Plasmonics is a promising counterpart of nanophotonics which has witnessed major progresses in the control and manipulation of electromagnetic (EM) fields at extreme-subwavelength scales [1–3]. Central to all studies is the surface plasmon’s unique capability to harvest, process, and concentrate light and convert it into energetic near-fields, thermal power, and hot carriers [4–7]. All these remarkable applications and developments of subsequent optical devices have been realized using resonant plasmonic nanostructures across the ultraviolet (UV) to the far-infrared spectra. Resonances in nanophotonics and nanoplasmonics are the fundamental phenomena that play a critical role in defining the operating mechanism, quality, and performance of the tailored tools. In recent years, several types of radiating and nonradiating resonances and spectral effects have been successfully excited and introduced by scientists such as Fano resonances [8–10], electromagnetically induced transparency (EIT) [11–13], Bormann and Kerker effects [14–20], toroidal multipoles [21–24], anapoles [23, 25–27], and charge transfer plasmons (CTPs) [28–30]. Excluding the later instance (CTP), other resonances can be induced based on robust coupling of optically driven modes between proximal metallic nanoparticles (NPs) with sharp protrusions in the near-field regime. Conversely, for the CTP, the charge transfer in particle plasmon resonances has been reported in (1) conductively linked NPs [31–35], (2) reversible electrochemical compounds [36], and (3) metallic systems with subnanometer atomic openings (Fowler-Nordheim (FN) and quantum tunneling principles) [37–40].

There have been ongoing theoretical and experimental advancements to understand the possibility of direct dynamic charge transport between plasmonic objects in the absence of capacitive coupling between plasmonic particles. Relatively, researchers have shown the possibility of plasmon-driven charge transfer process in DNA-mediated metallic NP dimers [31, 34]. In this context, one can fill the gap between metallic NPs with DNA (as a scaffold for the growth of the NPs within the gap), and by increasing the metallic NP concentration, the transition from capacitive to conductive coupling can be formed. Besides, for the conductively bridged NPs, it is confirmed that due to the shuttling of photoexcited electrons across the junction, the CTP spectral feature appears in the lower energies far from the superradiant dipole moment [41]. On the other hand, as an alternative route to this method, it is well-accepted that particles with atomic gaps in between (below ~0.5 nm) are able to sustain CTPs through molecular quantum tunneling principle [40, 42]. Theoretically, the electron tunneling across an atomic-scale gap at optical frequencies cannot be explained using classical electrodynamic theory and requires quantum mechanical description.

In spite of possessing interesting properties, CTPs are inherently suffering from lack of spectral tunability. Recent demonstrations that the functionality of CTP feature can be optimized by integrating bulk metallic systems with thermally and electronically controllable materials show why tunable CTP spectral features are important for implementing novel and advanced plasmonic tools [36, 43–46]. More precisely, this shortcoming has successfully been addressed by combining plasmonic structures with, for example, phase-change materials (PCMs) [43, 44, 47], graphene [45, 46], and thermally tunable substances (i.e., InSb).

In this review, we summarize recently introduced mechanisms that underpin the electron transition between plasmonic particles and trending applications of functional CTPs. One of the goals of this article is to present a comprehensive depiction of various modalities developed for the excitation of tunable CTPs. Besides, we explain the state-of-the-art approaches that have been proposed for enhancing the tunability of CTPs and the use of this spectral phenomenon in designing practical photonic tools such as modulators, switches, and nonlinear harmonic signal generators. Next, we illustrate how a careful integration of subwavelength plasmonic resonators with optically, thermally, and electronically functional materials leads to the emergence of multifunctional nanooptoelectronic and terahertz (THz) devices.

Direct manipulation of charges via redistribution and transfer has been introduced as a simple approach to excite CTP spectral features and studied both theoretically and experimentally in nanoscale plasmonic systems [31–36]. Possessing a different principle compared to capacitive resonance interference, the direct charge transfer between plasmonic resonators and particles enables tailoring advanced plasmonic tools. Additionally, this direct transition of charges between the metallic structures delivers exceptional abilities to actively control the charge shuttling by varying the intensity of the incoming beam, which could be used for the development of near-infrared and THz plasmonic devices.

