Recent Advances in the Design and Photocatalytic Enhanced Performance of Gold Plasmonic Nanostructures Decorated with Non-Titania Based Semiconductor Hetero-Nanoarchitectures

Abstract

Plasmonic photocatalysts combining metallic nanoparticles and semiconductors have been aimed as versatile alternatives to drive light-assisted catalytic chemical reactions beyond the ultraviolet (UV) regions, and overcome one of the major drawbacks of the most exploited photocatalysts (TiO₂ or ZnO). The strong size and morphology dependence of metallic nanostructures to tune their visible to near-infrared (vis-NIR) light harvesting capabilities has been combined with the design of a wide variety of architectures for the semiconductor supports to promote the selective activity of specific crystallographic facets. The search for efficient heterojunctions has been subjected to numerous studies, especially those involving gold nanostructures and titania semiconductors. In the present review, we paid special attention to the most recent advances in the design of gold-semiconductor hetero-nanostructures including emerging metal oxides such as cerium oxide or copper oxide (CeO₂ or Cu₂O) or metal chalcogenides such as copper sulfide or cadmium sulfides (CuS or CdS). These alternative hybrid materials were thoroughly built in past years to target research fields of strong impact, such as solar energy conversion, water splitting, environmental chemistry, or nanomedicine. Herein, we evaluate the influence of tuning the morphologies of the plasmonic gold nanostructures or the semiconductor interacting structures, and how these variations in geometry, either individual or combined, have a significant influence on the final photocatalytic performance.

1. Introduction

Since the first reported example of heterogeneous photocatalysis in 1911 applied to the degradation of Prussian Blue by ZnO powder and illumination [1], the degree of sophistication and complexity in photocatalyst design has experienced a huge development [2]. In heterogeneous photocatalysis, the process is initiated by the interaction between incident photons and the catalyst. The photon absorption by the catalyst (typically a semiconductor) leads to the promotion of valence band electrons into the conduction band, thereby creating electron-hole pairs. Those carriers can induce the subsequent generation of free radicals (e.g., hydroxyl (OH), superoxide (O₂⁻)) to target specific chemical reactions. As a requirement, the energy of the incident photon (hν) must be equal or higher than the energy band gap (E_g) of the catalyst, meaning that the incident electromagnetic wavelength must be energetic enough to overcome the barrier to excite an electron from the highest occupied energy levels to the lowest unoccupied levels. Well-established semiconductor-based photocatalysts, such as TiO₂ or ZnO possess high E_g values (3.05 [3] and 3.3 [4] eV, respectively) and are constrained to the more energetic ranges of the solar spectrum (i.e., UV window representing only up to 5% of the solar radiation) for an effective photoactivation. Abundant efforts have been devoted in the past decades to expand the photocatalytic response of heterogeneous semiconductor photocatalysts towards the visible and infrared ranges in order to maximize the absorption of the solar spectrum [5,6,7,8,9]. One of the most promising and explored strategies has consisted on the combination of semiconductor structures with noble-metal based nanoparticles [10,11,12]. Recent reviews available in the literature have deepened into the synergistic action of small metal nanoparticles decorating semiconductors and on how the controlled architecture of the latter may have a strong influence on the final photocatalytic outcome [13,14,15,16,17,18,19,20]. Another interesting aspect of metallic nanostructures correlates with their unique optical response that can be modulated upon variations of their specific size and morphology [19,20,21,22,23]. At the interface between the metal surface and other medium (with different dielectric properties), exist a phenomenon known as Localized Surface Plasmon (LSP) that consist on a coherent delocalized electron oscillation leading to the generation of an electromagnetic field both outside and inside the metal. An excitation with radiation of the right wavelength causes a resonance interaction and subsequent collective oscillation of conduction electrons, in the case of metallic materials, due to the restoring force between electrons and nuclei through Coulombic attraction (Figure 1). This phenomenon is called Localized Surface Plasmon Resonance (LSPR), and for metals like Au, Ag, or Cu, the LSPR may take place over a wide range of 400–1300 nm [24] as a function of their size and shape (Figure 1). Thus, metallic nanoparticles emerge as perfect candidates as visible near infrared (NIR) light harvesters to combine and improve the efficiency of semiconductor photocatalysts.

So far, TiO₂ has been set as one of the most explored semiconductors to form hetero-nanostructures in combination with metals to overcome its limited photo-response beyond UV window [25,26] that can overcome its one of the most widely used semiconductors to carry out photocatalytic reactions. Systematic evaluation of Au-TiO₂ hybrid systems exploring the role and influence of shape, specific configuration, heterojunction conformations, and so on, have been developed in the past years. Numerous and varied architecture designs have been successfully reported including core-shell (concentric and eccentric) [27,28], yolk-shell [29], Janus type structures [28], or even multi-component heterostructures [30,31] and their performance successfully tested towards energy and environmental applications [17,18,31,32,33]. The generation and assembly of these hetero-nanostructures offers multiple advantages but the number of alternative candidates to TiO₂ still remains as an open challenge. The present review intends to overview the most recent advances described in the literature involving the design of hybrid photocatalysts combining plasmonic Au nanoparticles and non-titania based semiconductor coatings organized in a wide variety of nanoarchitectures (vide infra). Herein, we paid special attention to plasmonic hybrids that involved the selection of anisotropic Au nanostructures (mostly nanorods (AuNRs) and nanostars (AuNSs)) and a controlled growth of semiconductors beyond the most typically studied (i.e., TiO₂ [27,34,35] or ZnO [5,36,37]). This approach allows a fine control of sizes for both metal and semiconductor, reduces the probability of recombination of carriers, and maximizes an intimate contact to form efficient heterojunctions [38]. In contrast, other methodologies lack sufficient control on the size and dispersion of metal nanoparticles or the corresponding supports. Furthermore, many times these semiconductor nanoparticle supports require additional tuning or post-treatments to ensure the exposure of preferential facets that do not necessarily prevent numerous bulk recombination events due to their inherent polydispersity [13]. Hence, recent research innovative trends, such as cancer therapy, require more accurate control of the photocatalysts dimension for proper internalization in cells and accurate reproducibility [39]. Therefore, we consider that the efforts made to improve the generation of novel plasmonic photocatalysts with a controlled size and distribution in a core-shell (or analogous nanoarchitecture) and/or semiconductor supports represents a very promising alternative to other metal-semiconductor configurations [13]. Herein, we have made special emphasis on highlighting the latest achievements with AuNRs or AuNSs as plasmonic cores and different oxides (CeO₂ and Cu₂O) or chalcogenide semiconductors (CuS, CdS) grown with core-shell, Janus, or dumbbell-like configurations. We have correlated the influence of the different architectures with their final photocatalytic response.

