Protonation tuning of quantum interference in azulene-type single-molecule junctions† †Electronic supplementary information (ESI) available: Synthesis and characterization of compounds, details of single-molecule conductance measurement and the calculation. See DOI: 10.1039/c7sc01014a Click here for additional data file.

The protonation of azulene cores offers significant conductance tuning in single-molecule junctions with quantum interference.


Introduction
The coupling of p-systems to each other is essential to their charge transport applications in organic and molecular electronics. [1][2][3][4] Recently, quantum interference effects due to different electron pathways through molecular systems with varied connectivities have been intensively explored via singlemolecule conductance studies of benzene, 5 oligo(phenylene ethynylene), 6 anthraquinone, 7 unsaturated carbon chains, 8 and azulene. 9,10 Destructive quantum interference brings antiresonances within their HOMO-LUMO gaps, which signicantly reduces their single-molecule conductance and leads to new control strategies for molecular-scale devices.
Among those molecules exhibiting destructive quantum interference, azulene derivatives have also been considered as responsive molecular devices and materials, [11][12][13] because the ve-seven ring system of azulene undergoes protonation in response to acid. 14 During protonation, the HOMO of the protonated azulene core with unpaired valence electrons is expected to move towards the Fermi level of the electrodes, which could create signicant electrical conductance changes. 15,16 Since the tuning of quantum interference is one of the major challenges for current molecular electronics, 10,17-20 protonation studies of azulene derivatives may offer a unique opportunity for tuning charge transport through single-molecule junctions.
Here we studied the protonation tuning of charge transport through single-molecule junctions of several azulene derivatives using a mechanically controllable break junction (MCBJ) technique. The single-molecule conductance measurements demonstrate that the azulene derivatives showed signicant conductance changes aer adding triuoroacetic acid (TFA). Furthermore, it is also found that the derivatives exhibiting destructive interference in the neutral state undergo more signicant changes upon protonation. These measured trends can be understood using a parameter free theory to model the molecular junction and compute the electron transmission of azulene derivatives in their neutral and protonated states.