A remarkable example for the excitation of CTP mode is observed in a two-member dimer consisting of a pair of proximal metallic NPs that are connected with a conductive junction (insets in Figure 1(a)) [32]. In the method developed by Wen et al. [32], a dimer antenna with a conductive junction has been considered as the intermediate case between a conventional metallic dimer with capacitive opening and a nanorod. This allows to understand the formation of new spectral features due to the presence of a metallic bridge between NPs. Figure 1(a) demonstrates the numerically obtained scattering efficiencies of three different structures with judiciously defined geometries that are stated in the figure caption. Under p-polarized beam illumination, in the dimer limit, the capacitive coupling between the two nanodisks leads to strong hybridization of plasmons and excitation of a superradiant mode around ∼1.95 eV. On the other hand, bridging the dimer with a conductive nanowire gives rise to the formation of two peaks in the spectrum correlating with the dipolar mode (at ∼2.1 eV) and a narrow CTP mode at lower energies far from the classical dipole (around ∼0.96 eV). Lastly, for the plasmonic nanorod, compared to the dimer structure, the dipolar resonance is red-shifted to ∼1.3 eV. As can be seen in the charge plots in Figure 1(b), the fundamental difference between the CTP extreme and dipolar mode is the distinguished oscillation of the electric current density in the junction of the conductive nanowire-bridged plasmonic dimer, while there is no such a feature in both dimer and nanorod structures. Moreover, the charge distribution at the energy of CTP explicitly represents the splitting of charges across the structure (see Figure 1(b), (2)).

To further study the role of the conductive junction on the excitation of CTP, Wen and teammates have shown that the junction geometry and conductance strictly determine the position and intensity of the CTP feature. To this end, the researchers employed CTP-resonant nanostructures based on aluminum and gold substances. Considering a nanowire with the frequency-dependent AC conductivity of \sigma \left(\omega \right), and specific length (l), width (w), and thickness (t), the corresponding conductance can be written as

As a critical parameter, in Figure 1(c), the junction conductance as a function of nanowire width is plotted, numerically and experimentally. Obviously, by increasing the width of the junction, the relative conductance increases slightly for gold structure, while this value sharply increases for the aluminum nanowire. Here, for structures with the equivalent geometries of the bridging nanowire, aluminum structures hold a higher junction conductance than gold structures. Further analysis for the junction properties and plotting the CTP position as a function of junction conductance reveal that the CTP mode exhibits a dramatic sensitivity to the junction conductance in the small conductance limit (Figure 1(d)). However, this sensitivity becomes strikingly less as the conductance increases beyond a specific value (~50) for all the studied cases.

Recent studies have also shown that CTP mode can be viewed as a spectral feature at much lower energies (e.g., far infrared, THz) when the coupling between neighbor resonators is in the weak regime. As a leading work in the THz regime, Ahmadivand et al. [33] have represented the transition from capacitive coupling to direct charge transfer using a symmetric cluster of V-shaped metallic microblocks. The insets in Figures 1(e)–1(g) illustrate the scanning electron microscopy (SEM) images of the unit cell, where the diameter of the central disk gradually increases. As obvious in the normalized transmission amplitude (NTA) profile (Figure 1(e)), the capacitive coupling plays a fundamental role in the emergence of two pronounced minima in the spectral response. The deeper minimum around 1.22 THz correlates with the dipolar resonant mode, while the quadrupole mode appears as a weaker dip around 1.85 THz. Here, the middle disk enhances the strength of the induced multipolar modes via the dipole-dipole interaction [48]. To continue, increasing the diameter of the disk monotonically and providing a touching regime between the disk and V-shaped assemblies lead to the elimination of the quadrupolar moment and emerging of the CTP dip around ~0.5 THz (Figure 1(f)). Due to the formation of insignificant touching spots between the blocks and the central disk, one can expect the direct flow of charges across the unit cell similar to the optical systems and quantum transitions. Further increases in the diameter of the central disk to realize the overlapping regime give rise to a damping in the energy of both dipolar and CTP modes projected from the spectral characteristics of the entire unit cell (Figure 1(g)). The cross-sectional electric-field (E-field) intensity along the unit cell axis helps to perceive the role of the conductive disk in the excitation of CTP feature. As plotted in Figures 1(h)–1(j), by inserting and increasing the size of the disk, the E-field confinement reduces dramatically owing to losing the capacitive coupling between the V-shaped blocks. Additionally, the overlapping areas between the V-shape and central resonators allow to induce sharper and boosted plasmonic dipolar and CTP modes in the THz limit. The geometry of the merging regions enables the shuttling of charges between the linked V-shaped pixels of a standalone assembly. This can be better understood by defining the resistance (R) of the disk as a function of the junction geometry [49]:

