Main text

Surface plasmons (SPs), namely the collective oscillations of free carriers in metal nanostructures that arise in response to incident electromagnetic fields, enable confinement and control over light at length scales well below the diffraction limit [1–4]. The localized SP energy, however, doesn’t remain confined in the metal nanostructure indefinitely. Both geometry/boundary induced and material specific radiative and nonradiative decay channels eventually dissipate the confined SP energy, either as free-space radiation (radiative loss) or as absorption. Besides the conventional Ohmic losses, absorption also includes the generation of electron-hole pairs in the material (nonradiative loss). The radiative loss of SPs is an attractive feature for several important applications that leverage the radiative decay channel for engineering the spectral and directional emission properties of quantum emitters [5–7]. On the other hand, nonradiative losses have long been thought to be deleterious to the performance of plasmonic (metallic) structures and different approaches have been explored to reduce the nonradiative losses [8–10]. This notion has gradually changed in the last couple of decades, and there has been a concentrated effort to develop useful applications by leveraging nonradiative losses. One example is photothermal therapy [11–13], where the temperature increase of the plasmonic structure from the absorption of plasmon energy is utilized in medical treatments.

In addition to the thermalized carriers that are under thermal equilibrium with the lattice, the nonradiative decay of SPs also produces highly energetic carriers (both electrons and holes) that are not in thermal equilibrium with the lattice. Such nonequilibrium “hot-carriers”, with distributions that substantially deviate from the equilibrium Fermi-Dirac distribution, are expected to enable efficient energy-harvesting applications, ranging from photodetection and photovoltaic schemes circumventing bandgap limitations [14–18] to catalyzing chemical reactions [19–24]. Below we provide a brief discussion of the recent developments in this field.

Among early suggested applications of plasmon assisted hot-carriers are photodetectors operating below the bandgap [15,16]. In this approach, hot-carriers are generated by the resonant SPs at telecommunications wavelengths in lithographically created gold plasmonic structures on silicon substrates. The generated energetic hot-carriers are distributed over a range of energies from $\{{{E_F} - \hbar \omega :\; {E_F} + \hbar \omega } \}$, where ${E_F}$ is the Fermi energy of gold and $\hbar \omega $ is the energy of the incident photon (those carriers above ${E_F}$ are the hot-electrons, while those below ${E_F}$ are the hot-holes). When $\hbar \omega $ is larger than the Schottky barrier height (${\mathrm{\Phi }_B}$), which for the Au-Si interface is ∼ 0.6 eV, some of the hot-carriers generated in the plasmonic structure have energies exceeding the barrier. Thus, they can flow to the semiconductor, provided the carriers have a positive momentum component that is directed from the plasmonic structure to the semiconductor (Fig. 1(A)). This flow of charge leads to the current, which is subsequently measured using suitable electronics, thereby detecting the incident photon flux. These generated photocurrents closely follow the SP absorption spectra, which along with the observed polarization dependence, strongly suggesting that the arising photocurrents are due to the generated nonequilibrium hot-carriers. A key advantage of this approach, as opposed to the conventional photodetection approach that employs pn semiconductor junctions, is that the generated photocurrent is no longer limited by the bandgap of the semiconductor or the interband transition energy in the metal. Instead, the Schottky barrier height is the limiting factor, which is usually much smaller than the constituent material bandgaps, thereby enabling the detection of the below-bandgap photons.

 

Fig. 1. A, (Left panel) Band diagram schematic of metal-silicon interface indicating the process of hot-carrier injection and photodetection. Schematic (Top right panel) and a scanning electron micrograph (Bottom right panel) of an Au nanoantenna device on Silicon [16]. B, Schematic showing the transfer of hot-carriers from a metal nanoparticle to nearby reactants surrounding the nanoparticle [19]. C, Schematic of a molecular junction-based light to electrical energy harvesting device, featuring a plasmonic structure (periodic metallic stripes) and a transparent electrode. (Figures (A) and (B) are adapted from Refs. [16] and [19], respectively)

Download Full Size | PDF

Subsequently, several groups explored the “below bandgap” photodetection with hot-carriers. For instance, Chalabi et al. [14], instead of the Schottky barrier, explored the possibility of using a tunnel barrier for photodetection and investigated the role of the external voltage bias on the photodetection quantum efficiencies. Additionally, the strong polarization dependence of the underlying plasmonic structures can be used to design polarization selective photodetectors. Li et al. [18] in turn used chiral plasmonic structures, which exhibit strong plasmonic response for one helicity of incident circular polarization and negligible sensitivity for the other helicity. This allowed to detect and distinguish the right and left circular polarizations following the same principles as those outlined above. While all these works conclusively showed photon detection below the bandgap, the demonstrated quantum efficiencies were rather small, about 0.1% or even less. Such small quantum efficiencies can be attributed to the mismatch between the energies of hot-carriers under steady-state conditions and the associated Schottky/tunnel barrier heights, which for most metal-semiconductor/metal-insulator-metal junctions are about 0.5 eV or larger. As per Fowler theory [25,26], the quantum efficiencies of the above outlined photodetectors can be approximated as $c\frac{{{{({\hbar \omega - {\phi_B}} )}^2}}}{{\hbar \omega }}$, where c is a device specific emission coefficient. Although substantial number of carriers with energies exceeding the barrier height are generated soon after the plasmons nonradiatively decay, they rapidly relax to lower energies (due to electron-electron and electron-phonon collisions) before they can cross the barrier. As a result, under continuous wave illumination and steady-state conditions, only a tiny fraction of the generated hot-carries have energies exceeding ${\mathrm{\Phi }_B}$ and contribute to the photocurrent thereby limiting the quantum efficiencies.