2. Metal-Semiconductor Hetero-Nanostructures: Different Configurations and Light-Driven Activation Mechanisms

A key feature of plasmonic photocatalysts is related with the photo-induced generation of highly energetic electrons (hot electrons) generated via LSPR [40]. The distribution of these hot electrons can be described by the Fermi equation using an elevated effective temperature [41]. After light absorption, LSPR decay may occur either radiatively, through re-emitted photons, or non-radiatively, for instance through transfer of hot electrons [42], generally through intraband excitations within the conduction band [40,43] thereby causing electrons from occupied energy levels to be excited above the Fermi energy. Hence, after coupling with metals, typical semiconductors used in photocatalysis can capture these hot electrons and generate reactive species by using visible-NIR light. The formation of metal-semiconductor heterojunctions allows hot electrons to be accepted into semiconductor conduction band and carry on the photocatalytic process (Figure 2a). The energetic barrier formed at the metal-semiconductor interface is called Schottky barrier [44]. Hot electrons are injected into semiconductor conduction band when their energy is superior to Schottky barrier energy (E_SB) which is lower than the bandgap of semiconductor (E_g) [45]. After the hot electron generation, holes are also generated in the plasmonic structure as illustrated in Figure 2b. For this mechanism to occur there must be a good interaction between the metal and the support, which make critical the synthesis step of these hybrid materials.

Nevertheless, several mechanisms could be involved independently or, most often, concurrently in photocatalysis using hybrid plasmonic materials [46]. Depending on the interaction of the plasmonic nanoparticle with the support and the electronic characteristic of the latter, plasmonic excitation can improve the photocatalytic properties of materials in several ways: (i) increasing absorption and scattering of light [47]; (ii) enhancing of the localized electric field [48]; (iii) hot charge carriers generation and transfer [49], already mentioned; (iv) dipole induction on non-polar molecules [50]; (v) local heat generation [46,51], depicted in Figure 2c.

Since nature, size, shape, and crystalline structure of the nanoparticle determine the energy of the LSPR [52] and in consequence the wavelength of the light used in photocatalysis, the control of those cited parameters is fundamental for the synthesis of suitable hybrid materials [53,54,55]. Concretely, the introduction of anisotropy in plasmonic-semiconductor systems adds a superior level of performance. Plasmonic Au cores exhibit a wide range of anisotropic nanostructures (nanoshells, nanorods, hollow spheres, nanoprisms, triangles, cubes, nanostars, urchins, etc.) with different LSPR [8,18,24,31,52,56] (Figure 3a). Typically, plasmon energy is concentrated on the sharpest edges of the anisotropic plasmonic nanostructures of high curvature [57,58,59] where light harvesting will take place more efficiently [60].

As shown in Figure 3b, it is possible to differentiate between core-shell, dumbbell, Janus, and yolk-shell configurations attending to the semiconductor distribution around the metallic cores [38] with several implications in their photocatalytic activity. Structures with exposed surfaces of both the metal and the semiconductors (i.e., dumbbell and Janus) exhibit a superior photocatalytic performance due to the continuous exposure of the reactants with the generated charge carriers (e⁻/h⁺) [27,40,60,61,62,63,64]. Different anisotropic heterostructures have been prepared with SiO₂ [65], Pt [66,67], Fe₂O₃ [68], Cu₂O [60], CeO₂ [63], and TiO₂ [27]. Yolk-shell structures (Figure 3b) are characterized by a hollow shell and an inner plasmonic core with several benefits for the photocatalytic process: (i) the presence of a hollow shell ensures higher specific surface area as it possesses an inner and external part. (ii) The small thickness of shells shortens charge diffusion distance, reducing possible bulk recombination processes, and (iii) void space present allows reflection of light inside the hollow shell, causing light scattering, and boosting the number of available photons [69].

3. Evaluation of Au-CeO₂ Anisotropic Hetero-Nanoarchitectures

The combination of plasmonic Au nanostructures with cerium oxide as complex hybrid architectures represents an interesting example of synergistic photocatalyst that maximizes the positive properties of each individual counterpart beyond the most explored TiO₂ or ZnO hybrids [70,71,72]. The redox properties of Ce endow this oxide with large capability to transport oxygen via Ce(IV)-Ce(III) pair redox cycles within its framework [21]. As a result, an extensive number of oxygen vacancies are present in ceria-based materials surface [73], favoring a continuous cycle of catalytic reaction-regeneration in oxidation reactions [21,74,75,76,77]. Nevertheless, CeO₂ exhibits two major drawbacks for photocatalytic purposes: (i) CeO₂ possesses a band gap of 3.2 eV burdening its photocatalytic response to the ultraviolet region; (ii) CeO₂ possesses low carrier mobility, hindering its transport to surface after the electron-hole pair photogeneration to react with the targeted substrates [78,79]^.

Regarding the generation of the plasmonic hybrid photocatalyst, (i) CeO₂ as an n-type semiconductor possesses high density of states in its conduction band, which confers a good electron-accepting capability [40] and forms a heterojunction and Schottky barrier with Au to allow a proper hot-electron injection [21,22] (Figure 4). Therefore, the synthesis and development of Au-CeO₂ photocatalysts to maximize the redox properties of ceria has boosted the research on optimizing the interaction of these materials through an extended variety of imaginative configurations. Thus far, CeO₂ has been successfully deposited onto nanorods (both forming dumbbell [22] and core-shell [21,22,80] structures), spheres [21,81], or hollow cages [82].

