Molecular switches and real-time ion sensing in pyridinium circuits via a single-molecule STM-break junction

Abstract

The functional electronic and spectro-electrochemical properties of two structural pyridinium isomers, Py_Down-BF₄ and Py_Up-BF₄, were studied at the single-molecule level using the STM-BJ technique. These isomers differ in the position of the redox-active pyridinium core. The aim was to identify the role of core's position in promoting reversible switching between electromers (redox isomers) in solution and at the gold–pyridinium–gold junction circuit. We measured the single-molecule conductance of each pyridinium isomer in various electrolyte environments using tetrabutylammonium salts (TBABF₄, TBAPF₆, TBABr, and TBACl). The choice of electrolytes played a crucial role in the histograms’ shapes—junction distribution, width, and peak position—which act as unique conductance fingerprints for each isomer. During STM-BJ measurements, a dynamic evolution in the conductance histograms was determined, particularly with the electrolytes TBAPF₆ and TBABF₄. This behavior was attributed to the real-time detection of interactions between the positively charged pyridinium core and the electrolyte anions within the gold–pyridinium–gold junction. The dynamic evolution in single-molecule conductance was rationalized by the Gibbs free energies (ΔG) for the anion–cation pairs obtained from density functional theory (DFT) calculations. Furthermore, the theoretical trend predicted by DFT combined with the Keldysh nonequilibrium Green's function (NEGF) formalism (DFT-NEGF) was consistent with the experimental results.

---

New concepts

This research presents a breakthrough in applying pyridinium redox molecules in a gold– molecule–gold circuit for two key functions: (1) molecular switches and (2) single-molecule sensors. Unlike previous studies in solutions, we demonstrate reversible switching between two stable states when the molecules are placed between gold contacts. This is significant, as metallic surfaces typically suppress switching behavior. Our results show that varying the structure of the redox center enables switching on gold surfaces, laying the foundation for active pyridinium-based switching components in molecular electronic devices, a key goal of nanotechnology. Using the STM-BJ technique, we measured the single-molecule conductance of two isomers in different electrolyte environments, enabling real-time anion sensing. While earlier studies highlighted general electrolyte effects on conductance, our work reveals the specific role of anion identity in modulating the dynamic behavior of pyridinium junctions. This real-time sensing behavior has not been reported for these molecules or their derivatives. Overall, this study opens new horizons in controlling redox switching and ion recognition at the single-molecule level. It advances the development of functional nanoscale devices for molecular electronics and sensing applications, tackling key challenges in nanoscience and nanotechnology.

1. Introduction

Pyridinium molecules are among the most exciting entities to be used as functional parts of electronic devices due to their inherent electronic properties and the strong correlation between their structure, activity, and functionality.^1–3 Pyridinium molecules and derivatives are well-known as dye compounds and functional redox-active molecules whose electronic properties can be modified to promote the desirable spectral and electrochemical properties for practical applications.^1,2,4–6 They are heterocyclic compounds consisting of 6-membered rings, with a positively charged nitrogen, the so-called pyridinium core, Fig. 1(a). The positively charged nitrogen atom confers electron-withdrawing, electron-accepting, and redox properties.^4,7,8 Due to these properties, pyridiniums and derivatives have been proposed as electrophorus molecular components for potential applications in capacitor devices, as active electron acceptor components in chromophores,^6,7 as building blocks for electrode materials for solid-state Li-ion-type organic batteries (LOBs),⁵ and as anolytes in redox flow batteries, among other applications.⁶ Moreover, their molecular structure and core position govern their electrochemical properties that influence the electron transfer (ET) mechanisms.^4,9 As reducible electrophores (electron acceptors),¹ these systems can undergo multi-electron redox processes by modulating their molecular architecture.^4,9 In particular, the pyridinium core can participate in a two-electron redox event, which induces a geometric transformation between distinct redox isomers—typically transitioning from an axially planar oxidized state to a pyramidalized reduced state.^4,9 Conversely, when structural changes hinder the pyridinium core in the molecular backbone, this switching between the two redox isomers is blocked. This behavior has been demonstrated by electrochemical and spectroelectrochemical studies carried out in structural pyridinium isomers dissolved in organic solutions.^2,9,10 This reversible switching between redox isomers is an exciting property that makes pyridiniums promising functional electronic elements; however, this behavior must occur when the pyridinium is confined to the surface for their electronic device applications. Conversely, Hromadová et al. reported deactivating the N-pyramidalization of the pyridinium redox center due to the adsorption process of molecular wires based on the expanded pyridinium compounds on the electrode surface (and electrified interface).⁴

Fig. 1 (a) Molecular structure of Py_Down-BF₄ (blue) and Py_Up-BF₄ (red); the pyridinium “core” is indicated by dashed dark circle. Cyclic voltammograms of (b) Py_Down-BF₄ and (c) Py_Up-BF₄. (d) Pictorial representation for one or two-electron single-step reduction processes. 0.1 M TBABF₄ in acetonitrile, the scan rate is 50 mV s⁻¹. Spectroelectrochemistry of (e) Py_Down-BF₄ and (f) Py_Up-BF₄ in acetonitrile TBABF₄ = 0.1 M.

In other aspects such as electrocatalysis research, pyridinium molecules have been used as molecular wires to anchor iron phthalocyanine (FePc) molecular catalysts to gold electrodes, building electrocatalytic self-assembled nanostructures that promote the oxygen reduction reaction, ORR.^11–13 This process plays a crucial role in fuel cell systems.¹⁴ Thus, the high activity of these self-assembled FePc systems for the ORR is supported by the pull effect of pyridinium's core on the metal iron center in FePc (promoting a hard Fe). This improves the catalysis process due to the optimal interaction between a hard O₂ molecule and hard Fe (supported by the Pearson principle).^11–13 Hence, due to their excellent properties as building blocks and molecular wires, their electrical properties have been studied at the single-molecule level by determining the molecular conductance using scanning tunneling microscopy based on the break-junction (STM-BJ) technique.^11,12,15 Obtaining molecular junctions makes it possible to evaluate the electron transport properties by determining the single-molecule conductance. It involves developing electrodes with a single molecule positioned between them based on a metal–molecule–metal circuit.^16,17 According to this, the structural changes and substituents in the pyridinium core ring and the molecular backbone can strongly influence their molecular conductance.^11,18,19 Besides, by combining STM-BJ and electrocatalysis fundamental concepts, pyridiniums have been used to develop efficient nanoscale wires by exploiting well-established electrocatalytic molecular platforms based on transition-metal phthalocyanine blocks.¹² On the other hand, as pyridiniums are salt compounds that suffer dissociation in organic solutions, yielding the pyridinium cations and, in most cases, the BF₄⁻ as a counterion (or another anion, e.g., ClO₄⁻, PF₆⁻ and Cl⁻), it is important to establish their functional properties as molecular wires, considering the effects of the counterions at the nanoscale level in the circuit. Thus, as promising building blocks to be applied in electronics and electrocatalytic devices, comprehension of the electron transfer process of the pyridinium/metal hybrid circuit at electrode–electrolyte interfaces is a crucial factor to consider for those applications, among others.¹⁵ In this direction, Li et al. demonstrated that the solvation environment affects molecular charge transport of bipyridinium molecules at electrode interfaces by measuring the single-molecule conductance in a series of different supporting electrolyte and counterion environments.²⁰ Molecular charge transport in bipyridinium molecules was found to strongly depend on the chemical identity of counterions and the solvation environment.²⁰

In this work, we investigated two structural isomers, Py_Down-BF₄ and Py_Up-BF₄, where the only difference in the molecular backbone is the core positions. These isomers were employed to study three key aspects: (1) the effect of core positioning (Up or Down) on single-molecule conductance, achieved by synthesizing both pyridiniums with anchoring groups capable of coordinating to gold electrodes to form single-molecule junctions; (2) the role of the core position in redox-switching behavior, as examined using spectroelectrochemical methods and scanning tunneling microscopy break-junction (STM-BJ) techniques; and (3) the influence of electrolyte counterions on the single-molecule conductance of each isomer, probed through STM-BJ experiments and supported by density functional theory (DFT) calculations. We have previously reported Py_Down-BF₄ and Py_Up-BF₄ as molecular wires for building electrocatalytic self-assembled systems of transition-metal phthalocyanines to promote energetic interest reactions.¹³ Due to their efficiency as ET wires, this study represents an experimental platform for advancing the rational design of nanostructured electrodes. Our results demonstrate that, unlike Py_Down-BF₄, Py_Up-BF₄ exhibits distinct switching behavior both in solution and when incorporated into an electrode–molecule–electrode junction, as assessed using the STM-BJ technique. Notably, this switching behavior is triggered under high potential bias conditions, even in the absence of an electrochemical gate or an electrolyte environment, highlighting a bias-induced mechanism intrinsic to the molecular structure.^20–22 On the other hand, using the STM-BJ technique, we have measured the single-molecule conductance of Py_Down-BF₄ and Py_Up-BF₄ in different electrolyte environments. The electrolytes used for the study correspond to four tetrabutylammonium salts (TBA) with anions having a wide range of coordinating strengths, comprising strongly coordinating species like the chlorine and bromide anions (Br⁻ and Cl⁻) to the intermediate anionic species like the tetrafluoroborate (BF₄⁻) and hexafluorophosphate (PF₆⁻) anions.²³ Thus, the experimental and calculated conductance values obtained for both isomers, Py_Up-BF₄ and Py_Down-BF₄, show that the TBACl increases the molecular conductance in both isomers compared to the other electrolytes. By recording and analyzing the single-pyridinium junctions in real-time measurements, we have detected the difference in the dynamic evolution of the single-molecular junction according to the structural pyridinium isomer and the chemical nature of the counterion from the electrolyte. Thus, the histogram built from thousands of single-molecule junctions displayed a different distribution over time, presenting a dynamic evolution until reaching a stable conductance histogram shape that served as a fingerprint for each electrolyte used. This dynamic behavior was not observed when the STM-BJ was carried out without electrolytes in the solution. Furthermore, this dynamic evolution in the histograms depends on the anion used as a counterion and the pyridiniums’ core positions; thus, the systems with PF₆⁻ have higher dynamic evolution in comparison with the Br⁻ and Cl⁻ systems, whose histograms always display the same shape in the time. This dynamic evolution in the single-molecule junctions was supported by the Gibbs free energy (ΔG) for all anion–cation (pyridinium) pairs obtained from DFT calculations. These findings lead to exciting possibilities for cutting-edge advancements in molecular electronic devices, electrode design, and single-molecule chemical sensors.