A closely related application to photodetection is photovoltaics [27,28]. Hot-carrier based photovoltaics offers an enticing prospect of harvesting the energy of below bandgap photons. However, so far, there have been no demonstrations of hot-carrier based below bandgap photovoltaics with meaningful efficiencies, which again can be attributed to the mismatch between the steady-state hot-carrier energies and the barrier heights. Improving the quantum yields by engineering the barrier heights with either conventional metal-semiconductor junctions or via molecular junctions with versatile transmission characteristics (more on this below) could be the crucial next step to improve the efficiencies of hot-carrier based light-to-electrical energy harvesting devices to the point where they are of commercial value.

Catalysis is yet another avenue that has been extensively explored in the context of plasmon assisted hot-carrier generation [20–24]. It has been suggested that the hot-carriers generated via surface plasmon excitation can transfer some of their energy to the reactants/adsorbates surrounding the nanoparticle that contribute to the reaction (Fig. 1(B)). This process will push the reactants up in the energy landscape, effectively reducing the reaction activation energy barrier and catalyzing the reaction. Among prominent applications of plasmon assisted photocatalysis are dissociation of hydrogen [20,29], ethylene oxidation [22], ammonia decomposition [21], water splitting [24,30,31] and methane dry forming [32], to name a few.

While the observed plasmon assisted enhancements in the reaction rates is widely accepted, the governing physical processes contributing to the enhancements in the reaction rates are still under discussion [33]. The nonradiative decay of plasmons not only generates hot-carriers but also increases the temperature. Consequently, both hot-carrier effects and thermal effects contribute to plasmon driven chemical reactions and quantifying their relative contributions in plasmon driven chemical reactions has been a topic of intense debate [34]. We direct the readers to the recent articles [33,35,36] that discuss these issues in great detail. While the role of non-thermal hot-carriers in plasmon-induced chemical reactions is still under debate, there is a long-standing precedence describing the role of non-thermal effects in surface chemistry, both experimentally and theoretically [37–39].

A key aspect that strongly influences the performance efficiencies of all these applications is the energy distribution of the generated hot-carriers. Different theoretical frameworks involving either ab-initio first-principle calculations or semiclassical approaches have been developed to gain insights into these distributions [33,40–43]. These frameworks however involve different assumptions on the material properties and carrier relaxation processes, which have led to starkly different predictions on the energy distributions of the hot-carriers. To resolve these discrepancies and to gain insights into the physical processes governing hot-carrier generation, researchers have developed special experimental methods to quantify the hot-carrier distributions. Heilpern et al. [44] used an optical pump-probe technique in conjunction with a double-inversion approach to infer the transient hot-carrier distributions and their relaxation dynamics, while Pensa et al. [45] used the information from the energy barrier reduction for a plasmon assisted polymerization reaction to gain partial information on the steady-state hot-carrier energies.

Until recently, experimental elucidation of plasmonic hot-carrier energy distributions under steady-state conditions was lacking. Progress towards this goal was made in recent experiments where transport measurments through single molecule junctions (SMJs) were combined with plasmonics [46,47] and the steady-state energy distributions of hot carriers in plasmonic thin films were obtained. Specifically, SMJs featuring sharp peaks in the transmission characteristics were trapped between an ultra-thin Au film with an integrated grating coupler, and the Au tip of a scanning tunneling microscope probe [46]. Such SMJs act as spectral filters, wherein those carriers in a narrow energy window near the transmission function peak predominantly contribute to the current. Additionally, the spectral position of the molecular filter was tuned by varying the voltage bias between the STM probe and the Au film underneath. By measuring the differences in the bias voltage dependent currents flowing through such tunable molecular filters with and without plasmonic excitation (excited with an 830 nm continuous wave laser), the steady-state energy distributions of the generated hot-carriers were obtained. The measured distributions revealed that most of the hot-carriers are localized in an energy window from {-0.3 eV: 0.3 eV} relative to the Fermi level and substantially fewer hot-carriers were found at higher energies.

While past works combining plasmonics and molecular junctions have explored several interesting regimes of light-matter interactions, including the room-temperature strong coupling [48] and quantum plasmon resonances [49], the recent developments in probing the transport properties of molecular [46,47,50,51] and vacuum tunnel [52] junctions under the plasmonic excitation offer new avenues for energy harvesting applications. A distinct advantage with molecular junctions, as opposed to regular tunnel junctions and Schottky barriers, is that one could tailor the transmission characteristics by modifying the chemical structure of the molecule creating the junction. Additionally, advances in nanofabrication have enabled the production of large area molecular junctions [53,54]. Recent experiments have already established that plasmon induced photocurrents through large area alkanedithiols [51] to be comparable to metal-semiconductor Schottky barrier based photodetectors. One can thus envision having large area molecular junctions with transmission characteristics akin to a Heaviside function and sandwiched between plasmonic structures on one side and a contact electrode on the other side (Fig. 1(C)). Such junctions will allow the transfer of nearly all carriers of one kind (either electrons or holes) to flow from the plasmonic structure to the second contact, which would potentially enable the next generation of below bandgap photodetectors and photovoltaics with efficiencies exceeding the state-of-the-art hot-carrier devices.

The progress made in the field of plasmon-assisted hot-carrier generation has already contributed to crucial fundamental insights into the role of nonradiative loss in light-matter interactions, as well as to practical advances in energy conversion applications. We anticipate the future endeavors in the field to be aimed towards understanding and distinguishing the role of non-thermal and thermal carriers in plasmon driven processes, as well as exploring designer junctions for separating the energetic carriers. Insights from such studies would be crucial for rational design and development of future light energy harvesting technologies.