In general, the synthesis of core-shell Au-CeO₂ nanostructures has been typically reported via hydrothermal treatment of a chelated Ce precursor [21,22,80,81,82]. Li et al. [21] reported the first synthesis of Au-CeO₂ nanostructures with core-shell configuration. One of the problems facing CeO₂ selective deposition on Au nanostructures was the rapid condensation of Ce precursors to form CeO₂. By using ethylenediaminetetraacetic acid (EDTA) as chelating agent, fast hydrolysis of Ce (III) ions could be prevented [83]. Thus, tuning Ce (III)/EDTA ratios allowed a fine control of CeO₂ deposition rates. Regarding the synthesis of dumbbell and Janus Au-CeO₂ systems, two outstanding contributions have been demonstrated in the recent literature by Pan et al. [63] and Jian et al. [22], respectively. It was possible to control the selective growth of CeO₂ on AuNR tips by controlling the amount of cetyltrimethylammonium bromide (CTABr) adsorbed onto the Au surface, which acted as a blockade of CeO₂ nucleation.

The first reported examples of Au-CeO₂ hybrids in photocatalysis were structurally analogous to the conventional counterparts traditionally used in heterogeneous catalysis for CO oxidation reactions. Kominaim et al. [84,85] reported the use of Au nanoparticles dispersed onto nanostructured CeO₂ supports to oxidize formic acid under visible light irradiation [84,85]. Nevertheless, Au-CeO₂ photocatalysts consisted in the random deposition of Au nanoparticles onto a relatively large CeO₂ support that prevented a deep understanding of the photocatalytic mechanism. As mentioned before, Li et al. [21] reported the first core-shell configuration for Au-CeO₂ nanoparticles. They tested the use of both spheres and rod-shaped Au plasmonic cores. The Au@CeO₂ catalysts exhibited different LSPR absorption maxima at different wavelengths within the visible-NIR region (530, 591, 715 nm, respectively) [21]. These hybrid configurations were employed towards the photooxidation of benzylic alcohol. The catalysts were subjected to calcination post-treatments to remove the excess of the directing surfactant employed to grow the ceria shell (CTABr), which did not possess any influence in the morphology/crystallinity of the sample. The influence of this calcination treatment was thoroughly evaluated upon irradiation with two different excitation sources (Xe lamp (λ > 420 nm) and laser (λ = 671 nm)). For Xe lamp, a systematic study of Au@CeO_{2, LSPR 530nm} was performed. It was found that the Au@CeO₂ calcined sample exhibited the highest photocatalytic activity, suggesting that the removal of CTABr enhanced the formation of Schottky barrier between Au and CeO₂ and facilitated a better charge carrier transfer in the catalyst. Upon laser irradiation, a similar trend was confirmed for the uncalcined and calcined samples, respectively. Calcined Au@CeO_{2, LSPR 591nm} showed the highest photocatalytic activity. The authors attributed the enhanced response to the better match between the LSPR absorption maximum of the calcined hybrid (redshifted after calcination to 680 nm) and the incident laser wavelength at 671 nm.

As previously pointed out in Figure 2c (vide supra), not only the hot electron injection pathway can supply electrons in the semiconductor conduction band. Interestingly, the generation of heat caused by one of the decay mechanisms caused by LSPR [86] can also promote the excitation of electrons from the valence band to the conduction band. As control experiments, Li et al. [21] studied the influence of the temperature in the catalytic activity in the absence of light. They found a considerable catalytic activity at the temperature that reached the system under illumination in darkness conditions, suggesting that part of the photocatalytic activity was thermally induced. As a potential mechanism (Figure 5a), the authors suggested a series of redox steps (vide infra): (i) electron injection into Ce(IV) conduction band, which generated Ce(III) species; (ii) O₂ adsorbed on previously generated Ce(III) sites forming Ce(IV)-O-O· (iii) that radical can remove α-H of benzyl alcohol yielding Ce(IV)-O-OH; (iv) the radical dehydrogenated benzylic alcohol combines with Ce(IV)-O-OH to produce the final benzaldehyde product and H₂O₂, and simultaneously, an electron returning to the Au bands to ensure electron-hole pair recombination.

Alternatively, Wang et al. [80] demonstrated that the plasmon-induced hot-electron injection under NIR illumination in CeO₂ coated AuNRs with Janus configuration accelerated photo Fenton-like reactions using H₂O₂ as substrate. Again, the hot electron injection in Ce(IV) induced the formation of Ce(III) that acted as active species in Fenton-like reactions [87] with the generation of highly reactive ·OH that subsequently facilitated the degradation of an organic model pollutant. As depicted in Figure 5b, a heterostructure with plasmonic exposed faces possess important photocatalytic advantages. Pan et al. [63] demonstrated that half-encapsulated AuNRs with CeO₂ possessed better activity in the catalytic reduction of 4-nitrophenol (4-NP) with NaBH₄, in comparison with their core-shell AuNR@CeO₂ counterparts. Hot electrons were transferred from the plasmonic AuNR into CeO₂, generating holes on Au surface. BH₄ can donate electrons to AuNR generating H species that further convert 4-NP into 4-aminophenol. On the other hand, CeO₂ is able to withdraw e⁻ from AuNR that absorbs H⁺ from H₂O yielding the same H active species [88]. This phenomenon could not occur in the case of core-shell AuNR@CeO₂ since Au facets remained unexposed to the liquid reaction media and yielded a lower catalytic activity.

