Visible-light responsive organic nano-heterostructured photocatalysts for environmental remediation and H₂ generation

1. Introduction

The energy shortages and environmental pollution have constitute a worldwide concern, and thus demand a ready solution to assure the sustainable development of human society. Among the various proposed solutions, semiconductor-based photocatalysis has gained increasing interest over the past few decades, because it can convert solar energy to chemical fuels, in the form of hydrogen and hydrocarbon fuels, or to address environmental issues by purification of harmful pollutants [[1], [2], [3], [4]]. Since the pioneering work on photocatalysis reported by Fujishima and Honda in 1972, many semiconductors have been explored and tested as photocatalysts [[5], [6], [7], [8], [9]]. Generally, a photocatalytic reaction mainly include five steps [10,11]: i) light absorption to form electron-hole pairs, ii) separation of electron-hole pairs and migration to the semiconductor surface, iii) adsorption of the reagents, iv) redox reaction on the semiconductor surface, v) desorption of the products. In this regard, current efforts are being made to maximize the efficiency of these processes, mainly through increasing the solar light utilization and preventing the recombination of electrons and holes. As the first requirement, visible-light response is often a must for the applicability of a photocatalyst.

Previous work has mostly concentrated on the exploration of inorganic semiconductors (ISCs) as photocatalysts, but the fixed- and narrow-spectrum response often limits its overall catalytic efficiency [[12], [13], [14]]. By comparison, the use of organic semiconductors (OSCs) constitutes an important alternative, considering their strong and wide absorption bands in the visible region [[15], [16], [17], [18]]. Another important merit concerning organic materials is that their molecular structure and photoelectric properties can be simply tuned through molecular design and tailoring. Moreover, the progresses made in the supramolecular assembly of OSCs have prompted their photocatalytic applications by enhancing the stability and possibility of reuse in heterogeneous media. The effect of shape, size, crystal type and self-assembly on the optoelectronic properties have been studied extensively for organic nanostructures. Among the candidates, porphyrins [[19], [20], [21]], phthalocyanines [[22], [23], [24], [25]], fullerenes [25,26], perylenetetracarboxylic diimide derivatives (PDIs) [27,28], g-C₃N₄ [29,30], and etc have stood out (Fig. 1), because they possess many exciting features like excellent semiconducting behavior, thermal stability, and the self-assembly ability into ordered nanostructures. Thus far, a series of papers have appeared dealing with nanoscale organization of them for different photocatalytic purposes, involving organic synthesis, remediation of water and air, and water splitting into hydrogen.

Although great advances have been made in nanostructured organic photocatalysts, a single component can still not do a good job because of the short photogenerated electron-hole pair lifetime. It is widely accepted the electron-hole pairs tend to recombine very fast, i.e., in the range of nanoseconds. In order to enhance the spatial charge separation efficiency, a rational design of organic nano-heterostructures is always a research focus [[31], [32], [33]]. The heterostructures can allow for not only retarding the charge recombination, but above all going beyond the features of each one due to the synergistic effects. As a matter of fact, the design of organic nano-heterostructures provides a solid basis for creating high-efficiency photocatalysts for practical use.

Since a systematical information over the development of the field is still missing, we herein make a comprehensive review on the recent efforts in engineering various organic nano-heterostructures, with emphasis on the structure/property relationships, and their use in photocatalysis, including hydrogen evolution, CO₂ reduction, and pollutant purification. Before that we give a brief introduction of their structural and optoelectronic properties. Some challenges and perspectives on the future developments are also be covered in this review.

2. Structural and optoelectronic properties of OSCs

OSCs are a class of materials that are primarily composed of carbon and hydrogen atoms, sometimes containing a few heteroatoms like nitrogen, oxygen, and sulfur. They unit the electronic, chemical and mechanical benefits with a material structure that can be easily adjusted via chemical synthesis. That is, it is possible to tune the electronic properties for desirable absorption/emission wavelength, to make it soluble/insoluble, or to render it mechanically robust, lightweight, and flexible. These properties imply that the techniques available for processing or patterning OSCs may be realized with a variety of solution-processing techniques or vacuum deposition methods [34,35]. For example, the methods for patterning organic materials can include printing, embossing, imprint lithography, and etc. These properties also enable the booming flexible applications by use of OSCs as active materials in electronic and optoelectronic devices.

OSCs have the electronic advantages typical of semiconductor materials, but differ strongly from ISCs [36]. The weak, non-covalent, van der Waals forces between organic molecules always result in a low intermolecular orbital overlap and low dielectric constant, taking a value of about ε_r≈3.5 [37,38]. This implies that coulomb interactions are significant enough to bind the electron-hole pairs created by optical excitation. In another word, photon absorption in OSCs affords the generation of an exciton, or coulombically bound electron-hole pair. In order to overcome the exciton binding energy for useful charge carriers, the photogenerated excitons must diffuse to a heterostructure interface that has a sufficient band energy offset. While for ISCs, free charges are created by optical excitation from a valence band (VB) to a conduction band (CB), because the high dielectric constant (ε_r≈11) makes the coulomb effects between electrons and holes trivial due to dielectric screening.

Compared with the electronic coupling between atoms in ISCs, the weaker coupling between molecules in OSCs also leads to relatively narrow VBs and CBs, more localized charge carriers, lower carrier mobilities (typically 10^-5-0.1 cm² V^-1·s^-1), along with small exciton diffusion lengths (L_{D ≈} 3-40 nm) [37]. As for the more localized carriers in OSCs, the terms of highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) are used here to replace VB and CB, respectively. The mobility limitations also emphasize the importance of building nanoscale interfaces, where the excitons can diffuse a short distance typically on the order of tens of nanometers before relaxing.

