The Control of Intramolecular Through-Bond and Through-Space Coupling in Single-Molecule Junctions

Abstract

Open AccessCCS ChemistryRESEARCH ARTICLE1 Feb 2022The Control of Intramolecular Through-Bond and Through-Space Coupling in Single-Molecule Junctions Zhibing Tan†, Wenlin Jiang†, Chun Tang†, Li-Chuan Chen†, Liangliang Chen, Junyang Liu, Zitong Liu, Hao-Li Zhang, Deqing Zhang and Wenjing Hong Zhibing Tan† State Key Laboratory of Physical Chemistry of Solid Surfaces, iChEM, College of Chemistry and Chemical Engineering, Xiamen University, Xiamen 361005 †Z. Tan, W. Jiang, C. Tang, and L.-C. Chen contributed equally to this work.Google Scholar More articles by this author , Wenlin Jiang† Beijing National Laboratory for Molecular Sciences, CAS Key Laboratory of Organic Solids, CAS Center of Excellence in Molecular Sciences, Institute of Chemistry, Chinese Academy of Sciences, Beijing 100190 †Z. Tan, W. Jiang, C. Tang, and L.-C. Chen contributed equally to this work.Google Scholar More articles by this author , Chun Tang† State Key Laboratory of Physical Chemistry of Solid Surfaces, iChEM, College of Chemistry and Chemical Engineering, Xiamen University, Xiamen 361005 †Z. Tan, W. Jiang, C. Tang, and L.-C. Chen contributed equally to this work.Google Scholar More articles by this author , Li-Chuan Chen† State Key Laboratory of Applied Organic Chemistry, College of Chemistry and Chemical Engineering, Lanzhou University, Lanzhou 730000 †Z. Tan, W. Jiang, C. Tang, and L.-C. Chen contributed equally to this work.Google Scholar More articles by this author , Liangliang Chen State Key Laboratory of Physical Chemistry of Solid Surfaces, iChEM, College of Chemistry and Chemical Engineering, Xiamen University, Xiamen 361005 Google Scholar More articles by this author , Junyang Liu State Key Laboratory of Physical Chemistry of Solid Surfaces, iChEM, College of Chemistry and Chemical Engineering, Xiamen University, Xiamen 361005 Google Scholar More articles by this author , Zitong Liu *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] State Key Laboratory of Applied Organic Chemistry, College of Chemistry and Chemical Engineering, Lanzhou University, Lanzhou 730000 Google Scholar More articles by this author , Hao-Li Zhang *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] State Key Laboratory of Applied Organic Chemistry, College of Chemistry and Chemical Engineering, Lanzhou University, Lanzhou 730000 Google Scholar More articles by this author , Deqing Zhang *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] Beijing National Laboratory for Molecular Sciences, CAS Key Laboratory of Organic Solids, CAS Center of Excellence in Molecular Sciences, Institute of Chemistry, Chinese Academy of Sciences, Beijing 100190 Google Scholar More articles by this author and Wenjing Hong *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] State Key Laboratory of Physical Chemistry of Solid Surfaces, iChEM, College of Chemistry and Chemical Engineering, Xiamen University, Xiamen 361005 Google Scholar More articles by this author https://doi.org/10.31635/ccschem.021.202000748 SectionsSupplemental MaterialAboutAbstractPDF ToolsAdd to favoritesDownload CitationsTrack Citations ShareFacebookTwitterLinked InEmail Electronic coupling between individual building blocks plays an essential role in charge transport through molecular materials and devices. However, the investigation of the transmission mechanism in charge transport via intramolecular coupling remains challenging. Herein, we demonstrate the transition of the intramolecular through-bond and through-space coupling in a single-molecule junction with a family of diketopyrrolopyrrole (DPP) derivative by varying intramolecular donor–acceptor (D–A) interactions. The transition is accomplished by regulating D–A interactions by inserting different aromatic rings inside, leading to two orders of magnitude difference of the single-molecule conductance. The flicker noise analysis demonstrates that the conductance difference arises from the control of the contribution between through-bond and through-space coupling. These findings are further supported by the calculation that the intramolecular coupling among molecular building blocks correlates with the D–A interaction, providing a promising way to regulate the contribution between through-bond and through-space coupling in the charge transport through molecular materials and devices. Download figure Download PowerPoint Introduction To fabricate advanced molecular devices, the fine-tuning of the electrical performance at the single-molecule level is of utmost importance. Through-space and