2. Results and discussion

2.1 Electrochemical and spectroelectrochemical studies for Py_Down-BF₄ and Py_Up-BF₄ molecular wires

Given the close relationship between the molecular structure and the electrochemical activity of pyridinium salts, we investigated the redox properties of the Py_Down-BF₄ and Py_Up-BF₄ isomers using cyclic voltammetry. These molecules have previously been reported as functional molecular wires within self-assembled electrocatalytic systems, primarily serving to connect the FeN₄ molecular catalyst to a gold surface.^11,13 Therefore, the molecular backbone has functional anchor groups with an affinity for a gold surface (–SCH₃) and FeN₄ (pyridine moiety). Thus, in this work's first stage, we have studied the impact of the core position variation in the two structural isomers on the electrochemical and spectroelectrochemical activities. The core corresponds to the positively charged nitrogen in the heterocyclic ring. Fig. 1(a) shows the molecular structure of both isomers, as mentioned above, where the core differs from the up or down positions with respect to the pyridine ring. Fig. 1(b) and (c) and Fig. S1 show the cyclic voltammograms of each isomer, where Py_Down-BF₄ presents two redox-step processes associated with two one-electron transfer steps placed in the pyridinium's core redox center. On the other hand, Py_Up-BF₄, Fig. 1(c) presents a one-redox-step process associated with a two-electron transfer (bi-electronic redox process) centered in the pyridinium's core. To assign the electronic nature of each process, we have used ferrocene (Fc/Fc⁺) as the redox reference. The charge obtained in micro coulombs (μC) for the reduction process of ferrocene was compared with the charge for the oxidation process of the Py_Down-BF₄ and Py_Up-BF₄ to estimate the number of electrons involved in the isomers’ redox process. Ferrocene (Fc) has a reversible one-electron oxidation/reduction process.²⁴ This methodology is widely used to study redox processes in molecules whose faradaic processes are not defined.^1,9,10,24 To obtain the charge corresponding to each faradaic process, the area under the curve of the reduction process was integrated (Fig. S3(c) and (d)). Subsequently, the charge of the faradaic process was determined using eqn (1).

Q is the process's faradaic charge, A_i is the integrated area under the curve, and dE/dt is the scan rate in Volts s⁻¹. The correlation between the integrated area under the cyclic voltammogram peak (A_i) and the number of electrons transferred for the reduction process, as well as the processing of data, is detailed in the SI, pp. S6–S8.

Single-step vs. two-step redox processes have been reported for pyridinium derivatives (polyarylpyridinium electrophores), where the two-electron transfer in the same potential value (bi-electronic process) has been explained due to a structural rearrangement in the molecular backbone.⁹ Therefore, the bi-electronic process that the Py_Up-BF₄ isomer undergoes would correspond to a conformational change during the redox process, giving rise to electromer formation (redox isomers), Fig. 1(d). Namely, Py_Up-BF₄ switches between two conformational stable states, axial geometry in the oxidized state and pyramidal geometry in the reduced state. In this way, the molecule can suffer a bi-electronic redox process by two electrons at the same voltage in the same redox-active site. Conversely, Py_Down-BF₄ contains higher volume substituents in N₁ (methyl thiophenyl moiety), hindering the pyramidal conformational changes.

According to the above, the conformational changes with the redox activity of Py_Down-BF₄ and Py_Up-BF₄ isomers in solution were evaluated by UV-vis spectroelectrochemistry. For better visualization of the spectroelectrochemistry experiments, the UV-vis spectra have been graphed in the cathodic scan direction (recorded every 0.1 V), Fig. 1(e) and (f). Since as the applied potential reaches −1.6 V for Py_Down-BF₄, the highest energy band experiences a decrease in absorbance and a hypsochromic shift of 14 nm.²⁵ Furthermore, the shoulder at 310 nm disappears completely, and new broadband emerges at 514 nm. After performing the potential sweep in the opposite direction, the UV-vis spectrum was recorded (gray line in Fig. 1(e)), which maintains the same shape as the spectrum at −1.6 V. In the case of the Py_Up-BF₄, when the applied potential reaches −1.6 V, the highest energy band experiences an increase in absorbance along with a bathochromic shift of 10 nm.²⁵ Moreover, the bands at 316 and 387 nm disappear, and new broadband appears at 520 nm. After performing the potential sweep in the opposite direction, the UV-vis spectrum was recorded (grey line in Fig. 1(f)), which returns to the form of the spectrum recorded before applying potential perturbation. The final spectra registered when the potentiodynamic perturbation ended (grey line in Fig. 1(e) and (f)) displayed significant differences between both isomers. Thus, the initial and final Py_Down-BF₄ spectra taken without potential applied (Fig. 1(e)) differ from those obtained for Py_Up-BF₄, which have identical initial and final spectra (Fig. 1(f)). The irreversibility of the Py_Down-BF₄ process is associated with steric factors because for the pyridinium core to be in the reduced state, the methyl thiophenyl moiety must first rotate and then leave the plane. However, it will be prevented from returning to the original position.⁹ In fact, theoretical calculations show that the torsion angle between the pyridinium core and the methyl thiophenyl moiety (Fig. S4 in the SI) is 72° in the minimum energy structure and must overcome an energy barrier of ∼26 kcal mol⁻¹ to reach a local minimum where the ligand is distorted to allow the same plane for the pyridinium core and the methyl thiophenyl moiety. Additionally, the energy barriers for the conformational transitions obtained by the ωB97X-D3BJ/def2-TZVP level via the relaxed energy surface scan of the torsion angle (Fig. S5 in the SI) displayed that when the local minimum is reached for the first time (after an intermediate structure formed), the high steric hindrance and distortion prevented returning to the original position of Py_Down-BF₄. Conversely, the Py_Up-BF₄ can distort the structural backbone reversibly by undergoing the redox process, which proves the switch behavior supported by the reversible formation of electrometers in this spectroelectrochemical experiment. On the other hand, two closely spaced reduction processes for Py_Down-BF₄ (Fig. 1b) can be explained by a potential compression mechanism, similar to what has been reported by Fortage et al., for other pyridinium derivatives.⁹ Energy surface scans indicate that there are two equivalent minima separated by high energy barriers (26–35 kcal·mol⁻¹; Fig. S5 in the SI), which limit structural reorganization after the first reduction. This sterically constrained environment allows only a partial relief of coulombic repulsion, resulting in the second electron being added at a potential very close to that of the first. Additionally, as mentioned above, spectroelectrochemical data support the presence of an irreversible structural distortion, Fig. 1(e), which is consistent with this mechanism.

2.2 Single-molecule switching in pyridinium molecular wires

Once the Py_Down-BF₄ (blue) and Py_Up-BF₄ (red) molecular wires were characterized by cyclic voltammetry to study the impact of the structural variation on their redox properties, single-molecule conductance was studied in both pyridiniums by the STM-BJ technique. The STM-BJ is a useful technique for determining the efficiency of electron transport through molecular wires and its relationship with their structural changes in the molecular backbone. Therefore, single-pyridinium junctions were achieved by breaking gold point contacts in 1 mM of each target of molecule in 1,2,4-trichlorobenzene solution; the single-molecule conductance was measured at a bias voltage of 100 mV to compare the influence of the core positions with that of pyridinium molecular wires. Thus, the conductance traces for each pyridinium showed plateaus near integer multiples of the quantum of conductance, G₀ = 2e²/h (where e is the charge of the electron and h is the Planck constant), with an additional plateau in a molecule-specific range below G₀, indicating the formation of a gold–pyridinium–gold junction¹⁵ (Fig. S6(d) in the SI). Measurements were carried out thousands of times and compiled into conductance histograms, which displayed a well-defined peak at a molecule-dependent conductance value (Fig. 2(a.1) and Fig. S6 in the SI). The molecular conductance obtained for Py_Up-BF₄ was higher than that obtained for Py_Down-BF₄, which is 3.93 × 10⁻⁵ and 2.02 × 10⁻⁵ G₀, respectively (Fig. 2(a.1)). Therefore, the core positions impact the conductance values slightly, where both values are in the same order of magnitude.

