Renaissance of elemental phosphorus materials: properties, synthesis, and applications in sustainable energy and environment

Bulk black-P exists in three crystalline phases, orthorhombic, simple cubic, and rhombohedral, and the most intriguing is the semiconducting orthorhombic phase belonging to the space group Cmca(64) with eight phosphorus atoms per cell.¹⁰⁹ Orthorhombic black-P can change its structure reversibly at a high pressure morphing into the denser rhombohedral or simple cubic forms.¹¹⁰ However, these two forms are unstable under normal conditions and are not suitable for use in practice. In this review, unless specified, black-P refers to the orthorhombic phase which is the focus of most research. The lattice constants a, b, and c of black-P crystals, determined by time-of-flight neutron powder diffractometry by Cartz et al., are 4.374(2), 3.3133(9), and 10.473(5) Å, respectively.¹¹¹ Owing to orbital hybridization, black-P has a nonplanar bilayer structure with two different bond angles (α = 96.34° and β = 103.09°). From the side view, there are two types of P–P bonds, a short bond (2.224 Å) and a long bond (2.244 Å) connecting two close atoms in different sublayers.¹¹² This results in anisotropy between the zigzag (y) and armchair (x) directions and the crystalline symmetry of single-layered black-P is two-fold rotational (D_2h), which is far less than the six-fold rotational (D_6h) symmetry observed in single-layered graphene, which has a flat hexagonal lattice.⁷¹ The reduction in crystalline symmetry is fundamental to the various chemical and physical properties. For instance, the band structure of black-P exhibits anisotropy irrespective of the thickness, giving rise to in-plane anisotropic properties in charge transport, thermal conductivity, and optics.^113,114 The interlayers are held together by vdW forces with a layer spacing of 3.214–3.729 Å,¹¹⁵ depending on different stacking orders. The relatively weak vdW forces make black-P suitable for exfoliation to prepare phosphorene and allow for property turnability.

The atomic structure of the blue-P monolayer, as discussed in Section 2, can be viewed as the flipping of specific atoms, transforming the armchair ridges into zigzag ridges in the black-P monolayer without altering the local bond angles. This leads to a non-planar but isotropic structure for single-layered blue-P, a significant difference from the anisotropic structure of its black counterpart. Zhu et al. noted the possibility of a new phosphorus phase resulting from a certain dislocation of black-P and conducted ab initio calculations to investigate the basic properties of the allotrope.⁴⁹ They dubbed this new allotrope blue phosphorus (blue-P) with the color defined by its fundamental bandgap. Blue-P atoms are covalently bonded in the layer with a bond length of 2.27 Å, and a weak interlayer interaction of 6 meV per atom keeps the layered structure intact with an interlayer distance of 5.63 Å.⁴⁹ The optimized blue-P monolayer has a hexagonal unit cell with lattice vectors of || = || = 3.33 Å. The interlayer interaction shows little effects on the in-layer structure, causing only a slight shift in the lattice vector from 3.324 Å in bulk to 3.326 Å in the monolayer. The free-standing bulk crystals of blue-P do not exist, although single-layered blue-P has been reported to grow on several noble metals.⁵²

Green-P is a metastable EPM that combines features of armchair-ridged black-P and zigzag-ridged blue-P. Han et al. were the first to report the structural and electronic properties in silico.⁵⁰ Bulk green-P has a monoclinic C2/m structure with six atoms per unit cell and in-plane lattice constants of a = b = 7.13 Å. Green-P exhibits the greatest layer thickness (2.91 Å) among the three allotropes due to the presence of three slightly curved sublayers in each layer, compared to 2.11 Å for black phosphorene and 1.24 Å for blue phosphorene. All the atoms in green-P are three-fold bonded with in-plane lengths of 2.23 Å and out-of-plane bond lengths of 2.26 Å. The AB stacking order is energetically preferred and the interlayer distance is 2.16 Å. Although green-P has a shorter interlayer distance than black-P and blue-P, the interaction between green-P layers at 74 meV per atom is similar to that of Black-P (79 meV per atom), indicating its potential for mechanical exfoliation. Moreover, the buckled structure of green phosphorene aligns well with corrugated Au and Ag substrates, suggesting that epitaxial growth is favored with this allotrope.⁵⁰

The structure of type I amorphous red-P has been the subject of debate and two main models are currently being considered.²⁸ Model 1 postulates that it consists of chain-like pentagonal nanotubes resembling monoclinic violet-P,^118,119 while model 2 proposes a layered structure similar to that of orthorhombic black-P.^120–122 Model 2, which explains the alterations in the intermediate-range order in phosphorus layers based on growth factors,¹²⁰ has been widely adopted and accepted by many studies. However, several structural studies using Raman spectroscopy and XPS have found evidence of the presence of pentagonal tubes in the microstructure of amorphous red-P,^100,118 suggesting that model 1 may also play a role.