Despite these limitations, most organic materials have high absorption coefficients of α > 10⁵ cm^-1, so at the absorption peak, a thin film with thickness of only 100 nm can absorb 90% of the light incident on it [18]. This means that the thicknesses of organic layer can be kept very thin but still highly absorptive and meantime the charge transport behavior can be preserved. As mentioned above, the absorption spectra of OSCs are broad and, in addition, can be modified by synthetic chemistry to give light absorption across the visible spectrum and even into the near infrared. A basic knowledge of these features establishes the rationale for the various device architectures, and also accounts for the rapid progress in the field of photocatalysis recently.

3. OSC-based nanostructured photocatalysts

For the past few decades, the unique optoelectronic properties of OSCs have intrigued great exploration on large aggregation of organic molecules, which are held together mainly by noncovalent intermolecular interactions, including π-π stacking, van der Waals forces, hydrogen bonding, coordination interaction, and etc. Accordingly, various methods have been put forward to fabricate organic nanostructures with different morphologies, crystallinities, and optoelectronic properties. Reprecipitation is the most extensively employed strategy to prepare zero-dimensional (0D) organic nanostructures [39]. Self-assembly of organic molecules in liquid phase, or via vapor deposition is a key strategy for fabrication of one-dimensional (1D) structures [40]. Beyond the isolated nanostructures, the scale-up of them into higher-ordered architectures with complexity and hierarchy can also be achieved via liquid phase or vapor phase strategies, or different printing strategies. During the assembly, morphological control can be attained by a fine control over the self-assembly process. It is also noticed that the optoelectronic properties display an obvious morphology dependence in the organic aggregates during these processes.

For instance, in the case of zinc meso-tetra(4-pyridyl)porphyrin (ZnTPyP), the resultant morphologies changed from nanoparticles (NPs) to nanofibers (NFs) just by tuning the surfactant concentration or the aging time during the reprecipitation process [41]. Moreover, the 1D NFs displayed a higher catalytic efficiency in photooxidizing rhodamine B (RhB) than the 0D NPs. Similar results are also gotten from the photocatalytic water splitting test using the three photocatalysts of meso-tetra(4-pyridyl)porphyrin (H₂TPyP) nanooctahedra, ZnTPyP nanowires (NWs), and an intermediate mixture of H₂TPyP nanooctahedra [42]. Among them, ZnTPyP NWs exhibited the highest hydrogen evolution rate of 47.1 mmol g^-1 h^-1. The higher reactivity of 1D nanostructures is the result of the strengthened intermolecular π-π coupling along the long axes. The long axes of π-π stacks have extensive delocalization, hence increasing the light absorption, as well as providing a preferential guide for electron without trapping effects, which enhances the lifetime of the charge carriers.

It is highly recognized that mobility can undergo an increase up to several orders of magnitude when the molecular packing gets improved [43]. As an example, electron mobility in C₆₀ single crystals was tested to be 2.1 cm² V^-1 s^-1, but it fell by at least 3 orders of magnitude by imperfect purification, uncontrolled crystallization, or by oxygen traps [38]. Another study revealed that, impurified pentacene had a mobility of 35 cm² V^-1 s^-1 at room temperature, which was promoted to 58 cm² V^-1 s^-1 at 225 K in the case of pentacene single crystals [44]. With this in mind, growing effort has been devoted to the synthesis of photocatalytic nanocrystals.

For example, the Sn (IV) meso-tetraphenylporphyrin dichloride (SnTPPCl) nanocrystals were prepared via an interfacial self-assembly driven microemulsion process [45]. By adjusting the reaction conditions, different nanocrystals were obtained, such as nanosheets (NSs), octahedra, and microspheres. The hierarchical nanocrystals displayed collective optical properties due to molecular ordering, and hence excellent photocatalytic behavior in platinization and decomposition of methyl orange (MO). Further examples were the use of acid-base neutralization method to give different morphologies of ZnTPyP [46]. With increased surfactant concentration, a series of morphologies were synthesized with controllable size and dimensions, including amorphous NPs, as well as crystalline nanodisks, tetragonal nanorods (NRs), and hexagonal NRs. The crystalline and porous nanodisks showed the best performance. In the two cases, the X-ray diffraction (XRD) patterns of SnTPPCl octahedra were assigned to a tetragonal space group, whereas the ZnTPyP tetragonal NRs indexed as a monoclinic space group. Here the improved mobility was the major reason for the enhanced photoactivity.

The above concepts have been proved to be successful to boost the mobility, but still restricted to some extent. Interfacial nano-heterostructuring is a more feasible strategy to overcome the mobility limitation. The built-in energy level offset within the heterostructure provides a driving force to separate the exciton for ready charge transfer. In addition, a good control of the interfacial properties plays a critical role in achieving higher efficiency. Recently, increasing studies have been reported on the design and preparation of organic nano-heterostructures to enhance the overall photocatalytic efficiency. Typically, there are four types of nano-heterostructured organic photocatalysts, including: organic semiconductor/semiconductor (OSC/SC), organic semiconductor/metal (OSC/M), organic semiconductor/carbon (OSC/C), as well as OSC-based multinary nano-heterostructures. The latest progresses over these heterostructures are covered in detail in the following sections.