where \sigma \left(\omega \right) is the frequency-dependent AC conductivity, d is the length between junctions, and \delta is the contact width at the junctions due to overlapping. Moreover, according to Ohm’s law, the conductance is defined by [50]

where n_m is the electron density of metal, e is the elementary charge, \tau is the mean-time between collisions, and m_e is the mass of electron. Enlarging the diameter of the overlapping disk lengthens the distance between junctions, and the corresponding resistance increases accordingly (Figure 1(m)). Although the dynamic charges would be able to transit across the trail, the dissipative losses can be significant because of the inherent disk resistance. When the incident THz beam is resonant with the spectral line shape, the fairly strong confinement of charges across the lossy metallic resonators leads to a drastic electron decay. Thus, by extending the size of the middle disk, the induced dipolar and CTP modes shift towards the lower energies due to the enhanced dissipative absorption losses [51, 52].

As discussed in Introduction, the general idea to model the light-matter interactions and the resulting electromagnetic field distribution is mainly based on solving coherent charge density oscillations using classical electromagnetic theory by applying Maxwell’s equations [4]. By shrinking the interparticle distance down to nanoscales, localization and intensity of the incident electric field can be enhanced [53–55]. Nevertheless, when the system requires to operate beyond the nanometer gap regime (i.e., subnanometer openings) or contains a quantum topology (e.g., quantum dots, molecules, or atoms), nonlocal screening effects (due to quantum nature of electron) modify the plasmonic response of structure [56]. In this limit, quantum mechanical effects (e.g., collective quantum electron tunneling, nonlocal transport) occur and the whole system must be described under nonlocal conditions [57–59], by implementing advanced theoretical approaches [60–62]. This enables the calculation of the confinement of the induced surface charges appropriately [63]. Such computations in quantum systems have successfully been done using a quantum-corrected model (QCM) [42]. Basically, this concept is an updated version of the time-dependent density-functional theory (TDDFT), which combines the classical electromagnetic framework with the collective quantum electron tunneling [38]. As declared in Savage et al. [64], when the interparticle distance is d\ge 0.4 nm, one can treat the system using classical Maxwell’s descriptions to understand the plasmonic interactions. For  nm, classical formulation starts to diverge as a result of the increased rate of critical electron tunneling between proximally located NP surfaces (~ 0.3 nm), where the plasmon interactions start to get into the quantum regime. For  nm, the quantum tunneling dominates the spectral response of the entire system. By considering a rectangular barrier, this phenomenon can be predicted as the following: {\tilde{d}}_{QR}=\ln \left(3q\alpha \lambda /2\pi \right)/2q, where {\tilde{d}}_{QR} is the critical gap size, q is the semiclassical electron tunneling wavenumber, \lambda is the optical plasmon wavelength, and \alpha is the structural constant. More realistically, the addition of “coherent quantum transport” parameter into the formula given above boosts the tunneling rate, which gives rise to the quantum-tunneling CTPs (more detailed information about the calculations can be found in Ref. [64]). In the upcoming part, we briefly review the recent advances in the quantum-mechanical charge shuttling process in nearly touching metallic NPs.