through-bond coupling are fundamental concepts to understand electronic coupling,1,2 which also determine the electronic properties of the single-molecule junction.3–5 In the charge transport through single-molecule junctions, the electronic coupling between molecular components and electrodes depends on the dominant transport channel, on whether it does or does not participate in the formation of a chemical bond.5,6 At the same time, the charge transport through π–π stacking dimer junctions is dominated by through-space coupling.7,8 Although electronic coupling is critical for the design and fabrication of molecular circuits with advanced functions,9,10 the fine-tuning of the through-bond and through-space coupling in a single-molecule junction is still a challenge. To achieve control of the electronic coupling between different building blocks, we focused on the regulation of intramolecular donor–acceptor (D–A) interaction, which has attracted considerable attention in the construction of high-performance organic electronic devices using conjugated molecules or polymers.11 Among these molecular derivatives, diketopyrrolopyrrole (DPP) derivative, characterized by electron deficiency, is one of the promising electron-accepting motifs to construct high-performance organic semiconductors.11–13 Meanwhile, the aromatic groups connecting to the DPP core also play an important role in determining the electronic properties.14,15 By regulating the adjacent aromatic rings to the DPP core, we can fine-tune the electronic coupling of the single-molecule junction. In this work, we investigate the charge transport through five different DPP derivatives using the scanning tunneling microscope break junction (STMBJ) technique.6,16–20 These DPPs have a D–A–D structure, and the DPP cores are flanked by different aromatic rings (Figure 1a). The five molecules, having similar structures and lengths, show a conductance difference of more than two orders of magnitude. By performing flicker noise analysis, we find that the intramolecular coupling of the five molecules represents a gradual transition from through-bond to through-space coupling, which originates from the changes in the D–A interaction between molecular building blocks. This is also shown by the combined density functional theory (DFT) calculations. Figure 1 | (a) Schematic of the relationship between the intramolecular D–A interaction and through-bond/through-space coupling in single-molecule junction. (b) Five DPP molecules with different aromatic rings. Download figure Download PowerPoint Experimental Methods Materials and characterization techniques The reagents and starting materials including compound M1 ( Supporting Information Figure S1) were commercially available and used without any further purification, if not specified elsewhere. Compounds M1– M6 ( Supporting Information Figure S1) were synthesized according to the previous reports.21–23 The melting point was collected on a Buchi B-540 analyzer (BUCHI, Switzerland). 1H and 13C NMR spectra were recorded on a Bruker AVANCE III 400 MHz spectrometer (Bruker, Switzerland), and mass spectra were determined with a Bruker Solarix-XR high-resolution mass spectrometer (Bruker, Germany) (see Supporting Information Figures S1–S11 for further details). Single-molecule conductance measurement and data analysis We used a homemade STMBJ setup to perform the single-molecule conductance measurement. As shown in Supporting Information Figure S14, a piece of gold wire (0.25 mm in diameter) with one end burned into a bead was used as a tip, which was fixed on the piezo, while the piezo adhered to the bottom of the stepping motor. A gold-plated silicon wafer washed by piranha solution was placed below the tip. Then, 50 μL mesitylene (TMB) solution containing 0.1 mmol target molecule was dripped on the substrate, and the tip was immersed in the solution. In this context, both the tip and the substrate were connected with the external current amplifier and the controller, and a 100 mV bias voltage was applied between them. When the current of the two electrodes reached the set upper limit, the tip was controlled to remove it from the substrate. And when the current was below the detection limit of the instrument, the tip was controlled to go down to contact the substrate. Molecular junctions were formed during the break junction process so that the single-molecule conductance could be detected. All the experiments were performed in air at room temperature (under air conditioning set at 25 ± 3 °C). The construction method of one-dimensional (1D)/two-dimensional (2D) conductance and relative displacement