Next-Generation Multifunctional Carbon–Metal Nanohybrids for Energy and Environmental Applications

Abstract

Nanotechnology has unprecedentedly revolutionized human societies over the past decades and will continue to advance our broad societal goals in the coming decades. The research, development, and particularly the application of engineered nanomaterials have shifted the focus from “less efficient” single-component nanomaterials toward “superior-performance”, next-generation multifunctional nanohybrids. Carbon nanomaterials (e.g., carbon nanotubes, graphene family nanomaterials, carbon dots, and graphitic carbon nitride) and metal/metal oxide nanoparticles (e.g., Ag, Au, CdS, Cu2O, MoS2, TiO2, and ZnO) combinations are the most commonly pursued nanohybrids (carbon–metal nanohybrids; CMNHs), which exhibit appealing properties and promising multifunctionalities for addressing multiple complex challenges faced by humanity at the critical energy–water–environment (EWE) nexus. In this frontier review, we first highlight the altered and newly emerging properties (e.g., electronic and optical attributes, particle size, shape, morphology, crystallinity, dimensionality, carbon/metal ratio, and hybridization mode) of CMNHs that are distinct from those of their parent component materials. We then illustrate how these important newly emerging properties and functions of CMNHs direct their performances at the EWE nexus including energy harvesting (e.g., H2O splitting and CO2 conversion), water treatment (e.g., contaminant removal and membrane technology), and environmental sensing and in situ nanoremediation. This review concludes with identifications of critical knowledge gaps and future research directions for maximizing the benefits of next-generation multifunctional CMNHs at the EWE nexus and beyond.

Graphical Abstract

1. Introduction

To meet growing energy demand, mitigate water scarcity, and address global environmental pollution at the important energy–water–environment (EWE) nexus,(1,2) there is an ever-increasing need to engineer next-generation “superfunctional materials” that possess enhanced and/or fundamentally new properties and exhibit multifunctionalities.(3,4) For example, the inherent optical and electronic properties of conventional single-component materials (e.g., TiO₂) may be insufficient to achieve or sustain adequate catalytic efficiency for energy harvesting (e.g., photocatalytic H₂O splitting for H₂ and O₂ evolution) and contaminant (e.g., recalcitrant perfluorochemicals) removal due to very limited use of solar energy and rapid recombination of photogenerated electron–hole pairs during catalytic processes.(5) Hence, there is a strong need to retain or enhance the catalytic capability, and in the meantime invoke new (e.g., sunlight harvesting, contaminant adsorption and redox, biocidal, and antifouling) properties through assemblage of guest material(s) to parent matrices, fabricating multicomponent nanohybrids.(5,6) To deliver such advantages sustainably, the multifunctional nanohybrids need to maintain environmentally benign attributes.(7)

Carbon-based nanomaterials (CNMs) have always been in the frontier of materials science including carbon nanotubes (CNTs),(8) graphene family nanomaterials (GFNs: graphene, graphene oxide; GO, and reduced graphene oxide; RGO),(9) and most-recently, emerging materials namely carbon dots (CDs)(10) and graphitic carbon nitride (g-C₃N₄).(5) CNMs are chemically stable and structurally diverse with prominent light-absorptive and electron transport properties, and have appealing catalytic, redox, fluorescence, and luminescence attributes. Despite these advantages, CNMs are not as efficient in delivering some of the key functions as compared to those delivered by metal/metal oxide nanoparticles (MNPs like TiO₂ and ZnO semiconductors) such as wide bandgap, ability to maintain high electron–hole pairs separation and transfer efficiency, exceptional heat transfer and electron transport properties, and capability to donate metal ions (e.g., Ag⁺) for biocidal applications.(11,12) Hence, rationally designing nanohybrids with at least two dissimilar NMs (such as CNMs and MNPs) with diverse properties and complementary functionalities holds a great promise for addressing issues and challenges faced by humanity at the EWE nexus.(2–6,13)

Multiple benefits for EWE applications can be harnessed through hierarchical assemblage of CNMs and MNPs. First, during synthesis of carbon–metal nanohybrids (CMNHs), CNMs can be used to precisely engineer the properties (e.g., size, shape, morphology, crystallinity, and dimensionality) of MNPs that are relevant to the light-absorptive and contaminant-adsorptive/(photo)catalytic functions by controlling particle nucleation and growth.(14–18) Particularly, CNMs (1D CNTs and 2D g-C₃N₄ or GFNs nanosheets) can distribute and effectively stabilize the anchored MNPs, and thus can result in reduced aggregation, photocorrosion, leaching, and/or surface passivation of the composite material (Figure 1a,b). These advantages are attributed to the uniqueness of CNMs featuring thermal and chemical stability, large specific surface area (SSA), abundant surface-active sites and defects, and rich oxygen-containing functional groups.(19–22) The hybridized MNPs, in turn, also facilitate achieving a high degree of dispersion for the CNMs (2D layered g-C₃N₄ and GFNs) through enhanced physical segregation, constructing few-layered CNMs with large SSA and abundant surface-reactive sites.(23) Furthermore, CNMs can preconcentrate contaminants(24,25) or lower contaminant reaction potentials,(26) and thus facilitate the subsequent catalytic/redox reactions at the hybridized carbon–metal interfaces (CMIs; Figure 1a). Also, the range of light and electromagnetic absorption of MNPs can be extended due to the inherent ultraviolet–visible-near-infrared (UV–vis-NIR) light responses of CNMs (g-C₃N₄ and CDs).(5,27–30) Most strikingly, the charge, electron, heat, and mass transfer and separation within the precisely assembled nanoheterostructures are optimized due to enlarged interfacial contact areas, intimate interfacial interactions, altered electronic properties (bandgap), formation of new local charge centers, and creation of internal electric fields (Figure 1a; detailed mechanisms are illustrated in Section 2.1 below), all of which can enhance the catalytic/redox performances of CMNHs toward more efficient energy harvesting, water treatment, and environmental sensing and remediation (Figure 1c).(5,6) For certain CNMs (g-C₃N₄ and CDs) and MNPs (ZnO(31) and Ag₂SO₄(32)), their redox potentials are magnified after hybridization, further facilitating the degradation of recalcitrant contaminants (e.g., perfluorochemicals). Additionally, some MNPs (e.g., Ag, Au, Bi, and Cu) demonstrate localized surface plasmon resonance (LSPR), which can be improved when hybridized with CNMs (g-C₃N₄ and CDs), and such attributes may further facilitate simultaneous and rapid catalytic degradation of multiple contaminants.(5,6)

Figure 1.

Altered and emerging attributes of carbon–metal nanohybrids (CMNHs) make them the next-generation promising materials for addressing multiple issues and challenges at the important energy–water–environment (EWE) nexus. (a) Schematic diagram illustrates the electronic properties of CMNHs, highlighting the formation of the well-contacted, intimate carbon–metal interface (CMI). For metal nanoparticles (MNPs), the generation of photoexcited electron–hole pairs upon light irradiation is responsible for the photocatalytic/redox degradation of contaminants. However, the rapid recombination of photogenerated electron–hole pairs significantly weakens their photocatalytic/redox performances. In comparison, for CMNHs, the photoexcited electrons can transfer from the conduction band (CB) energy level (E_C) of MNPs to the Fermi energy level (E_F) of carbonaceous nanomaterials (CNMs; e.g., graphitic carbon nitride; g-C₃N₄ or carbon dots; CDs), producing the unique attributes of new built-in electric field, local electromagnetic field, and fluorescence resonance energy transfer at the CMIs of the hybridized nanoheterojunctions. These newly emerging attributes at the CMIs can accelerate spatial separation and migration of photogenerated electron–hole pairs for enhanced photocatalytic/redox performances. E_V refers to the valence band (VB) energy level of MNPs. Replotted from Li and Antonietti.(20) Copyright: 2013 Royal Society of Chemistry. (b) TEM image of CMNHs (e.g., g-C₃N₄- or RGO–MNP1–MNP2), and their uniqueness and multifunctionalities. GFNs denotes graphene family nanomaterials. MNP1 and MNP2 refer to two different MNPs. UV–vis-NIR denotes ultraviolet–visible-near-infrared. (c) The uniqueness and multifunctionalities of CMNHs render them exciting candidates for energy, water, and environmental applications including H₂O splitting for H₂ and O₂ production, CO₂ conversion for energy storage, contaminant removal, microbial disinfection, membrane technology, environmental sensing, and in situ nanoremediation.

This frontier review presents a comprehensive summary of research efforts on CMNHs (particularly for the most appealing g-C₃N₄- and CDs-based CMNHs) that can be positioned for addressing multiple challenges at the EWE nexus. The altered and newly emerging attributes of CMNHs introduced through material hybridization are highlighted first. It is followed by a detailed discussion on opportunities for applications relevant to energy, water, and the environment. Example applications include H₂O splitting and CO₂ conversion, contaminant removal and membrane technology, and environmental sensing and in situ nanoremediation. Critical knowledge gaps and challenges in these fields are also systematically elucidated to identify future research strategies for this exceptional class of nanohybrids at the EWE nexus and beyond.

2. Altered and Newly-Emerging Attributes of CMNHs

2.1. Electronic, Optical, Band, and Interfacial Charge Transfer Properties

Hybridizing CNMs and MNPs tailors the electronic and optical properties of CMNHs including electronic energy levels (bandgap structure), charge-carrier density and lifetime, UV–vis-NIR-light absorption, and nonradiative paths,(33,34) all of which are strongly related to the intimately contacted carbon–metal interfaces (CMIs; Figure 1a).(5,6,17) Taking g-C₃N₄ as an example: assembling g-C₃N₄ and MNPs with well-matched band and electronic structures(35) produces new electronic structures; that is, band bending is generated at the CMIs and creates a new built-in electric field within a charge region for accelerated spatial separation and migration of photogenerated electron–hole pairs (Figure 1a).(5) Through first-principles calculations with charge density difference and Mulliken population identifications, Ma et al.(36) demonstrated the formation of new built-in electric field at the type-II band alignment g-C₃N₄–BiPO₄ interface. This new built-in electric field also frequently appears in other g-C₃N₄ nanohybrids decorated with noble metals (Ag, Au, Bi, Pd, and Pt), transition metals (Ce, Cu, Fe, and Ni), and transition metal compounds (metal hydroxides and sulfides),(5,6,37,38) as validated with density functional theory (DFT) and total density of states calculations.(39) Also, the light absorption efficiency at the UV–vis-NIR range is enhanced particularly for g-C₃N₄- and CDs–CMNHs that include Ag, Au, Cu, and Bi in their nanoheterostructures, because these MNPs act as electron reservoirs and plasmonic cocatalysts mediated by LSPR.(5,40) As evidenced by finite integration simulation technique,(37) the local electromagnetic field (Figure 1a) produced by LSPR is another newly emerging attribute of CMNHs. More importantly, the new built-in electric field and the local electromagnetic field are closely intertwined, since the former modifies the latter by tuning the spin-polarized band structure and the Fermi level of CMNHss.(41) Furthermore, when the LSPR absorption of MNPs is partially overlapped with the optical absorption of CNMs (g-C₃N₄), the LSPR triggers plasmon resonance energy transfer (PRET; Figure 2) and excites charge-carriers at the CMIs (Figure 1a).(42,43) Both LSRP and PRET effects play positive roles in boosting electron and charge separation, decreasing charge-carriers density, and prolonging lifetime of charge separation for enhanced catalytic performance.(5) Finally, two or more of the new built-in electric field domains and charge-carriers are created for ternary g-C₃N₄-bimetallic systems (e.g., g-C₃N₄–Ag–Ag₃PO₄; Figure 3); and as such, the catalytic efficiency, stability, selectivity, and durability of the ternary systems are maximized.(5) The above alterations also frequently emerge for CDs–, GFNs-, and CNTs–CMNHs.