4. Types of OSC-based nano-heterostructured photocatalysts

4.1. OSC/SC nano-heterostructures

In OSC/SC heterostructures, the most applicable for photocatalysis is the staggered bandgap type (type II, Fig. 2) [11]. For type II, the CB and VB levels of semiconductor A lie above the corresponding levels of the semiconductor B. when A and B come into contact, the built-in electric field across the interface can direct the electrons and holes to flow in the opposite direction. In this case, the collected electrons in the CB of A will be transferred into the CB of B, while the left holes will migrate from the VB of B to the VB of A upon light excitation, reaching an aim to separate the spatial electron-hole pairs.

For example, we have succeeded in the design of type II organic/inorganic nanojunctions for enhanced photoelectrochemical (PEC) water splitting [47]. The junction was composed of n-type TiO₂ nanotube (NT) arrays and n-type PDI organic overlayer [48]. Anodization was adopted to give TiO₂ NT arrays, which had a pore diameter of about 40 nm, wall thicknesses of about 30 nm, and length up to 4 μm. The PDI overlayer was coated on TiO₂ by physical vapor deposition (PVD). A control on the distance between the deposition and evaporation zones gave rise to PDI overlayer of different thickness (Fig. 3(A)), making it possible to optimize the interfaces and thus to improve its optoelectronic properties for photocatalytic use. After optimization, the charge transfer resistance could be reduced from the original 2738 Ω to 1290 Ω (Fig. 3(B)) fitted from the electrochemical impedance spectroscopy (EIS) results. The charge separation behavior was further monitored by FL emission spectra and the time-correlated single-photon counting (TCSPC) technique. A $\widetilde{9}$0% FL quench was observed for PDI after hybridizing with TiO₂ and its FL lifetime was increased from 0.9 ns to a maximum 4.6 ns after hybridization. The FL measurements were in line with the EIS results, which were responsible for an improved charge separation efficiency. This improvement was well collaborated by the band alignments of PDI and TiO₂ for smooth charge transfer, where TiO₂ had a lower-lying CB edge (E_CB =-3.8 eV) than the LUMO band of PDI (E_LUMO =-4.2 eV). Furthermore, a wide visible-light response was achieved over 400-600 nm as plotted in the incident photon-to-current-conversion efficiency (IPCE) figure (Fig. 3(C)). As a consequence, the junction displayed a photocurrent value as high as 0.74 mA cm^-2 at 1.23 V vs. RHE even when no co-catalyst or sacrificial reagents were added (Fig. 3(D)). This value was superior to that by most reported TiO₂- or organo-based photoanodes.

In another report, TiO₂ was coated on the PDI NFs through a facile in-situ solution deposition to get the 1D nanocomposites [27]. Of this composite, PDI worked as the major photocatalyst that responded to visible light, and TiO₂ functioned as electron transfer relay to influence the overall performance by accepting electron from PDI and further transferring it to the co-catalyst Pt. Electron transfer occurred from the excited PDI to lower-lying TiO₂, and then was trapped by the Pt to participate in the H₂ evolution reaction. Further by side chain modification, 1D π-π stacking between PDI molecules could be adjusted to afford extensive delocalization of the electron along the long axe of the NFs. When irradiated by visible light (λ > 420 nm), hydrogen production was detected for the system by photocatalytic water splitting in the presence of sacrificial reagent methanol or triethanolamine. Besides PDI/TiO₂, PDIs have been enrolled in constructing a series of heterostructured photocatalysts, for instance, PDI/ZnS [49], PDI/CdS [50], and PDI/g-C₃N₄ [51]. These heterostructures benefit much from PDI’s strong visible-light absorption, tunable energy levels by chemical functionalization, as well as matched energy levels. Increasing efforts have been made to tune the electron donor-acceptor structure of PDI molecules and fabricate 1D nanostructures of PDIs, so as to adjust the structural and electronic properties at the interfaces for improved charge transfer efficiency.

g-C₃N₄ is also an n-type OSC, and stands out as a visible-light photocatalyst due to its suitable bandgap ($\widetilde{2}$.7 eV), chemical stability, along with the ease of making thin-layer NSs and porous structures [52]. Many studies have been reported on hybridizing g-C₃N₄ with ISCs to construct the type II heterostructures. These ISCs covered TiO₂ [53], ZnO [54], Fe₂O₃ [55], MoO₃ [56], SnO₂ [57], V₂O₅ [58], CdS [59], ZnS [60], MoS₂ [61], Ta₃N₅ [62], SiC [63], BiVO₄ [64], Bi₂WO₆ [65], Ag₃PO₄ [66], AgBr [67], AgI [68], and etc.