In recent years, innovative methodologies have been employed to determine the microstructure of amorphous red-P. Zhang et al.¹²³ discussed the four possible linear molecular structures of amorphous red-P that were previously proposed by chemists (Fig. 7a). These included the double triangle P₄ (1) proposed by Pauling and Simonetta in 1952,¹⁰⁸ the zigzag ladder P₄ (2) proposed by Salzmann et al.,¹²⁴ as well as [P8]P2[ (3) and [P10]P2[ (4) suggested by Häser and colleagues.^125,126 Through their single-molecule studies, they discovered that amorphous red-P is a straight-chain inorganic polymer with a zigzag ladder structure, which aligns more closely with structure (2). However, Deringer et al. adopted ML-based simulations to visualize the local structure and pressure-induced changes in amorphous red-P at the atomic scale.⁵⁸ They inferred that amorphous red-P is an open network of covalently connected phosphorus clusters. They analyzed the types and counts of cluster fragments, finding that P₅ rings, P2[P3]P2 and P3]P2[P3 clusters, and P₁₀ cages are the dominant building units, whereas P₃ rings and P₄ tetrahedra are much less abundant (Fig. 7b). The precise molecular structure of amorphous red-P still awaits verification through further computational and experimental investigation.

	Fig. 7 Structure and characterization of type I amorphous red-P, type II crystalline red-P, and violet-P11. (a) Possible linear molecular structures of amorphous red-P. Reproduced with permission from ref. 123. Copyright 2019 John Wiley and Sons. (b) Counts of relevant cluster fragments in the simulated model. Reproduced with permission from ref. 58. Copyright 2021 John Wiley and Sons. (c) SEM image of type II crystalline red-P. (d) SAED pattern of type II crystalline red-P with interplanar spacings. (e) XRD peaks marked with the same colour as spots in (d) represent a set of corresponding lattice planes with primary and secondary diffraction peaks. Reproduced with permission from ref. 129. Copyright 2021 Royal Society of Chemistry. (f) Crystal structure of violet-P11 viewed along two directions. Reproduced with permission from ref. 136. Copyright 2023 American Chemical Society.

Roth et al.³⁴ were the first to observe type II and III crystalline red-P using differential thermal analysis. In 1966, Rubenstein and Ryan¹²⁷ verified the presence of type II crystalline red-P by annealing in an inert atmosphere and matching the powder XRD patterns with Roth's findings. Nechaeva et al.¹²⁸ later produced type II crystalline red-P by chemical vapor deposition and identified the monoclinic structure with specific lattice parameters. Recently, Yan and colleagues have produced type II red-P crystals with a platelet morphology using a solution phase synthesis method (Fig. 7c).¹²⁹ The high crystallinity of the product is evident from the XRD results, with the most intense peak at 15.66° and a plane spacing of 5.66 Å, which is in close agreement with the results obtained by Roth and colleagues in 1947.^34,129 The nearly hexagonal symmetrical SAED pattern of the crystal suggests a monoclinic crystal system, and the plane spacing is consistent with the XRD pattern (Fig. 7d and e). However, so far, limited structural information is available for type III crystalline red-P. It is imperative to develop reliable methods for synthesizing type III crystalline red-P to facilitate the analysis of its crystal structure.