Following a similar methodology, Jia et al. [22] combined CeO₂ with AuNRs forming a dumbbell heterostructure (Figure 5b) and evaluated their photocatalytic performance in the N₂ photofixation with NIR irradiation. CeO₂ rich surface in oxygen vacancies (OVs) (Ce(III) sites) chemisorbed N₂ that could be reduced to NH₃ by injected plasmonic hot electrons, breaking triple bond N-N. Due to the exposure of Au facets to the reaction media, as-generated hot holes could be consumed by a hole scavenger (in this particular case, methanol) to close the photocatalytic cycle. Core-shell AuNR@CeO₂ nanostructures prevented the availability of hot holes and the accessibility of CH₃OH to the active sites, which reflected in a huge difference in the NH₃ generation rate (114.3 vs. 18.44 µmol·h⁻¹·g⁻¹) [22]. Thus, CeO₂ emerges as a promising photocatalytic material in combination with Au plasmonic nanostructures, being the main highlights the regeneration of highly desired OVs due to hot electron transference from Au and the photothermo-induced oxygen mobility [21]. The high oxygen mobility has shown to have influence during the photothermo-catalytic process, as demonstrated by Li’s experiments with temperature and in the absence of light [21,89]. CeO₂ as n-type semiconductor [90] and as oxygen ion conductor possess a unique charge separation effect: when a photon is absorbed, an electron is excited from the valence band (orbital O_2p) to an empty conduction band (orbital Ce_4f), forming a Ce⁴⁺(e⁻)/O²⁻ (h⁺) pair. The as-generated electrons remain localized in Ce (IV) centers favoring the charge separation and consequently, reducing the recombination phenomena (Figure 6) [90,91].

Cited reports confirm an effective synergistic effect in the photocatalysis between LSPR and CeO₂ and an enhancement of the photocatalysis with Janus-type heterostructures [22,80]. Hence, synthetized Au-CeO₂ nanostructures interact with visible and NIR light, remarking the promising interest of these materials in environmental, biomedical applications and solar energy harvesting, among others [69,91,92,93,94,95,96,97,98].

4. Evaluation of Au-Cu₂O/Au-CuS Anisotropic Heteronanostructures

Cuprous oxide has attracted much attention as a photocatalyst due to its remarkable features as a semiconductor, and its extended variety of active roles in solar and energy applications of paramount interests for the Sustainable Development Goals targeted by most of the worldwide governments’ agendas. Copper (I) oxide materials have been widely used in the degradation of organic compounds in solution and their photocatalytic activity has been studied against a large number of potential pollutants, such as organic material [99,100], drugs [101] or nutrients [102]. In addition, studies are also emerging where copper (I) oxide is being used in the generation of molecular hydrogen [32,103]. It is a highly abundant p-type semiconductor on earth with an energy band gap (E_g) of ≈ 2.17 eV that has a low cost and good absorption capabilities in the visible range when compared with other semiconductors [50,104]. However, some drawbacks have been also consistently reported for this oxide such as: (i) rapid recombination of electron-hole pairs; (ii) limited absorption of the semiconductor in the visible range; or (iii) natural tendency to form larger structures or disproportionation (especially in liquid media). The combination with anisotropic plasmonic cores to generate metal-semiconductor heterostructures represents an appealing alternative to overcome these problems [56,105,106]. These hybrid nanostructures offer two remarkable features, the Schottky barrier and the LSPR that improves the photocatalytic response by decreasing the carrier recombination rates and expanding the absorption range to the visible.

The selective growth of copper oxide onto plasmonic Au cores has been a trendy topic widely studied by multiple researchers in recent years [32,56,105,106,107,108]. However, the final morphology and selective exposure of specific crystalline facets has also attracted great interest in the field of photocatalysis in terms of both selectivity and reactivity [106,107,109]. The generation of core shell Au-Cu₂O hybrid structures with high precision has been reported in the recent literature [107,109,110]. Several parameters such as the nature of the reducing agents, the concentrations of reducing agent or pH values have been systematically evaluated to modulate the morphology, shape, and length of the oxide shells and favor the preferential exposure of selected crystal facets [32,109] (see Figure 7). For instance, Kuo et al. [109] reported the controlled synthesis of different morphologies including cubic structures, truncated cubes, cuboctahedra, or truncated octahedra to octahedra upon tuning the variation of NH₂OH. On the other hand, Kuo et al. [106] explored the influence of the different morphologies of AuNPs coated Cu₂O shell on the final photocatalytic performance towards the methyl orange (MO) degradation. They showed how core-shell architectures improved MO degradation from 18 to 50 µmol·cm⁻² in comparison with pristine Cu₂O nanostructures and how the influence was higher in the case of the cube-shaped counterparts. This was an example where the plasmonic core helped in the transfer of carriers and promoted a better conductivity in a photo-electronic device. Yuan et al. [107] also developed hybrid AuNRs-copper oxide nanoparticles where they explored variations on the type of reducing agent to tune the preferential octahedral or cubo-octahedral morphologies (see Figure 7a). Both structures were generated by Ostwald ripening process. Smaller crystals were dissolved and redeposited onto larger crystals when different reductants were added. Hydrazine as stronger reductant was reported to promote the copper species to their more stable morphology. They also tested their photocatalytic activity towards the degradation of MO. It was observed how hybrid structures enhanced more activity in comparison with analogous bare semiconductor counterparts. The Au internal core performed in a dual way, LSPR promoted electron injection into the semiconductor and also acted as a charge sinker to increase charge migration and separation. This was a representative example on how the presence of the Au core considerably affects the activity of the system and how the morphology of the external shell (corner-truncated octahedral vs. regular octahedral Cu₂O) is also highly relevant in the absorption and degradation of MO. The truncated octahedral core-shell led to 94% MO conversion from an original 10 mg·L⁻¹ MO concentration in 80 min [107]. This influence was also reported in Cu₂O nanoparticles with different morphologies that also exhibited a much higher photocatalytic activity in rhombohedral shaped structures due to the more selective MO absorption over <110> faces for subsequent further oxidation [111] (see Figure 7a). Recent studies have also reported the use of hollow Au cores to build Au-Cu₂O hybrids have also reported a successful tunability of their optical properties depending on their final architecture (Janus vs. core-shell). They reported an effective plasmon-induced energy transfer favored by the hollow Au nature that improved the MO degradation rates [56]. Other recent report by Xu et al. [112] has established a strong influence of the final heterostructure configuration on the capability to generate photocurrent. In this case, the used solid gold seeds incubated with a ligand that shifted the core-shell configuration towards a Janus-like structure.

