed that the polymers within the 2D islands are markedly longer than isolated ones.
The DBCh-Ag(111) system gives valuable insight into the formation of organometallic polymers, as the intermediate structures are experimentally accessible.We could identify that the adsorption of both the intermediates and the finally formed organometallic polymers are driven by a preferred adsorption site of the native Ag adatoms incorporated in these structures.Since the polymers are achiral, whereas some of the intermediates are chiral, we can infer that the organometallic polymer has to be formed through the breaking and re-establishing of metalorganic coordination bonds, rather than by attaching the intermediates to one another.The breaking and re-establishing of metalorganic coordination bonds is thus a key component enabling the formation of the polymers from the organometallic intermediates.

Results and Discussion
Samples were prepared by depositing DBCh (Figure 1a) on a clean Ag(111) substrate under ultra-high vacuum conditions.DBCh is prochiral in the gas phase.That means, once it is adsorbed on a surface it becomes chiral due to the confinement in a 2D space.Upon deposition of DBCh on Ag(111) held at room temperature (RT, ≈293 K), four distinct molecular phases, labeled as i-iv in Figure 1b could be observed when studying the sample with STM (see also Figure S1, Supporting Information).The rel- ative coverage of the four phases with respect to each other may to some extent be tuned by slightly varying the substrate temperature during deposition, whereas deposition on a sample held at 306 K, for example, leads to an almost exclusive coverage of phase iv (Figure S2, Supporting Information).

Phase i: Organometallic Monomers
Figure 2a shows an overview STM image of phase i, which reveals bowtie-shaped objects arranged in a hexagonal pattern.A closeup STM image of the network is presented in Figure 2b.Upon closer inspection, it becomes apparent that the bowtie-shaped units consist of a central "stripe" and two bright protrusions.One such set is highlighted by the white outline in Figure 2b, whereas the stripe and bright protrusions are indicated by a gray line and white arrows, respectively.
Since the stripes observed in the STM images have the dimensions of chrysene, they are attributed to single chrysene molecules originating from the DBCh monomers.The bright protrusions, on the other hand, can be either i) Br atoms still attached to the chrysene units or ii) native Ag adatoms coming from the Ag(111) substrate.To determine the nature of the bright protrusions, we consider the following: i) the distance between the bright protrusions and the nearest neighbor molecules (d 1 = 4.9 ± 0.4 Å, Figure 2b) is too large for noncovalent Br⋯H interactions if the bright protrusions were Br atoms (typical H-bond bond lengths are between 1.5 and 3.5 Å) [34,35] and ii) the molecules are decorated on their edges by dim protrusions (light blue arrow in the Figure 2b), which cannot be accounted for if the bright protrusions are Br atoms.Therefore, we conclude that DBCh is debrominated upon adsorption on Ag(111) at RT and attribute the bright protrusions to native Ag adatoms.Taken together, the principal building block of this network is thus an organometallic monomer.This is a surprising result, as typically dehalogenated monomers react to either form covalently coupled or organometallic polymers. [6,8]So far, the formation of organometallic monomers has been described previously in the literature and it has been suggested that their formation sensitively depends on the applied heat treatment. [22,36]However, the formation of well-ordered 2D self-assembled networks with the organometallic monomer as a building block is uncommon and rarely reported. [22,37][43][44] The molecular islands contain organometallic monomers of both chiralities, whereas monomers of one chirality tend to align in rows (Figure S3, Supporting Information).In Figure 2b two monomers with opposite chirality are encircled in pink and green, respectively.The chirality of the monomers can be discerned from the STM images by identifying the protrusions, which line up asymmetrically with respect to the chrysene unit.An additional close-up STM image where the chirality of the monomers can be discerned is provided in Figure S4 (Supporting Information).
Although the organometallic monomers, like DBCh, are chiral on the surface (thus creating local chirality), the networks consist of an equal mixture of both enantiomers and are thus racemic and no overall chirality is present. [45] tentative molecular model for the network is presented in Figure 2c.The molecules are arranged in a hexagonal network with unit cell parameters a = b = 1.29 ± 0.04 nm and  = 60 ± 1°.Interestingly, the Ag adatoms associated with one monomer align with the [110]direction of the substrate and the distance between them is 8.3 ± 0.2 Å, ≈3 lattice spacings.