As a pioneer work, Wu et al. [39] demonstrated the first theoretical explanation of plasmon resonances between extremely close NPs due to FN tunneling. In spite of possessing certain conditions to satisfy the direct tunneling (e.g., subnanometer gap), this approach provides a novel outlook to induce CTP modes for a different set of gap sizes by applying high-intensity illumination. From the active nanophotonic device perspective, the proposed concept is significant owing to its ability to control the charge shuttling process by changing the intensity of the incident light. As discussed previously, approaching two NPs to each other (nearly touching regime) and the presence of a metallic path between them lead to transition of oscillating electric charges across the nanoplatforms, in addition to the dipolar mode of the dimer system. As the conductive path becomes narrower, the excited CTP resonance shifts to the longer wavelength and several higher-order modes (e.g., hybridized dipolar and quadrupolar modes) appear [65–67]. Nevertheless, when the conductive bridge turns out to a ~0.4-0.5 nm gap, the CTP mode disappears, meaning that a quantum mechanical framework should be taking into account to model possible electron tunneling effects. To this end, in Figure 2(a), the authors demonstrated possible tunneling mechanisms as direct tunneling (effective for small interparticle distance) and FN tunneling (dominant when a strong electric field is applied) in a simple schematic. For the direct tunneling, electrons tunnel through the square barrier (from A to C); while for the FN tunneling, electrons first tunnel through the triangular barrier (from A to B), then transport to C. In general, the electrons exist in B are called “space charges.” Hence, this process can also be termed as space charge-based charge transfer. Moreover, the researchers here utilized the microscopic version of Ohm’s law (see Equation (2)) to specify the tunneling electrons in the gap [68]. It is important to note that for the calculation of tunneling electron density in the gap, Wu and colleagues applied the Kohn-Sham density functional theory, to model exchange correlation and Coulomb interactions between electrons [69, 70]. Further, they successfully explained the resulting CTP mode in light of the obtained tunneling electron density (n_{\text{gap}}) and gap conductivity ({\sigma }_{\text{gap}}), which are determined based on a numerical methodology developed by the authors [39]. To briefly discuss, \epsilon \left(\omega \right)=1+i\left({\sigma }_{\text{gap}}/\omega {\epsilon }_0\right) is utilized as a part of QCM to derive the optical response of the system. As demonstrated in Figure 2(b), when {\sigma }_{\text{gap}} is larger than 1.145\times {10}^5 S/m, a CTP mode is emerged. For the “tunneling” part indicated in Figure 2(b), a large plasmon-enhanced electric field (10¹⁰ V/m) (or “gap field”) is required to bring enough electron in the space charge area to induce the CTP resonant mode. In the following, in Figure 2(c), to better understand the physical mechanism behind the process, {\sigma }_{\text{gap}} and n_{\text{gap}} are compared for the gap sizes of 0.4 nm, 0.6 nm, and 0.8 nm in three different gap field regimes. In the direct tunneling limit, the tunneling barrier can be controlled through the charge potential when the gap field is small, which means {\sigma }_{\text{gap}} primarily rely on the gap length. As expected, when the gap is 0.4 nm, electrons can straightly tunnel along the narrower barrier (see Figure 2(d), (i)), with the increased tunneling probability of electrons, but for larger gaps (e.g., 0.8 nm), the charge flow process could not be sustained because of the wider tunneling barrier. To address this issue, the employed electric field intensity can be amplified (~10¹⁰ V/m) to start the FN tunneling procedure, in which the tunneling barrier is reshaped as plotted in Figure 2(d), (ii). In this regime, {\sigma }_{\text{gap}} and n_{\text{gap}} are very responsive to the intensity of the external illumination, meaning that these parameters can be easily modified depending on the applied electric field intensity. For instance, when the field intensity is altered from 10⁹ V/m to 10¹¹ V/m, {\sigma }_{\text{gap}}is increased from 1 to 10⁷ S/m. Any further increase in the field intensity would lower the level of barrier where all tunneling electrons can stay over the barrier (see Figure 2(d), (iii)). In this regime, {\sigma }_{\text{gap}} is not dependent on the applied field intensity and gap size, as illustrated in Figure 2(c). This results in perfect transmission of electrons through the gap and extremely large {\sigma }_{\text{gap}} to retain the CTP mode for the studied gap dimensions. Lastly, in Figures 2(e)–2(g), simulated extinction spectra for different gap size (d), {\sigma }_{\text{gap}}, and the gap field are presented to prove the tunability of the CTP peak. With the gap size of 0.4 nm, the CTP mode always arise either via direct tunneling (Figure 2(g)) or via FN tunneling (Figure 2(f)). For larger gap sizes (e.g., 0.6 nm and 0.8 nm), high intensity external irradiation (~10¹⁰ or 10¹¹ V/m) is necessary to excite the CTP resonant mode (see Figures 2(f) and 2(g)).