Results and discussion
The azulene derivatives of 1,3Az, 4,7Az and 5,7Az are designed with -SMe anchors with different connectivities to the azulene core, and their synthesis followed the synthetic route reported in ref. 21 and 22. UV-vis measurements were carried out for the azulene molecules dissolved in THF/TMB solution. Aer adding TFA, signicant color changes from aquamarine blue to orange occurred ( Fig. 1b), which is reected in the appearance of the new peak at 540 nm and 745 nm, respectively, suggesting the presence of protonated 1,3Az in solution. The new peaks also appear in the protonation of 4,7Az (536 nm and 718 nm) and 5,7Az (480 nm and 695 nm). The protonation of azulene could lead to the formation of stable tropylium cations, which causes signicantly different properties compared with the neutral species. [23][24][25][26][27] The single-molecule conductance measurements of azulene derivatives were carried out in a solution containing 0.1 mM target molecule in a mixture of THF (tetrahydrofuran) : TMB (mesitylene) ¼ 1 : 4 (v/v) using the MCBJ technique (see ESI † for more details). [28][29][30] To trigger the protonation, 10 mL (13.5 M) TFA was added into a 190 mL 0.1 mM solution of azulene derivatives in the liquid cell of the MCBJ set-up. Fig. 2a shows several typical individual stretching traces from the measurements in pure solvent, with 1,3Az, and with 1,3Az and TFA, respectively. The solvent experiment shows a tunneling decay aer rupture of the gold-gold atomic contact (a plateau at conductance quantum G 0 ), while the measurement of the targeted molecules reveals additional clear molecular plateaus varying from 10 À2 to 10 À5 G 0 , which are assigned to the conductance features of the Au-molecule-Au junctions.
Around 2000 individual traces are used to construct conductance histograms without data selection for further analysis. Fig. 2b and c represent the two-dimensional (2D) conductance histograms of the initial neutral state and protonated state of 1,3Az. It is found that the conductance intensity cloud shis to a higher conductance regime aer protonation, while the relative stretched distance distribution (shown in the inset) showed quite similar lengths. These ndings suggest that the conductance changes come from the enhanced charge transport through protonated azulene cores rather than from the conguration changes. Fig. 2d shows the 1D conductance histograms of 1,3Az before and aer adding TFA. There is a pronounced peak located at 10 À3.8 G 0 for the neutral 1,3Az. Control experiments of 1,3Az with different concentrations suggested that the conductance feature comes from the single-molecule junctions and that the peak position is independent of the concentration in solution (see Fig. S2-5 in ESI †). Compared with the conductance of the initial neutral state, the conductance of 1,3Az in the nal state increased by more than one order of magnitude from 10 À3.8 G 0 to 10 À2.7 G 0 . The conductance of protonated 1,3Az is similar to that of conjugated tolane dithiol (10 À2.7 G 0 ) 29 and even higher than that of biphenyl dithiol molecules (10 À3.8 G 0 ), 31 even though 1,3Az has a signicantly longer length. These ndings suggested that the protonated azulene core has great potential as a highly conductive building block for future molecular electronics studies.
We also studied the conductance changes before and aer adding TFA. Plotted as black curves in Fig. 2d-f, the most probable conductances of 1,3Az, 4,7Az, and 5,7Az are identied to be 10 À3.8 G 0 , 10 À4.2 G 0 , and 10 À4.9 G 0 , respectively. The conductance of 5,7Az falls below that of 4,7Az although 4,7Az has a longer length, suggesting the existence of destructive quantum interference in the charge transport through neutral 5,7Az. 9 Aer adding TFA for protonation, signicant conductance enhancement is observed for both 4,7Az and 5,7Az (Fig. S2-2 and S2-3 in ESI †), while the changes of 5,7Az are the  most pronounced among the three derivatives. Interestingly, it is found that the conductance of 5,7Az increased from 10 À4.9 G 0 to 10 À3.6 G 0 aer protonation, which was even higher than the conductance of protonated 4,7Az. Table 1 summarizes the characterized molecular junction lengths from their plateau lengths (see ESI † for more details) and conductances with literature data for comparison. By adding the snap-back distance of 0.5 nm, 29 the characterized lengths of the molecular junctions are 1.35 nm, 1.53 nm and 1.32 nm for 1,3Az, 4,7Az and 5,7Az, and 1.39 nm, 1.51 nm, and 1.35 nm for protonated 1,3Az, 4,7Az and 5,7Az, respectively. The agreement with the calculated molecular lengths (see also Fig. 4a) suggests that all of the molecules are able to be stretched to a fully-elongated conguration during the singlemolecule conductance measurements. Furthermore, it is found that the conductances we determined for neutral azulene are qualitatively in good agreement with previous studies by Xia et al. 21 but quantitatively a difference of 10 À0.2 G 0 is obtained for each molecule. Control experiments in 1,2,4-trichlorobenzene (see Fig. S2-1 in ESI †) reveal that the quantitative difference comes from the solvent tuning effect in different solvents. 32,33 To evaluate the environmental effect of adding TFA to the junctions, we carried out the control experiment with a similar molecule in which the azulene core is replaced by a phenyl core (Fig. 3a). A slight conductance increase from 10 À3.6 G 0 to 10 À3.4 G 0 is observed as shown in Fig. 3b (see Fig. S2-6 † for 2D histograms), indicating that Fermi level shiing of the electrode 33-37 by adding the TFA is not the major contribution of the conductance enhancement. A previous theoretical study suggests that the protonation could tune the destructive quantum interference in charge transport by changing the energy level of the molecular device. 20 Thus, the signicant conductance enhancement we observed for azulene junctions may come from the protonation of the azulene core.