Phase ii: Organometallic Dimers
An overview image of phase ii is presented in Figure 2d.Phase ii exhibits long-range order and (at this scale) appears as a series of lines stacked both laterally and longitudinally.Close-up images of this network are presented in Figure 2e and Figure S5 (Supporting Information).The elementary building block of this network appears as a set of two stripes and three bright protrusions, one of which is highlighted with a white outline in Figure 2e.Analogous to the reasoning presented above for phase i, the stripes have the dimensions of chrysene units and are therefore assigned to single chrysene units.The bright protrusions (marked by the three white arrows in Figure 2e) may again be attributed to either Br still attached to the chrysene unit or Ag adatoms from the substrate.Following similar reasoning, as we presented above for phase i, the outer two protrusions (marked by the outer two white arrows) are again too far from each other (d 2 = 4.3 ± 0.3 Å, Figure 2e) for noncovalent Br⋯H interactions to occur. [34,35]Furthermore, the angle between the outer protrusions of two neighboring molecules (one such set marked by the green arrows) is not favorable for halogen bonding. [46]In addition, the presence of dim protrusions decorating the exterior of the building blocks (one marked by the light blue arrow) suggests that C─Br bond cleavage occurred.Thus, the outer two bright protrusions are attributed to Ag adatoms.A minority of the monomers exhibits divergent contrast at the position of the exterior protrusion, as marked by the yellow circle in Figure 2e.This contrast may be due to a Br atom still attached at this position.[43][44] Likewise, the chrysene-to-chrysene distance within one elementary building block is 9.8 ± 0.4 Å, which results in a distance of 2.4 ± 0.2 Å for a possible C─Ag bond (Figure S6, Supporting Information).This is in very good agreement with values reported for metal-coordinated structures and therefore, the central bright protrusion is also attributed to a Ag atom. [34,42]hus, the elementary building block of phase ii is a metalcoordinated dimer consisting of two chrysene monomers and three Ag atoms.The dimers are all composed of a single type of enantiomer, making the dimers chiral.The 2D dimer islands, in which the dimers assemble, only consist of one type of chirality.Consequently, the dimer islands are chiral.We observed both types of chirality for the dimers/islands, which are presented in Figure S7 (Supporting Information).The chirality of the dimers turns out to be important for the formation of the 2D dimer islands, as dimers with mixed chirality cannot form 2D islands with the same unit cell size due to steric hindrance between the dimers (Figure S7, Supporting Information).The unit cell of the network as determined by STM has dimensions with a = 2.1 ± 0.1 nm, b = 1.3 ± 0.1 nm, and enclosed angle  = 107 ± 2°(marked in blue in Figure 2e).A tentative structural model of the network is presented in Figure 2f.The line connecting two exterior Ag adatoms each belonging to a dimer which are neighbors of each other aligns with the [110] substrate direction and the distance between these two Ag adatoms is 4.3 ± 0.3 Å, ≈1.5 Ag(111) lattice spacings.

Phase iii: Organometallic Trimers
Phase iii appears as sets of three stripes with four bright protrusions, self-assembled into a chevron pattern (Figure 3a).Two such sets are outlined in white in Figure 3a.To gain insight into their lateral arrangement, we first have to identify the structure of a single building block.Analogous to the previous two phases, the stripes' dimensions match those of chrysene molecules and therefore, the stripes are attributed to chrysene units.Similarly, the monomer-to-monomer distance is in agreement with having metal-coordination occur between chrysene units.Therefore, we attribute the central two bright protrusions to Ag adatoms.Interestingly, unlike for the previous two phases where the Ag adatoms always had similar distances from one another, now there is a difference in the distances between the Ag adatoms within one trimer: the distance between two outer Ag adatoms (8.6 ± 0.3 Å) is shorter than between the two inner Ag adatoms (9.8 ± 0.6 Å) (Figure S8, Supporting Information).