The structures of two other types of crystalline red-P have been studied.^23,28 Type IV fibrous-P and type V violet-P have been confirmed to consist of repeating polymeric tubes with a pentagonal segment arranged in a regular pattern of [P9]P2 [P8]P2[. The key difference between fibrous-P and violet-P lies in the manner the double tubes are connected. Violet-P consists of two interpenetrating tube systems that run perpendicular to each other, forming double stacks held together by vdW forces in the c direction. The double tubes in fibrous-P, on the other hand, are parallel and held together by vdW forces in both the a and c directions, making it a member of the unique 1D vdW crystal family with distinct bonding behavior and anisotropic physical properties.³⁶ The structural parameters of violet-P and fibrous-P are available. By using single-crystal XRD, Zhang et al.⁴⁸ have determined the lattice constants of violet-P crystals to be 9.210, 9.128, and 21.893 Å for the parameters a, b, and c, respectively. Meanwhile, Du et al.³⁶ have used XRD to show that the lattice constants of fibrous-P crystals are 12.198, 12.986, and 7.075 Å for a, b, and c, respectively, which are consistent with theoretical simulation results. Although both fibrous-P and violet-P have similar plate-like shape and color, they can be differentiated through mechanical stress tests, in which fibrous-P crystals break down into thin filaments, while violet-P crystals shatter into smaller platelets.⁸⁸ Furthermore, high-resolution TEM shows distinct morphological differences between the two crystalline phases.

Bulk crystals of violet-P can be peeled into few- or single-layered violet phosphorene by applying external forces. DFT calculation has been used to compute the binding energy per layer needed for complete exfoliation of the bulk crystal into monolayers. The binding energy of violet phosphorene is 0.35 J m⁻², which is comparable to that of graphene (0.34 J m⁻²) but lower than that of black phosphorene (0.40 J m⁻²) or blue phosphorene (0.42 J m⁻²).¹³⁰ This indicates that exfoliating violet phosphorene is easier than other types of phosphorene. In 2020, violet phosphorene was synthesized and characterized as a 2D nanobelt composed of tubular strands with an interlayer spacing of 11 Å in AB stacking.⁴⁸ Regarding fibrous phosphorene, Hu and Guo have confirmed that fibrous-P bulk crystals have a layered structure along the (001) direction and can be exfoliated experimentally.¹³¹ Each layer consists of four [P9]P2 [P8]P2[ polytubes with 84 phosphorus atoms held together by vdW interactions. The average interlayer distances along the (001) and (200) facets are 5.83 Å and 5.57 Å, respectively. The binding energy for fibrous phosphorene, calculated to be 0.88 J m⁻²,¹³² is at least two times greater than that of violet phosphorene, making exfoliation of fibrous-P more challenging. However, fibrous phosphorene is held together by vdW interactions in two directions (a and c) similar to selenene and tellurene^133–135 and can be fabricated by epitaxy or other methods.

Very recently, Cicirello and colleagues synthesized a new two-dimensional allotrope of violet phosphorus, dubbed violet-P₁₁, using a salt flux method in the presence of bismuth.¹³⁶ The crystal structure of this allotrope from different views is shown in Fig. 7f. The structure of violet-P₁₁ is similar to that of violet-P. However, violet-P₁₁ possesses a higher structural symmetry compared to that of violet-P. The asymmetric unit of violet-P₁₁ is entirely occupied by 11 distinct phosphorus atoms, while that of violet-P comprises 21 different phosphorus atoms. Single-crystal XRD has revealed that violet-P₁₁ crystallizes in the monoclinic space group C2/c(15), with unit cell parameters of 9.166(6), 9.121(6), and 21.803(14) Å for a, b, and c, respectively.¹³⁶ Violet-P₁₁ exhibits substantial stability in ambient air even after being exfoliated for at least an hour.

	Fig. 8 Electronic band structures of black, violet, fibrous, blue, and green phosphorus. (a) Layer dependence of bandgaps for black-P as calculated by various methods. Reproduced with permission from ref. 139. Copyright 2014 American Physical Society. (b) Spectroscopic measurement of the bandgap for black-P. Reproduced with permission from ref. 140. Copyright 2014 American Chemical Society. (c) Theoretical band structure and (d) the mapping of the Fermi surface of bulk fibrous-P. Reproduced with permission from ref. 36. Copyright 2021 Nature Publishing Group. (e and f) Layer dependence of quasiparticle bandgaps for few-layered blue-P. Reproduced with permission from ref. 142. Copyright 2021 American Physical Society. (g) Layer-dependent bandgaps and relative energy of violet-P and black-P. The electronic band structure of the violet-P monolayer (h) and bulk form (i). Reproduced with permission from ref. 130. Copyright 2016 American Chemical Society.