Fig. 2 (a.1) One-dimensional conductance histograms for Py_Up-BF₄ (red histogram) and Py_Down-BF₄ (blue histogram) without the supporting electrolyte in 1,2,4-trichlorobenzene constructed from 5000 traces. Measurements were carried out at a bias voltage of +0.1 V. (b.1) One-dimensional conductance histograms for Py_Up-BF₄ constructed from 4000 traces at bias voltages of +0.5 V (green lines) and −0.5 V (red lines) and (c.1) constructed from 4000 traces at bias voltages of +1.2 V (green lines) and −1.2 V (red lines). (d.1) One-dimensional conductance histograms for Py_Down-BF₄ constructed from 4000 traces at bias voltages of +0.5 V (green lines) and −0.5 V (blue lines). (a.2) Single-pyridinium junction for Py_Up-BF₄ and Py_Down-BF₄. (b.2) Simple Py_Up-BF₄ binding without the conformational state change. (c.2) Single-pyridinium junction switching between two conformation states for Py_Up-BF₄. (d.2) Pyridinium-junction, Py_Down-BF₄, without reversible changes in their conformational state.

Once the conductance values for each pyridinium molecular wire were determined, the STM-BJ was used to detect the switch behavior of Py_Up-BF₄ in a gold–pyridinium–gold junction circuit but without an electrochemical gate, namely, only with bias voltage applied. As demonstrated in the previous section, Py_Down-BF₄ was used as a control system because the switch behavior is blocked. In order to determine the influence of the bias voltage to promote the switch behavior in Py_Up-BF₄ into the junction, different bias values were applied from 0.1 to 1.2 V. In addition, the bias's polarity was changed from positive to negative (+/−) at each bias value. Approximately four thousand traces were collected and compiled with the conductance histograms at each bias voltage polarity. According to this, the histograms for the Py_Up-BF₄ were indifferent to the changes in the polarity of bias voltage (+/−)0.5 V, presenting a well-defined conductance peak, Fig. 2(b.1) and Table 1. However, at a bias value of +1.2 V, the histograms obtained for every four thousand traces (green line in Fig. 2(c.1)) displayed a change in the shape peak, widening it and shifting the conductance peak to a higher value when −1.2 V is applied (red line in Fig. 2(c.1) and Table 1). Interestingly, this behavior is reversible, can be controlled only with the changes in polarity bias voltage (+/−) and can be attributed to the switch behavior between two stable states in the pyridinium junction due to electromer formation (axial and pyramidal conformation, Fig. 2(c.2)). Despite (+/−) 1.2 V being a high potential, it is important to mention that the formal potential value for a bi-electronic process in electrochemical response is 1.264 V vs. Ag/AgNO₃, corresponding to the inversion potential value.²⁶ Thus, the switching in the single-conductance behavior was supported by three (+/−) cycles, each one by alternating four thousand traces at +1.2 V and four thousand traces at −1.2 V, successively. Because the structural changes in the molecular backbone of the single-molecule conductance impact directly the configuration and distribution of metal–molecule–metal junctions, the broad histograms obtained at a bias of −1.2 V would correspond to the pyridinium core to be pyramidalized conformation. Thus, the gold tip could interact with two nitrogen atoms (reduced core and pyridine) in this conformation, Fig. 2(c.2).

Table 1 Molecular conductance values were determined from the peak fitting performed to the 1D-histograms for Py_Down-BF₄ and Py_Up-BF₄ obtained at different polarity bias voltages (+/−)

Py	Bias potential/V	Cycle	Molecular conductance G0/×10^−5
Up	+0.1	—	3.93 ± 0.58
Down	+0.1	—	2.02 ± 0.16
Up	+0.5	1	9.57 ± 0.18
	−0.5	1	5.88 ± 0.34
	+0.5	2	7.00 ± 0.18
	−0.5	2	5.40 ± 0.34
Up	+1.2	1	4.43 ± 0.35
	−1.2	1	8.57 ± 0.87
	+1.2	2	4.44 ± 0.35
	−1.2	2	4.92 ± 0.87
	+1.2	3	4.55 ± 0.35
	−1.2	3	7.11 ± 0.87
Down	+0.5	1	1.95 ± 0.12
	−0.5	1	1.68 ± 0.06
	+0.5	2	4.10 ± 0.86
	−0.5	2	2.00 ± 0.63

---

On the other hand, the “control” pyridinium junction, Py_Down-BF₄, presented no reversible changes in the peak conductance histograms at a bias voltage of (+/−) 0.5 V, where the peak shape was widening each time the polarity bias was changed, regardless of polarity (+/−), Fig. 2(d.1). This behavior could denote the decomposition of the molecule, or, considering the spectroelectrochemical response (Fig. 1(d) and (e)), at this bias voltage, the ring with the methyl thiophenol group could rotate in the molecular backbone plane without returning to the original position, Fig. 2(d.2). Thus, the single-pyridinium conductance presents a progressive broad distribution to end the experiments (second cycle), Fig. 2(d.1), where the initial peak did not restart with the polarity, Fig. 2(d.1). Thus, the switching behavior displayed for Py_Up-BF₄ molecular wire into the circuit metal–molecule–metal junctions makes it an excellent candidate for applications such as molecular switches and molecular machines developing different electronic devices. As mentioned above, it has been reported that the adsorption process of pyridinium redox centers on the surface promotes the deactivation of the switching process.⁴ Therefore, it is important to highlight that this reversible switching between redox isomers is promoted without an electrochemical gating or an electrolytic environment and when the molecule is in direct contact with the surface (substrate and STM tip).

2.3 Impact of the counterions from the electrolyte environments in the single-molecule conductance of pyridinium isomer wires: experimental and computational studies

Because the pyridinium molecular wires dissociate in solution in Py⁺ and BF₄⁻ (as a counterion), we studied the influence of the electrolyte on the single-molecule conductance for each isomer, Py_Up-BF₄ and Py_Down-BF₄. The main objective was to form molecular junctions in electrolyte environments with different anions able to interact with the pyridinium cation, Py⁺. Thus, the isomers worked as molecular “models” to explore anions’ effects on the molecular junctions. The electrolytes used for the study correspond to four tetrabutylammonium salts (TBA) with anions in a wide range of coordinating strengths (BF₄⁻, PF₆⁻, Br⁻, and Cl⁻).^23,27 First, we studied pyridinium Py_Down-BF₄ because this molecule does not form electromers. Therefore, as shown in the above section, any change in single-molecular conductance below a bias voltage of (+/−) 0.5 V can be attributed in the first instance to the effects of the counter-ion. The experiments were performed immediately after preparing the solutions to detect the anion exchange dynamics at the molecular junction, Fig. 3(a.1); thus, the 1D and 2D-real-time histograms (D = dimensional) were built continuously every thousand measurement traces (Fig. 3(a.2)–(a.4) and Fig. S7 and S8). The histogram obtained using TBAPF₆ as an electrolyte distinguishes two conductance peaks during the first thousand traces, Fig. 3(a.2) and (a.3). After the two thousand traces, those peaks converge to a single-conductance peak that is maintained over the measurement time (black line in Fig. 3(a.2)). This peak is similar to that obtained in the absence of an electrolyte at 2.02 × 10⁻⁵ G₀, Fig. 2(a.1) and Fig. S6 in the SI. This dynamic evolution behavior is also displayed in 2D real-time conductance histograms (Fig. 3(a.3) and (a.4) and Fig. (S7 and S8) in the SI). Furthermore, three kinds of conductance traces are observed in the first two thousand measurements according to the number of steps below G₀. Namely, two steps in the same trace are observed at 2.43 × 10⁻⁴ and 1.32 × 10⁻⁵ G₀ (black line in Fig. 3(a.3) and blue line in Fig. S3(c) in the SI) and one-step traces are observed at 10⁻⁴ and 10⁻⁵ G₀, respectively (green and red lines in Fig. S7(c) in the SI). After two thousand traces, one-step traces at 10⁻⁵ G₀ are predominant (black and red lines shown in Fig. 3(a.4) and Fig. S7(c) in the SI, respectively). On the other hand, the 1D and 2D-real-time histograms for Py_Down-BF₄ obtained to form single-molecule junctions in TBACl, Fig. 3(b.1), display just a single conductance peak from the beginning of the measurement time, Fig. 3(b.2–b.4) and Fig. S9 and S10 in the SI. Furthermore, this single peak conductance is maintained over the measurement time, directly related to one-step traces, evidencing a stable single-molecule junction configuration (black line shown in Fig. 3(b.3 and b.4) and Fig. S9(c) in the SI). In addition, the 1D and 2D-real-time histograms obtained for TBABr displayed similar behaviour to that when TBACl is used as an electrolyte support, with a sharp conductance peak (Fig. (S11 and S12)). This behavior can be explained by the cation–anion coordination strengths in the Au–Py-Down-Au junctions for Cl⁻ and Br⁻ compared to PF₆⁻ and BF₄⁻.^23,28. In our models, the original anion–cation pairs (Py_Down-BF₄) undergo evolution to form a new pair with the anion from the electrolyte. Thus, the BF₄⁻ present as a counterion of the pyridinium cation could be shifted by the PF₆⁻ surrounding (which is 100 times more concentrated than native BF₄⁻) in the Au–Py-Down-Au junctions, and this shifting behaviour is detected at a single molecule level by the shape of the histograms. Namely, two conductance peaks for PF₆⁻ at the beginning converge to one stable conductance peak with time. Conversely, Cl⁻ and Br⁻ shift the native BF₄⁻ faster than PF₆⁻, exhibiting just one sharp peak in the time. We have carried out the STM-BJ measurements of the Py_Down-BF₄ with TBABF₄ as electrolyte support to support this behaviour. Thus, the 1D and 2D-real-time histograms displayed a stable conductance peak in this case (Fig. (S13 and S14)). Although this stable peak is broader than the peak of conductance observed for Py_Down-BF₄ without an electrolyte in the solution (Fig. 2(a.1)), this behaviour can be evidence of the addition of BF₄⁻ anions in the environment junctions, increasing the distribution of the molecule's geometric configuration between the two gold contacts.²⁹