In another recent study, Kulkarni and Manjavacas [71] investigated the quantum effects related to the charge transfer process by examining the spectral response of gold dimer with a two-level system (TLS) (e.g., an atom or a molecule). In this understanding, fully quantum-based computations showed that CTPs are only recognizable if one of the energy levels of the studied two-level system is in resonance with the Fermi level of the dimer system, by enabling the electron transition along the junction. At resonance, the absorption spectrum of the system and the conductance of the junction are calculated, and the outcomes indicated that the conductance of the junction is equivalent to one quantum of conductance, which isG_0=2e^2/h. In Figure 3(a), a schematic for the studied subwavelength system is illustrated. This system consists of two identical gold spheres of 32 au in diameter and a small sphere with a diameter of 6 au to model the TLS. In all calculations, the whole platform is assumed to be in a vacuum. To explore the capabilities of the system, TDDFT in the adiabatic local density approximation [72, 73] is utilized. To this end, the authors only take into account the conduction electrons of the metal and the gold NPs are modelled based on jellium approximation, where the ionic background charge has a uniform charge density (n_0). Here, n_0 is judiciously selected to correlate with the density of gold, which is equal to a Wigner-Seitz radius of 3 au. In a similar fashion, TLS is formed by having a Wigner-Seitz radius to make sure that TLS only adds a single electron to the system. Without applying any external field, the one-electron potential of the platform is defined as V_{\text{eff}}\left(\mathbf{r}\right)=V_0\left(\mathbf{r}\right)+V_{\text{H}}\left[n\left(\mathbf{r}\right)-n_0\left(\mathbf{r}\right)\right]+V_{\text{xc}}\left[n\left(\mathbf{r}\right)\right], where V_{\text{H}} is the Hartree potential, n is the electronic density, V_{\text{xc}} is the exchange-correlation potential (also known as the Perdew-Zunger functional as a part of local density approximation [74]), n_0 is the background ionic charge density, and V_0is the uniform background potential which incarcerate the electrons within the NPs and the TLs. For the gold NPs, this potential is fixed at -4.6 eV (5.2 eV lower than the vacuum level) to be able to have a proper electron spill-out. As plotted in Figure 3(b), the energy levels of the TLS are tuned according to the Fermi level of the NPs. Additionally, the background and the resulting equilibrium one-electron potentials are represented as gray and red colors, respectively, for V_{\text{TLS}}=-4.6 eV. It is evident from Figure 3(b) that the potential barrier of the junction (indicated as dashed black line) is reduced by virtue of the TLS. In the following, the associated equilibrium electronic density is demonstrated in Figure 3(c), which clearly indicates Friedel oscillations due to the discretization of the electronic levels [74]. Besides, the electronic density in the junction is a proof of the single energy level of the TLS. Although the TLS decreases the potential barrier, the electronic density in the gaps between the NPs does not reach to zero. In Figure 3(d), the authors plotted the absorption spectrum for two different cases (by considering, the incident field is polarized along the z-axis): (a) the bare metallic dimer (gray line) and (b) the dimer with the TLS (red line). For the latter case, the background potential of the TLS is selected as equal to the NPs. In both cases, a bonding dipolar mode (BDP) [75, 76] appears around 5 eV. For the low-energy part of the spectrum, three new peaks are formed, due to the addition of TLS into the dimer system. To comprehensively evaluate the origin of these modes, the corresponding charge distributions are studied on the surface of the NPs (see Figure 3(e)). The results clearly show that the mode around 5 eV has a dipolar nature, and the other three modes indicate a monopolar distribution pattern, which is the solid evidence of the charge transfer through the junction. What is more, the modes at 1.1 eV and 1.55 eV display an oscillation of charges inside the NPs owing to the finite size effects. Based on these results, one can state that the only way to induce a CTP in this system is contingent upon the use of the levels of the TLS as the conductive paths (see the schematic in Figure 3(f)). Thus, the authors analyze the effect of background potential of the TLS on the response of the system. As demonstrated in Figure 3(g), V_{\text{TLS}} is varied to manipulate the position of the levels of the TLS. In this plot, each dot represents an energy level of the whole platform. Depending on the localization rate of the TLS levels, the size of the dot can be bigger, which is beneficial to differentiate localized levels in the TLS. A group of states are localized close to the Fermi level of the system, shown in black dashed line, for small numbers ofV_{\text{TLS}}. These states become more localized as the background potential is getting deeper and the states move to lower parts in the energy diagram. However, because of the strong localization, the interplay between the NPs is diminished. Particularly, when V_{\text{TLS}} is lower than -16.2 eV, an extra level starts to localize on the TLS. Lastly, in Figures 3(h) and 3(i), the absorption spectrum of the system (at a low-energy part) is presented for theV_{\text{TLS}} values considered in Figure 3(g). The results explicitly verify the requirement of having a localized state in the TLS, whose energy is near the Fermi level of the system, to be able to generate a CTP mode.