Additional studies have also exploited the synergy of Au- Cu₂O hybrid materials. For instance, Yu et al. developed [32] a hot-dog-like hybrid configuration where they selectively deposited copper oxide on the transverse part of large aspect ratio AuNRs, leaving the edges completely exposed [32]. It was possible to change the kind of Cu₂O-AuNR architecture from core-shell to tips-exposed hot-dog type structure by reducing the amount of copper precursor added to the synthesis. The preferential epitaxial growth of Cu₂O over the transversal part of AuNR promoted a first nucleation over these sides, leaving the edges completely uncoated under low copper concentration conditions. In a sequential manner, they even incorporated very thin TiO₂ layers on top of the copper oxide and established a critical evaluation of the photocatalytic activity for each hybrid configuration towards hydrogen production. The charge separation efficiency was significantly higher for the hot-dog configuration than for the bare Cu₂O or the fully-covered core-shell counterparts. The hot-dog configuration also exhibited an enhanced photocatalytic response almost 24-fold higher than the Cu₂O nanoparticles (80 µmol H₂ g⁻¹ h⁻¹ vs. 3.4 µmol H₂ g⁻¹ h⁻¹) and four times higher than the core-shell configuration (19.3 µmol H₂ g⁻¹ h⁻¹). The authors partially attributed the outperformance of the hot-dog configuration to the generation of four hot-spots at the interface of the non-coated AuNR and the semiconductor shells (see Figure 7b) of the hot-dog configuration. They concluded that this interface was more efficient in the transport of carriers due to the direct interaction of holes with glucose (used as sacrifice agent) and to the more efficient generation of hot electrons to induce the reduction step toward H₂ generation (see Figure 7b) [32]. The results were even better in the case of the ternary shell due to the more efficient p-n junction formed with the addition of a titania layer.

Even though ternary heterostructures are not the main focus of this review, it is worth mentioning recent efforts to build Z-schemed TiO₂-Au-Cu₂O photoelectrodes [30] or the addition of a third component of small Au NPs to the Au-Cu₂O hybrids reported by Yu et al. [113]. They designed a more complex core-shell AuNR@Cu₂O (octahedral) architecture additionally decorated with AuNPs (Figure 8). This ternary composite was able to increase the photocatalytic activity of the binary material by incorporating AuNPs, improving the transfer and separation of displaced charges through the semiconductor. Using higher wavelengths where Cu₂O is not able to absorb, light would penetrate through the structure reaching AuNR. Hot electrons could be generated via LSPR effect and injected into the conduction band of the semiconductor. Hot electrons and holes generated in the conduction band of Cu₂O could be promoted from bulk Cu₂O to the surface of the catalyst where the electrons are mainly trapped by Au nanoparticles to form singlet oxygen radical from the O₂ present on the surface. Meanwhile, holes places mainly in the surface of Cu₂O p-type semiconductor are used for the direct MO degradation. An electron transfer “push–pull” synergetic effect was enhanced with the help of both AuNR core and AuNPs on the surface of Cu₂O.

Sang Woo Han and co-workers conducted a comprehensive and systematic study where they developed hybrid plasmonic structures and explored how the variation of configuration affected the photocatalytic performance for hydrogen production using methanol as hole scavenger [60]. They synthesized plasmonic gold nanostars (Au NSs) and partially deposited copper monoxide (Cu₂O) tuning the stabilizing agent (sodium dodecyl sulfate (SDS), (CTABr), or Polyvinylpyrrolidone (PVP) (see Figure 8b). The Janus-like structures with <111> vertices completely covered by the semiconductor were achieved by using PVP as a stabilizing agent. A complete core-shell type configuration was achieved with the aid of sodium dodecyl sulfate (SDS) as a stabilizing agent and the structure with the exposed vertices was achieved by combining two surfactants, CTABr and SDS. The nature of each stabilizing agent was crucial for the selective deposition of the semiconductor (Cu₂O) onto the plasmonic Au NSs. Another important parameter to consider is the curvature of the surface of the plasmonic Au cores. In this systematic study, other Au seeds were used to prepare Janus-like architectures. It was observed how more shaped seeds could satisfactorily enhance the proposed architecture. However, it was not possible for less shaped seed structures using PVP as a stabilizing agent. In contrast, samples with less sharp vertex led to the preferential formation of core-shell type structures.

Yu et al. [60] also found that the use of PVP as a capping agent allowed a stronger interaction with low curvature sites from the Au NSs apexes, thereby leading to a preferential nucleation and growth of Cu₂O tips at those positions (see Figure 8b). Homogeneous core-shell heterostructures required the use of SDS instead of PVP. The random assembly and weaker protecting capability of SDS enabled a complete overgrowth of Cu₂O. In vertex-exposed Cu₂O-Au NSs a binary mixture of CTABr/SDS was used. The electrostatic interaction between both surfactants induced the generation of more stable micelles in aqueous solution that preferentially absorbed in the high-curvature sites, leaving the rest of Au nanostructure more accessible for Cu₂O nucleation. To further analyze the influence of the angle and the surfactants in the synthesis of anisotropic structures, AuNRs with different aspect ratios and binary mixtures of two surfactants (CTAB and SDS) were also evaluated by Yu and coworkers. They observed that the formation of Janus-type structures was preferentially induced as the aspect ratio of the AuNRs increased [60]. The authors studied the photocatalytic response of the different AuNSs-copper oxide configurations. A clear correlation between the efficiency of conversion of solar energy and the topology could be established during the hydrogen generation using methanol as sacrificing agent. Better conversions were achieved in partially covered structures compared to fully covered structures. Furthermore, no catalytic response was observed in the absence of the catalyst or in the presence of non-coated plasmonic cores. Furthermore, copper oxide of physical mixtures of both Cu₂O and AuNSs did not render any significant conversion, thereby reinforcing the synergistic role of the plasmonic photocatalyst with a well-defined architecture (especially for the Janus-type configurations).