histogram refer to some of our recent work.20,24,25 Flicker noise measurements For the conductance noise measurement, the tip was controlled to stabilize for 100 ms after being removed from the substrate surface and when the molecular junction was formed. To get the noise power density spectra (PSD), we cut out the conductance feature to perform discrete Fourier transformation, and the data were squared. Then the PSD was integrated from 100 to 1000 Hz ( Supporting Information Figure S17). The 2D histogram of normalized flicker noise power versus average conductance was constructed from thousands of conductance traces. The 2D Gaussian distribution fitting was applied to determine the scaling power. The conductance power was normalized from G1.0 and G2.0, and the correlation power was determined by the correlation coefficient, which showed the smallest absolute value in the fitted Gaussian distribution equations. The value of “n” was determined by the correlation between the average conductance and noise power. When the through-bond coupling dominated the charge transport through the single-molecule junction, the “n” would be 1.0. This means when the noise power is normalized by G1.0 (G is the average conductance), there is no correlation between the noise power and G, leading to an orthogonal noise distribution. In contrast, when the through-space coupling is dominating, the “n” would be 2.0. This means that even when the noise power is normalized by G1.0, there is still a strong correlation between the noise power and the average conductance. Thus, the noise distribution is a slant ellipse, suggesting that a larger conductance would lead to larger noise power and vice versa. Theoretical calculations The molecular devices used in the electron transport simulations are calculated with a combination of DFT and nonequilibrium Green’s function (NEGF), using the Atomistix Tool Kit (ATK) Q2019.12 packages with the Slater–Koster tight-binding method.26 The two gold (111) electrodes consisted of a 4 × 4 unit cell with an extension region thickness of five. A trimer gold atoms cluster was employed to form the stable contact with the anchors on both sides of the gold electrodes ( Supporting Information Figure S19). The geometries of molecules and the gold clusters are optimized with k-points sampling (1 × 1 × 200) with a force threshold of 0.05 eV/Å, while the position of gold surface atoms is fixed. With the same parameters used for the molecular device geometries optimization, the zero-bias transmission coefficient and spectra are calculated further. According to the Maxwell–Boltzmann distribution, the molecular configuration with different dihedral angles in the thermal equilibrium state also obeys the energy distribution law. The statistical relation is shown as: G = G θ ⁢ ⁡ exp ⁡ ( − E θ k B T ) Σ θ ⁢ ⁡ exp ⁡ ( − E θ k B T ) G is the thermal averaged conductance, Gθ and Eθ are the conductance and potential energy at dihedral angle θ, T is the experimental temperature (298 K), and kB is the Boltzmann constant (1.38 × 10−23 J/K). Results and Discussion Single-molecule conductance measurements As shown in Figure 1b, we synthesized five different soluble DPPs with p-methylthiobenzenes at both ends, including difuranyl-DPP (F-DPP), dithienyl-DPP (T-DPP), dithiazolyl-DPP (Thia-DPP), dipyridyl-DPP (Py-DPP), and diphenyl-DPP (B-DPP) (see Experimental Methods section for details). The DPP cores have strong electron-withdrawing abilities. Changing the aromatic ring adjacent to the DPP core provides different levels of intramolecular D–A interaction. Thiomethyl group is used as an anchoring group to connect with the gold electrode to form single-molecule junctions, and the carbon chains are attached to the N-positions in the DPP core to ensure solubility. UV–vis absorptions of all DPP molecules are shown in Supporting Information Figure S12. The spectrum onsets of the absorption edge of the DPP derivatives are detected at λ = 658, 642, 632, 597, and 555 nm, respectively. Simultaneously, cyclic voltammetric ( Supporting Information Figure S13) measurements demonstrate that, from F-DPP to B-DPP, their electrochemical bandgap increases from 1.46 to 1.80 eV, suggesting that their bandgap increases gradually from F-DPP to B-DPP (see Supporting Information Table S1 for a detailed analysis). A homemade STMBJ setup ( Supporting Information Figure S14) is used to measure the single-molecule conductance of DPPs in the mesitylene (TMB) solution containing 0.1 mmol/L of the target molecules.20,27 Two gold electrodes are controlled to approach and repeatedly separate the trapping molecules, and a 100 