Contrary to phases i and ii, the exterior two bright protrusions are at 3.4 ± 0.4 Å from the nearest neighbor molecule and thus, a (weak) hydrogen bond could be formed if the bright protrusions were Br atoms. [34]However, from the observed contrast, which matches with the contrast observed for phases i and ii, we identify the exterior two protrusions with Ag adatoms.The elementary building block of phase iii is thus a trimeric metal-coordinated unit.The trimers consist of chrysene units of both chiralities, i.e., one trimer has either L-R-L (left-right-left) or R-L-R chirality and both types of trimers are found within the same island making phase iii heterochiral.Thus, unlike phases i and ii, for which the elementary building blocks are chiral, the trimers are achiral.In addition, a glide plane exists within the network (Figure S8, Supporting Information).Similar to what was found for phase ii, several trimers appeared with alternate contrast at the terminating bright protrusion (yellow circles in Figure 3a), which we attribute to Br atoms still attached to the chrysene unit.The trimers are decorated by dim protrusions which we attribute to the split-off Br atoms and which are on average 2.7 ± 0.2 Å from the nearest neighbor molecule.These split-off Br atoms act as a glue for forming 2D trimer islands through attractive Br─H interactions with the H atoms of the chrysene units.The unit cell of phase iii is highlighted by the blue rectangle in Figure 3a

Phase iv: Organometallic Polymers and Their Growth
For the last phase observed for deposition on an Ag surface held at RT (phase iv) an overview STM image is presented in Figure 4a.The phase appears as a series of stripes and bright protrusions, which we attribute to chrysene molecules and native Ag adatoms, respectively (Figure 4b).Thus, phase iv is an organometallic polymer.The organometallic polymer consists of both left-and righthanded enantiomers.Similar to the other phases discussed so far, the exterior of the polymers is decorated by dim protrusions which we attribute to split-off Br atoms (see also Figure S9 Supporting Information).The distance between the Br atoms and the nearest molecule is 2.7 ± 0.3 Å and thus, fits well with values reported for noncovalent Br⋯H interactions.][43][44] However, that does not necessarily result in well-ordered islands.Instead, the polymers have different lengths and can align in different directions.
Upon annealing the sample at 363 K, phase iv is exclusively obtained (Figure 4c,d).However, in comparison to the organometallic polymers obtained at room temperature, those that are obtained upon annealing are longer and align parallel to form 2D islands.Similarly, depositing on a substrate held at 373 K during deposition (Figure 4e,f) exclusively yields phase iv as well.This sample preparation also leads to an increase in island size compared to postdeposition annealing at 363 K (vide infra).The organometallic polymers obtained upon deposition at both room temperature and annealing to 363 K and deposition at 373 K are presented at the same scale in Figure S10 (Supporting Information).
In addition to the parallel arrangement of the polymers, they are also related to each other by an offset in the longitudinal direction, i.e., a chrysene unit of one polymer aligns (in the direction perpendicular to the polymers) with the Ag adatoms of the neighboring polymers and vice versa (Figure S9, Supporting Information).This relation is also reflected in the Low-Energy Electron Diffraction (LEED) data obtained for such samples (Figure S11, Supporting Information), which indicates long-range order in 2D.From the LEED and STM data, we determined that the polymers grow at an angle of 17 ± 2°with respect to the [110] substrate direction and have an oblique unit cell with a = 1.91 ± 0.07 nm, b = 1.60 ± 0.08 nm and  = 109 ± 3°.The unit cell of phase iv is presented in Figure 4g.
We observed that for the heat-treated samples on average (82 ± 6)% of the chrysene─Ag─chrysene bonds occurred between chrysene units of opposite chirality.That is significantly more than the expected 50% if the polymer was stochastically formed.Thus, there is a clear preference for heterochiral coupling during the formation of the polymer.The statistical analysis for the probability of heterochiral coupling was performed for several sample preparation conditions (Figure S12, Supporting Information).The origin of this preference for heterochiral coupling will be discussed later.