The comparison of the photocatalytic activities among core shell-type structures with different plasmonic structures (spheres and stars) in their core can be explained considering the excitation capacity of both materials at different wavelengths. Using excitation wavelengths of 700/750 or 800 nm, the sharp surfaces of AuNSs caused a much higher plasmonic excitation than their spherical counterpart. Janus-like architectures present higher catalytic activity than Core-Shell structures. Not covered Au highly increased the final photocatalytic activity of hybrid structure letting the oxidation reaction of hole scavengers to occur for methanol, whereas core-shell architecture dramatically deprives the accessibility of Au surface to hole scavengers.

It is shown how using identical AuNSs cores with a Cu₂O overgrowth forming anisotropic heterostructures, induced completely different photocatalytic outcomes. It was observed a strong influence of Au-Cu₂O plasmon energy transfer. Coated vertex Au nanostar enhanced higher plasmon excitation in comparison to vertex exposed Au NSs configurations and leading the photocatalytic activity of hybrid structures under those experimental conditions [60]. To better understand the interaction mechanism between the plasmonic core and the outer semiconductor layer, the AuNSs were completely coated with a thin insulating SiO₂ layer prior to the subsequent deposition of copper oxide. The hybrids with the SiO₂ interlayer exhibited a considerable decrease in activity, pointing out that the mechanism of action was carried out by plasmon-induced photocatalysis [60].

Au-Cu₂O hybrid structures are being also currently explored for biomedical applications [114,115,116,117] such as photodynamic therapy treatment of cancer cells. Xu et al. [118] performed in vitro studies with AuCu-based materials, taking advantage of their ability to generate toxic reactive oxygen species (ROS) and its highly efficient phototherapy. AuNRs were used to selectively deposit copper on one of the sides of the Au structure using hexadecylamine as a protective ligand and sequentially making a coating with Au, which modified the structure inducing a galvanic replacement reaction and decreasing the amount of copper in the sample.

Interestingly, there are additional copper-based heterostructures where the semiconductor deposited over the plasmonic structure of Au is a Cu-based chalcogenide. These semiconductors are not expensive and more stable p-type semiconductors with a well-defined LSPR in the visible region. Furthermore, the amount of free charges in the material can be tuned upon modification of experimental parameters. Generating non-stoichiometric copper deficient structures of CuS induces the formation of transporters in the p-type semiconductor and exhibits subsequent properties such as photothermal catalysis or photocatalysis through the generation of ROS [119]. This type of structure can be made from previously formed copper oxides, giving rise to yolk shell-type structures where the plasmonic core was enclosed in an external copper chalcogenide shell and this can move freely within the voided structure [120]. To carry out the synthesis, it is typically necessary to obtain a copper (I) oxide structure through the conventional protocol, growing copper oxide following an epitaxial growth process. Later, the sample is sulfurized by introducing a sulfide precursor, such as Na₂S in an acidic medium, and it is allowed to evolve until the yolk-shell type structure is formed, maintaining the crystalline faces of the original Cu₂O structure. The catalytic ability of Au-CuS yolk shell structures to form ROS has been used in cancer chemotherapeutic therapies. Zhang et al. used the photothermal and photodynamic properties of this particular hybrid to perform a co-therapy using the resonance energy transfer (RET) mechanism [116].

Wang et al. [121] synthetized hybrid core-shell Au-CuS nanostructures directly without the need to use Cu₂O as a sacrificial agent in the formation of the nanostructure and modified the final properties of the photocatalyst through the non-stoichiometric growth of CuS at different amounts of copper. They carried out a systematic study in which they evaluated the photocatalytic properties at different wavelengths, ranging from UV to Infrared through the visible spectrum and determined the influence of the different parts of the hybrid nanoparticle on the activity of the catalysts. The plasmonic characteristics of the Au nucleus could be measured by changing its dimensions and modifying the thickness of the CuS coating deposited around it. On the other hand, the plasmonic characteristics of the copper sulfide shell could be modified by changing the concentration of free holes, which is closely linked with the crystalline phase of copper sulfide and the Cu:S ratios found.

Nanostructured Au can act as a photosensitizer (see Figure 2), being able to generate more electron-hole pairs in the semiconductor by means of RET mechanisms and thus improve the photocatalytic activity in the visible spectrum. Moreover, charge recombination in the surface of the semiconductor can be partially suppressed with the help of metals and charge migration through the bulk of Cu₂O could be enhance more easily in the presence of Au. The controlling deposition of copper around Au nanostructure enables an overall downsize of the hybrid nanoparticles compared with conventional Cu₂O nanostructures. It also allows a more selective deposition of Cu₂O. All these encountered possibilities appearing with hybrid structures have a positive influence to enhance the photocatalytic response of the hybrid.

5. Evaluation of Au-CdS Anisotropic Heteronanostructures

Other semiconductors commonly used in photocatalysis reactions are metal chalcogenides. However, some of their most remarkable features, such as the absorption of light at certain wavelengths or the separation of charges are far from optimal. For this reason, alternatives have been proposed to improve the photocatalytic properties of these systems when we use a range of wavelengths that encompasses the visible or infrared spectrum while achieving a better and more effective charge separation. An interesting alternative that arises from the need to solve this problem is to create hybrid metal-semiconductor nanostructures that allow the use of the visible-NIR spectrum thanks to the metal’s plasmonic properties. Moreover, a wisely metal incorporation in semiconductor to form hybrid structure also improve the charge separation and its photocatalytic activity. Metal chalcogenides have been in the spotlight for the past years due to their versatility and potential number of applications in the energy field (i.e., photovoltaics, imaging). In addition, chalcogenides semiconductors have been also studied for photocatalytic emerging applications in solar energy conversion and renewable energy development [34,122,123,124,125,126,127,128,129].

Great interest has been devoted in recent years to the development of photocatalysts able to exploit the sun’s energy to generate combustible, thus trying to reduce the amount of fossil fuels used [130,131]. One of the most interesting alternatives is H₂ generation as fuel using solar energy and semiconductor-based photocatalysts are postulated as one of the most promising alternatives to achieve this objective [132]. Some interesting features for solar-assisted photocatalysis in H₂ generation could be: [133] (i) It is necessary to find a semiconductor that has a narrow band gap and is highly efficient in absorption light; (ii) the band edge has to be more negative than standard hydrogen reduction potential; (iii) it should have good enough charge separation and migration to provide enough active sites.