For example, we have synthesized a bimodal macroporous g-C₃N₄/SnO₂ nanohybrid for photooxidation of methyl blue (MB) [57]. The hybrid was facilely synthesized through heat treatment of the homogeneous mixture of thiourea and SnCl₄ that were dried by rotatory evaporation. The involvement of SnCl₄ could not only regulate the pore structure of g-C₃N₄ into NTs to render it highly macroporous, but also yield SnO₂ to couple with g-C₃N₄ for formation of charge transfer interfaces. An improvement on the macropore structures in the g-C₃N₄/SnO₂ hybrids gave rise to an enhanced specific surface area (SSA) of 44.3 m² g^-1 (Fig. 4(A) and (B)) that could keep up with most nano/mesoporous g-C₃N₄ reported. As a result, the abundance of the in-plane macropores and active sites contributed to a higher light absorption, as well as a significantly increased charge separation efficiency, which was confirmed by the large quench of photoluminescence (PL) intensity (Fig. 4(C)) and higher photocurrent generation (Fig. 4(D)) for the hybrid. Mechanistic insight into the higher charge separation was gained from the band structures. SnO₂ has a CB (E_CB = -0.11 eV) and VB (E_VB =3.59 eV) that lie below that of g-C₃N₄ (E_CB = -1.12 eV, E_VB =1.57 eV). The matched band alignment and the large SSA thus provided a great impetus to separate the photogenerated charges. Therefore, the g-C₃N₄/SnO₂ hybrid exhibited a 2.4 times higher photocatalytic activity than pure g-C₃N₄ in photooxidizing MB (Fig. 4(E)), and long-term stability was also ensured for repeated use (Fig. 4(F)).

As covered in most reviews, the g-C₃N₄ was reported to hybridize with ISCs. While, its coupling with OSCs also constitutes an important part in the photocatalytic applications, which is yet not well dealt with in the previous reviews. The structural similarity between polymeric g-C₃N₄ and small-molecular OSCs makes their coupling more versatile, which can either performed through π-π conjugated interaction or through covalent bonds [[69], [70], [71], [72], [73]].

In one study, g-C₃N₄ was coupled with a zinc phthalocyanine derivative (designate as Zn-tri-PcNc) to induce efficient photocatalytic H₂ production under near-infrared light [71]. The strong intramolecular interactions came from the formation of -CO-NH- bonds between the -NH end group of g-C₃N₄ and the COOH groups in Zn-tri-PcNc, in addition to the π-π staking (Fig. 5(A)). Such intimate forces led to an effective charge transfer behavior, which was evidence by a series of tests like photoluminescence (PL) spectra, time-resolved photoluminescence spectra (TRPS), as well as band alignment analysis. The decreased PL emission was a good probe into the occurrence of electron transfer between g-C₃N₄ and Zn-tri-PcNc. In thermodynamic aspect (Fig. 5(B)), the CB level of g-C₃N₄ was -1.12 eV, which was sufficiency positive than the LUMO of Zn-tri-PcNc (-1.40 eV). It indicated that the electron transfer favorably took place from the excited Zn-tri-PcNc to g-C₃N₄. In dynamic aspect, the electron transfer process was detected by TRPS (Fig. 5(C)). The FL lifetime was calculated to be 3.97 ns for single Zn-tri-PcNc, but prolonged to be 6.23 ns after interaction with g-C₃N₄. The longer FL lifetime allowed the photogenerated electrons to be efficiently transferred from Zn-tri-PcNc to g-C₃N₄, and then trapped by the cocatalyst Pt. The cascading electron transfer finally resulted in a largely enhanced and stable visible-light photoactivity (125.2 μmol h^-1) and turnover number (TON, ∼5008 h^-1) for H₂ production after adding proper coadsorbent (Fig. 5(D)).

In another example, tetracarboxyphthalocyanine (ZnTcPc) was covalently bonded to g-C₃N₄ to exhibit excellent visible-light photooxidation of phenols [72]. The efficient electron transfer enabled by their matched energy structure was responsible for the generation of three active species like ·O₂^-, hole, and ·OH, which were separately detected after the addition of corresponding scavengers BQ, KI and IPA. These active species could thus participate in the oxidation of phenos to give the products such as benzoic acid, hydroxybenzoic acid, and phthalic acid.

However, the type II heterostructure is not very competent to impede the ultrafast electron-hole recombination. In order to guarantee an efficient charge separation, p/n junction is often proposed. The synergy between the internal electric field and the band alignment make p/n junction more effective in prompting charge separation, transfer and prolonging the lifetime of the charge carriers. Particularly, a photocatalysis system composed of OSCs and a p/n bilayer differs apparently from conventional systems of ISCs in view of the response range in visible-light region.

Abe’s group have conducted great studies on the construction of organic p/n bilayers for PEC water splitting and pollutant removal [[74], [75], [76], [77], [78], [79], [80]]. In particular, significant efforts have been devoted to designing p/n bilayers responsive to wide or full visible-light energy, or even near-infrared energy. In one work, an organic p/n bilayer consisting of p-type zinc phthalocyanine (ZnPc) and n-type fullerene (C₆₀) was fabricated by two-step vapor deposition, and used to reduce H⁺ into H₂ [79]. The PEC measurement revealed an efficient evolution of H₂ from ZnPc/C₆₀ bilayer, and the action spectra for photocurrent measured at ITO/ZnPc (75 nm)/C₆₀ (125 nm)-Pt further showed that H₂ evolution took place across the full visible-light region of <750 nm (Fig. 6(A)). From the action spectral characteristics, it could be seen H₂ evolution was induced by absorption of both ZnPc and C₆₀ for ITO/ZnPc/C₆₀-Pt electrode (Fig. 6(B)). Since a photovoltage of about 535 mV was present at the ZnPc/C₆₀ interface, it was enough to dissociate the exciton formed within the bilayer into free electrons and holes at the p/n interface (Fig. 6(C)). Thus, the p/n bilayer was able to reduce H⁺ into H₂ at the C₆₀ side.

