Surface-mediated assembly, polymerization and degradation of thiophene-based monomers

Temperature mapping of the different molecular phases of tribromoterthienobenzene on (111) coinage metals.


TBTTB on Au(111) -additional STM and
identification of the molecules can be obtained by assigning the chiral character to every molecule on the basis of its appearance in STM images, as shown in Figure S2a, which allow us to identify an ordered racemic mixture of the elf-assembled layer. To confirm this point we performed gas-phase DFT calculations as discussed in the main text, which show that the racemic mixtures are more stable, while pure enantiomeric ones are more energetic. In Figure S1 we identify all the possible interactions. By examining their symmetry, we can reduce the number of possible intra-row configurations to two, one relative to an enantiopure row, labeled P S or P R depending of the enantiomer, and the other relative to racemic composition M ( Figure S1a). The possible inter-row interactions depend on the composition of the rows and the shift between them ( Figure S1a).
Due to symmetry considerations, there are thus four non-degenerate configurations for the self-assembled structure of TBTTB, based on the composition of the single rows. In Figure S1b, we report the highest cohesive energy values for each of the calculated combinations, showing that the phases with enantiopure S2 rows are the less stable. In addition to being more stable, the MM phase also matches the enantiomer recognition from the STM images, hence this structure is reported in Figure 1 of the MS. Annealing the RT phase reduces the long-range ordering of the molecules, and above 100 °C two shortrange repeating structures (shown in Figure S2c

S3
The appearance of the Br-Au peak proceeds in conjunction with the reduction of the Br 3p doublet peak at 183.7 eV and 190.2 eV assigned to Br-C.
This observation, together with the slight shift of C 1s (from 284.8 to 284.6 eV at 100 °C), suggests the simultaneous presence of dehalogenated molecules forming short OM chains and intact molecules, with a pattern ( Figure S2d) that has changed from the close-packed structure obtained at RT ( Figure S2a Figure S15). At sub-monolayer coverages, the Au herringbone reconstruction is completely twisted, with the network present only on the enlarged FCC portion ( Figure S3a, b). Based on previous reports, 2-5 this is due to the presence of the electronegative bromine atoms, which lifts the Au surface reconstruction, so that the Au atoms release the stress.

Dosed @ 200 °C
When dosed on Au(111) at 200 °C, TBTTB forms an ordered hexagonal network, as explained in the main text. This structure is identified as an OM structure from both XPS and STM data.

Hot-dosed vs Annealed -C1s peak
Dosing TBTTB on pre-heated metal surfaces yield highly ordered structures with respect to dosing at RT and subsequently annealing, as discussed in the MS. While STM suggest that in both case we have OM structures, and the difference therefore relies on kinetic factors, we also used XPS to confirm our hypothesis. Figure S5 shows the comparison between the C1s peaks when hot-dosing or dosing+annealing, for Au. While the resolution is lower for the hot-dosed, due to both a lower integrating time and to a lower quantity of molecules (due to the lower density of the ordered phase), it is still possible to compare the spectra. . The different signal-to-noise ratio of the red curve in a is due to a shorted acquisition time, while the larger FWHM of the red curve in b is due to a higher pass energy used (20eV instead of 15eV).

Organometallic vs polymer network size
In Figure S5 the DFT calculated distances of both an OM and a polymeric structure are reported. The experimentally measured distance is equal to 1.1±0.1 nm (Figure 4c), which is in agreement with the calculated OM distance of 1.11 nm ( Figure S5a).  When dosed on Cu, the thiophene rings of the TBTTB molecule are already partially opened at RT and only S-Cu bonds are observed at 420 °C. When deposited on Au and Ag, the molecules are intact at RT, and S-M (sulfur-metal) bonds are observed only when the surface is heated at 300 °C. While on these surfaces the degradation is small in percentage, each molecule has three thiophene rings, and therefore a mere 10% of S-M bonds can affect up to 30% of the molecules, hence drastically reducing the structural order. This data shows that bromine atoms undergo chemical shifts, due to making bonds with Au atoms, as described in the MS, before they entirely desorb from the Au surface. On Ag and Cu, the chemical status of bromine atoms remains the same as the adsorption upon dehalogenation, throughout the annealing steps. At 420 °C no Br atoms remain on the Au surface, while only a partial desorption is undergone on Ag and Cu ( Figure S8).

Additional XPS data and temperature mapping of the chemical state of TBTTB
Putting together all data presented in the MS and recalled here in Figures S7 and S8 we can produce a map of the chemical state of the monomer's thiophene rings at each studied temperature, which is reported in Figure 9 in the main text, in the TOC and in Table S2. However, from Figure S7 we do observe a consistent molecular desorption, and we therefore include in Figures S9 the same composition vs temperature map but accounting for the desorption. In both cases, from our data it is not possible to discern between a completely broken molecule and several with only one broken thiophene. Moreover, the molecules could be in a mixed state, i.e. one lobe could be in an OM state, another in a polymeric one. Therefore, the composition vs temperature maps are referred to single thiophene rings, and not to TTB monomers. The amount of intact molecules at each temperature is obtained from the percentage of de-halogenated molecules, while the amount of broken thiophenes from the percentage of S-Au. The S9 amount of OM and polymers instead has been obtained by analyzing the shift of the C 1s peak, and the line profile distances of the STM images, which showed the absence of polymers at RT and the absence of OM bonds at 400°C for all surfaces. Table S2. For each sample, the monomer's thiophene rings chemical state percentage is reported at different temperatures. Figure S9. Fraction of monomers in each observed chemical state for the three studied surfaces at each temperature; differently from Figure 9 in the main text, these graphs take into account molecular desorption.