Figure 2.

(a–h) Transmission electron microscope (TEM) images of Ag@SiO₂ (core@shell) nanostructures in which the thickness of the SiO₂ shell corresponds to 0 (a; no SiO₂ capping), 8 nm (b), 12 nm (c and h; 12 nm nanogap), 17 nm (d), and 21 nm (e). (f–g) TEM images of pristine g-C₃N₄ and Ag core NPs, respectively. (i–n) Finite difference time domain (FDTD) simulations of the field enhancements for AgNPs (i), Ag@SiO₂-8 nm (j), Ag@SiO₂-12 nm (k), Ag@SiO₂-17 nm (l), Ag@SiO₂-21 nm (m), and hollow SiO₂ sphere with 12 nm shell thickness (n) using a 3D FDTD solver (openEMS). (o) Schematic diagram illustrates the photocatalytic H₂ evolution by the g-C₃N₄–Ag@SiO₂ heterojunction in which the plasmon resonance energy transfer (PRET) effect and Förster resonance energy transfer (FRET) effect gradually become weakened with the widening of the nanogap shown in (a–e), as well as the photocatalytic H₂ evolution rate for each g-C₃N₄–Ag@SiO₂. The optimized photocatalytic H₂ evolution rate (11.4 μmol/h) is achieved at the nanogap of 12 nm in which the PRET and the FRET are perfectly balanced. TEOA refers to triethanolamine. Replotted from Chen et al.(94) Copyright: 2015 Wiley.

Figure 3.

(a) Formation rate of different products during photocatalytic CO₂ reduction by the direct Z-scheme g-C₃N₄–SnO_2–x composite under simulated sunlight irradiation. The m wt % SC refers to g-C₃N₄–SnO_2–x with different SnO_2–x loading capacities. P25 refers to Degussa TiO₂. (b) Reactions and mechanisms of photogenerated electron–hole pairs separation and transfer at the visible-light-driven Z-scheme g-C₃N₄–SnO_2–x CMI. The photogenerated electrons in the CB of SnO_2–x are injected into the VB and annihilate the holes of g-C₃N₄ via CMI, which facilitates the electron–hole pair separation and suppresses the charge recombination. Replotted from He et al.(114) Copyright: 2015 Elsevier. (c) Formation rate of different products during photocatalytic CO₂ conversion by the Z-scheme g-C₃N₄–Ag–Ag₃PO₄ heterojunction under simulated sunlight irradiation. The m AC refers to g-C₃N₄–Ag₃PO₄ with different molar ratios of Ag₃PO₄ vs g-C₃N₄. The optimal CO₂ conversion rate (57.5 μmol/h/g) of the g-C₃N₄–Ag–Ag₃PO₄ is 6.1- and 10.4-folds higher than that of g-C₃N₄ and Degussa P25 TiO₂, respectively. (d) Reaction and mechanisms of photogenerated electron–hole pairs separation and transfer through the Z-scheme g-C₃N₄–Ag–Ag₃PO₄ heterojunction (e.g., CMIs). Because the CB edge of Ag₃PO₄ is more negative than the Fermi level of metallic Ag, the photoinduced electrons of Ag₃PO₄ CB flow to the metallic Ag. The holes in the VB of g-C₃N₄ then shift to metallic Ag and combine with the electrons. One of the advantages of the Z-scheme photocatalytic system is that the Z-scheme electron–hole pairs separation and transfer can retain the strong redox potential of CMNHs for CO₂ photoreduction. Replotted from He et al.(115) Copyright: 2015 American Chemical Society.

CDs exhibit strong UV–vis-NIR-light absorption due to π → π* transition of C=C bonds and n → π* transition of C=O bonds.(28,44) One of the intriguing features of CDs, that is, photoluminescence can be efficiently quenched either by electron acceptor or donor, evidencing that CDs are both UV–vis-NIR-light- and redox-responsive photocatalysts.(28,44) Integrating CDs with MNPs creates unique up- and down-converted photoluminescence properties, rendering CDs–CMNHs exciting photocatalysts. Especially, the up-converting photoluminescence attribute of CDs, that is, the ability to emit shorter wavelength (higher intensity) of lights (300–530 nm) compared to the excitation wavelengths (700–1000 nm), can be utilized to excite lower energy photons (sunlight) into a higher energy level. Hence, the UV–vis-NIR-light absorption range of CD–metal nanohybrids is broadened in photocatalytic applications.(45–48) Additionally, for CDs and g-C₃N₄ that process energy transfer properties, overlapping between the emission spectra of CDs/g-C₃N₄ and the absorption spectra of MNPs causes excitation of plasmon resonance and creates strong local electric fields around MNPs. The newly created local electric fields perturb the inherent exciton states of CDs/g-C₃N₄, and thus induce fluorescence resonance energy transfer (Figures 1a and 2).(49–51) The fluorescence resonance energy transfer and other energy transfer-related (luminescence and photoelectrochemical) attributes make CMNHs promising sensors for environmental sensing.

While the CMIs can prolong the charge-carriers lifetime and thus enhance catalytic performance of CMNHs (Figure 1a), there is a maximum number of charges (saturation point) that the CMIs can store until the band bending terminates the current flow.(20) This indicates the presence of an optimized metal loading capacity (MLC), below and above which the catalytic performance of CMNHs can deteriorate. MNPs displaying a higher work function provide an enlarged Schottky barrier and elevate the charge separation efficiency; both of which are ultimately reflected by the enhanced catalytic performance of CMNHs.(20,52)

2.2. Particle Size, Shape, Morphology, Crystallinity, and Dimensionality

During in situ synthesis of CMNHs, the size of MNPs is often tuned because CNMs control the nucleation and growth of MNPs.(14–17) Due to high thermal conductivity, CNMs can stabilize small MNPs by suppressing their growth during crystallization and phase transformation via heat sink effect.(53) The heat sink effect enables formation of small (~5 nm) CdS quantum dots onto g-C₃N₄ nanosheets; whereas, CdS can grow into ~100 nm particles without g-C₃N₄.(54) Roughly 10.5, 7.4, and 7.1 nm TiO₂ NPs are formed when 0, 1, and 2 wt %, respectively, of GO nanosheets are added during GO–TiO₂ synthesis, validating the inhibiting effect of CNMs in MNP growth.(55) Varying carbon/metal ratio also produces nanohybrids with different sizes; for example, upon addition of 10% Ag, the particle size of CDs–Au_1.0 decreases from ~30.0 to ~4.1 nm (CDs–Au_0.9Ag_0.1).(56)

MNPs’ shape and morphology are also tailored due to the strong influence of CNMs.(14–17) g-C₃N₄ nanosheets can anchor differently shaped TiO₂ (0D NPs, 1D nanowires, 2D nanosheets, and 3D mesoporous nanocrystals)(57) and CeO₂ (rods, cubes, and octahedrons).(38) CDs drive the morphology change of Cu₂O, evolving from cubes to spheres through modulation of the Ostwald ripening step during nucleation and surface reconstruction processes.(58) MNPs’ crystallinity can also be tuned; for example, KBr/KI and HCHO/Na₂C₂O₄ are common crystal phase-controlling agents for stabilizing (100) and (111) facets of Pd NPs.(59) High-resolution transmission electron microscope (HRTEM) images show that the Pd nanocrystals display cubic and tetrahedral profiles enclosed by (100) and (111) facets onto g-C₃N₄ nanosheets.(59) The higher energy (001) facet of anatase TiO₂ NPs can also be decorated onto g-C₃N₄ nanosheets using a solvent evaporation process to boost photocatalytic performance.(60) The (110), (100), and (111) facets of CeO₂ NPs are anchored onto g-C₃N₄ nanosheets based upon HRTEM observations. And the g-C₃N₄–CeO₂ (110) nanohybrids exhibit the highest photocatalytic activity as demonstrated by H₂O splitting.(38)

Acting as supporting templates (Figure 1b), the larger-sized g-C₃N₄, GFNs, and CNTs can control nucleation and growth of smaller MNPs (Supporting Information (SI) Figure S1a–d,g,h). Small CDs (1–10 nm) can also act as templates, but in most cases, are attached onto MNPs surfaces, constructing a “dot-on-particle” (CDs-on-MNPs) heterodimer structure (SI Figure S1e,f).(61) Using physical mixing and hydro/solvothermal techniques (SI Tables S5–S7), small CDs are intimately deposited onto Cu₂O with lattice spacings of 0.32 and 0.25 nm for CDs (002) and Cu₂O (111) planes (SI Figure S1f).(62) The surface-coarsened TiO₂ nanobelts have large SSA and abundant nucleation sites for CDs growth (SI Figure S1e).(63) Using a solvothermal method, a lattice-spacing of 0.211 nm CDs is also achieved and the CDs are decorated onto Na₂W₄O₁₃ flakes.(64) Conversely, under a chemical reduction approach, MNPs prefer to grow onto CDs surfaces, forming a core–shell (CDs–MNPs) structure because CDs’ oxygen-containing groups facilitate the formation of MNPs via chemical reduction pathways.(65) A shell of Pd(66) and Ag(65) NPs are formed onto CDs surfaces, as confirmed by HRTEM and selected-area electron diffraction analyses.