Like the case of ZnPc/C₆₀, the full visible-light response was also found in the catalyst system composed of 29H,31H-phthalocyanine (H₂Pc, p-type) and C₆₀, or of H₂Pc (p-type) and 3,4,9,10-perylenetetracarboxylic bisbenzimidazole (PTCBI, n-type). H₂Pc/C₆₀ bilayer with different thickness was studied for H₂ and O₂ evolution. Likewise, the action spectra for photocurrent showed that H₂ evolution arose from the absorption of both H₂Pc and C₆₀, over the entire visible-light region of < 750 nm (Fig. 7(A)). While, the whiskered H₂Pc/PTCBI bilayer showed an enhanced photoanodic output, because the formation of H₂Pc whisker led to a rough and enlarged p/n interface (Fig. 7(B)) [81]. The H₂Pc whisker was formed in a vertical direction towards the surface, to give an intensified absorption. While in the case of thick H₂Pc whisker, the oxidation kinetics at the H₂Pc/water interface was higher by near 3 orders of magnitude than that with planar H₂Pc. The energy diagram was illustrated in Fig. 7(C) to show the electron transfer process. Upon excitation, the photogenerated electrons flew from p-type H₂Pc to n-type PTCBI, and the holes moved from PTCBI to H₂Pc then to the H₂Pc/water interface. The sacrificial thiols then rapidly were oxidized by these holes to make electron-hole effectively separated. When H₂Pc/PTCBI bilayer was deposited to flexible substrate like Nafion [82], it featured an all-organic photocatalyst, and worked efficiently for the degradation of trimethylamine, taking a value of ca. 40% of external quantum efficiency for CO₂ formation.

In another example, the organic p/n bilayer composed of PTCBI and p-type lead phthalocyanine (PbPc) was demonstrated to be responsive to widespread photoenergy in the visible-near infrared regions of 400-1100 nm. Such wide response came from the absorption of both PTCBI and PbPc [83]. In addition, a photovoltage of ca. 400 mV accounted for the occurrence of charge separation at the PTCBI/PbPc interface, considering that the potentials of PTCBI and PbPc were 0.05 V and +0.34 V, respectively. The wide photoresponse and the photovoltage thus induced a series of photophysical reactions where Fe^II(CN)₆^4- was oxidized and H⁺ was reduced into H₂.

In our work, PVD method was adopted to give 1D organic single-crystal p/n nano-heterostructure, comprising of p-type H₂TPP and n-type N,N-(dicyclohexyl) perylene-3,4,9,10-tetracarboxylic diimide (CH-PTCDI) [84]. The advantages of this p/n nano-heterostructure could be supported by the following aspects: a more effective charge separation driven by the large donor-acceptor interface; directed charge transport along the 1D long-range axes; increased mobility by crystallization; and wide and strong absorption by both H₂TPP and CH-PTCDI. SEM (Fig. 8(A)) and TEM images (Fig. 8(B)) revealed the double-layer morphology where H₂TPP layer had a thickness of ca. 500 nm, and the thickness of CH-PTCDI layer was ca. 200 nm. The marked fluorescence (FL) quench of CH-PTCDI well proved the efficient electron transfer between CH-PTCDI and H₂TPP when they were combined (Fig. 8(C)). This was thermodynamically supported by their matched band structures, inducing the electron to flow from the excited H₂TPP to CH-PTCDI, and the holes vice verse (Fig. 8(D)). The properties were responsible for the final enhancement on the photoactivity in degrading MB (Fig. 8(E)). The stability was also ensured for successive uses (Fig. 8(F)).

4.2. OSC/M nano-heterostructures

An alternative method has also been proposed by designing OSC/M heterostructures to construct heterogeneous interfaces. When they come into contact, the potential difference will be created due to their different work functions, which drives the exciton to dissociate into free charge carriers. The potential difference here is called Schottky barrier. Moreover, the metallic NPs are also appealing for their unique catalytic, optical and electric characteristics, which can coordinate with OSCs to improve overall photocatalytic performance. The catalytic properties of some noble metals (e.g., Pt, Ru) make them an efficient cocatalyst. The surface plasmon resonance (SPR) effect is an important optical characteristic of some metallic NPs (e.g., Au, Ag), which contribute to a large enhancement of light absorption.

The low overpotential for water reduction makes Pt the most efficient cocatalyst for OSCs, and the Schottky barrier between them can promote the electron transfer from OSCs to Pt. Many reports have appeared dealing with Pt as cocatalyst to help enhance the photocatalytic H₂ evolution. Yu’s group made great use of Pt NPs as cocatalysts to enhance the photoactivity of g-C₃N₄ in CO₂ reduction and H₂ evolution [85,86]. He also pointed out that the shape of Pt NPs had an important impact on the catalytic activity and selectivity. As in one work, different shapes of Pt NPs were precisely synthesized to have different exposed facets, and were then ex situ anchored onto the surface of g-C₃N₄ (Fig. 9(A)-(C)) [86]. The photocatalytic H₂ evolution experiments were conducted to test its visible-light response, in which 10 vol% TEOA was added as the sacrificial agent. The photoactivity of the obtained g-C₃N₄-Pt hybrids was compared, following an increasing order of g-C₃N₄-cubic Pt<g-C₃N₄-octahedral Pt<g-C₃N₄-spherical Pt (Fig. 9(D)). As the light absorption and FL emission didn’t show much difference by changing the shape of Pt cocatalysts, the reason for the catalytic difference could be explained by different crystallographic facets. It is thought that defects are more likely to occur at some places like edges, steps or corners, and the atoms at these defect sites are more thermodynamically active to initiate physicochemical reactions. In another word, the catalytic reactivity is largely determined by the atoms at these defect sites. For instance, the (100) surface has a square atomic arrangement, which differs from (111) surface that has a hexagonal atomic arrangement. Such difference greatly affects the number of surface atoms at the defect sites, which is thereby responsible for the catalytic difference.

