Molecular Junctions Based on Aromatic Coupling

Molecules investigated between two electrodes usually are designed as rod-like systems with two anchor groups. Single molecules with two anchor groups can be bridged between the electrodes of an MCBJ setup (as already discussed in 2.3.2). Sulfur forms a covalent bond to gold which is more stable than the Au-Au bond.[168] Therefore if a molecule is immobilized in an MCBJ and the electrodes are further bended apart, the gold atoms migrate towards the tips until the junction breaks. For molecules with only one anchor group rather than two, one would not expect a stable metal-molecule-metal junction because the molecule can only attach to one side of the junction (B in Figure 48). However, this assumption ignores intermolecular interactions and, as we shall see, these interactions can be relevant to the behavior of molecular junctions. We investigate the formation of metal-molecule-molecule- metal junctions consisting of two molecules, each binding to one of the electrodes.[59]

Figure 48: Molecules with two anchoring groups form metal-molecule-metal junctions (A). Molecular rods lacking a second anchor group are not expected to form a stable junction (B). However, we investigate whether it is possible to connect two electrodes with two molecules both fixed to one of the electrodes. It is postulated that intermolecular interactions such as π-π stacking might be the origin of a stable junction formation (C).

Here we report MCBJ investigations on molecules with anchor groups on both ends, and the situation when one anchor group is removed. The influence of aromatic π-π coupling between adjacent molecules on the formation of molecular junctions will be discussed.

2.3.3.1 Molecular Signals of OPEs with One Anchor Group

The rigid OPE systems are considered as molecular rods. While keeping a thiol anchor group on one side, we systematically vary the group on the other side of the rod. As a reference molecule we look at molecule 1 with two thiol anchor groups, one on each side, providing strong binding to both electrodes. This molecule does form single molecule junctions (chapter 2.3.1). In the OPE 2 one thiol anchor group is replaced by a pyridine unit in the para-position. Pyridine can form a coordination bond to gold, and one can still expect that OPE 2 forms single molecule junctions, although less stable with a lower conductance (chapter 2.3.2). In OPE 4 the position of the pyridine unit is changed to the ortho-position and therefore the pyridine anchor group is hidden. Providing very comparable electronic properties as in OPE 2, the pyridine can hardly form a bond to gold due to its position. As a terminal atom of 4 the hydrogen atom of the terminal phenyl ring Ht has to be considered. In the

molecule 5 the pyridine unit is completely removed, leaving a bare phenyl unit. Also here the opposite group to the thiol anchor group is a hydrogen atom Ht. No

formation of a single metal-molecule-metal junction is expected for molecule 5 as it has only one anchor group on one side of the rod. As another control molecule the OPE comprising one thiol unit but only consisting of two phenyl rings 9 was designed.

Molecule Length (Å) No. of samples Conductance (G0)

1 20.7 (S-S) 4 (1.2 ± 0.1)·10-4

2 18.7 (S-N) 3 (5.7 ± 2.4)·10-5

4 19.8 (S-Ht) 3 (6.6 ± 1.3)·10-6

5 19.8 (S-Ht) 3 (5.9 ± 2.4)·10-6

8 12.9 (S-Ht) 4

Table 2: The molecule 1, 2, 4, 5 and 8 were synthesized in their acetyl protected form (R = COCH3) In

situ deprotection leads to the free thiols which then bind to the gold electrodes (R = Au). Ht denotes

the terminal hydrogen atom. The length of each molecule is obtained after energy minimization using the MM2 force field (ChemDraw 3D). The single-molecule conductance value G was deduced from the peak that appears in the log(G) histograms of 100 opening G(z) curves obtained from three to four different samples.