Dimensionality is a fundamental parameter that defines the atomic structure of material, and thus determines material properties and functions.(67) Assembling CNMs and MNPs creates equivalent or higher dimensional nanohybrids.(68) TiO₂ nanobelts retain their original 1D nature after surface-loading of small CDs (SI Figure S1e). This holds true for 2D g-C₃N₄/GFNs and 1D CNTs when surface-loaded with spherical MNPs (SI Figure S1a–d,g,h); because the nanoheterojunctions mask the 0D nature of MNPs.(68) However, the dimensionality is increased when vertically stacking 2D g-C₃N₄ or GFNs with 2D MNPs like WS₂ and MoS₂, constructing new 3D graphene-WS₂(69) and graphene-MoS₂.(70,71) Coupling 1D CNTs and 2D GFNs with MNPs (Co, Fe₃O₄, and FeCo) also yields 3D CNTs–GFNs-metal nanohybrids (SI Table S1). Increased dimensionality of CMNHs makes them more accessible to contaminants and can guide their applications at the EWE nexus and beyond.(72)

2.3. Carbon/Metal Ratio and Hybridization Mode

Carbon/metal ratio or metal loading capacity (MLC) is another key attribute (SI Figure S1g,h) of CMNHs, because MLC determines particle dispersion state in aqueous solutions, sunlight utilization, and catalytic performance. The absorption edge of g-C₃N₄–CdS nanohybrids can be tailored by varying the mass ratio of g-C₃N₄/CdS, yielding a shift in the optical absorption toward higher wavelengths in the visible-light region.(73) The optimum activity of g-C₃N₄–CdS (7:3) is ~21- and ~42-times higher than that of bare g-C₃N₄, as shown via degradation of methyl blue and 4-aminobenzonic acid, respectively.(73) The photocurrent density of CDs–TiO₂ nanohybrids is enhanced and reaches a maximum with increasing content of CDs (0–0.4 mg/mL), but a further increase in CDs content decreases the photocurrent due to CDs aggregation (which compromises electron and charge transfer).(74) The optimal MLC also exists for CNTs– and GFNs–CMNHs (CNT–TiO₂,(75) GO–Ag–Ti,(76) and RGO–MoS₂–ZnS(77)). Besides MLC, the distribution of MNPs onto CNMs can be tuned by controlling synthesis conditions; that is, by controlling electrochemical deposition potential and time, or by varying the amount of nucleation in the dispersion of metal precursors (g-C₃N₄–TiO₂).(57)

Depending upon the hybridization mode, ex-situ and in situ strategies can be used to hybridize CNMs and MNPs.(78) The ex-situ approach utilizes covalent, noncovalent, π–π stacking, and electrostatic interactions to combine parent NMs via interlinkers.(78) The SiO₂ interlinker is used to covalently bind silane-functionalized CDs and AuNPs, forming a core–shell unit.(79) For comparison, the in situ approach involves direct nucleation, growth, and deposition of MNPs onto CNMs using electrochemical, sol–gel process, hydrothermal/solvothermal, and gas-phase deposition techniques (SI Tables S5–S10).(78) The advantages of the in situ approach mainly include (1) CNMs can stabilize uncommon or novel crystal phases of MNPs; and (2) continuous amorphous or single-crystalline films with controlled thickness or discrete units of NPs, nanorods, nanobelts, and nanobeads can be fabricated with the presence of CNMs.(78) Compared to the ex-situ methods, the in situ method produces well-contacted CMIs, which is the key in catalytic and redox reactions of CMNHs (see Section 3 below).(5,6)

3. Energy, Water, and Environmental Applications of CMNHs

While some reviews exist for CMNHs since 2015 including those for CNTs-,(80–82) GFNs-,(80,82–84) CDs-,(28,45,46) and g-C₃N₄-based CMNHs,(5,6,52,85,86) the altered and newly emerging attributes of CMNHs have not been highlighted, particularly with respect to how these can be harnessed for EWE applications (Figure 1). In this section, unique properties of the hierarchical nanoheterojunctions are identified for their use in energy harvesting, water treatment, and environmental remediation.

3.1. Energy Harvesting

3.1.1. Overall H₂O Splitting for H₂ and O₂ Production

The appealing attributes of CMNHs (Section 2) bring in exceptional advantages, which can be harnessed for transformational applications in energy harvesting, such as for H₂O splitting (H₂ and O₂ evolution) and CO₂ conversion using solar radiation.(5,6) H₂ is a clean and renewable fuel with the highest energy density (140 MJ/kg).(87) Functioning as an economically feasible photocatalyst, the g-C₃N₄-based CMNHs can effectively split H₂O to produce H₂ and O₂ under solar irradiation.(5,6) In the pioneering work of Wang et al.,(88) the photocatalytic H₂-evolution rate under visible-light illumination is elevated by 1–2 orders of magnitude (from 0.1–4 to 10.7 μmol/h), when 3 wt % Pt is decorated onto 2D g-C₃N₄ nanosheets. Such advantages could be imparted on the nanohybrids by the introduction of CMI (with new built-in electric field and local electromagnetic field; Figure 1a) that prolongs charge-carriers lifetime and accelerates the separation and transfer of photogenerated electron–hole pairs.(88) The pivotal role of new built-in electric field in speeding charge separation and transfer during photocatalytic H₂-evolution is also demonstrated for g-C₃N₄–SrTiO₃ with XPS and DFT analyses.(89) To minimize cost and make practical applications possible, research directions have later been directed toward Pt-free inexpensive MNPs-incorporated g-C₃N₄–CMNHs (including TiO₂, ZnO, MoS₂, Fe₂O₃, CdS, and BiVO₄).(5,6,90) One of the highest photocatalytic H₂-evolution rates upon visible-light irradiation is reported at 31400 μmol/h/g for ultrathin g-C₃N₄-α-Fe₂O₃ due to the large SSA, optimized light absorption, and accelerated transfer of photogenerated electron–hole pairs at the well-contacted g-C₃N₄-α-Fe₂O₃ CMI, whose quantum efficiency also reaches up to ~44.4% at 420 nm.(91)

Benefiting from the wide-ranged light response, high light-absorption efficiency, and low charge-carriers recombination rate, the LSPR-responsive MNPs (Ag, Au, Cu, and Bi) profoundly broaden the application scope of CMNHs in plasmonic photocatalysis (H₂O splitting), surface-enhanced Raman scattering, and plasmon-enhanced fluorescence.(6) In photocatalytic H₂O splitting, the visible-light-responsive photocurrent density is 1000-folds higher for 1 wt % Au-decorated g-C₃N₄ compared to the bare g-C₃N₄ due to LSPR, yielding a 23-times higher H₂-production reactivity for g-C₃N₄–Au (vs g-C₃N₄).(92) The LSPR frequency and contribution in photocatalytic applications vary, depending on the size, shape, crystallinity, and dimensionality of MNPs, as well as MLC and hybridization mode between MNPs and CNMs (Section 2); because these factors affect light absorption, CMI, and the plasmon resonance energy transfer (PRET; Section 2.1) of CMNHs.(6,93) Evidence shows that there is an optimal physical distance between MNPs (e.g., AgNPs) and photocatalysts (e.g., g-C₃N₄) when the LSPR absorption of MNPs partially overlaps with the optical absorption of photocatalyst.(5,6,94) This is because LSPR-induced PRET effect shortens the charge-carriers transfer distance and inhibits the charge-carriers recombination. However, nonradiative energy transfer—Förster resonance energy transfer (FRET) occurs when the distance between MNPs and photocatalyst is too close, which adversely quenches the photogenerated charge-carriers. Using an engineered-nanogap strategy, that is, by loading plasmonic Ag@SiO₂ (core@shell) nanostructures (nanogap = 8–21 nm) onto g-C₃N₄, Chen et al.(94) demonstrated that the optimized nanogap of 12 nm can balance the positive PRET and negative FRET effects based upon finite difference time domain (FDTD) simulations,(93) which yields the maximum H₂-production activity (11.4 μmol/h) under solar irradiation. These findings open-up doors for nanoengineering of efficient CMNHs by precisely tuning architectures (distance between CNMs and MNPs) of CMNHs for H₂O splitting.

Recently, the Z-scheme photocatalytic system (two different photocatalysts are coupled by an appropriate shuttle redox mediator to form Z-shape catalytic system; for example, Figure 3) has attracted tremendous attention since it speeds up electron–hole pairs separation/transfer spatiotemporally (Section 2), and concurrently retains or enhances redox capability of CMNHs.(5,6,86,95) Compared to g-C₃N₄ and g-C₃N₄–CdS, the Z-scheme g-C₃N₄–Au-CdS is >34-times (3.1 vs 106 μmol/h) more active in photocatalytic H₂-evolution. Such an enhancement occurs primarily due to AuNPs’ role as “electron-bridges”(95) that promote electron transfer between g-C₃N₄ and CdS.(96) Notably, the Au@CdS (core@shell)-assembled g-C₃N₄ nanosheet shows ~126-folds (0.15 vs 19 μmol/h/g) higher H₂-production rate than bare g-C₃N₄ due to enhanced light absorption and Z-scheme separation of charge-carriers.(97) Other Z-scheme systems showing excellent photocatalytic H₂O splitting efficiency and high stability and selectivity include g-C₃N₄–Ag₃PO₄–Ag₂MoO₄,(98) g-C₃N₄–Au-TiO₂,(95) g-C₃N₄–TiO₂,(99) and g-C₃N₄–NiTiO₃(100) (SI Table S4).

Attributed to the highly porous nanostructures, large SSA, wide-spectrum light absorption, fast electron–hole separation at CMIs, and favorable π–π interactions (enhanced charge-carriers generation and transfer) between metal organic frameworks (MOFs) and triazine rings of g-C₃N₄, the hierarchically arranged g-C₃N₄-MOFs (ZIF-8, UiO-66, and MIL-53)(101) demonstrate remarkable H₂O splitting performance. The g-C₃N₄-ZIF-8 composites present a high H₂-production rate of 309.5 μmol/h/g due to the synergy among photoluminescence, electron–hole separation and charge transportation, and redox capabilities.(102) Using time-resolved transient fluorescent spectroscopy measurements, Wang et al.(103) found that the photoluminescence lifetime of charge-carriers is shorter in g-C₃N₄–UiO-66 vs g-C₃N₄ (2.26 vs 2.88 ns), yielding a 17-times increase in H₂-evolution rate, since the shorter photoluminescence lifetime reveals a more rapid transfer of photogenerated electrons. The H₂-production rate of 905.4 μmol/h/g is achieved for g-C₃N₄-MIL-53(Fe), which is 335- and 47-folds higher than that of MIL-53(Fe) and g-C₃N₄, respectively. The greater catalytically active sites and expedited electron–hole migration at g-C₃N₄-MIL-53(Fe) CMIs are responsible for such enhancement.(104)

Besides photocatalytic activity, the stability and selectivity of CMNHs in H₂O splitting are optimized when bimetallic NPs are decorated onto CNMs.(6) The 1.0 wt % PtCo-loaded g-C₃N₄ nanohybrids show greater H₂-evolution rate (960 μmol/h/g) and stability (~28 h) compared to monometallic g-C₃N₄–Pt (330 μmol/h/g), because bimetallic PtCo NPs increase surface defect density and alter the Fermi level of CMNHs (both of which promote photoinduced electron–hole pairs separation).(105) A 3.5- (g-C₃N₄–Au) and 1.6-folds (g-C₃N₄–Pd) increase in H₂-evolution rate is reported for 0.5 wt % AuPd-decorated g-C₃N₄ (326 μmol/h/g), which can maintain high photocatalytic activity after four cycles by sustaining visible-light absorption and transfer of electron–hole pairs from the AuPd alloy.(106) It is also remarkable to note that the g-C₃N₄–PtCo nanohybrids possess a long-term stability after 510 h of reaction with no noticeable deactivation in photocatalytic H₂O splitting.(107)