Another interesting example can be seen with porphyrin nanostructures, as their self-metallization ability provides a facile means of preparing porphyrin-metal heterostructures. This is simply done by exposing the mixed solutions of porphyrin nanostructures, a proper metal complex along with a sacrificial electron donor to visible light irradiation. On this basis, Shelnutt’s group has succeeded in obtaining a series of self-platinized porphyrin nanostructures, including NPs [87], NTs [88], NSs [89] and NFs [90], four-leaf clover-like [91]. As a matter of fact, these self-platinized porphyrin nanostructures were demonstrated to be promising visible-light photocatalysts for hydrogen evolution.

When the noble metals are downsized to clusters or even to single atoms, we can get a maximum utilization of these atoms for an enhanced photocatalytic reactivity. But the uniform dispersion of single-atom catalysts is often a difficult task because these atoms suffer from serious agglomeration with a loss of efficiency during repeated reactions. In this regard, Li et al reported the use of isolated single Pt atoms as co-catalysts to couple with g-C₃N₄ (Pt-CN) [92]. The reaction between the tri-s-triazine units in g-C₃N₄ and H₂PtCl₆ resulted in a rich formation of Pt-N/C bonds, making it possible to synthesize single-atom Pt-CN system under relatively low temperature. The single Pt atoms had a uniform distribution and configuration on g-C₃N₄. g-C₃N₄ with 0.38 wt% of Pt clusters exhibited a moderate photoactivity in H₂ generation, g-C₃N₄ with Pt nanoparticles (Pt-NPs-CN) gave the worst activity. By comparison, the turnover frequency of Pt-CN (0.16 wt% Pt loading) reached 775 h^-1, approximately 9 times higher than that of g-C₃N₄ with Pt NPs (Fig. 10(A)). The remarkable enhancement could be ascribed to the presence of the isolated single Pt atoms that caused a change on the surface trap states of g-C₃N₄ (Fig. 10(B)). This was elucidated by the ultrafast transient absorption (TA) spectroscopy (Fig. 10(C)), a technique to track the the real-time photoexcited carrier dynamics. The average recovery lifetime for Pt-CN was determined to be ca. 433 ps, which was twice longer than the ca. 117 ps for Pt-NPs-CN. The lengthened lifetime of the charge carriers hence revealed the modulated electronic structure of g-C₃N₄ due to the involvement of the single Pt atoms. Yu’s group also gained a deep insight into the mechanisms of single-atom Pt in the performance enhancement through studying the single-atom Pt-embedded g-C₃N₄ system.

Besides the function as an electron relay, the novel metallic NPs could absorb light to generate a strong electric field and a high concentration of hot electrons the NP surface (called SPR effect). The electrons harvest light energy and then flow to the CB of the semiconductors to fulfill the electron transfer. Thus, these plasmonic metallic NPs have been widely anchored onto the ISC or OSC support to enhance the sunlight utilization and charge separation efficiency. Plamonic Au NPs are especially used to construct plasmonic photocatalysts due to its strong visible-light absorption. For example, a calcination-photodeposition technique was used to construct plasmonic photocatalysts of Au/Pt/g-C₃N₄ for antibiotic degradation [93]. TEM images revealed that the spherical Au and Pt NPs with a size of 7-15 nm were uniformly formed on the surfaces of g-C₃N₄ (Fig. 11(A)). It was seen from the UV-vis diffuse reflectance spectra that, g-C₃N₄ alone displayed an absorption edge shorter than 460 nm, whereas a broadened absorption peak was observed for both the Au/g-C₃N₄ and Au/Pt/g-C₃N₄ samples at around 550 nm arising from SPR effect of Au NPs (Fig. 11(B)). When tested for photodegrading tetracycline hydrochloride (TC-HCl), a representative antibiotic, the catalytic rate followed an order of Au/Pt/g-C₃N₄ (93.0%) > Au/g-C₃N₄ (78.6%) > Pt/g-C₃N₄ (67.2%) > g-C₃N₄ (52.8%) after the same irradiation time (Fig. 11(C)). Here, the performance enhancement could be attributed to the electronic and catalytic benefits from the metallic NPs, where the SPR effect of Au resulted in an extended responsive photoenergy range, and the electron-sink function of Pt gave rise to a promoted charge separation efficiency (Fig. 11(D)). Apart from Au, Ag and some other non-nobel metals like Bi are also capable of displaying SPR effect in the visible-light region, which have been applied in SPR-enhanced photocatalysis.

4.3. OSC/C nano-heterostructures

Hybridization of semiconductors with carbon materials is another way to enhance the photoenergy conversion. Among the candidates, carbon nanofibers (CNFs), carbon nanotubes (CNTs), and graphene are the most widely employed supports owing to their pleasant physical and electrical properties. The advantages of the carbon materials lie in the following aspects: i) the good conductivity renders them good acceptors and transfer/transport pathways, which leads to increased charge separation and transport efficiency; ii) the high SSA enable plentiful of active sites for the adsorption of the reactive species; iii) the similar π-conjugated and flexible structure between these carbon supports and OSCs afford them more compatible and diverse in coupling.