Depositing on samples held at temperatures between 363 and 423 K also exclusively leads to the formation of phase iv (Figure S13, Supporting Information).Upon deposition on samples held at temperatures ≥423 K, the sample quality starts to deteriorate (Figure S13, Supporting Information): the island size as well as the parallel alignment deteriorates and the molecules start to dissociate/undergo uncontrolled covalent bond formation.In particular, at temperatures ≥533 K covalent bond formation became more likely (Figure S14, Supporting Information).However, such structures lack long range, in contrast to our observations on Au(111). [27]hen the sample was held at elevated temperatures during deposition, we additionally found a dependence of the molecular coverage on the molecular deposition flux with lower flux resulting in lower molecular coverage while keeping the nominal thickness according to the simultaneous quartz microbalance measurements the same (Figure S15, Supporting Information shows the example for keeping the sample at a temperature of 423 K during deposition).This indicates that during the formation of phase iv molecules may desorb from the substrate.[49] Thus, the following scenario can be deduced: there is a certain nonzero probability that individual molecules will desorb once they are deposited on the surface.As the molecules land on the surface, only a limited amount of time is available for them to diffuse and meet another molecule for undergoing noncovalent interactions with other molecules.The noncovalent bond formation is most likely the requirement that no desorption occurs.In the low flux case, the surface molecular density at any time during the deposition process is lower than the surface molecular density at the same time in the high flux case.Consequently, the molecules in the low flux case require on average a longer time to diffuse over the surface before bond formation and nucleation can happen.Therefore, at lower molecular flux molecules are more likely to desorb before nucleating, leading to an overall lower molecular coverage.
Finally, to gain further insight into the growth of the organometallic polymers, we analyzed the distribution of polymer lengths, as observed with STM, in dependence on sample preparation (see Experimental Section).Figure 4i summarizes the polymer length distribution in dependence of the substrate temperature during deposition.Additional length distributions are provided in Figure S16 (Supporting Information).While polymer length hardly exceeds 15 nm when depositing on a sample kept at 306 K, we observed polymer lengths >50 nm when the substrate was heated to ≈373-393 K during deposition.In general, keeping the Ag substrate at moderate temperatures above RT during deposition let the average length of the polymers increase with the distribution becoming more shifted to the right, i.e., on average the polymers increased in length and outliers became more common.When increasing the sample temperature to 423 K upon deposition, the distribution became narrower and the average length of the polymers decreased.This is in line with the observation that sample quality started to deteriorate at this temperature (vide supra).Similar results were obtained for postdeposition annealing at 363 and 423 K (Figure 4k; Figure S16, Supporting Information), although the length distribution for these samples is less broad.
The origin of the broadening of the distribution upon deposition on substrates held at 373-393 K may be rationalized as follows.The polymers on these samples exist in two distinct environments: isolated from other polymers (the environment of the polymer is mostly bare Ag surface) and inside a well-ordered 2D polymer island.Comparing the lengths of the isolated polymers versus the ones inside the 2D islands (see Experimental Section), we observed that the ones inside the 2D islands grew larger (Figure 4j).The broadening of the length distribution of all chains on the sample can then be rationalized by the presence of these longer polymers that appear inside the 2D islands, which cause a long tail in the distribution.
To explain the increased length of the polymers inside the 2D islands we propose the following mechanism: the split-off Br atoms enable the parallel alignment of the polymers via attractive and energetically favorable Br⋯H interactions.That means the growth next to or inside one of the 2D islands is energetically favored over isolated growth.Thus, during polymer formation and while the molecules still diffuse over the sample surface they are more likely to connect to polymers arranged in 2D islands instead of joining isolated polymers.From the long-range order of both the polymers and their 2D islands, we can deduce that, under reaction conditions, the monomers' surface mobility has to be high and their reactivity low. [50,51]Thus, the monomers can explore the potential energy landscape before making a C─Ag─C bond.Taken together, the Br-mediated growth of the well-ordered 2D islands comprises the following two important steps: i) DBCh monomers are more likely to couple to organometallic polymers already arranged in islands due to the favorable Br⋯H interactions and ii) since the polymers attach parallel to each other they are less likely to hinder 1D growth of neighboring polymers.The co-adsorbed Br is thus an important factor in the growth of the polymers as well as the 2D islands, enabling a Br-enhanced growth mode.