mV bias voltage is applied between the two electrodes. The current through the two electrodes is recorded and converted into conductance-displacement traces for the statistical analysis. Figure 2a gives the conductance-displacement curves that show that the conductance decreases exponentially with increased displacement when there is no molecule in the solution, while when the target molecule exists, a visible plateau appears at the range from 10−3.5G0 to 10−5.5G0. The latter indicates that both thiomethyl groups of the molecule are anchored on the gold electrode to form a molecular junction, and the conductance of the plateaus provides the conductance of single-molecule junctions, suggesting that the conductance reduces gradually from F-DPP to B-DPP. Figure 2 | (a) Typical individual conductance-displacement curves and (b) 1D conductance histogram of STMBJ measurement for F-DPP (pink), T-DPP (turquoise), Thia-DPP (light green), Py-DPP (light orange), and B-DPP (blue) obtained from 972, 1144, 930, 992, and 1185 traces, respectively. The light grey curves in (a) and (b) represent the blank experiments. (c and d) 2D conductance histograms of F-DPP and B-DPP. (e) The relative displacement distributions of all DPPs ranging from 10−0.3G0 to 10−4.8G0, 10−5.0G0, 10−5.5G0, 10−5.8G0, and 10−6.5G0 respectively. Download figure Download PowerPoint We also constructed the conductance histogram from ∼1000 individual traces without data selection to get the quantitative value of molecular conductance (Figure 2b). The sharp peak at G0 represents the conductance of a single gold atom in contact (G0 is quantum conductance, equal to 2e2/h). Besides the G0 peak, the most probable single-molecule conductance values of F-DPP, T-DPP, Thia-DPP, Py-DPP, and B-DPP can be determined from the position of the molecular conductance peak to be 10−3.63G0, 10−3.97G0, 10−4.45G0, 10−4.95G0, and 10−5.65G0 respectively. For the curve of B-DPP, the conductance signal also appears between 10−4G0 and 10−5G0, and we consider that this signal may come from the interaction between the electrode and the molecular π system.4 It is suggested that the electrode may have experienced the process of first interacting with the π system and then with the sulfur atom during the opening process. However, the main conductance peak should be the result when the molecule is completely extended. The 2D conductance histograms are acquired by superimposing these conductance traces to reveal the origins of molecular configurations contributing to the conductance. Both Figures 2c and 2d show the distinctive conductance intensity cloud for F-DPP and B-DPP. (The 2D conductance histogram of T-DPP, Thia-DPP, T-DPP, and the blank experiment are shown in Supporting Information Figures S15 and S16.) The relative displacement distribution (Figure 2e) shows that the conductance features of all DPPs is located at a similar displacement of 1.7∼1.8 nm. Considering the snapback distance of gold–gold atomic junctions (∼0.5 nm),19,20,25,27,28 the lengths of the molecular junctions are determined to be 2.2–2.3 nm for all DPPs, which match the calculated lengths of DPP molecules ( Supporting Information Table S2). These results show that the single-molecule conductances vary by more than two orders of magnitudes from F-DPP, T-DPP, Thia-DPP, Py-DPP to B-DPP while the lengths of the molecules remain almost the same. The conductance variation of DPPs is even more significant than that of molecular junctions induced by the quantum interference effect or configuration changes.2,29–31 Flicker noise analysis of the molecular junctions The flicker noise measurements5–7,20,32–35 of these DPPs were performed to reveal the origins of the significant conductance variation among DPPs. We paused the tip for 100 ms when a stable molecular junction is formed during the break-junction process. Thousands of conductance plateaus measured within the fixed displacement period were analyzed to obtain the conductance noise PSD (see Experimental Methods section for details of flicker noise analysis), which was normalized by the mean conductance G. The distributions of the normalized PSD (normalized by G) against G are shown in Figures 3a–3e. As shown in Figure 3a, the normalized PSD of F-DPP shows a noncorrelation to G, which is associated with the domination of through-bond coupling in the charge transport through F-DPP. Figure 3 | (a, b, c, and d) 2D histogram of normalized flicker noise power versus average conductance for F-DPP, T-DPP, Thia-DPP, Py-DPP, and B-DPP derivatives obtained from 5776, 6339, 4220, 6998, and 5427 traces, respectively. The noise power scales as