Fig. 3 (a.1) General scheme for the STM-BJ configuration of Py_Down-BF₄ in an environment with the TBA-PF₆ electrolyte (Py_Down-BF₄/PF₆⁻) and (b.1) Py_Down-BF₄ in an environment with the TBA-Cl electrolyte (Py_Down-BF₄/Cl⁻). (a.2) 1D-histograms obtained in real-time for the system Py_Down-BF₄/PF₆⁻ in the trace ranges of 1001–3001 (blue line), 3001–5001 (green line), and 5001–13 001 (red line) and the entire trace range of 1001–13 001 (dark line). (b.2) 1D-histograms obtained in real-time for the system Py_Down-BF₄/Cl⁻ in the trace ranges of 3001–5001 (blue line), 5001–8001 (green line), and 8001–11 001 (red line) and the entire trace range of 3001–11 001 (dark line). (a.3) 2D-histograms obtained in real-time for Py_Down-BF₄/PF₆⁻ in the initial trace range of 1001–3001 and (a.4) the final trace range of 5001–13 001. (b.3) 2D-histograms obtained in real-time for Py_Down-BF₄/Cl⁻ in the initial trace range of 3001–5001 and (b.4) the final trace range of 8001–11 001. The black lines in the 2D histograms correspond to the most representative trace of each counterion exchange. All measurements were carried out at a bias voltage of +0.1 V.

In the case of the Py_Up-BF₄ isomer at the single-molecule junctions in an electrolyte environment with TBAPF₆, Fig. 4(a.1), the 1D and 2D-real-time histograms displayed a similar behavior to those of Py_Down-BF₄. Namely, the histograms suffer a dynamic evolution over time. In the first two thousand traces, one peak is elucidated at high conductance, 4.99 × 10⁻⁴ G₀, and a broader low-conductance peak is observed at 2.22 × 10⁻⁵ G₀, as illustrated in Fig. 4(a.2) and (a.3) and Fig. S15 in the SI, Table 2. The presence of this broader peak indicates a larger distribution of junction geometries and molecular conformations which contribute to transport within this regime.³⁰ As the measurements progress, a new, well-defined peak emerges at 5.33 × 10⁻⁵ G₀. The final histograms reflect this configuration, represented by the purple line in Fig. 4(a.2). The conductance traces obtained during the real-time STM-BJ measurements help illustrate the dynamic evolution of the peaks observed in the histograms, which transition from two peaks to a single peak. Specifically, the conductance traces (the black line in Fig. 4(a.3) and the red line in Fig. S15 in the SI) indicate the presence of two conductance configurations in the single-molecule junctions at the initial stage (evidenced by two distinct steps in the trace). Over time, these configurations converge into a more stable state, as shown by the elongated diagonal step in the trace (black line in Fig. 4(a.4) and Fig. S15 in the SI). Therefore, like the Py_Down-BF₄ isomer, the native BF₄⁻ can be shifted by the PF₆⁻ anion (present from the electrolyte) in the molecular junction, and the differences in the histogram shapes between both isomers are a result of the molecular backbone changes in the pyridinium's core (charged positively) position.¹³

Fig. 4 (a.1) General scheme for the STM-BJ configuration of Py_Up-BF₄ in an environment with the TBA-PF₆ electrolyte (Py_Up-BF₄/PF₆⁻) and (b.1) Py_Up-BF₄ in an environment with the TBA-Cl electrolyte (Py_Up-BF₄/Cl⁻). (a.2) 1D-histograms obtained in real-time for the system Py_Up-BF₄/PF₆⁻ in the trace ranges of 1001–3001 (red line), 3001–6001 (green line), and 6001–9001(black line) and the entire trace range of 1001–9001 (purple line). (b.2) 1D-histograms obtained in real-time for the system Py_Up-BF₄/Cl⁻ in the trace ranges of 1001–4001 (red line), 4001–8001 (green line), and 8001–12 001(black line) and the entire trace range of 1001–12 001 (purple line). (a.3) 2D-histograms obtained in real-time for Py_Up-BF₄/PF₆⁻ in the initial trace range of 1001–3001 and (a.4) the final trace range of 6001–9001. (b.3) 2D-histograms obtained in real-time for Py_Up-BF₄/Cl⁻ in the initial trace range of 1001–4001 and (b.4) the final trace range of 8001–12 001. The black lines in the 2D histograms correspond to the most representative trace of each counterion exchange. All measurements were carried out at a bias voltage of +0.1 V.

Table 2 Molecular conductance values obtained from the peak fitting performed to the real-time 1D-histograms for each Py_Down-BF₄ and Py_Up-BF₄ dissolved in the electrolytes based on TBA with different anions (PF₆⁻, BF₄⁻, Br⁻, and Cl⁻)

Real-time per min		40		80		160^a		160^a
Traces/1 × 10^3		1–3		3–5		5–9^a		1–9^a
Molecular conductance/G0
Py	Anion	G0/×10^−4	G0/×10^−5	G0/×10^−4	G0/×10^−5	G0/×10^−4	G0/×10^−5	G0/×10^−4	G0/×10^−5
		Peak 1	Peak 2	Peak 1	Peak 2	Peak 1	Peak 2	Peak 1	Peak 2
a This value represents the average of the measurements performed in the different electrolyte environments. See the SI.
Down	PF6^−	2.43 ± 0.40	1.32 ± 0.01	—	3.32 ± 0.06	—	3.47 ± 0.13	—	3.20 ± 0.01
	BF4^−	—	7.35 ± 0.77	—	7.14 ± 0.54	—	7.32 ± 0.74	—	7.15 ± 0.01
	Br^−	—	4.62 ± 0.55	—	5.87 ± 1.33	—	4.88 ± 0.68	—	4.80 ± 0.63
	Cl^−	—	8.28 ± 0.24	—	8.12 ± 0.28	—	8.56 ± 0.11	—	8.92 ± 0.36
Up	PF6^−	4.99 ± 0.87	2.22 ± 0.25	—	5.14 ± 0.09	—	5.67 ± 0.46	—	5.33 ± 0.17
	BF4^−	3.57 ± 0.13	2.66 ± 0.12	3.64 ± 0.74	4.25 ± 0.60	4.43 ± 0.74	4.55 ± 0.29	3.89 ± 0.03	4.15 ± 0.18
	Br^−	—	4.11 ± 0.02	—	4.58 ± 0.01	—	4.08 ± 0.15	—	4.11 ± 0.10
	Cl^−	—	5.63 ± 0.10	—	5.65 ± 0.01	—	5.80 ± 0.13	—	5.68 ± 0.04

---

Moreover, the real-time 2D histograms compiled displayed the dynamic evolution of the molecular junctions (Fig. 4(a.3) and (a.4) and Fig. S15(d.1–d.4) and S16 in the SI). In contrast, when the single-molecule junctions formed for Py_Up-BF₄ in the TBACl electrolyte environment, only one peak was well-resolved at 5.68 × 10⁻⁵ G₀, Fig. 4(b.2). However, in comparison to the Py_Down-BF₄-TBACl system, this peak is broader than the one obtained from that isomer due to the conductance traces exhibiting a high distribution between one and two steps (black line in Fig. 4(b.3) and (b.4) and Fig. S17(c) in the SI). The real-time 2D histograms compiled displayed this behavior, Fig. 4(b.3) and (b.4) and Fig. S17. Interestingly, 1D and 2D-real-time histograms obtained for Py_Up-BF₄ in the TBABr environment also presents one well-resolved peak at 4.11 × 10⁻⁵ G₀ (Fig. S19 and S20 in the SI). However, the histogram's shape differs from that obtained when TBACl is used as an electrolyte and for the Py_Down-BF₄-TBABr system. This difference in the molecular junction is expressed in the conductance traces, which exhibit an elongated diagonal one-step (Fig. S19(c)), unlike the other two systems mentioned. Finally, for Py_Up-BF₄ in the TBABF₄ environment, the 1D and 2D-real-time histograms obtained displayed two peaks at the beginning at 3.57 × 10⁻⁴ and 2.66 × 10⁻⁵ G₀ which converge to a prominent peak at 4.15 × 10⁻⁵ G₀ with a shoulder at 3.89 × 10⁻⁴ G₀. The conductance traces and the real-time 2D histograms reflect this behavior, as shown in Fig. S21 and S22. Thus, this behavior showed that the backbone's core position directly influences the junction's anion–cation pair. Unlike the Py_Down-BF₄-TBABF₄ system, which displayed just one peak conductance during the measurement (Fig. S13 and S14 in the SI), the Py_Up-BF₄-TBABF₄ displayed a dynamic evolution during the STM-BJ measurements.