Both g-C₃N₄ and CDs can convert NIR-light to visible-light, making them useful as universal energy-transfer materials for photocatalytic energy conversion. Particularly, the ternary g-C₃N₄-MNPs-CDs nanoheterojunctions excel in H₂O splitting. A 53-times higher H₂-evolution rate (212.4 μmol/h/g) is reported for the Z-scheme g-C₃N₄–MoS₂–CDs with excellent photostability than that of g-C₃N₄–MoS₂. Enhanced light absorption, accelerated charge transfer at two CMIs (g-C₃N₄–MoS₂ and CDs–MoS₂), and more catalytically active sites rendered by MoS₂ are responsible for the observed results.(108) The ultrastable g-C₃N₄–UiO-66-CDs photocatalyst achieves a H₂-production rate of 2930 μmol/h/g upon visible-light illumination, which is 32.4-, 38.6-, and 17.5-folds higher than that of g-C₃N₄, UiO-66, and g-C₃N₄–UiO-66, respectively.(109) Other types of CMNHs also perform well in photocatalytic H₂O splitting. The commonly used CDs–CMNHs, for example, CDs–BiVO₄ and CDs–NiP photocatalysts show the optimal H₂-evolution rates of 4.02 and 398 μmol/h/g, respectively, under visible-light illumination, which are much higher than the H₂-evolution rate of parent component materials.(45,48,110) Recent findings on photocatalytic H₂O splitting by CNTs–, GFNs–, CDs–, and g-C₃N₄–CMNHs are detailed in SI Tables S1–S4.

3.1.2. CO₂ Conversion for Energy Storage

Converting the major greenhouse gas CO₂ into energy-bearing products (CO, CH₄, HCOOH, HCHO, and CH₃OH) offers a feasible means not only in combatting climate change but also in alleviating energy crisis.(111,112) Again, the appealing electronic/optical/catalytic/redox attributes of g-C₃N₄–CMNHs make them the next-generation of robust photocatalysts, which facilitate CO₂-conversion for energy storage.(5,6) High yield (107 μmol/h/g) and selectivity (94%) are reported for 43 wt % Co₄-decorated g-C₃N₄ nanohybrids in CO₂ photoreduction under visible-light irradiation (425–700 nm), which occurs due to facilitated charge transfer at the CMIs and excellent surface oxidative capability of Co₄.(113) Using the Z-scheme g-C₃N₄–SnO_2–x photocatalyst at MLC = 42.2 wt % SnO_2–x, the CO₂ photoreduction rate reaches 22.7 μmol/h/g, which is 4.3- and 5-folds higher than that of g-C₃N₄ and P25 (TiO₂), respectively (Figure 3a).(114) This occurs because under the direct Z-scheme system, electrons at the CB of SnO_2–x interact with photoexcited holes at the g-C₃N₄ VB, creating a strong reducing capability for the excited electrons in g-C₃N₄, which can reduce CO₂ to CO, CH₄, and CH₃OH (Figure 3b). A higher CO₂ conversion rate (57.5 μmol/h/g) is later reported by the same group for the Z-scheme g-C₃N₄–Ag–Ag₃PO₄ nanoheterostructures, in which AgNPs function as electron mediator and charge transmission bridge to construct the Z-scheme system (electrons flow through Ag₃PO₄ CB to g-C₃N₄ VB; Figure 3c–3d).(115) Not only that, the synergy between Z-scheme electron/charge transfer and LSRP effect of AgNPs (energize more electrons) causes 12.7-, 7.9-, and 2-times enhancement in electron consumption rate (87.3 μmol/h/g) for the g-C₃N₄–Ag–TiO₂ than TiO₂, g-C₃N₄, and Ag–TiO₂, respectively.(116) These findings provide a head start for nanoscale engineering of highly efficient Z-scheme photocatalyst to convert CO₂ into energy-bearing chemical products.

The altered attributes including particle size, shape, and morphology of MNPs in CMNHs (Section 2) can significantly impact the efficiency and product selectivity in CO₂ conversion. Larger AuNPs (100–150 nm in size) function as electron/charge bridges in the Z-scheme g-C₃N₄–Au-BiOBr composite and enhance CO₂ reduction (CO production rate = 6.67 vs 2.63 μmol/h/g for g-C₃N₄–Au-BiOBr vs g-C₃N₄–BiOBr). Whereas smaller 10–20 nm AuNPs promote CO₂ reduction largely due to LSRP.(117) The uniformly decorated PdNPs with different preferentially exposed facets (cubic (100) and tetrahedral (111)) onto g-C₃N₄ nanosheets exhibit varying degrees of CO₂ reduction. For these catalysts, the Pd(111)-g-C₃N₄ performs better due to higher adsorption energy (E_A = 0.230 vs 0.064 eV) of CO₂ by Pd(111) compared to Pd(100), determined via first-principles calculations.(59) After CO₂ adsorption, the activation barrier (E_B) is lowered from 7.15 to 3.98 eV and from 6.79 to 4.15 eV, respectively, for Pd(111) and Pd(100) facets, again validating that Pd(111) is more active for CO₂ reduction (3.98 eV < 4.15 eV).(59) Using DFT calculations, Cao et al.(118) also showed that the tetrahedral Pd(111) facet is more active than cubic Pd(100) in CO₂ photoreduction by g-C₃N₄–Pd. The underlying cause is identified as the electron sink effect, CO₂ adsorption, and CH₃OH desorption capability of Pd(111). These findings bring forward exciting new opportunities for tailoring MNPs’ size and structure in CMNHs to achieve better CO₂ reduction.

Besides g-C₃N₄–CMNHs, other CMNHs with various catalytically active sites and defects (Stone–Wales defects and vacancies) and oxygen-containing groups, along with stable particles in suspension (nonaggregating due to surface coating of negatively charged CDs; Section 2.2) and inhibited surface passivation of MNPs (Figure 1b) also exhibit efficient CO₂ reduction. Examples include: CDs–Cu₂O nanohybrids which show efficient production of CH₃OH (56 μmol/h/g) compared to that of Cu₂O only (≤38 μmol/h/g). In these systems, CDs function as photosensitizers and electron donors/acceptors, which prevent charge-carriers recombination.(119) The diffuse reflectance spectroscopic measurements further indicate that the CDs–Cu₂O can adsorb higher amount of light in the 640–2500 nm wavelength region compared to Cu₂O, demonstrating that the CDs–Cu₂O is more NIR-sensitive and can better utilize a wider portion of the electromagnetic spectrum for CO₂ photoreduction.(119) The NIR-light-driven CO₂ reduction is also observed for 1 wt % CD–-Bi₂WO₆, which shows 9.5- and 3.1-folds increment in CH₄ production over Bi₂WO₆ nanoplatelets and nanosheets, respectively.(120) The full-spectrum UV–vis-NIR-driven CO₂ photoreduction also frequently occurs for other CDs–CMNHs (CDs-TiO₂,(121) CDs–CdS,(122) and CDs–ZnO(123)). These findings present a potentially new platform for developing highly efficient and inexpensive CDs–CMNHs for CO₂ conversion using the full-spectrum of inexhaustible sunlight.

Regardless of CNMs type, an optimal MLC always exists for CMNHs (Section 2.3), which renders catalytic performance (for H₂O splitting, CO₂ conversion, and contaminant removal). Compared to other MLCs, the 1 wt % MgO^–1 wt % CuNi-loaded CNTs present the highest catalytic efficacy. This is due to the promoted dispersion of Ni (larger SSA and more catalytically active sites), restrained reduction of NiO, and lowered activation energy of NiO toward catalytic reaction.(124) The C₂H₅OH catalytic yield is reported at 49.1% and 92.2%, respectively, for two Pd-CNT nanohybrids (Pd/PdO ratio = 90/10 vs 60/40), depending on the architecture and dispersion status of Pd NPs that control electron transport and mass transfer processes.(125) During CO₂ photoreduction, the optimal 23 wt % Ni-graphene reaches the maximum CH₄-evolution rate (642 μmol/h/g) and quantum yield (1.98%) due to excellent charge separation at the C–Ni CMI.(126) The optimum MLC is also present for other CNTs– (CNTs–Pd(125) and CNTs–Ni–Zr(127)), GFNs–CMNHs (graphene-MoS₂–TiO₂(128) and RGO–Pt–TiO₂(129)) and CDs–CMNHs (CDs–TiO₂(121) and CDs–ZnO(123)). Consequently, more research in this area is essential to maximizing the catalytic performance of CMNHs by optimizing MLC.

The multifunctionality of CMNHs in energy harvesting sector (both H₂O splitting and CO₂ conversion) is demonstrated by concurrent H₂O splitting and CO₂ conversion by g-C₃N₄–Au-TiO₂,(130) H₂O and CO₂ photoreduction by RGO–BiWO₆-g-C₃N₄,(19) CO and CH₄ production by graphene–TiO₂,(131) CO₂ reduction in generating syngas (CO and H₂) by CDs–Co₃O₄–C₃N₄(132) and g-C₃N₄–Ag,(133) and many others shown in SI Tables S1–S4. Additionally, the multifunctional CMNHs also show great potentials in other energy-related applications including electrocatalytic reactions (e.g., oxygen reduction reaction, electrocatalytic oxidation of alcohols, electrochemical reduction of CO₂ and H₂O₂, methane reforming, and others; see SI Tables S1–S4). Interested readers can find more detailed information regarding CMNHs applications in electrocatalytic reactions in the literature.(134–138)

3.2. Water Treatment

3.2.1. Contaminant Removal and Microbial Inactivation

Simultaneous, fast, and effective removal of multiple inorganic/organic pollutants and inactivation of microbes have been at the forefront for developing water treatment technologies. Benefiting from the uniqueness and multifunctionality, the CMNHs have already shown outstanding ability for contaminant removal (adsorption and photocatalytic/redox degradation). CMNHs can quickly (minutes to several hours) and effectively remove a range of contaminants (generally >90% degradation), including organic contaminants such as dyes, phenols, and persistent organic pollutants (e.g., polycyclic aromatic hydrocarbons and polychlorinated biphenyls), emerging contaminants (pharmaceuticals and personal care products, PPCPs; endocrine disrupting compounds, EDCs; and perfluorochemicals), and inorganic toxins such as heavy metals (As, Cd, Cr, Hg, and Pb) and radionuclides (Am, Eu, La, and U) (SI Tables S1–S4). Inactivation of microbes is also achieved effectively by the CMNHs (SI Tables S1–S4). Selected examples and associated mechanisms for contaminant removal are briefly presented here (detailed mechanisms particularly those for microbial inactivation are given in SI Tables S1–S4).