The results of the corresponding conductance measurements are shown in Figure 49. It is evident that for all the molecules 1 – 4 a pronounced conductance peak is obtained. It is surprising that both the height and the width of these peaks are similar for molecules 1 – 4. The conductance value of molecule 2 is slightly reduced by approximately a factor of 2-3 as compared to the reference OPE 1. This smaller value is due to weaker electronic coupling between nitrogen and Au atoms as discussed in section 2.3.3. Although we expect a peak in log(G) for molecule 2, a similar peak of equal magnitude for molecule 4 comes as a surprise. Because the nitrogen atom of the pyridine structure in molecule 4 is “hidden”, it may provide, if at all, a much less probable binding site to the electrode.

Figure 49: Conductance histograms for the different molecules. Molecule 1 and 2 have two anchor groups, whereas molecule 4, 5 and 8 only comprise one anchor group. The histograms are formed from 100 G(z) curves and the blue solid lines are the Gaussian fit curves.

In contrast to this, the identical magnitude and width of the peak in log(G) suggest that molecule 4 also binds in the junction with a similar probability as molecules 1 and 2. Hence, the observed strong binding must have another origin. This is further supported by the measurements for molecule 5. Although the anchor group at one

end is now removed completely, we obtain very similar results for 3 and 4. These results lead to the question why molecular junctions form, even with molecular rods having only a single linker group on one side of the rod.

We believe that the connection between the electrodes is made possible by a π-π stacking interaction between a pair of molecules.[59, 169-171] If one molecule is anchored by means of its thiol linker group to, for example, the left electrode, another one bound to the right electrode can complete the mechanical assembly of the junction through π-π-coupling through the phenyl rings. This interpretation is supported by the shift of the conductance peak to lower values by more than an order of magnitude. In this picture, a reduced G value is expected, because the pair of molecules will be longer than a single dithiol molecule anchored between gold electrodes. We observe that the junctions form with a similar probability, whether for dithiolated compounds or monothiolated compounds. As the strength of π-π stacking depends strongly on the conjugation extent of the system and the overlap between adjacent compounds, one straightforward control experiment, without modifying the electronic structure of the molecules to a great extent, is to investigate an OPE compound with only two phenylene units (molecule 8). Although this peak is shallow, single conductance traces are still clearly visible but noisier plateaus at values larger than the plateau values for molecules 4 and 5. We believe that the peak for molecule

8 is less pronounced because of the reduced π-π-interaction for these shorter

molecules, leading to a mechanically less stable junction. A pair of molecules of 8 tends to have a higher conductance value than a pair of molecules of 5 because of the reduced distance the electrons have to tunnel between the gold electrodes through the molecular bridge. We also checked that for molecules where no aromatic stacking is possible, such as in monothiolated alkane chains, we do not observe the formation of molecular junctions. We also emphasize that without deprotecting the thiol function, no molecular signature can be detected. This permits the elimination of other unanticipated interactions except for intermolecular interactions to explain the signal observed for compounds 4, 5 & 8.

The different conductance values obtained for the metal-molecule-molecule- metal junctions compared to the metal-molecule-metal junctions can be qualitatively attributed to the difference in the distance that electrons have to tunnel between the gold electrodes. This distance is determined by the length of a π-π stacked pair of molecules. The stacking of a pair of monothiol molecules 4 is illustrated in Figure

50 A. Figure 50 B-D present different stacking configurations of the same molecule and Figure 50 E shows the molecule with two anchor groups fixed on both sides to the electrodes. Due to electrostatic repulsion of the negative π-electron clouds, the aromatic rings will not sit directly on top of each other, but rather in a staggered configuration.[172] Despite working with diluted solutions, the molecules self-assemble at the surface of the gold electrodes, leading to a high local concentration. We therefore expect the molecules to form coplanar stacks as illustrated in Figure 50 A. Still, there are three possible arrangements, either one pair of phenyl rings stack (D in Figure 50), two pairs of phenyl rings stack (B in Figure 50) or three pairs of phenyl rings stack (C in Figure 50). We believe that Figure 50 B represents the actual stacking configuration of a pair of molecules of compound 5.