Because the core position in the molecular backbone determines the shape of the 1D and 2D real-time histograms for the Py_Up-BF₄ and Py_Down-BF₄ in each electrolyte environment as a fingerprint, and the anion's electrolyte determines the dynamic evolution of the peak conductance, we have obtained by the DFT calculations the adsorption energies E_ads, adsorption enthalpy ΔH_ads, and the adsorption free energies ΔG_ads for all anion–cation pairs. First, all the anion–cation pairs are stable in solution, forming non-bonded structures (Fig. 5(a)), with adsorption energies E_ads in the range of −5.7 to −4.3 kcal mol⁻¹ (Table 3). The PF₆⁻ anion displays the strongest interaction in both “Down” and “Up” pyridinium conformations (E_ads = −5.7 kcal mol⁻¹), where the higher affinity with the cation could be related to their high volume and higher polarizability (27.1 a.u.) compared to other anions (<19.8 a.u). The stability of the Cl⁻–cation pair is also remarkable, with an E_ads value of at least −5.4 kcal mol⁻¹. Although there are slight differences between the “Up” and “Down” configurations, the “Up” configuration could be slightly more favorable for certain anions, such as Cl⁻. Nevertheless, the similar adsorption energies between “Up” and “Down” configurations show that the geometric orientation has a minimal impact on the stability; thus, the adsorption process is mainly controlled by intermolecular interactions (e.g., electrostatic and dispersion). Accordingly, the ALMO-EDA decomposition shows that electronic stability emerges due to a balanced contribution of non-bonded interactions, i.e., electrostatic (ΔE_ELEC) and dispersion effects (ΔE_DISP), accounting for up to ∼92% of the stabilizing energy (Fig. 5(b)).

Fig. 5 (a) Minimum energy adsorption conformations of anion–cation pairs. (b) The relative single percentage contributions of ALMO-EDA terms (%ΔE_i, in %) of stabilizing energies to the adsorption energy E_ads.

Table 3 Stability of the cation–anion pairs: adsorption energy (E_ads) for the anion–cation pair formation, Gibbs free energy for the counter-ion exchange of the native BF₄⁻ by an anion X (ΔG), and solvation energy (ΔE_SOLV). Values are presented in kcal mol⁻¹

Py	Anion	Eads	ΔG	ΔESOLV
Down	BF4^−	−4.5	0.0	70.1
	Br^−	−4.8	−6.0	74.1
	Cl^−	−5.4	−7.5	77.9
	PF6^−	−5.7	−2.8	65.6
Up	BF4^−	−4.9	0.0	68.3
	Br^−	−4.8	−6.0	73.4
	Cl^−	−5.7	−7.2	76.9
	PF6^−	−5.7	−2.4	64.2

---

We also analyse the spontaneity of the ion-pair formation on the basis of the counter-ion exchange of the native BF₄⁻ bound to Py⁺ by an anion X⁻ supplied by the electrolyte, i.e.,

P_y⋯BF₄⁻ + X⁻ ⇄ P_y⋯X⁻ + BF₄⁻ (2)

Under this assumption, the entropy change resulting from the release of BF₄⁻ favors the spontaneity of the process. The Gibbs free energy ΔG of the ion exchange process is negative in all cases, showing that the ion exchange is thermodynamically spontaneous under standard conditions. Note that ΔG = 0 for the BF₄⁻ case because the ion-exchange is in equilibrium in this case. The trend in the ΔG values is Cl⁻ ≈ Br⁻ > PF₆⁻, fully consistent with the real-time histograms: rapid stabilization for Cl⁻/Br⁻ vs. slower evolution with PF₆⁻. In addition, it must be pointed out that, experimentally, a bias potential is applied to molecular junctions, perturbing the nano-environment surrounding the junction, which influences the formation of cation–anions pairs and stability. Moreover, the cation–Cl⁻ pairs show the lowest ΔG values among the studied anions, indicating a more favorable pair formation. The high solvation energy (ΔE_SOLV, Table 3) of the cation–Cl⁻ pairs penalizes their stability in the free state, but it also results in a significant energy release upon adsorption, shifting the equilibrium towards the cation–anion pair formation. This behavior is associated with the small size and high charge density of Cl⁻, causing strong electrostatic/dispersion interactions with the cation while interacting effectively with the polar solvent (propylene carbonate). Therefore, Cl⁻ anions are preferred in processes where fast and stable adsorption is required.

2.4 Charge transport properties of the isomers

Employing the molecular junction architectures described in the computational results, we probed the transport properties of the systems through the DFT-NEGF methodology. The transmission spectra and the projected density of states (PDOS) on the molecular unit are shown in Fig. 6, and the conductance values computed for each system are summarized in Table 4.

Fig. 6 Transmission spectra and projected density of states (PDOS) on the molecular central unit computed for each of the Py_Up-BF₄ and Py_Down-BF₄ systems through the DFT-NEGF methodology. The red arrows indicate the position of the LUMO calculated through the molecular projected self-consistent Hamiltonian (MPSH) states.

Table 4 Zero bias molecular conductance values (G₀) for the Py_Up-BF₄ and Py_Down-BF₄ isomers calculated through the DFT-NEGF methodology

Py	Anion	Molecular conductance/G0
a No electrolyte.
Down	—^a	2.01 × 10^−3
	PF6^−	1.04 × 10^−2
	Cl^−	1.56 × 10^−2
Up	—^a	1.82 × 10^−3
	PF6^−	2.09 × 10^−2
	Cl^−	2.42 × 10^−2

---

As can be seen, both pyridinium isomers exhibit practically the same zero bias conductance value, around 2 × 10⁻³ G₀. In both cases, the main contribution to the conductance stems from a broad transmission peak above the Fermi level. This last fact, added to the corresponding broad peaks in the PDOS centered at the same energy values, suggests a resonance with one of the molecule frontier virtual orbitals. To obtain further insight into the transport mechanisms, we computed the isomer's frontier orbitals under the influence of the electrodes through the molecular projected self-consistent Hamiltonian (MPSH) states. These states are obtained by diagonalizing the molecular part of the full self-consistent Hamiltonian of the molecular junction device. We observe that, although both systems present similar zero-bias conductance, the features of the transmission peaks are quite different. On the one hand, their intensity in the Py_Up-BF₄ isomer junction is almost three times greater than that in the Py_Down-BF₄ one; nevertheless, the former is 0.1 eV farther from the Fermi level than the latter. The effects of the decrease in intensity and position from the Fermi level appear to compensate, resulting in similar zero bias conductance values for both systems, though much greater current values for the Py_Up-BF₄ are expected when applying a bias voltage.

Regarding the addition of the TBAPF₆ and TBACl electrolytes, it can be observed that, for both isomers and, regardless of the electrolyte used, the zero bias conductance increases nearly in one order of magnitude. As in the previous case, the conductance value chiefly depends on the resonance of the central molecular unit's LUMO (Fig. 6), but this time, the peaks in the transmission and PDOS, as well as the position of the LUMOs, are shifted towards the Fermi level around 0.1 eV for the Py_Up-BF₄ and 0.05 eV for the Py_Down-BF₄ isomer, for both electrolytes. The smaller displacement of the main transmission peak in the Py_Down-BF₄ case is now compensated with a threefold increase in intensity, yielding a close conductance value between both isomers, in analogy to the effect observed in the systems without electrolytes. It is worth mentioning that, after the addition of the electrolytes, the wave function of the LUMO experienced no important changes in its main features (Fig. 7).

Fig. 7 Wave functions corresponding to the LUMO of each of the studied isomers computed through the molecular projected self-consistent Hamiltonian (MPSH) states.

3. Experimental

3.1 Materials and reagents

The structural isomers of pyridinium molecules, 1-(4-(methylthio)phenyl)-2,6-diphenyl-4-(4-pyridyl)pyridinium tetrafluoroborate (Py_Down-BF₄) and 1-(4-pyridyl)-2,6-diphenyl-4-[4(methylthio)phenyl]pyridinium tetrafluoroborate (Py_Up-BF₄), were synthesized following the experimental methodology previously reported by our research group.¹³

3.2 Electrochemical measurements

Cyclic voltammetry was performed using a bipotentiostat (Pine instrument, model WaveDriver 200 EIS). The working electrode was a platinum electrode obtained from CH instruments (model CHI102, diameter 2.0 mm). Before each experiment, electrodes were electrochemically cleaned for 10 min using electrolysis in 0.5 M H₂SO₄. The electrolysis was performed at −320 mV vs. Ag/AgCl (3.0 M KCl). The electrode surface was mechanically polished for 5 minutes using a cloth with a drop of 0.1 μm alumina suspension (Struers, AP-A suspension) followed by ultrasonic treatment in Milli-Q water for 2 minutes. The electrodes were then dried under nitrogen flow. Platinum wire was used as the counter-electrode. The Ag/AgNO₃ reference electrode was used and prepared as indicated for the fabricant (Pine instrument).³¹ 0.1 M tetrabutylammonium tetrafluoroborate (TBABF₄, AK Scientific, <98%) in acetonitrile (ACN) was used as the electrolytic solution. The concentration for Py_Down-BF₄ = 1.125 × 10⁻⁴ M and ferrocene = 2.25 × 10⁻⁴ M; the concentration for Py_Up-BF₄ = 7.5 × 10⁻⁵ M and ferrocene = 1.5 × 10⁻⁴ M. All CV experiments were conducted at a scan rate of 50 mV s⁻¹ and potential values are given versus Ag/AgNO₃.