CNTs–TiO₂ is used for photocatalytic degradation of a mixture of 22 PPCPs and EDCs in wastewater effluents at low concentrations (μg/L) under UV and simulated solar irradiation.(139) The CNTs–TiO₂ performs better (9–96% vs 9–87% degradation efficiency, and 0.05–0.43 vs 0.05–0.17 min^–1 degradation rate constant) compared to conventional photocatalysts (Degussa P25 TiO₂). Mechanisms responsible for such performance are likely enhanced dispersion of TiO₂, preconcentration of contaminants on the surfaces of both CNTs and TiO₂ NMs, rich surface-active sites of both, and rapid separation of photoinduced electron–hole pairs. These findings underscore that CNTs–TiO₂ has promise in removing emerging organic pollutants from wastewater.(139) Furthermore, complete and fast (minutes to several hours) removal (adsorption, catalysis, and redox) of a diverse set of contaminants including heavy metals, radionuclides, dyes, phenols, PPCPs, EDCs, and polychlorinated biphenyls, as well as inactivation of pathogens from water and wastewater have been frequently reported for GFNs–CMNHs like GO–Ag,(140) GO–Ag₃PO₄,(141) RGO–PdAg,(142) RGO–Ag–Fe₃O₄,(143) and GO–MnFe₂O₄ (e.g., maximum adsorption capacities for La and Ce are as high as 1001 and 982 mg/g).(144) Also, coremoval of ciprofloxacin (88%), rhodamine B (97%), tetracycline (67%), and bisphenol A (60%) is also observed for CDs–BiOBr within 1–3.5 h under visible-light irradiation due to enhanced light absorption and excellent active centers for charge-carriers separation at the CMIs.(145) Fabricated with a biogenic green and cost-effective approach, the g-C₃N₄–Ag composite shows a high dye degradation efficiency (~100% and ~89% degradation of methylene blue and rhodamine B within 4 h) and a strong performance toward inactivation of pathogens (Escherichia coli, Staphylococcus aureus, and Pseudomonas aeruginosa) under visible-light illumination.(146) Enhanced AgNP dispersion, larger SSA, prolonged visible-light absorption due to LSPR, suppressed charge recombination, and greater production of reactive oxygen species (ROS) such as ^•O₂^– and ^•OH and release of Ag⁺ ions collectively contribute to the greater photocatalytic performance and reactivity of g-C₃N₄–Ag than the parent NMs.(147) For CMNHs that include antimicrobial MNPs (Ag, Au, CuO, TiO₂, and ZnO), their antimicrobial performance is always higher than the parent MNPs. The greater antimicrobial activity of the MNPs in the nanoheterostructures is likely caused by enhanced particle dispersion, larger SSA, enhanced direct-interaction between MNPs and microbes, and additional antimicrobial activity from the CNMs(148) (SI Tables S1–S4). Inactivation of P. aeruginosa has been successfully achieved by harnessing microwave radiation and generating ROS with CNT–Er₂O₃ nanohybrids.(149) For MNPs with magnetic properties like Fe, Ni, Co, Fe_xO_y, and MFe₂O₄ (M denotes metal),(150,151) the magnetically separable CMNHs can be easily recycled using an external magnetic field, suppressing the likelihood of generating a secondary waste from the release of the nanoscale treatment agents. Moreover, the nanohybrids exhibit high stability, selectivity, and reusability with no appreciable deterioration in reactivity after several consecutive cycles of use (n ≥ 6),(152–155) due to the strong mechanical strength of the carbon scaffolds. These advantages can greatly minimize operational cost, while enhancing the removal efficiency of multiple contaminants from water and wastewater.

Recently, CMNHs-enabled single-atom catalyst(156) has attracted significant attention due to high catalytic activity (maximized atomically catalytic efficiency due to complete exposure of surface sites), stability, and selectivity. Using a facile confined-interface-directed route, the Pd atoms are anchored onto the interfaces of double-shelled hollow RGO (inner shell)-amorphous carbon (outer shell) nanospheres, as shown by DFT calculations.(157) The resulting RGO–amorphous carbon–Pd nanohybrids show a significantly higher turnover frequency (602 min^–1) than that of RGO–Pd (106 min^–1) and amorphous carbon-Pd (97 min^–1) in 4-aminophenol reduction due to the ideally dispersed Pd atoms allowing for access to all catalytic surface sites. Furthermore, the nanohybrids exhibit high stability in 4-aminophenol reduction (100% conversion during five repeated cycles and >95% conversion after the eighth repeated cycle).(157) These findings offer a new direction in maximizing the atomic efficiency, activity, and stability in metal-based heterogeneous catalysis for contaminant removal.

Through Raman, XPS, and EPR characterization, the dual-reaction-centered g-C₃N₄–Al₂O₃–Cu and CDs–Al₂O₃–Cu nanohybrids (electron-rich Cu center and electron-deficient Al site) are reported to significantly facilitate electron transfer and ^•OH generation, showing remarkable promise in catalytic degradation of organic pollutants under mild Fenton-reaction conditions.(158) A similar multiple-reaction-centered RGO–CoAl (layered double hydroxide; LDH)-g-C₃N₄ is recently reported to exhibit ~100% photocatalytic removal and mineralization of congo red and tetracycline within 30 min, which is ~20- and ~15-times higher than that of CoAl and g-C₃N₄, respectively (Figure 4).(159) This is due to the beneficial 2D stacking of RGO, CoAl, and g-C₃N₄ that results in multiple intimate CMIs (RGO–CoAl and g-C₃N₄–CoAl) and hinders direct recombination of electron–hole pairs, thereby accelerating interfacial charge transfer. Similar findings have also been reported for 2D-2D g-C₃N₄–MoS₂ and 2D-0D g-C₃N₄–Pt in photocatalytic degradation.(160) These results highlight the significance of dimensionality in controlling photocatalytic reactivity of CMNHs, which also brings forward a new rationale for nanoscale engineering of multiple-reaction-centered heterojunctions as high-performance photocatalysts for water treatment.

Figure 4.

(a–h) Transmission electron microscope (TEM) (a) and high-resolution TEM (b) images of the ternary 2D-2D-2D RGO–CoAl (LDH)-g-C₃N₄ composite photocatalyst, and energy-dispersive X-ray (EDX) mappings of component elements of total (c), C (d), N (e), Co (f), Al (g), and O (h), respectively. (i–j) Photocatalytic degradation and total organic carbon (TOC) removal of congo red (CR; (i) and tetracycline (TC; j) by the ternary RGO–CoAl-g-C₃N₄ photocatalyst under visible-light illumination. (k) Schematic diagram illustrates the photocatalytic mechanisms for CR and TC degradation by the ternary RGO–CoAl-g-C₃N₄ nanoheterojunction. The appealing photocatalytic performance is mainly due to the large intimate interfacial contacts between 2D RGO, 2D CoAl, and 2D g-C₃N₄, which accelerates the interfacial charge-transfer processes at the RGO–CoAl CMI, g-C₃N₄–CoAl CMI, and RGO–g-C₃N₄ interface. LDH refers to layered double hydroxide. Replotted from Jo and Tonda.(159) Copyright: 2019 Elsevier.

Besides the strong influence of particle size, shape, morphology, dimensionality, and MLC discussed earlier, the mode of material hybridization also impacts the performance of CMNHs for contaminant removal. The in situ synthesized CDs–TiO₂ (is-CDs–-TiO₂) shows much higher photocatalytic activity for benzene, pesticides, and phenol over the 3 synthesized CDs–TiO₂, due to greater up-converted photoluminescence, enhanced particle dispersion, and faster transfer of photogenerated electron–hole pairs at the closer CMIs of is-CDs–TiO₂.(161) These results indirectly demonstrate the dominant role of CMIs, which control the catalytic performance of CMNHs (with intimately contacted CMIs within the nanoheterojunctions; Figure 1).

3.2.2. Membrane Technology

Membrane-based water treatment and desalination technologies are popular treatment options in many parts of the world.(162–164) CNMs (CNTs and GFNs) have been extensively used to modify membranes and impart mechanical strength, improve water permeability and flux (e.g., aligned CNTs), enable tunability of hydrophobicity, introduce selectivity and antifouling capability, and deliver flexibility toward functionalization.(4,165–167) Recently, CNTs– and GFNs–CMNHs have shown great potential in various membrane technologies including reverse osmosis (CNTs–TiO₂,(168) GO–Ag,(169) and GO–Fe(170)), forward osmosis (CNTs–SiO₂-polyvinylidene fluoride(171) and GO–Ag(172)), ultrafiltration (CNTs–Al(173) and GO–TiO₂(174)), nanofiltration (CNTs–Al(175) and RGO–UiO-66(176)), microfiltration (GO–Al₂O₃(177)), capacitive deionization (CNTs–MnO_x(178) and GO–Ag–Cu(179)), electrochemical deionization (RGO–FePO₄(180)), membrane distillation (RGO–Bi₂WO₆(181)), and organic solvent nanofiltration (GO–ZIF-8(182)). For example, the RGO@Fe₃O₄ nanofiltration membranes present high water permeance (~300 L/m²/h/bar) and dye (Rhodamine B and bisphenol A) and ion (CuSO₄, CdSO₄, MnSO₄, and CoSO₄) rejection by utilizing the expanded interlayer spacings and nanochannels between the ordered laminar RGO layers due to uniform loading of Fe₃O₄ NPs (Figure 5).(183) The RGO@Fe₃O₄ nanofiltration membrane system can be easily scaled up for wastewater treatment, and its sufficient mechanistic strength and stability under high-pressure and cross-flow operations will enable these applications (Figure 5).(183)

Figure 5.