Gn, and Figure (f) gathers the n values of DPPs. The n value is determined by the smallest absolute value of the correlation coefficient between normalized PSD and average conductance, and the correlation coefficients are shown in Supporting Information Table S5. Download figure Download PowerPoint On the contrary, as shown in Figure 3e, the correlation between the PSD of B-DPP and G is strong, suggesting that the higher G leads to the higher PSD and vice versa. When the PSD of B-DPP is normalized by G2.0 ( Supporting Information Figure S18d), the normalized PSD shows a noncorrelation to G, suggesting that the through-space coupling dominates the charge transport through B-DPP. In between, as shown in Figure 3f, we show that when the PSD of T-DPP, Thia-DPP, and Py-DPP (Figure 3f and Supporting Information Figure S18) is normalized by G1.2, G1.4, and G1.6 (the n of Gnis called as scaling exponent in this work), the noncorrelation between the normalized PSD and G can be observed, suggesting that both through-bond and through-space coupling are participating in the charge transport but to different degrees.5,7,20 Thus, the change of the D units in the series of DPP molecules can not only modify the corresponding single-molecule conductance but also lead to delicate control of charge transport from through-bond to through-space coupling. Theoretical calculations It is worth noting that the aromaticity of the D units used in the five DPP target molecules shows the following trend, furan < thiazole < thiophene < pyridine < benzene,36,37 which is approximately consistent with the conductance order of DPP molecules. But for T-DPP and Thia-DPP, even the aromaticity of thiophene is weaker than thiazole. The intramolecular hydrogen bond O···H (between the oxygen of the DPP and the β-hydrogen of the thiophene)14,38 of T-DPP reduces the dihedral angle between the two aromatic planes, leading to a stronger intramolecular D–A interaction and higher conductance than F-DPP. The bandgap order of DPPs determined from the theoretical calculation is also consistent with the experiment results (see details in Supporting Information Tables S1 and S3). To understand the transition from through-bond to through-space coupling, we optimized the molecular configuration of the DPP molecules using Gaussian 09 software with the DFT/B3LYP-6-311+G (d,p) method (the alkane chains are replaced by methyl for simplification) and calculated the transmission spectra using ATK software with the Slater–Koster tight-binding method.26 As shown in Figure 4a, we found that the delocalized highest occupied molecular orbital (HOMO) isosurface of F-DPP becomes a localized pattern in B-DPP. (The same is true for the HOMO-1 orbital; see Supporting Information Figure S20 for more detail.) For the charge transport through molecular wires, the coupling strength between the two electrodes is determined by the dominant molecular level.7 Since the HOMO isosurfaces of B-DPP is distributed at both ends of the molecule, a weak coupling (in other words, the through-space coupling) between two electrodes is expected. On the contrary, we expect strong coupling (or through-bond coupling) in F-DPP, while such coupling of T-DPP, Thia-DPP, and Py-DPP fall in between B-DPP and F-DPP. We further calculated the HOMO and lowest unoccupied molecular orbital (LUMO) energy levels of the D part (furan, thiophene, thiazol, pyridine, and benzene) and the A part (the DPP core) with the same basis set in Gaussian software (see Supporting Information Table S4 for more details). From the comparison of the HOMO level of the D fragment, the energy level order is furan < thiophene < thiazol < pyridine, which is also consistent with the conductance order of the single-molecule conductance experiment, except for the benzene one with poor flatness. In other words, the closer the energy levels are (HOMO of the D fragment and LUMO of the DPP core), the stronger the intermolecular interaction. These results indicate that the D–A interaction in the molecule gradually weakens from F-DPP to B-DPP. Figure 4 | (a) The HOMO orbital patterns of DPPs. (b) Calculated potential energy curves as a function of the dihedral angle between the D and the A fragment. (c) Calculated proportion curves as a function of the dihedral angle between the D and the A fragment. The total area under each curve is 1. (d) Energy level diagram of the DPP junctions. Download figure Download PowerPoint Furthermore, we try to analyze the source of flicker noise in DPP junctions. According to previous report,5 the flicker noise not only originates from the fluctuation of the electrode-molecule interface