3.3 UV-vis spectroelectrochemistry measurements

Optical spectroelectrochemical experiments were performed using a spectroelectrochemical bundle from Pine Instruments, featuring a “Honeycomb” cell kit with Pt working and counter electrodes and an Ag/AgNO₃ reference electrode.³¹ 0.1 M tetrabutylammonium tetrafluoroborate (TBABF₄, AK Scientific, <98%) in acetonitrile (ACN) was used as the electrolytic solution. The concentration for Py-Down-BF₄ was 1.125 × 10⁻⁴ M, and the concentration for Py-Up-BF₄ was 7.5 × 10⁻⁵ M.

3.4 Scanning tunneling microscopy break-junction (STM-BJ) measurements

The STM-BJ measurements were carried out using a custom-made scanning tunneling microscope to perform a break-junction technique. The setup and method have been previously described.³² The gold substrates used are commercial substrates of diameter 10 nm (PELCO®, Gold Coated AFM/STM Metal Specimen Discs, TED PELLA, INC).³³ The gold substrates were subjected to a cleaning treatment before use. For this, the substrates were immersed in a fresh 3 : 1 piranha solution of H₂SO₄ (J.T. Baker®):H₂O₂ (Supelco, Perhydrol® 30%) for 30 s, after the substrates were rinsed for 1 min with ultrapure water (Milli-Q system, 18.2 MΩ cm). This last step was repeated five more times. Finally, the gold substrates were dried with a constant flow of ultrapure nitrogen (99.99%, LINDE). Subsequently, the Au substrates were introduced for 15 min into a constant O₃ flow system (O₃ generator model CE EMC LVD FCC, voltage = 223 V; current = 0.21 A; force = 34.7 W).³² The STM tip was a gold metal wire (diameter: 0.25 mm and purity: 99.99%; American elements) and cut diagonally with Tronex pliers (7112).

3.5 STM-BJ switch behavior

1 mM solutions of Py_Down-BF₄ and Py_Up-BF₄ dissolved in 1,2,4-trichlorobenzene (Sigma-Aldrich, 99.0%) were prepared. Two drops of solution were deposited on the clean gold substrate, and the STM-BJ measurements were performed by breaking gold point contacts between the substrate and the STM tip; the conductance was measured at different bias voltage values and polarities (+/−). For obtaining each of the conductance traces of the Au-Py_Down-BF₄-Au and Au-Py_UpBF₄-Au systems, the gold tip was continually brought close enough to the gold substrate (to a conductance value greater than G₀) and pulled at a pull rate of 20 nm s⁻¹; the tip excursion was limited to 15 nm from the substrate surface. The conductance trace data were obtained and recorded using Igor version 6.0 (WaveMetrics Inc.). From these data, 1D-histograms (D = dimensional) were constructed without any data selection. The histograms were grouped into quantities between 5000 and 4000 traces and adjusted to a Lorentzian curve to obtain the conductance value and compare it with each of the systems. 2D-histograms were constructed based on the conductance value as a function of displacement, also grouping the same number of traces as the 1D histogram.

3.6 STM-BJ counterion behavior

1 mM solutions of Py_Down-BF₄ and Py_Up-BF₄ dissolved in propylene carbonate (Sigma-Aldrich, <99.7.0%) with 0.1 M of each N-tetrabutylammonium (TBA) salts, i.e., tetrabutylammonium hexafluoro-phosphate (TBAPF₆, Sigma-Aldrich, <99.0%), tetrabutylammonium tetrafluoroborate (TBABF₄, AK Scientific, <98%), tetrabutylammonium chloride (TBACl, AK Scientific, <98.0%) and tetrabutylammonium bromide (TBABr, AK Scientific, <98.0%), were prepared. In the case of TBACl, this compound was previously dried in an oven at 60 °C for 48 h to eliminate the water of hydration, and then it was subsequently stored in a vacuum desiccator. Each of the eight solutions: Py_Down-BF₄/TBAPF₆, Py_Down-BF₄/TBABF₄, Py_Down-BF₄/TBACl, Py_Down-BF₄/TBABr, Py_Up-BF₄/TBAPF₆, Py_Up-BF₄/TBABF₄, Py_Up-BF₄/TBACl, and Py_Up-BF₄/TBABr, were prepared and measured immediately. Thus, two drops of each target solution were deposited on the clean gold substrate, and the STM-BJ measurements were performed immediately by breaking gold point contacts between the substrate and the STM tip. The STM tip was covered with Apiezon Wax to avoid leakage currents.¹³ Continually, the histograms were registered, and every thousand conductance traces were complied. The time required for measuring one thousand traces was 20 minutes. Data processing was conducted in the same way as mentioned in the previous section.

3.7 Theoretical methodologies of adsorption conformations and stability

Metropolis Monte Carlo search found the initial adsorption conformations between 100 and 1.0 × 10⁵ K in the Adsorption Locator module of Material Studio.³⁴ Further geometry optimization without constraints at the ωB97X-D3BJ/def2-TZVP level in ORCA 5.0.1 was performed for the five lowest energy minimum candidates.^35–38 Vibrational analyses were performed on the optimized geometries to confirm the true minimum nature (lack of imaginary frequencies) and obtain thermochemical data under room conditions (T = 298 K and P = 1 atm). Implicit solvent effects were included with the conductor-like polarizable continuum model (CPCM) and using propylene carbonate as solvent.³⁹ The adsorption energy (E_ads) was computed as follows:

E_ads = E_{Ligand–Anion} − (E_Ligand + E_Anion) + Δ_ZPE (3)

where E_Ligand, E_Anion, and E_{Ligand–Anion} are the total energies of the free ligand, the free anion (PF₆⁻, BF₄⁻, Cl⁻, or Br⁻), and the cation–anion pair, respectively; Δ_ZPE is the energy contribution from the zero-point energies. Hence, the more negative the E_ads values, the more stable the cation–anion pair is. The geometrical counterpoise correction (gCP) was used to avoid the E_ads underestimation due to the basis set limit.⁴⁰ E_ads values were decomposed into physicochemical contributions [electrostatics (ΔE_ELEC), dispersion (ΔE_DISP), charge transfer (ΔE_CT), polarization (ΔE_POL), and solvation (ΔE_SOLV)] in the Q-CHEM 6.2 program via the ALMO-EDA method at the ωB97X-D3BJ/def2-SVP level.⁴¹

3.8 Charge transport properties of the Py_Up-BF₄ and Py_Down-BF₄ isomers

The charge transport properties of the Py_Up and Py_Down structural pyridinium isomers, as well as all geometry optimizations involved in the construction of the junctions, were obtained using the QuantumATK software package.^42,43 In all of the calculations, all atoms were represented by a double-ζ-polarized (DZP) basis set, along with PseudoDojo norm-conserving pseudopotentials,⁴⁴ employing the Perdew–Burke–Ernzerhof (PBE)⁴⁵ exchange–correlation (XC) functional, widely used in the determination of the transport properties in molecular junctions. The mesh cut-off energy was set to 210 Rydberg for both the real and reciprocal space grids in all calculations, with the self-consistency tolerance for the convergence of the Hamiltonian and density matrices being 1.0 × 10⁻⁴ eV.

The molecular junctions were constructed starting with the full geometry relaxation of the Py_Up and Py_Down isomers accompanied by the [BF₄]⁻ counterion, as well as the aforementioned systems with the addition of the TBAPF₆ and TBACl electrolytes, anchored to two Au(111) surfaces by means of the thiol and pyridine terminal groups bonded to the top Au atom of two 4 atom pyramids attached to each Au(111) surface. For all systems, the unit cell consisted of the molecular kernel between two Au(111) slabs formed by three 9 × 7 layers of Au atoms. During all geometry optimizations, the unit cell size was allowed to relax in the transport direction and kept fixed in transverse ones. Subsequently, we construct the molecular junctions employing the optimized geometries described above as the scattering zone, adding four more 9 × 7 Au layers on each side. Fig. S23 and S24 show the molecular junctions employed for the Py_Up-BF₄ and Py_Down-BF₄ isomers, respectively, with and without electrolytes (TBAPF₆ and TBACl). Then, the transport properties of the junctions were computed using the combination of the DFT with the Keldysh nonequilibrium Green's function (NEGF) formalism, known as the DFT-NEGF methodology.^46–48

4. Conclusions

We investigated the electronic, electrochemical, and spectroelectrochemical properties of two pyridinium isomers, Py_Down-BF₄ and Py_Up-BF₄, which differ only in the position of the redox-active core. Using STM-BJ, we examined how the core positioning influences single-molecule conductance in the absence and presence of various electrolytes (TBABF₄, TBAPF₆, TBABr, and TBACl). Our findings show that the core position governs the redox behavior, with Py_Up-BF₄ exhibiting a bi-electronic redox process that enables switching between two stable states. The switching behavior was characterized both in solution and within the Au-Py-Up-BF₄-Au molecular junction, using spectroelectrochemical and STM-BJ techniques, respectively. Moreover, at the single-molecule level, the position of the core was found to influence the conductance of the pyridinium isomers, where Py_Up-BF₄ exhibits a higher conductance than Py_Down-BF₄. On the other hand, the shape of the 1D and 2D conductance histograms—junction distribution, width, and histogram's peak position—obtained for both pyridinium isomers in different electrolyte environments was susceptible to the nature of the anion component of the electrolyte used and the pyridinium's core position. Furthermore, molecular conductance underwent dynamic evolution in the single-molecule junction circuit for Py_Up-BF₄-TBAPF₆, Py_Down-BF₄-TBAPF₆, and Py_Down-BF₄-TBABF₄. This behavior was attributed to the real-time detection of interactions between the positively charged pyridinium core and the anions from the electrolyte solutions. These interactions determine the dynamic molecule junction distribution, i.e., the evolving shape of conductance histograms over time. DFT calculations confirmed that all anion–cation pairs are stable, forming non-bonded structures where the geometric orientation (i.e., Up vs. Down) has a slight impact on stability; thus, the adsorption process is predominantly governed by intermolecular interactions (electrostatic and dispersion forces). In addition, the formation of cation–anion pairs is not spontaneous under standard conditions due to solute–solvent interactions and entropy-related effects.