(a–c) Schematic diagrams and (d–f) SEM images of cross-sectional morphology of (a and d) RGO, (b and e) RGO–Fe₃O₄, and (c and f) RGO@Fe₃O₄ membranes. Note that the RGO–Fe₃O₄ is obtained by simply mixing Fe₃O₄ NPs with RGO suspension using the intercalation method, in which the Fe₃O₄ NPs are disorderly and loosely attached onto RGO. In contrast, the RGO@Fe₃O₄ is synthesized via in situ solvothermal strategy, in which size- and density-controllable Fe₃O₄ NPs are uniformly grown on the regularly stacked RGO nanosheets through precise coordination. (g) Water permeance and rejection of RGO, RGO–Fe₃O₄, and RGO@Fe₃O₄ membranes shown in (a–f). (h) Water permeance of RGO@NPs membranes using nonporous NPs (Fe₃O₄ and TiO₂) and porous NPs (UiO-66) having different particle sizes (Feed solution: 50 mg/L of rhodamine B and bisphenol A; 10 mM of CuSO₄, CdSO₄, MnSO₄, and CoSO₄; Pressure: 2 bar; and Temperature: 25 °C). (i) photographs (black part of the inner surface is covered by sealing epoxy) and (j) SEM cross-sectional image of the RGO@Fe₃O₄ membrane deposited on the inner surface of a ceramic tube (ZrO₂ and Al₂O₃ supporting layers). (k) The cross-flow nanofiltration system using the RGO@Fe₃O₄ membrane deposited on the inner surfaces of tubular ceramic tubes can be easily scaled up for effectively treating wastewater for real-world applications due to the stability for surviving the high pressure and cross-flow operations. Replotted from Zhang et al.(183) Copyright: 2017 Wiley.

g-C₃N₄ (tris-triazine) is an ideal material for membrane technologies because of its geometry and the triangular nanopores (~3.11 Å)(184) that exist on the 2D nanosheets, which allow for easy passage of water molecules with kinetic diameter of 2.6 Å.(185) The g-C₃N₄–Ag₃PO₄-polyvinylidene fluoride,(186) g-C₃N₄–Fe(OH)₃-Al₂O₃,(187) and g-C₃N₄–CNTs–GO–TiO₂(188) nanofiltration membranes, g-C₃N₄–Ag₃PO₄-poly(ether sulfone) microfiltration membranes,(189) and g-C₃N₄–Ag–nafion ultrafiltration membranes(190) exhibit improved fouling resistance, photocatalytic degradation efficiency, antibacterial activity, stability, reusability, and water flux. For example, the g-C₃N₄–CNTs–GO–TiO₂ nanofiltration membranes exhibit enhanced water flux (~16 L/m²/h/bar) while maintaining an increased dye (~100% for methyl orange) and salt (67% for Na₂SO₄) rejection efficiency. CNTs are known to expand the interlayer spacing between neighboring graphene nanosheets and thus enhance the stability and strength of the membrane, while the g-C₃N₄ and TiO₂ NPs deliver the desired catalytic-activity.(188) The g-C₃N₄–CNT–GO–TiO₂ membranes also display multifunctionalities in photocatalytic coremoval of ammonia (50%), sulfamethoxazole (80%), and bisphenol A (82%) in wastewater from aquaculture.(188) Integrating membrane filtration with photocatalysis (incorporating g-C₃N₄–CMNHs) opens doors for fabricating the next-generation antifouling membranes in water treatment and desalination applications.

In addition to g-C₃N₄, the facile production and appealing physicochemical attributes (small size, good biocompatibility, tunable hydrophilicity, rich surface functional groups, and antifouling properties) of CDs enable these materials to modify desalination and water treatment membranes.(191) CDs–CMNHs can be readily integrated with various membrane materials (thin film nanocomposites and polymers) in reverse osmosis, nanofiltration, and pressure retarded osmosis applications.(191) The CDs–CMNHs-modified membranes (CDs–TiO₂–SiO₂)(192) have shown to outperform unmodified membranes in terms of treatment performance (salt rejection and contaminant degradation time and efficiency), stability, and reusability. The enhancement in hydrophilicity, permeability, and antifouling property results in biofilm reduction (medicated by electrostatic repulsion between negatively charged CDs and bacteria, physical interaction, and enhanced oxidative stress).(148) Because the CDs–CMNHs are nonselective toward bacteria, the membranes functionalized with these materials present a proof-of-concept which can be used for developing novel antibacterial membranes in the future.

3.3. Environmental Sensing and Remediation

3.3.1. Environmental Sensing

CMNHs have shown advantages in sensing of multifarious environmental species such as pollutants (heavy metals, antibiotics, pesticides, phenolics, and microcystins) and biomacromolecules (enzymes, proteins, RNA, and DNA) (SI Tables S1–S4). Appealing attributes in effective electron and electrochemical charge transfer (Section 2.1 and Figure 1), abundant surface functional groups, high sensitivity, strong photostability, and favorable biocompatibility can enhance the sensing performance.(50,193–195) The dual-emission CDs–CdSe–ZnS@SiO₂ fluorescent probe is developed for in vivo imaging of Cu²⁺ (0.2–1 μM linear detection range) in living cells with high degrees of specificity and sensitivity.(196) A graphene-Bi framework is assembled for in situ detection of multiple heavy metal ions (1–120 μg/L of Pb(II) and Cd(II), 40–300 μg/L of Zn(II), and with a detection limit of 0.02–4 μg/L). The controllable nanoarchitecture, large SSA, and fast mass and electron transfer are unique attributes of the nanohybrids.(197) Based on blue photoluminescence and excitation-wavelength-dependent emission, the CDs–Eu sensor can selectively detect tetracycline (with a linear range of 0.5–200 μM and a detection limit of 0.3 μM) for lake water samples.(198) Despite interferences from methyl parathion, pentachlorophenol, and carbaryl, the RGO–BiPO₄ (RGO/BiPO₄ optimal mass ratio = 0.03) photoelectrochemical sensor can selectively detect chlorpyrifos within 0.05–80 ng/mL with a low detection limit of 0.02 ng/mL (mediated by reduced particle agglomeration).(199) The highly selective CNTs–TiO₂ photoelectrochemical sensor shows ultrasensitive detection range (1.0 pM–3.0 nM) for microcystin-LR.(200) These findings manifest the robustness of CMNHs in sensing environmental pollutants at ultratrace levels with high degrees of selectivity and stability.

In addition to environmental pollutants, CDs–Au-poly(amidoamine) immunosensor can identify an important cancer biomarker (alpha-fetoprotein) with a wide linear detection range of 100 fg/mL–100 ng/mL and a low detection limit of 0.025 pg/mL for serum samples.(201) An innovative dual-channel CDs–Au biosensing system is recently fabricated to concurrently monitor multiple nucleotide sequences (breast cancer and thymidine kinase RNA/DNA) with a linear range of 4–120 nM and a detection limit of 1.5–4.5 nM (Figure 6).(202) Excellent visible-light response and fluorescence resonance energy transfer, novel hairpin structure, and strong interactions between AuNPs and DNA account for the ultrahigh selectivity, sensitivity, specificity, and multifunctionality of the biosensors toward RNA/DNA (Figure 6).(202) The CDs–Au-based biosensing model presents a prototype for nanoengineering similar or more sophisticated CMNHs-based monitoring systems to analyze any possible gene sequence or aptamer–substrate complex in environmental matrices.

Figure 6.

A dual-channel CDs–Au biosensing system for assaying RNA and DNA due to the LSPR effect of AuNPs, strong interaction between AuNPs and ssDNAs, and appealing fluorescent attribute of CDs. (a) Schematic illustration of the dual-channel sensing model for assaying multiple nucleotide sequences including BRCA1 (breast cancer 1) and TK1 (thymidine kinase 1) RNA/DNA. (b) Schematic illustration for identifying the amount of hairpin structure hybridized with AuNPs-DNA. (c) Fluorescence spectra of the sensing model in the presence of different concentrations of BRCA1 RNA excited at λ = 345 nm (0–120 nM; c1), and in the presence of varied concentrations of TK1 RNA excited at λ = 535 nm (0–120 nM; c2). A good relationship (R² > 0.99) occurs between the fluorescence intensity and the concentration of BRCA1/TK1 RNA. The CDs–Au biosystem can detect BRCA1 RNA/DNA in the linear range of 4–120 nM with a detection limit of 1.5 nM and 2.1 nM, respectively, for RNA and DNA. Also, it can detect 10–120 nM TK1 RNA/DNA with a detection limit of 3.6 nM and 4.5 nM, respectively, for RNA and DNA. Replotted from Zhong et al.(202) Copyright: 2017 Elsevier.

The g-C₃N₄–CMNHs also show promising advantages in environmental sensing of biomacromolecules due to fast response and high detection sensitivity arising from their unique electrical and optical attributes (Section 2.1 and Figure 1) and abundant surface functional groups.(52,203–205) The suppressed charge recombination and improved photocurrent conversion efficiency make the g-C₃N₄–TiO₂-graphene PEC biosensors highly sensitive to pcDNA3-HBV in the linear range of 0.01 fM–20 nM with a 0.005 fM detection limit.(206) Such a biosensor also exhibits a high degree of selectivity (no obvious interferences with presence of pcDNA3, pcDNA3-His, pCMV5, pCMV-N-HA, and pCMV-C-HA), stability (for 14 days), and reproducibility (relative standard deviation = 2.3–4.5%).(206) Owing to the novel exciton–plasmon interactions in the p–n heterojunction, enhanced resonance energy transfer and photocurrent, and tunable signal change modulation, the g-C₃N₄–CdS@Au–Ag photoelectrochemical biosensor can trace sub-fM level (0.05 fM) microRNA-21 in complex biological samples with good specificity, reproducibility, and stability.(207) A similar g-C₃N₄–CdS photoelectrochemical immunosensor is also reported to exhibit a wide linear detection range (0.01–10 nM) and a low detection limit (3.53 pM) for N⁶-methyladenosine (m⁶A; methylated RNA) for the blood serum from breast cancer patients (SI Figure S2).(208) The dynamic monitoring of the m⁶A methylated RNA expression in vivo provides the g-C₃N₄–CdS-based biosensors with the capability of early cancer detection abilities.

3.3.2. In Situ Nanoremediation

Restoration of contaminated sites remains to be challenging due to inefficiency in contaminant removal alongside with high cost of conventional remediation technologies, for example, the U.S. Environmental Protection Agency estimates that approximately US$209 billion is needed to clean up 294 000 US contaminated sites from 2004–2033.(209) Currently, nanoscale zerovalent iron (NZVI) is the only nanomaterial that has been used in pilot- and field-scale demonstrations for in situ nanoremediation purposes.(210) However, the high aggregation propensity of NZVI (generally ≤1 m transport distance from injection point) and its lack of selectivity toward contaminants greatly limit its remedial performance for contaminated site remediation.(210) Because CMNHs may provide improved stability, potentially longer travel distances, and an ability to remove multiple contaminants effectively and simultaneously (Section 3.2 and SI Tables S1–S4), the next-generation CMNHs hold a potential for in situ nanoremediation.(211) Compared to the large volume of literature on CMNHs’ applications in energy harvesting and water treatment fields discussed earlier, no research has been reported to explore the promising applications of CMNHs for contaminated site remediation. Because aggregation and transport propensities of NMs (e.g., NZVI) dictate their performance in contaminated site remediation, this section focuses on the aggregation and transport of CMNHs in aquatic environments to direct the development of next-generation multifunctional CMNHs for in situ nanoremediation.