To further investigate the connectivity-dependence of conductance enhancement between the neutral state and nal protonated state, we introduced a simple parameter free theory, [38][39][40][41] in which the Hamiltonian H of the PAH core was identied with a connectivity matrix C, whose elements C ij were equal to unity for ij nearest neighbors and zero otherwise. This theory correctly predicts conductance ratios for a range of PAHs and therefore here we use the same theory to investigate how the conductance of azulenes with different connectivities changes from the initial state to the nal state. Starting from the azulene core Hamiltonian, the resulting room temperature electrical conductances for each connectivity 40 are shown in Fig. 4b and c in the initial and protonated states. The neutral-state conductance (Fig. 4b) of the 5,7Az connectivity shows destructive quantum interference near the gap center (E F z 0) over a wide energy range between À0.7 and 0.6. In contrast, the 4,7Az and 1,3Az connectivities show no such feature near the gap center and therefore their neutral-state conductance is higher than that of 5,7Az. Comparing Fig. 4b with our measured conductances suggests that the Fermi energy in the neutral state is located just below the gap center (see the dotted black vertical line in Fig. 4b). With this choice of Fermi energy, the conductance order is 1,3Az > 4,7Az > 5,7Az at the initial state (Fig. 4b).
In the protonated state, density functional theory reveals that the spin-up and spin-down energy levels of the azulene in a junction are not equal (Fig. S3-2 in the ESI †) and therefore the spin-independent transmission peaks associated with the neutral-molecule HOMO will split into separate spin-up and spin-down transmission resonances, leading to a reduction in the HOMO-LUMO gap of the protonated molecules. In   addition, protonation is expected to introduce a negative electrostatic potential in the vicinity of the molecule, which causes the HOMO to increase in energy and move closer to the Fermi energy. (shown by the black vertical dashed lines in Fig. 4b and  c). As shown in Fig. S3-2 in the ESI, † in the initial state the system is not spin polarised. However, as shown in Fig. S3-3-S3-5, † in the protonated state within the junction, due to the charge transfer from the molecule to the gold, spin-splitting occurs.
The HOMO-LUMO gap in the protonated state arises from the spin-up HOMO and the spin down LUMO and consequently the new HOMO-LUMO gap in the protonated state is smaller than the initial one. In our tight-binding model, this splitting is taken into account using site energies which differ by 0.4 eV for spin up and down, respectively. By choosing the Fermi energy in the tail of the LUMO, as shown in Fig. 4c, the conductance of all connectivities increases upon protonation, and the order is changed compared with that in the neutral state. In agreement with the experiments as shown in Fig. 4d, the conductance order is 1,3Az > 4,7Az > 5,7Az in the neutral state and it changes to 1,3Az > 5,7Az > 4,7Az in the protonated state. Uniquely, the neutral 5,7Az possesses a sharp interference dip near the Fermi energy, and consequently exhibits the largest conductance increase upon protonation. In contrast, for 4,7Az and 1,3Az, the neutral state is not affected by destructive interference and therefore the shi towards the HOMO produces a smaller increase in conductance. Thus, the protonation of 5,7Az provides more signicant tuning than that of 4,7Az and 1,3Az (see ESI † for more detailed discussions).

Conclusions
To conclude, we studied the protonation tuning of azulene derivatives with quantum interference effects via measuring the single-molecule conductance using the MCBJ technique. It is found that the single-molecule conductance of protonated azulene is more than one order of magnitude higher than the conductance of the initial state. More importantly, the molecule with destructive quantum interference near the Fermi energy at the neutral state could provide more signicant conductance tuning. These experimental observations are supported by our recently developed parameter free theory. Our nding suggests that the protonated azulene derivatives could be considered as a promising building block for the construction of new conjugated materials with stimuli-responsive and enhanced charge transport ability, and reveals the potential of protonation as a novel approach to tune the quantum interference effect in charge transport at the single-molecule level.

Conflicts of interest
There are no conicts to declare.