Theoretical transport calculations for Py_Up-BF₄ and Py_Down-BF₄ anchored between Au(111) electrodes show that both isomers have similar zero-bias conductance. Both isomers present a close zero-bias conductance value. In both cases, the main contribution to conductance stems from a transmission peak product of the resonance of the bridging molecule's LUMO. However, the features of the signal differ for both isomers: for Py_Up-BF₄, it is three times stronger than in the Py_Down-BF₄ case, but the latter is located 0.1 eV closer to the Fermi level than the former. These effects compensate each other, yielding a close zero-bias conductance value in the two junctions featuring each isomer. In the presence of the TBAPF₆ and TBACl electrolytes, for both isomers, the zero bias conductance increases one order of magnitude due to the displacement of the transmission peak mentioned above toward the Fermi level, which suggests that both isomers exhibit higher conductivity. The theoretical results are consistent with experimental findings, where the Py_Up-BF₄ exhibits a higher conductance by a factor of 2 than Py_Down-BF₄. Moreover, the conductance increases progressively in the presence of electrolytes, with no electrolyte showing the lowest conductance, followed by TBAPF₆, and then TBACl exhibiting the highest conductance.

Author contributions

Ana María Méndez-Torres: formal analysis, funding acquisition, investigation, methodology, visualization, writing – original draft, and writing – review and editing. Rubén Oñate: conceptualization, methodology, supervision and visualization. Dany S. Monje: writing – original draft and writing – review and editing. Ana Pizarro: methodology. Nicolás Montenegro-Pohlhammer: investigation and software. Nadim Darwish: formal analysis and writing – review and editing. Diego Cortés-Arriaga: investigation and software. Gloria Cardenas-Jiron and Ingrid Ponce: conceptualization, formal analysis, funding acquisition, investigation, methodology, software, supervision, validation, visualization, writing – original draft, and writing – review and editing.

Conflicts of interest

There are no conflicts of interest to declare.

Data availability

All data supporting the findings of this study are included within the manuscript and the SI. Supplementary information includes characterization methodologies for electrochemical, spectroelectrochemical, and scanning tunneling microscope (STM) break junction measurements, along with the determination of electrochemical parameters. Additionally, computational details related to energy barriers for conformational transitions and transport properties are also provided. See DOI: https://doi.org/10.1039/d5nh00422e

Raw data are available from the corresponding author upon reasonable request.

Acknowledgements

This research was supported by ANID FONDECYT Regular 1251260 (I. P.), 1251156 (D. C.-A), and 1221072 (G. C.-J.); ANID FONDECYT iniciación 11251644 (A. M. M.-T); ANID FONDECYT postdoctoral 3210181 (N. M.-P); ANID becas/doctorado nacional 21191379 (A. P.); Beca de Excelencia para Extranjeros 2021: modalidad regular (D. S. M). Powered@NLHPC: this research was partially supported by the NLHPC (ECM-02) supercomputing infrastructure of the Universidad de Chile and FONDEQUIP EQM180180. N. D. thanks the Australian Research Council (FT240100140). We are most grateful to Professor Latha Venkataraman from the Institute of Science and Technology Austria for her exceptional contribution to the development of the STM Break-Junction Setup at the University of Santiago de Chile.