Lastly, we compare the polymer length distribution for samples obtained by deposition on a hot substrate versus deposition on a substrate held at room temperature followed by annealing (Figure 4k).Interestingly, post-deposition annealing leads to a narrower distribution of polymer lengths and shorter polymers on average.Most noticeably, the long tail of the distribution, which we observed for samples prepared by deposition on a hot substrate, is missing.This can be explained by the lower prevalence of the polymer islands on the samples annealed after deposition, i.e., the reaction conditions for the two different temperature treatments are not the same and thus, are the key for clarifying the differences.For the samples annealed after deposition, all the molecular precursors are available at the start of the reaction, whereas for the samples prepared by deposition on a hot substrate, the precursors are gradually introduced through a constant molecular flux (via the evaporator).54] Figure 5. a-d) Schematic depiction of the Ag adatom positions with respect to the underlying Ag(111) substrate for phases i-iv.The Ag adatoms associated with the organometallic networks are depicted in yellow, whereas the substrate atoms are depicted in gray.d) A single homochiral coupling is depicted in blue.An atom on an unfavorable adsorption site as a result of only heterochiral coupling is depicted in red.

Formation of Phases i-iv
At this point, it is interesting to discuss the formation of the different organometallic structures that were observed in our experiments.To better understand the formation mechanism of the organometallic polymers (phase iv), comprehension of the following three aspects is essential: i) formation of the organometallic monomers, dimers, and trimers, ii) their respective selfassemblies (phases i-iii) and iii) their conversion into phase iv upon heat-treatment.
Concerning the first aspect, the formation of the organometallic monomers, dimers, and trimers only occurred when the substrate was held at room temperature during deposition.The organometallic polymers (phase iv) co-existed with the monomers, dimers, and trimers but they represented a minority phase.In addition, the polymers observed at room temperature were much less ordered compared to the polymers obtained upon heat treatment.This is an indication that the C─Ag bond is not reversible at room temperature, as reversible bonding to lead to more ordered structures. [50,51]The most likely irreversible nature of the C─Ag bond (at room temperature) makes the Ag atoms act as "end-caps," preventing the organometallic monomers, dimers, and trimers from coalescing into organometallic polymers, explaining their formation on samples that have not been subjected to heat-treatment.It should be noted that the monomers, dimers, and trimers require more adatoms for their formation than the polymers, and therefore the growth of the smaller species is not due to a lack of adatom availability.
With respect to the second aspect, from the absence of isolated organometallic monomers, dimers, and trimers on the surface we can deduce that phases i-iii are formed by diffusion of the organometallic monomers, dimers, and trimers on the surface.The self-assembly of such organometallic building blocks has been reported previously. [15,16,22,55]In addition, the organometallic monomers, dimers, and trimers do not have functional groups to guide their self-assembly, and the principal interaction between the organometallic monomers, dimers, and trimers is mediated by the co-adsorbed Br atoms.44] To gain insight into the formation of these self-assemblies, we schematically depicted the adsorption of the Ag adatoms on the Ag(111) surface for each of these phases (Figure 5).Since isolated Ag adatoms preferentially occupy the hollow site, [56] we placed the first Ag adatom on a hollow site and constructed the position of the other atoms according to the unit cell parameters.It appears that the Ag adatoms preferentially adsorb on a hollow or bridge site of the Ag(111) substrate.Several of our experimental results corroborate the existence of a preferential adsorption site: i) in phase i, the distance between two Ag adatoms is ≈3 lattice spacings (along a principal Ag direction).This distance is too short for a relaxed chrysene with two Ag atoms attached to it.Therefore, in order to accommodate this, the structure has to be strained, signifying that there is a preference for adsorption sites.ii) Similarly, for phase ii, we found a distance of 1.5 lattice spacings along the [110] substrate direction between two Ag adatoms each terminating a neighboring dimer, and each most likely adsorbing on a bridge site, iii) for phase iii we observed a modulation in Ag─Ag distance within each trimer, which also indicates a preferred adsorption site for each Ag adatom.We conclude that phases i-iii are stabilized by i) the co-adsorbed Br atoms providing noncovalent Br⋯H interactions and ii) the preferred adsorption site of the Ag adatoms in these phases.The preference for the adsorption site is such that the structure is strained in order to adapt to the underlying Ag substrate.