Figure 50: Staggered conformation of a molecular junction consisting of two molecules of 5 (A). Simplified schematic representation of different stacking configurations with either two pairs of phenyl rings stacking (B), three pairs of phenyl rings stacking (C) or one pair of phenyl rings stacking (D), respectively. Configuration B is the most probable one whereas the other configurations are believed to be less favored due to steric interactions (C) and reduced conjugation and overlap (D), respectively. The molecule bearing two anchor groups fixed to both electrodes is schematically represented in E. The lengths given are sulfur to sulfur distances calculated by MM2 minimization (ChemDraw 3D).

The structures in Figure 50 C & D are postulated to be less probable because of steric hindrance and because of weaker π-π interactions due to the reduced overlap between the molecules, respectively. The two sulfur atoms in dithiol functionalized OPE 1 are 20.7 Å apart. In the proposed configuration B (in Figure 50), the two sulfur atoms are separated by 29.1 Å. In a simple tunneling picture where electrons tunnel through an effective medium over a distance d, the conductance G can be written as

d Ae

G= −β . The decay constant ß is determined by the electronic parameters of the effective medium (here, the molecules). For a given electrode material (here, gold), the prefactor A depends on the electron density-of-states at the point where the molecules contact the gold electrode. Because this is determined by thiol anchors both in a single molecule and a stacked pair of molecules, this factor can be taken as a constant. The longer tunneling distance of electrons in a junction based on two stacking molecules leads to a 12 times smaller conductance by estimation. This is in reasonable agreement with experiment which shows a 20-fold difference.

π-π Stacking is stronger with a larger difference in electron density of the two stacking phenyl units. Therefore the π-π stacking is enhanced when an electron rich phenyl faces an electron poor phenyl. As pyridine reduces the electron density of its aromatic core, molecule 4 favors the overlap of two phenyl pairs and forces the molecules to stack in a slightly more planar fashion than in pairs of molecule 5. This small shift slightly reduces the tunneling distance leading to a small increase in conductance by a factor of 1.11.

As a further control experiment molecule 6 comprising the pyridine unit as a central ring was designed and synthesized. The pyridine on the central ring disfavors stacking with two pairs of phenyl moieties, but it could still stack with one ring (D in Figure 50). However, no molecular signature was obtained for molecule 6 when the previously described immobilization conditions in an MCBJ were applied. These results further corroborate that the molecular junctions are best described by the schematic representation B in Figure 50.

Figure 51: Molecule 6 comprises a pyridine ring as central unit. π-π Stacking is therefore disfavored. On the other hand OPE 7 comprises a perfluorinated phenyl ring which should enhance the stacking properties.

To enhance stacking properties, OPE 7 with a perfluorinated terminal phenyl ring was synthesized. The fluorine atoms reduce the electron density of the ring which should increase the stacking properties. However, no pronounced peak was obtained for 7. This might be attributed to the high local density of molecule leading to repulsion of the sterically more hindered perfluorinated phenyls when the two electrode tips come close together.

These results were further confirmed by calculations in a systematic first- principle study.[173]

In conclusion, the role of intermolecular interactions in molecular junctions was investigated. Whereas molecular rods commonly bear two anchor groups to provide stable binding to both electrodes, the molecules for these investigations were designed such that molecular junctions were solely formed due to π-π-interactions. The intermolecular π-π stacking interaction between monothiol molecules composed of alternating phenylene and ethynylene units was strong enough to induce the formation of molecular junctions. The conductance of the metal-molecule-molecule- metal junction was smaller than the conductance of reference OPE 1 due to longer electron tunneling distances, but interestingly the junction formation probability and the height and width of the peak in the logarithmic histogram were very comparable to OPE 1. We also showed that π-π stacking can be used as the dominant guiding force for the formation of molecular bridges in few-molecule electronic junctions.

2.3.4 From Single Junctions to Two-Dimensional Gold Nanoparticle