References

1. A. Gosset, L. Wilbraham, Š. N. Lachmanová, R. Sokolová, G. Dupeyre, F. Tuyèras, P. Ochsenbein, C. Perruchot, H.-P. J. de Rouville, H. Randriamahazaka, L. Pospíšil, I. Ciofini, M. Hromadová and P. P. Lainé, J. Am. Chem. Soc., 2020, 142, 5162–5176.
2. Š. Lachmanová, G. Dupeyre, J. Tarábek, P. Ochsenbein, C. Perruchot, I. Ciofini, M. Hromadová, L. Pospíšil and P. P. Lainé, J. Am. Chem. Soc., 2015, 137, 11349–11364.
3. S. Li, J. Li, H. Yu, S. Pudar, B. Li, J. Rodríguez-López, J. S. Moore and C. M. Schroeder, J. Electroanal. Chem., 2020, 875, 114070.
4. Š. N. Lachmanová, G. Dupeyre, P. P. Lainé and M. Hromadová, Langmuir, 2018, 34, 6405–6412.
5. X. He, L. Chen and T. Baumgartner, ACS Appl. Mater. Interfaces, 2024, 16, 48689–48705.
6. S. Samaroo, A. L. Pattillo, D. Servinski, W. R. Kruper, D. D. Carter, T. F. Guarr and D. P. Hickey, ACS Appl. Energy Mater., 2024, 7, 7640–7648.
7. J. Fortage, F. Puntoriero, F. Tuyèras, G. Dupeyre, A. Arrigo, I. Ciofini, P. P. Lainé and S. Campagna, Inorg. Chem., 2012, 51, 5342–5352.
8. J. Fortage, G. Dupeyre, F. Tuyèras, V. Marvaud, P. Ochsenbein, I. Ciofini, M. Hromadová, L. Pospísil, A. Arrigo, E. Trovato, F. Puntoriero, P. P. Lainé and S. Campagna, Inorg. Chem., 2013, 52, 11944–11955.
9. J. Fortage, C. Peltier, C. Perruchot, Y. Takemoto, Y. Teki, F. Bedioui, V. Marvaud, G. Dupeyre, L. Pospísil, C. Adamo, M. Hromadová, I. Ciofini and P. P. Lainé, J. Am. Chem. Soc., 2012, 134, 2691–2705.
10. J. Fortage, C. Peltier, F. Nastasi, F. Puntoriero, F. Tuyèras, S. Griveau, F. Bedioui, C. Adamo, I. Ciofini, S. Campagna and P. P. Lainé, J. Am. Chem. Soc., 2010, 132, 16700–16713.
11. C. Gutiérrez-Ceron, R. Oñate, J. H. Zagal, A. Pizarro, J. F. Silva, C. Castro-Castillo, M. C. Rezende, M. Flores, D. Cortés-Arriagada, A. Toro-Labbé, L. M. Campos, L. Venkataraman and I. Ponce, Electrochim. Acta, 2019, 327, 134996.
12. I. Ponce, A. C. Aragonès, N. Darwish, P. Pla-Vilanova, R. Oñate, M. C. Rezende, J. H. Zagal, F. Sanz, J. Pavez and I. Díez-Pérez, Electrochim. Acta, 2015, 179, 611–617.
13. A. Pizarro, G. Abarca, C. Gutiérrez-Cerón, D. Cortés-Arriagada, F. Bernardi, C. Berrios, J. F. Silva, M. C. Rezende, J. H. Zagal, R. Oñate and I. Ponce, ACS Catal., 2018, 8, 8406–8419.
14. R. Z. Snitkoff-Sol, A. Friedman, H. C. Honig, Y. Yurko, A. Kozhushner, M. J. Zachman, P. Zelenay, A. M. Bond and L. Elbaz, Nat. Catal., 2022, 5, 163–170.
15. H. Yu, J. Li, S. Li, Y. Liu, N. E. Jackson, J. S. Moore and C. M. Schroeder, J. Am. Chem. Soc., 2022, 144, 3162–3173.
16. T. Fu, K. Frommer, C. Nuckolls and L. Venkataraman, J. Phys. Chem. Lett., 2021, 12, 10802–10807.
17. J. Dalmieda, W. Shi, L. Li and L. Venkataraman, Nano Lett., 2024, 24, 703–707.
18. Š. Nováková Lachmanová, V. Kolivoška, J. Šebera, J. Gasior, G. Mészáros, G. Dupeyre, P. P. Lainé and M. Hromadová, Angew. Chem., Int. Ed., 2021, 60, 4732–4739.
19. Š. N. Lachmanová, J. Šebera, V. Kolivoška, J. Gasior, G. Mészáros, G. Dupeyre, P. P. Lainé and M. Hromadová, Electrochim. Acta, 2018, 264, 301–311.
20. J. Li, S. Pudar, H. Yu, S. Li, J. S. Moore, J. Rodríguez-López, N. E. Jackson and C. M. Schroeder, J. Phys. Chem. C, 2021, 125, 21862–21872.
21. T. Hines, I. Díez-Pérez, H. Nakamura, T. Shimazaki, Y. Asai and N. Tao, J. Am. Chem. Soc., 2013, 135, 3319–3322.
22. B. Capozzi, Q. Chen, P. Darancet, M. Kotiuga, M. Buzzeo, J. B. Neaton, C. Nuckolls and L. Venkataraman, Nano Lett., 2014, 14, 1400–1404.
23. M. Potangale, A. Das, S. Kapoor and S. Tiwari, J. Mol. Liq., 2017, 240, 694–707.
24. A. L. Eckermann, D. J. Feld, J. A. Shaw and T. J. Meade, Coord. Chem. Rev., 2010, 254, 1769–1802.
25. C. Reichardt and T. Welton, Solvents and Solvent Effects in Organic Chemistry, Wiley, 2010.
26. A. Kick, T. Weyhermüller, M. Hölscher, N. Kaeffer and W. Leitner, Angew. Chem., 2024, 63, e202408356.
27. P. D. Beer, M. G. B. Drew and K. Gradwell, J. Chem. Soc., Perkin Trans. 2, 2000, 511–519.
28. C. F. P. Miranda and L. M. N. B. F. Santos, J. Mol. Liq., 2025, 423, 126916.
29. J. M. Brisendine, S. Refaely-Abramson, Z.-F. Liu, J. Cui, F. Ng, J. B. Neaton, R. L. Koder and L. Venkataraman, J. Phys. Chem. Lett., 2018, 9, 763–767.
30. L. Venkataraman, J. E. Klare, C. Nuckolls, M. S. Hybertsen and M. L. Steigerwald, Nature, 2006, 442, 904–907.
31. Pine Research, Pyrolytic Graphite Electrode Information, https://pineresearch.com/shop/electrodes/working-electrodes/e1e/#documentation.
32. L. Scarpetta-Pizo, R. Venegas, P. Barrías, K. Muñoz-Becerra, N. Vilches-Labbé, F. Mura, A. M. Méndez-Torres, R. Ramírez-Tagle, A. Toro-Labbé, S. Hevia, J. H. Zagal, R. Oñate, A. Aspée and I. Ponce, Angew. Chem., Int. Ed., 2024, 63, 1–10.
33. L. Acuña-Saavedra, A. M. Méndez-Torres, G. Cárdenas-Jirón, R. Oñate, B. Sánchez-Allende, R. Venegas, R. Bernal, F. Melo, E. Imbarack, J. H. Zagal and I. Ponce, ACS Catal., 2025, 15, 719–737.
34. Dassault Systèmes, 2022, 22.1.0.3462, https://www.3ds.com/products/biovia/materials-studio.
35. S. Grimme, S. Ehrlich and L. Goerigk, J. Comput. Chem., 2011, 32, 1456–1465.
36. F. Weigend and R. Ahlrichs, Phys. Chem. Chem. Phys., 2005, 7, 3297.
37. J.-D. Chai and M. Head-Gordon, Phys. Chem. Chem. Phys., 2008, 10, 6615.
38. F. Neese, WIREs Comput. Mol. Sci., 2022, 12(5), e1606.
39. M. Cossi, N. Rega, G. Scalmani and V. Barone, J. Comput. Chem., 2003, 24, 669–681.
40. H. Kruse and S. Grimme, J. Chem. Phys., 2012, 136, 154101.
41. E. Epifanovsky, A. T. B. Gilbert, X. Feng, J. Lee, Y. Mao, N. Mardirossian, P. Pokhilko, A. F. White, M. P. Coons, A. L. Dempwolff, Z. Gan, D. Hait, P. R. Horn, L. D. Jacobson, I. Kaliman, J. Kussmann, A. W. Lange, K. U. Lao, D. S. Levine, J. Liu, S. C. McKenzie, A. F. Morrison, K. D. Nanda, F. Plasser, D. R. Rehn, M. L. Vidal, Z.-Q. You, Y. Zhu, B. Alam, B. J. Albrecht, A. Aldossary, E. Alguire, J. H. Andersen, V. Athavale, D. Barton, K. Begam, A. Behn, N. Bellonzi, Y. A. Bernard, E. J. Berquist, H. G. A. Burton, A. Carreras, K. Carter-Fenk, R. Chakraborty, A. D. Chien, K. D. Closser, V. Cofer-Shabica, S. Dasgupta, M. de Wergifosse, J. Deng, M. Diedenhofen, H. Do, S. Ehlert, P.-T. Fang, S. Fatehi, Q. Feng, T. Friedhoff, J. Gayvert, Q. Ge, G. Gidofalvi, M. Goldey, J. Gomes, C. E. González-Espinoza, S. Gulania, A. O. Gunina, M. W. D. Hanson-Heine, P. H. P. Harbach, A. Hauser, M. F. Herbst, M. Hernández Vera, M. Hodecker, Z. C. Holden, S. Houck, X. Huang, K. Hui, B. C. Huynh, M. Ivanov, Á. Jász, H. Ji, H. Jiang, B. Kaduk, S. Kähler, K. Khistyaev, J. Kim, G. Kis, P. Klunzinger, Z. Koczor-Benda, J. H. Koh, D. Kosenkov, L. Koulias, T. Kowalczyk, C. M. Krauter, K. Kue, A. Kunitsa, T. Kus, I. Ladjánszki, A. Landau, K. V. Lawler, D. Lefrancois, S. Lehtola, R. R. Li, Y.-P. Li, J. Liang, M. Liebenthal, H.-H. Lin, Y.-S. Lin, F. Liu, K.-Y. Liu, M. Loipersberger, A. Luenser, A. Manjanath, P. Manohar, E. Mansoor, S. F. Manzer, S.-P. Mao, A. V. Marenich, T. Markovich, S. Mason, S. A. Maurer, P. F. McLaughlin, M. F. S. J. Menger, J.-M. Mewes, S. A. Mewes, P. Morgante, J. W. Mullinax, K. J. Oosterbaan, G. Paran, A. C. Paul, S. K. Paul, F. Pavošević, Z. Pei, S. Prager, E. I. Proynov, Á. Rák, E. Ramos-Cordoba, B. Rana, A. E. Rask, A. Rettig, R. M. Richard, F. Rob, E. Rossomme, T. Scheele, M. Scheurer, M. Schneider, N. Sergueev, S. M. Sharada, W. Skomorowski, D. W. Small, C. J. Stein, Y.-C. Su, E. J. Sundstrom, Z. Tao, J. Thirman, G. J. Tornai, T. Tsuchimochi, N. M. Tubman, S. P. Veccham, O. Vydrov, J. Wenzel, J. Witte, A. Yamada, K. Yao, S. Yeganeh, S. R. Yost, A. Zech, I. Y. Zhang, X. Zhang, Y. Zhang, D. Zuev, A. Aspuru-Guzik, A. T. Bell, N. A. Besley, K. B. Bravaya, B. R. Brooks, D. Casanova, J.-D. Chai, S. Coriani, C. J. Cramer, G. Cserey, A. E. DePrince, R. A. DiStasio, A. Dreuw, B. D. Dunietz, T. R. Furlani, W. A. Goddard, S. Hammes-Schiffer, T. Head-Gordon, W. J. Hehre, C.-P. Hsu, T.-C. Jagau, Y. Jung, A. Klamt, J. Kong, D. S. Lambrecht, W. Liang, N. J. Mayhall, C. W. McCurdy, J. B. Neaton, C. Ochsenfeld, J. A. Parkhill, R. Peverati, V. A. Rassolov, Y. Shao, L. V. Slipchenko, T. Stauch, R. P. Steele, J. E. Subotnik, A. J. W. Thom, A. Tkatchenko, D. G. Truhlar, T. Van Voorhis, T. A. Wesolowski, K. B. Whaley, H. L. Woodcock, P. M. Zimmerman, S. Faraji, P. M. W. Gill, M. Head-Gordon, J. M. Herbert and A. I. Krylov, J. Chem. Phys., 2021, 155, 084801.
42. S. Smidstrup, T. Markussen, P. Vancraeyveld, J. Wellendorff, J. Schneider, T. Gunst, B. Verstichel, D. Stradi, P. A. Khomyakov, U. G. Vej-Hansen, M.-E. Lee, S. T. Chill, F. Rasmussen, G. Penazzi, F. Corsetti, A. Ojanperä, K. Jensen, M. L. N. Palsgaard, U. Martinez, A. Blom, M. Brandbyge and K. Stokbro, J. Phys.: Condens. Matter, 2020, 32, 015901.
43. QuantumATK version Q-2021.06. Synopsys QuantumATK (https://www.synopsys.com/silicon/quantumatk.html), version Q-2021.06, https://www.synopsys.com/silicon/quantumatk.html.
44. M. J. van Setten, M. Giantomassi, E. Bousquet, M. J. Verstraete, D. R. Hamann, X. Gonze and G.-M. Rignanese, Comput. Phys. Commun., 2018, 226, 39–54.
45. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1997, 78, 1396.
46. J. Taylor, H. Guo and J. Wang, Phys. Rev. B: Condens. Matter Mater. Phys., 2001, 63, 245407.
47. J. Taylor, H. Guo and J. Wang, Phys. Rev. B: Condens. Matter Mater. Phys., 2001, 63, 121104.
48. M. Brandbyge, J.-L. Mozos, P. Ordejón, J. Taylor and K. Stokbro, Phys. Rev. B: Condens. Matter Mater. Phys., 2002, 65, 165401.

---