There is a wide variety of photocatalysts but CdS is posited as a promising candidate for the generation of hydrogen. It has an optimal band gap of 2.4 eV that allows it to have photocatalytic activity under wavelengths of up to 500–600 nm in addition to a conduction band edge suitable for the generation of H₂ [134]. On the other hand, this semiconductor suffers from a rapid recombination of charges that make it inefficient. As an alternative to improve the migration and separation of charges, the generation of metal-semiconductor hybrids appears again as an alternative recently explored in the field [61,133,134,135,136,137]. In this particular hybrid, the rich existing chemistry of quantum dots has boosted the search for a wide variety of architectures searching the optimal coupling of energy levels necessary to improve the photocatalytic outcome towards H₂ production [34,136]. Developing anisotropic structures such as dumbbell type bimetallic particles [135,136], Janus-type particles [62], or yolk-shell structures [34,138] have reported considerable increases in the activity of the systems in comparison with more conventional core-shell type structures. No considerable activity enhancements have been shown in this latter configuration in comparison with the bare CdS semiconductor.

Zhao et al. [137], developed core-shell type hybrid structures and verified their efficiency in photocatalytic reactions for the production of hydrogen. They were able to synthesize concentric nanostructures where the plasmonic Au-core was completely covered by the semiconductor. Alternatively, they were also able to generate eccentric structures where the core of Au was shifted, having part of internal Au core exposed (see Figure 9a). For the synthesis of these nanostructures, a two-step multistage was necessary in which, first, silver-coated Au structures were generated in a core-shell configuration. Subsequently, the silver shell was sulfurized and a mixture of these nanoparticles were allowed to evolve with the cadmium precursor and certain conditions to enhance Au-CdS Janus-like/core shell-like nanoparticles. Zhao et al. reported [62] the need of eccentric nanoparticles of silver sulfide with partial crystallinity for further evolve into the generation of Janus-type structures. The catalytic studies showed that anisotropic materials with partially exposed Au core possessed much higher photocatalytic activity than completely coated core-shell configurations. A remarkable 730-fold enhancement in the H₂ evolution rates was detected for the anisotropic dimer (7.3 mmol H₂ g⁻¹h⁻¹ vs. 0.9 µmol H₂ g⁻¹h⁻¹, respectively). The authors claimed that increasing the temperature was necessary to ensure the proper presence of a well-defined crystalline phase. This phase was the niche that enabled a rational control of the anisotropy degree (see Figure 9a). In addition the partial exposure of the plasmonic Au was also determinant to maximize the SPR effects [62].

The photocatalytic activity of these hybrid structures was proved to have a highly symmetry dependence [137]. Janus-type particles were the most active, followed by analogous semiconductor particles and ending with core-shell Au-CdS nanoparticles. Furthermore, the Janus hybrids with the more pronounced anisotropy also exhibited the highest photoactivity using 300 W Xe lamp with application of a 400 nm cut-off long pass filter for water splitting reaction. This was attributed to the supporting role of Au acting as electron sink that could delay e/h⁺ recombination rates and perform water reduction reaction. Upon excitation at wavelengths higher than 400 nm, a rapid electron transfer occurs between the conduction layers of the semiconductor (CdS) towards the Au surface thereby enabling a more efficient water reduction to generate hydrogen. However, the nanostructured bare CdS or core-shell Au-CdS configurations prevent the separation of charges in such an efficient way and in the case of core-shell particles, the electrons remain trapped in the core of the particle and precluding the photocatalyst from an efficient hydrogen evolution reaction [137].

In an attempt to expand the photocatalytic response towards the less energetic NIR range and enhance core anisotropy, Li et al. [139] explored the possibilities of plasmonic AuNRs as inner plasmonic core. In this case, AuNRs were completely covered with silver to generate silver sulfide in a second step that would end up giving rise to CdS under certain conditions. Wang et al. [140] also developed more complex core shell-like nanostructures generating multilayered chalcogenide shells containing different combinations of Bi₂S₃ and CdS (Figure 9b,c). These samples were tested for dye photodegradation under visible region being the ternary combinations the most active in comparison with the binary counterparts (Au-Bi₂S₃ or Au-CdS). The enhancement observed in the ternary component was attributed to the transfer of electrons from the exposed semiconductor (CdS) with a 2.4 eV band gap passing through the Bi₂S₃ semiconductor intermediate layer with a 1.32 eV band gap until reaching the internal Au core. However, holes as positive charges, take longer to move from the valence layer from one semiconductor to another, enabling charges separation and, therefore, increasing photocatalytic activity. Moreover, hot electron injection from visible activated dumbbell-like AuNR to semiconductor can considerably increase the photoactivity compared with bare semiconductor or semiconductor coated conventional GNRs (see Figure 9b,c).

Xu et al. [136] developed different metal-CdS hybrid bimetallic structures and evaluated their photocatalytic activity to form benzaldehyde and H₂. They synthesized three types of plasmonic heteronanostructures: CdS coated AuNRs, CdS coated-Pt coated AuNRs, and CdS coated-Pt tipped AuNRs and compared their photocatalytic activity. It was observed an increasing order of activity: Au@CdS < Au@Pt@CdS < Au-Pt@CdS (50 vs. 100 vs. 150 µmol H₂ g⁻¹ h⁻¹, respectively). They also carried out controls with the CdS semiconductor and metallic nanoparticles, but the activity decreased considerably in both cases. In this study, the use of AuNRs improved the harvesting of visible-NIR light and induces a local electric field (Figure 2) that is capable of promoting the generation and separation of charges from the CdS semiconductor (Figure 10). The anisotropically deposited platinum at the edge of the nanorod acted as an electron sinker or electron reservoir enabling the electrons to be directed to the places where the reduction reaction of the protons takes place to obtain H₂. In addition, the asymmetric separation of both metals served as a highway for an efficient transport of electrons that partially prevent electron-hole recombination (see Figure 10). This represents another good example of how smart designs and proper understanding of the carrier transport mechanisms facilitates an improved photo-response outperforming the simpler configurations.