Finally, upon annealing, all three phases i-iii were converted into organometallic polymers (phase iv).To do this, the C─Ag bond of the Ag-terminated monomers, dimers, and trimers has to be broken and re-established, i.e., the C─Ag bond must be reversible.Moreover, as noted above, the organometallic polymers are predominantly heterochiral, whereas phase ii consists of homochiral dimers.Therefore, not only does the exterior C─Ag bond have to be reversible, but also the internal C─Ag bonds have to be broken and re-established because otherwise chiral phase ii cannot be converted into the predominantly heterochiral phase iv.Therefore, the formation of phases i-iii occurs due to the irreversibility of the C─Ag bond formation at room temperature, and they represent kinetically trapped phases, whereas upon annealing, the C─Ag bond formation becomes reversible and phase iv can form.
Phase iv also appears to preferentially accommodate Ag adatoms on a hollow site (Figure 5d).This adsorption site preference is a clear indication of the origin of the observed preference for heterochiral coupling (vide supra).Specifically, an exclusively heterochiral polymer eventually causes the adsorption of Ag adatoms on a top site (Figure 5d, red atom).Conversely, a polymer having a few homochiral couplings interspersed can accommodate all Ag adatoms on hollow sites (Figure 5d, blue atom) and consequently, should be from an energetical point of view favored.Thus, the observed preference for heterochiral coupling can be attributed to the preferential adsorption sites of the Ag adatoms.

Conclusion
We studied the formation of organometallic polymers from DBCh with native Ag adatoms on Ag(111) in dependence on the substrate temperature either during or after deposition.At room temperature, DBCh got dehalogenated and a delicate energy balance enabled the formation of three different organometallic reaction intermediates (monomers, dimers, and trimers).The reaction intermediates self-assembled intoachiral (for the monomers, trimers and polymers) and chiral (for the dimers) 2D islands and their experimental characterization provided valuable insight into the formation of the organometallic polymers.We could reveal that the reversibility of the C─Ag bond formation is a vital element in the formation of the organometallic polymer.Organometallic monomers, dimers, and trimers formed at room temperature because of the irreversibility of the C─Ag bond: the Ag adatoms act as an "end-cap" to polymer growth.By observing the amount of homo-and hetero coupling before and after the transformation of the reaction intermediates into the organometallic polymer, we find direct experimental evidence for the reversibility of the C─Ag-C bonds.Both the polymers and the organometallic reaction intermediates adapted to the surface in such a way as to accommodate the Ag adatoms on preferred adsorption sites on the substrate.The polymer's growth is path-dependent: where postdeposition annealing led to the formation of relatively short polymers, deposition on a substrate held at temperatures > room temperature unlocked a Br-enhanced growth mode which yielded more extended polymers.In addition, we identified a clear preference for heterochiral coupling for the organometallic polymers which we suggest is due to preferential adsorption sites of the Ag adatoms.This opens a pathway to steer the chiral selectivity of on-surface reactions.
We anticipate that these results will contribute to a greater understanding of the on-surface Ullmann-type coupling reaction, which is important for the rational design of technologically relevant 1 and 2D surface-adsorbed structures.Furthermore, our results can serve as valuable input for theoretical modeling to determine the energetical path from starting monomers to final reaction product(s).

Experimental Section
Sample Preparation: All experiments including sample preparation were conducted under ultra-high vacuum conditions.The Ag(111) single crystal was cleaned with repeated Ar + sputtering and annealing cycles.The monomer DBCh (commercially available, Sigma-Aldrich), was thermally evaporated from a four-pocket Knudsen cell evaporator (Omnivac) at 440 K.During the deposition, the sample was held at a constant temperature (room temperature or heated through resistive heating).The molecular flux during deposition was measured with a quartz crystal microbalance.