3.3.2.1. Aggregation of CMNHs

Hua et al.(212) first examined aggregation of graphene-TiO₂ in water under environmentally relevant pH (4–10) and salt type (0–200 mM NaCl and 0–8 mM CaCl₂). The nanohybrids display Derjaguin–Landau–Verwey–Overbeek (DLVO)- and Schulze–Hardy-type aggregation,for example, greater aggregation occurs at lower pH, higher ionic strengths, or with presence of Ca²⁺ (vs Na⁺). Our recent findings also demonstrate that DLVO theory and Schulze–Hardy rule well predict the aggregation behaviors of RGO–Ag, RGO–Fe₃O₄, and RGO–Ag–Fe₃O₄ nanohybrids in NaCl and CaCl₂.(213) Das et al.(214) probed the “part-whole” question of nanohybrid aggregation using CNT–TiO₂, which is a function of MLC (C/Ti molar ratio = 1:0.1, 1:0.05, and 1:0.033). The aggregation of CNTs–TiO₂ increases with increasing MLC (aggregation order: 1:0.033 < 1:0.05 < 1:0.1), which is inconsistent with DLVO theory’s prediction because the negative zeta-potential of CNT–TiO₂ follows the order of: 1:0.033 < 1:0.05 < 1:0.1 (electrostatic repulsion is the greatest for 1:0.1).(214) The authors explained that MLC-dominated properties such as fractal dimension and asphericity, charge heterogeneity, and surface roughness likely result in aggregation behavior of nanohybrids that cannot be captured by that of its parts.(214) These findings highlight the significance of pH and MLC in dominating CMNHs’ aggregation in aqueous solutions.

pH controls electrostatic double layer interactions, and thus NMs’ aggregation. NMs and CMNHs are expected to be stable when solution pH is far away from their pH_PZC (pH of point of zero charge); but tend to aggregate at their pH_PZC where net surface charge approaches zero. The pH_PZC of CNTs, GFNs, CDs, and g-C₃N₄ is reported at 3–4,(215,216) 2.5–3,(72,217–219) 2.0–2.5,(220) and 4–5,(221,222) respectively (Figure 7). Compared to CNMs, MNPs have higher pH_PZC, for example, Al₂O₃ (pH_PZC 8–10), Fe_xO_y (pH_PZC 7–9.5), TiO₂ (pH_PZC 6–7.8), and ZnO (pH_PZC 7.5–10.2) (Figure 7).(223) Hybridizing less negatively charged MNPs with CNMs causes a shift of CNMs’ pH_PZC toward a higher pH (Figure 7); for example, the pH_PZC of GO–MnFe₂O₄ nanohybrids (pH_PZC 4.85)(224) is higher than that of GO (pH_PZC 2.5–3).(219) Moreover, the pH_PZC of CMNHs shifts toward a higher pH with increasing MLCs. For example, pH_PZC of GO–TiO₂ follows the order: 4.1 > 4.0 > 3.5 > 3.2 > 3.0 for 1.0, 1.4, 2.9, 3.3, and 6.0 wt % GO-loaded nanohybrids, respectively.(225) Therefore, pH and MLC codetermine CMNHs’ aggregation via controlling their surface charges in aqueous suspensions.

Figure 7.

Reported pH of point of zero charge (pH_PZC) values of CNMs (marked within the red ellipse), CMNHs (marked within the black ellipse), and MNPs (marked within the pink ellipse). Starting from the left x-axis, the first four columns within the red ellipse denote: CNTs (3–4),(215,216) GFNs (2.5–3),(72,217–219) CDs (2–2.5),(220) and g-C₃N₄ (4–5),(221,222) respectively. The middle 12 dots/columns within the black ellipse denote: RGO–Fe₃O₄ (3.7),(257) GO–MnFe₂O₄ (4.85),(224) RGO–TiO₂ (5.2),(212) g-C₃N₄–BiOCl–Cu₂O–Fe₃O₄ (5.3),(258) graphene-Co₃O₄–Fe₂O₃ (5.3),(259) CDs–CeZrO₂ (5.8),(260) GO–Co₃O₄–Au (5.95),(261) RGO–Fe₃O₄ (5–6),(213) GO–Al–Fe (6.58),(262) RGO–Fe–Mn (7.47–7.56),(218) CDs-γ–Al₂O₃ (7.6),(260) and RGO–Zn-Fe (7.8)(263) nanohybrids, respectively. The last six columns within the pink ellipse denote: TiO₂ (6–7.8), Cu₂O (8–8.3), Fe_xO_y (7–9.5), Al₂O₃ (8–10), ZnO (7.5–10.2), and Zn–Fe oxides (8.8–10.4), respectively.(223) The numbers shown in the parentheses are the pH_PZC values of the materials.

Besides MLC, other newly emerging attributes of CMNHs (Section 2) also impact their aggregation. The separation distance between 2D g-C₃N₄ nanosheets is enhanced upon surface-anchoring of MNPs, which increases physical separation (also for RGO–Fe₃O₄; Figure 5). Increased separation distance weakens the van der Waals (vdW) attractions between g-C₃N₄-metal nanohybrids (vs individual g-C₃N₄), as demonstrated by higher dispersion of g-C₃N₄ and TiO₂ in the g-C₃N₄–TiO₂ suspension.(226) However, the localized vdW interaction at the CMIs is likely strengthened, due to the coupling of new built-in electric field and local electromagnetic field (Figure 1a). The coupling becomes more pronounced when magnetic MNPs (Fe_xO_y) are introduced, since magnetism synergistically intensifies the coupling and yields large CMNHs aggregates. For example, even at a high dispersant concentration (2 wt % carboxymethylcellulose), the hydrodynamic diameter (D_H) is much larger for the magnetic CNTs–Fe₃O₄ vs CNTs (3264 nm vs 3052 nm; ΔD_H = 212 nm), given that the anchored Fe₃O₄ is only 20–30 nm.(211) Therefore, the overall and localized vdW interactions collectively determine the aggregation of CMNHs in aqueous solution. Surface roughness,(214) charge heterogeneity,(214) dimensionality, and anisotropy are also likely to influence CMNHs aggregation. Diffusion- or reaction-limited cluster aggregation (DLCA or RLCA)(227) occurs, depending on the aggregation state (fast or slow) and dimensionality of particles. The aggregates are observed to be monodispersed in the DLCA regime, but become more compact in the RLCA regime.(227) Higher dimensional CMNHs call in the DLCA regime of aggregation, forming more compacted clusters and thus minimizing overall system entropy.(228–230) Surface defects including oxyanion functional groups introduced during CMNHs synthesis (e.g., chemical oxidation and ultrasonication) also influence CMNHs aggregation in aqueous suspension via altering electrostatic double layer (EDL) repulsive interactions.(23,231)

3.3.2.2. Transport of CMNHs

Transport capability of NMs determines their efficacy in remediating contaminated sites.(211) The transport of CNTs–Fe₃O₄ in laboratory-scale sand columns is recently reported.(211) The as-synthesized CNTs–Fe₃O₄ nanohybrids are highly aggregated due to strong hydrophobic, localized vdW, and magnetic attractions, but 2 wt % carboxymethylcellulose can effectively disperse CNTs–Fe₃O₄ particles by electrosteric repulsions. A novel transport feature was characterized by an initial lower effluent peak, followed by a sharp, higher effluent peak, probably due to the interplay between the variability of fluid viscosity (water and viscous carboxymethylcellulose) and size-selective retention(232) of CNTs–Fe₃O₄. The predicted maximum transport distance of CNTs–Fe₃O₄ using the Tufenkji-Elimelech model(233) ranges between 0.38–46 m, supporting the feasibility of applying the magnetically recyclable CNTs–Fe₃O₄ for in situ nanoremediation.(211) Our recent modeling efforts(234) reveal that conventional colloid transport model can capture the transport and retention of RGO–Fe₃O₄, RGO–TiO₂, and RGO–ZnO nanohybrids under a range of NaCl, CaCl₂, and NOM concentrations. Possible transport scenarios of the RGO–metal nanohybrids are forecasted via inverse fitting under environmentally relevant physicochemical conditions (flow velocity, porosity, and collector size) using Hydrus-1D software.(235)

The altered and newly emerging attributes affecting CMNHs’ aggregation (described above) also influence their transport in porous media. Other attributes that may alter CMNHs transport are discussed below. The small (~5.5 nm) negatively charged CDs (−21.2 to −38.2 mV in 1 mM NaCl at pH 6–9) show high mobility in sand columns even at very high ionic strengths (>50% breakthrough in 700 mM NaCl at pH 6).(236) Thus, the highly mobile CDs will enhance the mobility of CDs–CMNHs, when CDs are anchored onto MNPs surfaces (SI Figure S1e,f). In terms of potential retention mechanisms in porous media, straining likely dominates CNTs– and RGO–CMNHs retention as large CNTs–Fe₃O₄, RGO–Fe₃O₄, RGO–ZnO, and RGO–NZVI (≥1 μm) aggregates are frequently found at environmentally relevant conditions.(211,234) Straining has a dynamic role during the transport of CMNHs in porous media, since it progressively narrows down the pore-throat, and thus enhances subsequent particle retention, particularly near the column inlet, as has been demonstrated by the hyperexponential retention profiles in the transport studies of parent NMs and CMNHs.(211,234) More systematic studies are necessary to understand aggregation and transport of CMNHs (particularly g-C₃N₄ and CDs–CMNHs) in aquatic environments for their effective use as in situ nanoremediation agents.

4. Challenges and Perspectives

The advantages of CMNHs in harnessing solar energy for H₂ and O₂ evolution and CO₂ conversion stem from their appealing light harvesting capability, photocatalytic activity, stability, and selectivity, which are dependent on material type, composition (MLC), structure, crystallinity, morphology, dimensionality, and size that tailor electronic, optical, and band structure, and ultimately the charge transfer properties of the nanoheterostructures. However, fundamental knowledge on photoinduced electron–hole pairs, electron separation, and charge transfer dynamics at CMIs in the nanoheterostructures remains largely unexplored. This is particularly true for more complicated Z-scheme and MOFs-based CMNHs that exhibit high photocatalytic reactivity. Another key issue associated with Z-scheme and MOFs systems relates to the substantial energy loss and thus the low quantum yield during electron transfer processes.(237) State-of-the-art in situ characterizations like atom probe tomography, ion scattering, EPR and photoluminescence spectroscopy, and X-ray absorption spectroscopy (XAS) measurements combined with theoretical simulations (electronic structure modeling and first-principles DFT) are vital for unravelling electron–hole pair transfer pathways and charge cascading processes at molecular and atomic levels. This knowledge will enable designing more efficient, targeted, and economically feasible CMNHs with higher quantum efficiency by optimizing utilization of full-spectrum sunlight. To this end, prioritizing the development of g-C₃N₄–CDs–MNPs is necessitated owing to the excellent full-use of sunlight by CDs and their abundant reactive sites toward hybridization with other materials.