Wu et al. [135] explored an analogous combination of Au-Pt-CdS components structured in core-shell (Au@CdS) or AuPt-CdS dumbbell like configurations with different coverage degree (see Figure 11a). These hybrids were systematically tested in the photocatalytic hydrogen generation reaction under visible or NIR light irradiation [135]. The photocatalytic activities were compared when systems were faced with wavelengths belonging to the visible or the near infrared light and it was seen that the order in the activity depended not only on the type of structure used, but also on the wavelength used because of the mechanisms enhancement. When the catalysts were excited with wavelengths within the visible range, the order in the catalytic activity was as follows: Au-Pt-CdS >> Au@CdS > CdS >> Au-Pt. In contrast, when NIR illumination was used, the activity dropped significantly, and the order of the catalytic activity was: Au-Pt > Au-Pt@CdS > Au > Au@CdS. Still, the H₂ evolution rates (in the micromolar range) were less remarkable than those reported for the heterodimers (vide supra).

Under sunlight illumination, CdS semiconductor deposited on the sides of Pt-Au bimetallic nanoparticle induced both the RET and hot electron transfer processes and synergistically boosted the photocatalytic H₂ generation. However, when catalysts were activated by infrared light, Au-Pt bimetallic structure became the most active architecture indicating that excited hot electrons can further transfer to tipped Pt, but it is more complicated for electrons to enhance further migration from AuNR to CdS semiconductor (see mechanisms in Figure 11b). It was also observed how the catalytic activity of CdS coated metal tipped AuNRs (Au-Pt-CdS) was more active than its concentric counterpart Au@Pt@Cds when nanocatalyst were activated with visible light. The transversal LSPR induced electric field cannot reach the outside CdS semiconductor due to the presence of Pt NPs between the inner Au core and the external CdS layer (see Figure 11b). Pt nanoparticles around AuNR prevented the charge transfer between CdS and Au internal core. It has been observed how strategically modulating the metallic core of the structure by incorporating anisotropy or integrating several metals in a controlled way, the activity of the system was highly increased.

Likewise, yolk-shell structures have been recently reported as promising alternative hybrids with enhanced photocatalytic outcome provided by the selective control of the shell properties [138,141]. Han et al. [138] developed yolk-shell structures containing a plasmonic AuNR. The photocatalytic activity towards the generation of H₂ was compared to analogous core-shell type architectures and hollow CdS nanoparticles [138]. Using hole scavengers such as Na₂S or Na₂SO₃ to avoid photocorrosion of the CdS and exciting upon visible wavelengths, the yolk-shell structure generated a hydrogen conversion over 27 (1.7 mmol H₂ g⁻¹h⁻¹) or 12 (0.5 mmol H₂ g⁻¹h⁻¹) times greater than the core-shell type and hollow semiconductor structures (0.02 mmol H₂ g⁻¹h⁻¹). The authors also compared the catalytic activity of the yolk-shell Au-CdS hybrid system with an analogous system containing an insulating intermediate layer of silicon oxide Au@SiO₂-CdS [138]. No significant differences were observed when compared both photocatalysts under the same experimental conditions. For further explanation, several mechanisms were proposed.

The thermal activity of the rod was tentatively considered as one of the possible causes of the increase in activity, but the thermal evolution of the system for all photocatalysts was very similar. Since higher activity was achieved by using Au-coated plasmonic structures covered with SiO₂, it was concluded that plasmon energy transfer or plasmon-induced RET (see Figure 10) was not the mechanism by which these systems increased their activity. The researchers determined that the main cause for this process was radiative relaxation of the SPR of the AuNRs, such as near-field enhancement and resonant photon scattering (see Figure 2 and Figure 11c). More recent studies have concurred in finding higher activity for the yolk-shell type [34]. Yolk-shell type structures have the ability to generate a very large surface-to-volume ratio, increasing the amount of surface exposed for photocatalysis and decreasing the amount of bulk material to reduce charge recombination probabilities. In addition, plasmonic Au is capable of generating better charge separation as well as potential light scattering of the light inside the nanoparticle (see Figure 11c).

We have seen several examples of hybrid architectures, where not only the type of metal incorporated (Au, Pt) was important, but its arrangement with respect to the CdS semiconductor played a crucial role in enhancing the catalytic activity. Yolk-shell type structures provided a better optical response thanks to the light scattering generated in the internal part of this void structure. Dumbbell-like Au-Pt type bimetallic structures allow better migration and separation of charges, solving one of the most important problems of this type of semiconductor. It is important to highlight the development of CdS-based complex hybrid structures where a good design and configuration of the constructed architecture allows us to considerably improve the activity of these systems when it is compared with other less complex hybrid structures, such as bare CdS or core-shell architectures.

6. Conclusions

TiO₂ and ZnO assembled with plasmonic metals are still the most explored and developed heteronanostructures obtained using semiconductors. The search for novel alternatives that can expand and take advantage of the full-solar irradiation spectra has stimulated the search for novel alternatives that can provide additional features in terms of more efficient harvesting properties, more effective carrier transport and enhanced selectivities towards specific end-products. In this regard, the need to find newer and cleaner routes for energy production, the need to improve the efficiency of decontamination of harmful pollutants and the willingness to promote the revalorization of wastes or unwanted byproducts has boosted the research in the area of photocatalysis. Nowadays, the innovation in the design of smart combination of metal-semiconductors continue to be a very promising alternative. In this review, we presented some of the most promising and versatile alternative gold-semiconductor heterostructures. The main results highlighted throughout this revision of the most recent literature suggests an important morphology dependence in the final outcome and how less-conventional designs beyond core-shell configurations may pave the way for important breakthroughs in forthcoming years. Furthermore, the selection of anisotropic metals also represents an appealing strategy to maximize the virtues of plasmon excitation. Finally, we would like to make emphasis on the need to establish exquisite control and definition of these hetero-nanoarchitectures at the nanoscale level in order to establish a clear progress in biomedical applications, where size control and reproducibility are key for clinical translation.