but also the intermolecular interaction. In this study, we use the same anchor group (–SMe) to guarantee the same electrode-molecule interaction, suggesting that the different PSD distributions should only come from the molecular backbones. Specifically, the rotational freedom of molecules could bring about different dihedral angles between the D and A fragments and different electronic couplings between different building blocks. To address the rational freedom of molecules, we conduct the flexible scanning of potential energy surfaces with different dihedral angles. As shown in Figure 4b, the calculation reveals that the rotational energy barriers, relative to the dihedral angle between D and A fragments, decreases gradually from F-DPP to B-DPP, indicating that the weaker D–A interaction enhances rotational freedom. This rotational freedom makes the intramolecular coupling more significantly varied and leads to a PSD distribution with the through-space coupling feature. To get an intuitive understanding of the charge-transport mechanism through the five DPP molecules, we plot the diagram, as shown in Figure 4d. We divide the molecular wires into three parts, which are the two Ds connected by an A (the DPP core). The coupling strength between the electrode and the molecular backbone is defined as τ. The δ is the coupling strength between D and A in the molecular backbone. Since the anchor groups of all five molecules are the same, we expect a similar value of τ for the five molecules. The δ is related to the energy difference between the corresponding energy levels; for example, a smaller energy difference leads to a stronger coupling strength. Meanwhile, the LUMO of the A can also be taken as the barrier for the charge tunneling from the HOMO of the D to another D. When the D unit shows stronger aromaticity, we expect to see a lower energy HOMO, which leads to a larger energy difference between the HOMO of D and the LUMO of DPP, showing a larger intramolecular barrier to tunneling. Combined with the above experimental results, we provide a benchmark example to use PSD distribution to evaluate the intramolecular barrier. We also consider the influence of the molecular twist angle on molecular conductance. As has previously been demonstrated both theoretically and experimentally,31 molecular conductance is directly proportional to the square of cosθ. (The θ is the dihedral angle.) Based on this principle, using the conductance of F-DPP as reference, we calculate the conductance of the other DPPs (see Supporting Information Figure S21 for the molecular twist angle of DPPs). Supporting Information Figure S22 gives the comparison of the experimental conductance values and the values calculated from the molecular twist angle of DPPs based on the cos2θ relation. It can be seen from the calculated conductance value that with F-DPP as the reference, the conductance of T-DPP and Py-DPP has almost no change, while Thia-DPP and B-DPP have only changed two to three times. However, our experimental results show that from F-DPP to B-DPP, the conductance changes nearly two orders of magnitude. Therefore, the change of molecular conductance caused by the dihedral angle can be neglected. To further exclude the effect of the dihedral angle on conductance, we calculate the conductance value of DPPs at the same dihedral angle. As shown in Supporting Information Figure S23, dihedral angles between D and A are all fixed at 0°, 15°, and 30°, and the transmission spectra traces show that the DPPs have the same conductance order under different dihedral angles. And from F-DPP to B-DPP, the conductance shows a variation of two orders of magnitude. Although there is a negligible difference of molecular lengths (the maximum difference between F-DPP and B-DPP is ∼0.2 nm), their conductance order does not follow the order of molecular lengths ( Supporting Information Table S2), and such a small length difference can not lead to a significant change of conductance.39–41 The calculated proportion curves are obtained from the theoretical rotational energies in Figure 4b by the above statistical relations. The dihedral angle range for statistical analysis is 0∼60, where the energy barrier is suitable for molecular rotational freedom (∼20 kJ/mol). As shown in Figure 4c, the proportions of DPP serial molecules at ∼0° (except for B-DPP at ∼30°) are the highest. At the same time, the proportion shows an exponential decay as the molecular rotation barrier increases. Moreover, the sequence of the distributed concentration (F-DPP < T-DPP < Thia-DPP < Py-DPP < B-DPP) is consistent with the experimental PSD data (F