Experimental Techniques: STM imaging was performed in constant current mode with a commercial low-temperature STM (Scienta Omicron) operated at 77.8 K.All STM images were processed using the WSxM software package. [57]LEED patterns were simulated with the LEEDpat software package. [58]olymer Length Statistics: Polymer lengths were determined from STM images.Polymer lengths were measured on 4-12 images.Image sizes ranged from 30 × 30 to 100 × 100 nm 2 .The obtained histograms of polymer lengths were fitted with a log-normal distribution.For ease of comparison between samples, for each sample, the relative frequency was calculated by dividing the number of counts for a given polymer length by the total number of polymers.The number of polymers per sample was 432 (deposited at 293K), 375 (deposited at 373 K), 390 (deposited at 393 K), 479 (deposited at 423 K), 252 (annealed at 363 K), and 396 (annealed at 423 K).The statistical analysis was performed using OriginPro. T distinguish the two environments (2D islands and isolated polymers), the 2D islands were defined as at least ten polymers stacked laterally.To consider only the length of the polymers surrounded by other polymers, the outermost 5 rows of each 2D island were disregarded.

Figure 1 .
Figure 1.a) Schematic of the two on-surface enantiomers of 6,12dibromochrysene (DBCh).b) Overview STM image of the molecular structures obtained upon deposition of DBCh on an Ag(111) sample held at room temperature.Four molecular phases are indicated by the white outlines and labeled as i-iv.The set of three white lines (top right) indicates the Ag(111) principal directions.Scanning parameters: V bias = −1.5 V, I set = 20 pA.

Figure 2 .
Figure 2. a,b,d,e) Overview and close-up STM images of phase i a,b) and phase ii d,e), respectively.In (b,e), the primitive building blocks of each of the two phases are outlined in white, the bright protrusions are marked by the white arrows, the dim protrusions are marked by the light blue arrows, and one molecule with divergent contrast is marked by the yellow circle.c,f) Tentative structural models of phases i and ii, respectively.The red arrows indicate the unit cell vectors and the black arrow indicates an Ag(111) principal direction.For phase i, the unit cell is indicated in blue in (b) and (c), whereas for phase ii it is indicated in (e) and (f).Scanning parameters: a) V bias = −1.5 V, I set = 100 pA b) V bias = −1.0V, I set = 20 pA d) V bias = −1.5 V, I set = 20 pA e) V bias = −2.0V, I set = 200 pA.c,f) Color scheme: C, gray; H, white; Br, red; Ag, cyan.

Figure 3 .
Figure 3. a) Close-up STM image of phase iii.Two trimer building blocks of the network are outlined in white, the bright protrusions are marked by the white arrows and several molecules with divergent contrast are outlined by the yellow circles.Scanning parameters: V bias = −1.5 V, I set = 200 pA.b) Tentative structural model of phase iii.The unit cell vectors are depicted by red arrows.The unit cell is outlined in blue in both (a) and (b).The Ag(111) substrate directions are indicated by black arrows and one building block of the network is outlined in black.Color scheme: C, gray; H, white; Br, red; Ag, cyan.
. A tentative structural model of this phase is presented in Figure 3b.The unit cell parameters of this network are a = 1.6 ± 0.1 nm, b = 4.3 ± 0.2 nm and  = 91 ± 3°.The two unit cell vectors align with the [ 110] and [112] principal Ag directions, respectively.

Figure 4 .
Figure 4. a,b) Overview and close-up STM images of phase iv obtained upon deposition of DBCh on Ag(111) held at RT. c,d,e,f,h) Overview (c,e,g) and close-up (d,f) STM images of phase iv upon heat treatment of the samples.The images were obtained on samples annealed at 363K c,d), heated at 373K during deposition e,f) and annealed at 423K h).g) Tentative structural model of phase iv.The unit cell vectors are depicted by the red arrows in the image.One principal Ag(111) direction is depicted by the black arrow.i) Length distribution of the polymers formed depending on sample treatment.The symbols indicate the observed (relative) frequency of a certain polymer length, whereas the solid lines of the same color indicate a log-normal fit of the data.j) Histograms displaying the length distribution of the polymers of sample (e) in dependence on their location (inside or outside a molecular island).k) Comparison of length distributions obtained by annealing versus depositing on a hot substrate.Scanning parameters: a) V bias = −1.5 V, I set = 100 pA, b) V bias = −1.5 V, I set = 200 pA, c) V bias = −2.0V, I set = 20 pA, d) V bias = 1.0 V, I set = 200 pA, e,h) V bias = −1.5 V, I set = 20 pA, f) V bias = −1.0V, I set = 200 pA.g) Color scheme: C, gray; H, white; Br, red; Ag, cyan.