Graphene sheets have been highly recognized as an ideal two-dimensional (2D) scaffold due to its extremely high theoretical SSA ($\widetilde{}2$600 m² g^-1), promising charge transfer properties and tunable band-gap energy [94]. In some cases, porphyrin nanoassemblies have been included onto graphene scaffold for photocatalytic applications. In one case, Rajesh et al have fabricated a composite made of 5, 10, 15, and 20-tetrakis (4-carboxyphenyl) porphyrin (TCPP) NRs and reduced graphene oxides (rGO) to facilitated the spatial charge separation [95]. The strong coupling between rGO and TCPP NRs came from the electrostatic interaction between the positively charged NRs and negatively charged surface of the rGO. An electron transfer process was found for the composites and the electron transfer rate was estimated to be 10.0 × 10^-4 ps^-1. As a result, 1.9 fold enhancement of photocurrent was obtained for the composites under visible light illumination compared with that of pure TCPP NRs. Duong et al also reported the facile preparation of TCPP NRs well-dispersed on the surface of graphene nanoplates [96]. The resulted composites were also demonstrated to be good visible-light photocatalysts in degrading RhB. Another report by Guo et al has studied the inclusion of 1D porphyrin nanoassemblies on graphene oxide (GO) as photocatalysts, and tested its visible-light response [97].

We have integrated meso-tetra(p-hydroxyphenyl) porphyrin (p-THPP) NPs into rGO to get a monolithic p-THPP/rGO nanohybrid film (Fig. 12(A)) [98]. Because of coupling, the average FL lifetime of p-THPP was prolonged from ca. 362 to 473 ps (Fig. 12(B)), and meantime the charge transfer resistance was decreased from 176.2 to 46.7 Ω (Fig. 12(C)). These results indicated a greatly increased electron transfer in the hybrids, which was the main reason for the performance improvements (Fig. 12(D)). It was noteworthy that, the free-standing feature made the repeated use of p-THPP/rGO film much easier, and only a slight decrease was observed in the recycling experiment after seven repeated runs.

The similar 2D network and π-conjugated structure between graphene and g-C₃N₄ have inspired the great work on their coupling into 2D/2D hybrid heterojunction [99]. Liao et al have employed sonochemical approach to modify g-C₃N₄ with GO overlayer, to give efficient photocatalytic capability under visible light irradiation. Ong et al have made use of rGO and protonated g-C₃N₄ (pCN) to construct 2D/2D heterojunction photocatalyst with effective interfacial contact [100]. The synthesis was done by combing ultrasonic dispersion, electrostatic self-assembly with a subsequent NaBH₄-reduction process. Another example was provided by the formation of the 3D porous g-C₃N₄/GO aerogel (CNGA). The aerogel was synthesized via hydrothermal coassembly of g-C₃N₄ and GO NTs. The highly interconnected porous network of CNGA afforded plenty of pathways for rapid mass transport, large reactant adsorption as well as multireflection of incident light (Fig. 13(A)). In addition, the large interfacial contact between g-C₃N₄ and GO NTs enabled a highly promoted electron transfer (Fig. 13(B) and (C)). The resulted CNGA aerogel thus exhibited excellent photocatalytic activities for MO removal, as well as for the CO₂ reduction (Fig. 13(D)) under visible-light irradiation.

1D CNFs are also a good candidate as electron transfer/transport channels. Electrospun CNFs are appealing for their monolithic feature, good conductivity (ρ=(3-7) × 10^-3 Ω cm), as well as the high SSA due to its richly porous structure. Many studies have revealed that CNFs are able to capture and transport the photogenerated electrons along the 1D CNFs [[101], [102], [103], [104]]. In one study, CNF/g-C₃N₄ composite photocatalysts were synthesized via a two-step approach including electrospinning and a followed calcination process [105]. HRTEM image (Fig. 14(A)) indicated the existence of g-C₃N₄ along the surface of CNFs. The interfacial interaction between CNF and g-C₃N₄ was further evidenced by the strong PL quench (Fig. 14(B)). The involvement of CNFs rendered the composites a SSA of 35.9 m² g^-1 and pore volume of 0.17 cm³ g^-1, which were much higher than those of pure g-C₃N₄ (11.9 m² g^-1 and 0.1 cm³ g^-1). Because of the synergic effects of increased SSA, pore volume and charge separation efficiency (Fig. 14(C)), the CNF/g-C₃N₄ composites had a photocatalytic hydrogen production rate as higher as 1080 mol h^-1 g^-1 under visible light irradiation (≥420 nm), which was about 4.6 times higher than that of pure g-C₃N₄ (Fig. 14(D)).

Amorphous carbon and carbon spheres can also work as good cocatalysts to improve the photocatalytic activity. Examples can be seen with the amorphous carbon/g-C₃N₄ nanocomposites, and g-C₃N₄-encapsulating carbon spheres [106]. Likewise, the introduction of them can contribute greatly to higher visible-light absorption and larger SSA, hence benefiting the photocatalytic performance.

4.4. OSC-based multinary heterostructures

Core to photocatalysis is the multistep electron transfer in the photocatalysts. A well-organized multistep electron transfer system can allow for efficient photoenergy conversion. To this end, various attempts have been made in constructing ternary or multinary heterostructures on the basis of visible-light absorption, energy band match, as well as the more decisive electron transfer cascade [65,103,107].