Single-atom catalysis-based, dual-/multiple-reaction-centered, and double Z-scheme CMNHs have attracted significant interests due to their ultrahigh photocatalytic activity and stability that can be harnessed for energy harvesting, particularly for degrading recalcitrant contaminants (perfluorochemicals). Taking single-atom catalysts as an example: 2D GFNs and g-C₃N₄ offer ample supporting sites for accommodating single-atom metals with prefect dispersion.(238) Nonetheless, key thermodynamic parameters affecting the photocatalytic activity and quantum yield including charge-carrier mobility time, diffusion length, and lifetime are currently unknown. Coupling in situ microscopic (subangstrom-resolution aberration-corrected high-angle annular dark-field scanning transmission microscope), spectroscopic (X-ray absorption fine structure), and advanced modeling (DFT) measurements can be valuable for probing the oxidation state, bonding structure, and coordination environment of single-atom in the nanoheterojunctions. This information will direct the development of next-generation CMNHs-based photocatalysts with maximized metal-catalytic reactivity toward recalcitrant contaminant degradation. MLC should be considered for fabricating CMNHs to achieve optimum performances.

While CNTs– and GFNs–CMNHs underperform in H₂O splitting and CO₂ conversion compared to g-C₃N₄- and CDs–CMNHs, the large SSA, rich surface-reactive sites and defects, and remarkable electron transfer properties render them as powerful candidates for contaminant removal via adsorption and catalysis due (partly) to accelerated electron–hole pair separation for MNPs.(67) Additionally, CNTs and GFNs (e.g., RGO) acting as “electron-transport-bridge”(239) can facilitate the construction of Z-scheme g-C₃N₄-RGO–MNPs, resulting in well-contacted CMIs, short charge-transfer distance, and superior photocatalytic performances. Introduction order of parent materials, synthetic conditions, and nanoscale assembly can tune CMNHs’ properties (morphology, band alignment, layer arrangement, defect density, vacancy, and porosity) that affect the light utilization and photocatalytic efficiencies. Designing rational nanoheterojunctions by orderly assembling g-C₃N₄, GFNs/CNTs, and MNPs with well-matched energy levels warrants further exploration toward more efficient H₂O splitting, CO₂ conversion, contaminant removal, and microbial disinfection.

Tremendous progress has been made in advancing CMNHs-modified membranes for water treatment and desalination using reverse osmosis, forward osmosis, microfiltration, ultrafiltration, and nanofiltration technologies. The physicochemical properties of CMNHs along with their loading amount, assembling strategy, dispersion state, and orientation into the composite membrane modulate the mechanical stability, contaminant removal efficiency, solute rejection and antifouling ability, selectivity, and reusability of the membrane. Understanding the structure–property relationships of CMNHs in the composite membranes is pivotal for optimizing their performances, but how these parameters (e.g., loading amount and assembling strategy) tailor membrane structures, properties, and functions remains poorly understood. Systematic studies are essential to understanding the structure–property relationships for developing next-generation advanced membranes. Furthermore, pilot- or field-scale testings must be conducted to evaluate the reliability of long-term use of CMNHs-modified membranes for future industrial applications.

CMNHs (particularly CDs- and g-C₃N₄–CMNHs) show the benefits to sensitively and selectively monitor multifarious environmental contaminants and biomacromolecules (RNA/DNA) at ultratrace (pM–fM) levels due to their unique fluorescence and luminescence attributes. However, explanations on fluorometric and luminometric mechanisms for CMNHs-based sensors are rather empirical and far from clear because size, structure, crystallinity, morphology, surface states and defects, MLC, and hybridization mode all likely impact the absorbance and excitation/emission processes involving fluorescence/luminescence. New techniques like XPS, EPR, electrochemical impedance, and surface-enhanced Raman spectroscopy, two-photon fluorescence imaging, and up-converted fluorescence imaging coupled with theoretical modeling can facilitate a deeper understanding of the fluorometric and luminometric mechanisms toward designing the next-generation (bio)sensors for a broader and more effective environmental and biological applications (e.g., cancer diagnosis).

While DLVO theory and Schulze-Hardy rule can be used to qualitatively or semiquantitatively describe the overall aggregation behaviors of certain CMNHs (e.g., CNT–TiO₂,(214) RGO–Ag,(213) RGO–Ag–Fe₃O₄,(213) and RGO–-TiO₂(212)), challenges still remain for quantitative description of the aggregation kinetics and morphology/structure evolution of CMNHs aggregates, particularly when taking into account the newly emerging attributes of CMNHs (e.g., shape, dimensionality, MLC, and anisotropy). State-of-the-art in situ real-time identifications may enable more accurate and direct observations of CMNHs aggregation (e.g., growth kinetics and morphology/structure evolution). For example, using direct-imaging cryogenic transmission and scanning electron microscopy (cryo-TEM/SEM) with a nanoscale resolution, Kleinerman et al.(240) observed various stages of liquid crystalline phase evolution and domain morphology development of CNTs in aqueous suspensions, which is controlled by the aspect ratio, diameter, and purity degree of CNTs. Similarly, polarized light microscopy (PLM) allows direct imaging of CNTs(241) and GFNs(242) at the micrometer scale. Particularly, in situ real-time imaging with atomic force microscopy (AFM)(243) has the power to directly observe anisotropic aggregation kinetics of CMNHs. Small-angle scattering (SAS) technique can provide particle size distribution, dispersity (mono- and poly dispersity), and structural information on dimensionality of NPs in aqueous suspensions. Consequently, combining cryo-TEM/SEM, PLM, real-time AFM and SAS techniques along with DLVO theory calculations using surface element integration (SEI) technique(244–246) (SI Text S1) can advance a mechanistic understanding of CMNHs aggregation (growth kinetics and morphology/structure evolution) in aqueous systems.

The aforementioned future research directions can facilitate accurate descriptions of CMNHs aggregation. This can pave the way for more reliable description and prediction of CMNHs transport in environmental media. Nonetheless, natural soil and sediment matrices are highly complex and heterogeneous in terms of compositions, structures, properties, and functions. For example, physical and hydrodynamic properties of porous media such as pore structure, particle size distribution, porosity, preferential flow/pathway, geometry, connectivity, and tortuosity can strongly affect the transport of NPs (e.g., CMNHs).(247) New approaches of coupling mass transport measurements of CMNHs at the mesoscopic scale with direct measurements of physical and hydrodynamic proeprties of the porous media (e.g., by using a 3D X-ray computed tomography (CT) technique)(247,248) are needed to unravel the pore-scale processes.(249) Particularly, the nondestructive 3D X-ray CT technique can enable accurate characterization of pore network structure with regard to particle size distribution, porosity, and tortuosity of the porous media, all of which can influence CMNHs transport and retention in environmental media.

Mathematical modeling is also necessary to simulate and predict the transport and retention of CMNHs in environmental media. This is particularly important when the hydrodynamic size of CMNHs aproaches the pore size/throat of porous media in which straining and ripening effects on colloid retention become significant.(232) The modified MODFLOW-,(250) continuum-,(251) and artificial neutral network-based models(252) are good candidates, since these model formulas mechanistically account for colloid retention mechanisms including particle aggregation, straining, ripening, site-blocking, and size-exclusion. Alternatively, physically based mechanistic models like two-region, two-domain, and dual-permeability models can explicitly account for preferential flow and local/bulk transport in porous media at different scales (e.g., pore, representative elementary volume, and field scales).(247) To better correlate CMNHs’ retention mechanisms with their newly emerging attributes (e.g., morphology, dimensionality, and MLC), porous media properties (e.g., physical and hydrodynamic properties), and environmental conditions (e.g., water content, water chemistry, and pore-water velocity), machine learning technique (i.e., decision tree) can be advantageous for identifying factors influencing CMNHs transport and retention in porous media.(253,254) This may facilitate the development of quantitative relationships between CMNHs mobility in porous media and their physicochemical properties using X-ray CT techniques, mathematical modeling, and machine learning techniques.

Optimizing the transport capability of CMNHs for in situ nanoremediation remains challenging, as these CMNHs (e.g., CNTs–Fe₃O₄) are likely highly aggregated even at low ionic strengths.(211) Stabilizers like surfactants, polymers, biomacromolecules, or their combinations should be examined for identifying their contributions in enhancing CMNHs transport in soil and groundwater. Unlike conventional colloid transport studies that neglects fluid viscosity variability, fluid viscosity should be monitored continuously, especially with presence of high concentrations of stabilizers whose viscosity could be several-orders of magnitude higher than that of water.(211) Monitoring fluid viscosity helps unravelling CMNHs transport mechanisms which, in turn, enables more accurate estimation of operational time and remediation efficiency for in situ nanoremediation.

Practical application of noble MNPs can be impeded by the high price and low abundance. Fortunately, inexpensive synthesis of CNMs and noble-metal-free MNPs has been scaled-up to kilogram scale.(255,256) Moreover, most highly efficient CMNHs are obtained via in situ strategy in which affordable metal salts are used as precursors. One can thus prepare large-scale CMNHs at reasonable price for real-world applications such as wastewater treatment and in situ nanoremediation. Benefiting from the versatility of CNMs, inexpensive magnetic MNPs (Fe_xO_y) can be facilely and strongly anchored onto CMNHs to render recyclable and reusable benefits, which lowers down the usage cost and minimizes potential environmental risks of CMNHs (e.g., toxicity from MNPs like AgNPs). While research, development, and application of CMNHs are still in the infancy, the invigorating properties/multifunctionalities of the new type of material can afford fascinating new opportunities for scientists and engineers to devote more research activities into this rapidly growing field. Particular focus should be devoted to developing next-generation environmental benign CMNHs. We are most optimiztic that more revolutionary applications of CMNHs along with the advancements of both fundamental physics and chemistry and practical techniques will expand the horizon and open new doors for simultaneously achieving better human life, cleaner energy, and environmental sustainability.

Abbreviations

CNTs

carbon nanotubes

GFNs

graphene family nanomaterials

CDs

carbon dots

g-C₃N₄

graphitic carbon nitride

CMNHs

carbon–metal nanohybrids

EWE

energy–water–environment

CNMs

carbonaceous nanomaterials

MNPs

metal nanoparticles

CB

conduction band

CMI

carbon–metal interface

DFT

density functional theory

EDCs

endocrine disrupting compounds

EPR

electron paramagnetic resonance

GO

graphene oxide

HRTEM

high-resolution transmission electron microscope

LDH

layered double hydroxide

LSPR

localized surface plasmon resonance

MLC

metal loading capacity

MOF

metal organic framework

PPCPs

pharmaceuticals and personal care products

PRET

plasmon resonance energy transfer

RGO

reduced graphene oxide

ROS

reactive oxygen species

SSA

specific surface area

UV–vis-NIR

ultraviolet–visible-near-infrared

VB

valence band

XAS

X-ray absorption spectroscopy

XPS

X-ray photoelectron microscopy