As discussed in the previous section, in the quantum tunneling of energetic optically driven electrons, possessing an active control over the CTP spectral feature is limited to modifying the incident field intensity and/or morphological variations [42, 45]. On the other hand, the induced CTPs via the direct transition of optically driven electrons across the bulk metallic paths between NPs suffer from the inherent lack of tunability. Recently, these challenges have effectively been addressed by using optoelectronically and optothermally tunable components in the purpose of the excitation of functional CTPs [43–47, 77]. Such an active tunability allows for the exploration of several integrated plasmonic instruments and applications owing to its great potential for the next-generation multifunctional technology. Beyond the fundamental studies, now research in CTP devices has been focused on the experimental attempts to efficiently transform its capabilities into the real-world applications, where two fundamental issues must be addressed: functionality and scalable fabrication. While the former has been successfully realized by employing optoelectronically and thermally tunable compounds, in considering the later concern for industrial applications, one of the major challenges is scalable cost-effective fabrication of the functional metastructures. This can be done, for instance, by developing robust techniques based on nanolithography-free techniques. In this section and following subsections, we briefly review the recent techniques that have been utilized to optimize the tunability of CTPs. We will demonstrate how the integration of the plasmonic nanostructures with optoelectronically and optothermally controllable components improves the tunability of CTP spectral features.

In recent years, the field of plasmonics has experienced rapid progresses in understanding the dynamics of conductively bridged and nearly touching metallic NPs. As we described in this focused review, many intriguing physical effects have been predicted in the systems supporting distinguished CTP excitations. So far, various theoretical and experimental investigations have been performed to realize oscillating electric current along the conductive junction, as a key principle of the CTP excitation, depending on the type of the nanosystem. However, related investigations have been limited to the excitation approaches and verifying this principle by standard measurements. Now that the CTP spectral feature has been reached in a broad range of platforms in modern nanophotonics, we believe that it is time to numerically and experimentally explore the modalities towards enhancing the functionality of this spectral phenomenon, eventually, to find its useful and practical applications. We also speculate that CTP states possess a strong potential to make a profound impact in the development of coming generation multifunctional nanophotonic instruments. More precisely, the transition between hybridized dipolar and distinct CTP modes have been explicitly demonstrated under capacitive and conductive coupling regimes, respectively. Highly attractive features of CTP resonances have opened the door for novel active devices and applications, including but not limited to surface-enhanced IR absorption (SEIRA), surface-enhanced Raman scattering (SERS), optical switching, modulation, and waveguiding; however, active manipulation of these resonances using functional materials is overlooked.

In this review, we briefly presented the recent accomplishments in the use of versatile materials towards actively tunable CTP-based nanoscale tools. As indicated in the considered studies, having an active control on the CTP mode is possible using functional materials like graphene and GST, rather than applying morphological variations. The exceptional optical and electrical characteristics of these materials have allowed researchers to design active CTP-resonant devices. We envisage that this review will provide detailed understanding for the evolution from passive to active CTP-based platforms and pave the road for developing next-generation nanophotonic devices to reach new functionalities.