For example, Hayashi et al have developed a multistep electron transfer system made of ternary composites (porphyrin, ZnO NPs, and rGO) [108]. The photocurrent density by this hierarchical electron transfer system was remarkably enhanced with an IPCE value as high as 70%. The difference between the LUMO of TCPP to the CB of ZnO provided an impetus for charge separation, and moreover rGO worked as an electron mediator that could accept electrons from ZnO or porphyrin moiety, hence resulting in higher photocurrent generation. Following the same strategy, Duong et al have reported the fabrication of a ternary nanohybrid using graphene/TiO₂/TCPP, and it was tested to be an efficient and stable catalyst for photooxidation of RhB under solar light [109]. The SEM (Fig. 15(A)) and TEM (Fig. 15(B)) images showed that TCPP NRs and TiO₂ NPs were successfully formed on the graphene surfaces. In this system, TCPP NRs absorbed visible light, and TiO₂ NPs could make use of UV light to generate electron-hole pairs (Fig. 15(C)). The wide light utilization and the multistep electron transfer processes finally made this ternary system highly efficient and durable in dye photodegradation (Fig. 15(D)). Further examples are the CdTeSe-porphyrin-rGO hybrid system, to give 240-fold increase on photocurrent intensity under visible light irradiation. In this case, electrons were generated by the major photoabsorber porphyrin, and then were transferred from the LUMO of porphyrin to the CB of adjacent CdTeSe NRs due to the potential difference, and finally flowed to RGO for a cascading electron transfer.

We have extended the cascading electron transfer system to the free-standing organic/inorganic ternary nanohybrids (CNF, ZnO NRs, porphyrin) [103,104,110]. ZnO has a low-lying CB edge (E_CB = -4.2 eV) relative to the LUMO band of TCPP (E_LUMO = -3.6 eV), that all lie above the Fermi level of CNFs ($\widetilde{-}$4.50 eV). Cascading band positions are thereby established across them when coupling. Under light illumination, excitons are generated in TCPP, and the potential gradient across the three components provides the driving force to separate the excitons and transfer the electrons from TCPP to ZnO and then to CNFs in a cascading way. The electron transfer cascade system was responsible for an enhanced photocurrent generation, and also worked more efficiently in photodegrading RhB.

Spatial integration of two or more light absorber and an electron relay into a multinary heterostructure is also a strategy to enhance solar energy conversion, since a collective light absorption from them can give rise to a broadened responsive light range. Li et al have demonstrated a facile synthesis of organic/inorganic ternary nanohybrids made of tin(IV) prophyrin (SnPor), Ag NPs, and rGO (Fig. 16(A)) [111]. The UV/Vis absorption spectra of the ternary hybrid combined the Soret band of SnPor, the absorption band of rGO, and the plasmon resonance band of Ag NPs (Fig. 16(B)). The nanohybrids exhibited high photocurrent generation (Fig. 16(C)) and good catalytic activity in the degradation of RhB (Fig. 16(D)), as a result of the photosensitization effect of SnPor and the SPR excitation in the AgNPs. A similar methodology has been employed to fabricate ternary nanocomposites of Sn-porphyrin (SnP), angular Au NPs and rGO [112]. In this system, SnP and Au NPs served as photoactive, visible-light absorbers, and the synergistic effect of SPR excitation in the AuNPs and SnP led to a 27-fold enhancement in the photocurrent density when irradiated at 550 nm.

5. Summary and perspective

To sum up, this paper has outlined the recent advances on the organic nano-heterostructured photocatalysts. OSCs unit the electronic, chemical and mechanical benefits with a material structure that are easily tuned via molecular design. Assembly of them into nanostructures can lead to an improved photostability and light utilization. But the intrinsic low charge mobility still limits the overall photoconversion efficiency. Interfacial heterostructuring seems a logical step forward by prompting the charge transfer/transport process, in particular when the ease of molecular functionalization is concerned, making it convenient to adjust the interfacial or surfacial properties. This paper discussed the general strategies in making organic nano-heterostructures for developing highly efficient and stable photocatalysts. They are mainly divided into four types of OSC/SC, OSC/M, OSC/C, and OSC-based multinary nano-heterostructures, in different mechanisms to enhance the light utilization, charge transfer, and mass transport.

Despite these efforts, there are several challenges to be tackled in future. First, compared with the inorganic nano-heterostructured photocatalysts, the generation of free charges is still limited, as the size of organic nanostructures is usually larger than the inorganic counterparts so that the generated charges tend to recombine before they reach the interface. Meanwhile, the organic nanostructures are organized mostly via π-π stacking, and the weak interactions are not favorable for perfect molecular packing. So, it is commonly hard to get a good crystallization of organic nanostructures, which is not advantageous for the electron transport either. Second, the interface geometry and properties need to be clarified, and an in-depth insight into them will help to understand and determine the roles of the different components. Moreover, a quantitative prediction of the charge behaviors requires more rational calculations and simulations, especially in light of the electronic properties, excitation dynamics, and charge transport of the heterostructures. Even more importantly, the question whether the heterostructured materials are useful in the devices requires more exploration of the spatial geometry. After these issues are well addressed, the organic nano-heterostructured photocatalysts will show more promise in photocatalytic pollutant removal or water splitting by use of a wide or even full spectrum of sunlight, and they will show more feasibility in the device fabrication and environment adaptability, either in portable or deformable shapes.

Declaration of Competing Interest

The authors declare that they have no conflict of interest.

Acknowledgements

This research was financially supported by the National Key Research and Development Program of China (2016YFB0700300), the National Natural Science Foundation of China (Grant No. 51503014 and No. 51501008) and financial support from Fundamental Research Funds for the Central Universities (No. 230201818-002A3)

---