OM network chirality
The enantiomer identification of the TBTTB molecule was obtained by processing the STM images. In the case of self-assembly of intact molecules on Au, we assigned the chiral character to every molecule on the basis of its appearance in STM images. For the OM network, the molecules are dehalogenated and interact more strongly with the surface. Hence, the overall appearance in STM images can be related to the lattice site where the molecules are located; therefore, we used different procedures to identify the enantiomers. We applied three different identification methods to identify the features. An application of the three approaches on a case image is reported in Figure S8.
For the first method, we focused only on complete hexagons, on which we imposed triangles on top of every hexagon side, with the same orientation (with the base of the triangle parallel to hexagon side). We then proceeded to rotate these triangles to match the STM appearance and reported the shift as an angle.
However, this method allows recognition of closed polygons only, as the open pores lack the necessary symmetry.
The second approach focuses rather on the C-Ag bonds. Every molecule is in fact bound to three silver adatoms belonging to a network. We therefore built a triangle between the Ag atoms around every molecule, then imposed a triangle on top of the molecule, again trying to match the appearance in STM.
Finally, we evaluated the shift between the two triangles. However, this method allows recognition of molecules which belong to the OM networks only and does not work for single molecules.
Finally, we tried to identify the molecules by STM appearance only, as done in the Au case, and we compared the result with the other two methods. The results are quite satisfactory, and in 97% of the cases the identification was the same between the third and the first two methods. We therefore decided to use this last approach, more convenient to apply and valid for all the visible molecules.
Applying this method to all the different cases studied (TBTTB on Au and Ag at different temperatures) states that while we always observed a racemic mixture, the enantiomers where arranged in an ordered phase only in the case of intact molecules (TBTTB on Au(111) at RT), and they were randomly mixed in all the other cases (OM and polymers). S12 Figure S10. 15 15 nm 2 STM image of TBTTB on Ag(111). In all the three cases, we attempted to identify the chirality of the × molecules, using different approaches described above. The graphs in a, b report the angular differences evaluated in the relative figures. In both cases, it can be noted that two peaks are present, meaning that we have a mixture of both the enantiomers. In the last case (c) two zoom-in of the image (1.5 1.5 nm 2 ) are reported. × S13

STM unit cells for DFT calculation
By calibrating the images on both Ag and Au, it was possible to estimate both the pore-to-pore distance between two adjacent hexagons, and the molecule-to-molecule distance (which is taken as each side of the hexagon). Because of the intrinsic defected structure of our network (due to the presence of a racemic mixture), the dimension was measured as an average of the distance across several hexagons (results are shown in Figure S7b and Figure S8). The experimental measured distance is similar in both cases, and is equal to 2.0±0.1 nm for Ag(111), and 1.9±0.1 nm for Au(111). The DFT structure was also built starting from these analyses. The superimposition of the unit cell used for DFT simulation with an experimental image is shown in Figure S9c. This adsorption site and orientation of the TBTTB molecule was used to calculate the OM and polymer structures ( Figure 6, S12 and S13).

DFT calculation of TBTTB polymers on Ag(111)
The 5-layer Ag(111) slab for simulating the polymer contained 135 Ag atoms, with the dimension of a=15.35 Å, b=15.35 Å, c=30.00 Å, and the angle between a and b vectors=60 degrees. The results for both polymers made by S-only or a mixture of S-and R-enantiomers are shown in Figure S13. Although S16 our XPS data confirms the formation of a polymer, the simulation of ordered polymer was aimed to identify the difference between the dimensions of OM hexagons from the polymeric structures.

Study of order
STM images can show the overall appearance of the obtained phases, yet they do not yield a measure of the order of each phase. While we can easily infer (by visual inspection) that the hot dosed phase shows a more ordered pattern than the RT one, we wanted to use a standardized way to measure the deviation with respect to a perfect hexagonal structure, to quantify the improvement, if any, arising from the procedure.
To this end, we used the procedure successfully exploited by Ourdjini et al. 7 In their work, they statistically analyzed the network properties of the minimum spanning tree (MST) of the network created by taking the closed cells in the surface as its nodes. This procedure allows to compare the tessellation generated by the 2D polymer with respect to perfectly ordered networks with different primitive geometries. To obtain these statistical data, it is first necessary to process the STM images to detect the different regions enclosed by the molecules, i.e. the different "cells", and identify their centers. As the next step, we constructed an undirected network, determining the neighbours of each node with the aid of a Voronoi tessellation process. As it can be seen in Figure S16(b), we did not include in the network the cells contiguous to the image borders, to avoid the undesired bias that the framing would introduce. Finally, the MST, which will be the subject of the statistical analysis, is found. We seek to extract the area-normalized average (m) and standard deviation (σ) of the distribution of edge lengths in the MST, as it has been shown that these two values can characterize a pattern in terms of their geometrical primitives (i.e. squared, hexagonal, etc.) and its degree of order. 8,9 Figure S17 shows the data for our systems, together with data from Ourdjini et al., 7 where one can see the comparatively high order of our organometallic polymers in a quantitative way, as well as how the hot dosed system offers a higher degree of order.