Fibrous-P and atomically layered blue-P have indirect bandgaps. Du et al. have plotted the electronic band structures of bulk fibrous-P and shown an indirect bandgap of approximately 1.57 eV (Fig. 8c).³⁶ The conduction band minimum (CBM) and valence band maximum (VBM) can be found around the L and Γ/Y points of the Brillouin zone, respectively. To correct for the underestimated value, they applied the modified Becke–Johnson functional, resulting in an indirect bandgap range of ∼1.80 to 2.11 eV. The 3D mapping of the Fermi surface reveals weak dispersion in the valence band (Fig. 8d), which is caused by electron confinement in one dimension. Lu et al. used the HSE06 hybrid functional and determined the indirect bandgap of fibrous-P bulk crystals and monolayers to be 2.25 and 2.46 eV, respectively.¹³² The bandgap of a fibrous-P bilayer is 2.34 eV, which is only slightly narrower than that of the monolayer, indicating that the number of stacked layers has minimal effects on the bandgap of fibrous-P. This makes fibrous-P based materials ideal for catalysis because bulk crystals can be directly used without the need for extra exfoliation or precise control over layer numbers. Zhou et al.¹⁴² have assessed the evolution of quasiparticle bandgaps in few-layered blue-P (Fig. 8e and f). The indirect bandgap diminishes as the layer count rises starting at 3.41 eV for the monolayer, reaching 1.09 eV for the penta-layer, and finally changing into a metallic bandgap in the bulk form. The bandgap calculated using PBE is significantly smaller than that determined using quasiparticle approximation, yet the degree of band dispersion remains largely unchanged. This suggests that the bandgap width of blue-P is larger than that of black-P and green-P, and bulk blue-P even exhibits metallic characteristics. However, whether bulk blue-P is semiconducting or metallic is controversial,^49,142,143 and further theoretical studies on the electrical structure of blue-P crystals are required. Experimentally, the bandgap of 4 × 4 blue phosphorene on Au(111) is measured to be approximately 1.10 eV by angle-resolved photoemission spectroscopy,¹⁴⁴ which is significantly lower than the theoretical bandgap of free-standing blue phosphorene. This results from the strong in-layer strain caused by the interaction between phosphorene and the Au(111) substrate,^52,145 which disturbs the lattice symmetry of free-standing blue-P, hence impacting the electrical characteristics.

Violet-P differs from the other four allotropes as it undergoes an indirect-to-direct bandgap cross-over when downscaling from the bulk to monolayers. This cross-over is also found in many semiconducting transition metal dichalcogenides (TMDs),^146–148 making it a useful way to tune the electronic properties of layered semiconductors. Schusteritsch et al.¹³⁰ have developed several calculation methods for determining the electronic band structure of Violet-P (Fig. 8g). The PBE method with dispersion corrections shows the bandgap of bulk violet-P to be 1.45 eV, while monolayer violet-P exhibits a direct bandgap of 1.76 eV at the Z point in the Brillouin zone (Fig. 8h–i).¹³⁰ A more accurate HSE(TS) hybrid calculation gives wider bandgaps of 2.17 and 2.52 eV for the violet-P bulk form and monolayer, respectively. UV/vis diffuse reflectance spectroscopy reveals an optical bandgap of 1.7 eV for violet-P crystal powder, similar to 1.77 eV obtained in photoluminescence of single-crystal violet-P.⁴⁸ The optical bandgap is narrower than the calculated value because it is measured on a substrate-supported sample rather than in vacuum.¹¹² Both fibrous-P and violet-P have similar bandgaps that are not greatly affected by the number of layers due to similarities in the atomic structures.⁸⁸ Further research is needed to fully explain why violet-P has an indirect-to-direct cross-over instead of an indirect bandgap like fibrous-P.

The bandgaps of the aforementioned EPMs are influenced by the number of layers, and the bandgap becomes smaller with increasing number of layers due to the influence of vdW interactions between the layers. As the layer number increases, the interlayer interaction becomes stronger to increase band dispersion and produce a narrower band gap.^11,149,150 Single- and few-layered phosphorene displays pronounced quantum confinement, and the valence and conduction bands mimic those of a quantum-well structure. However, in the bulk counterpart, energy dispersion becomes densely packed, causing the discrete energy bands to turn into quasi-continuous sub-bands.¹¹ In addition, the bandgaps of phosphorus allotropes are sensitive to electric fields,¹⁵¹ molecule adsorption,¹⁵² and tensile strain.¹⁵³ These external factors offer significant control over the bandgap, particularly in the single-and few-layered structures. As shown in Fig. 9, EPMs with tunable and layer-dependent bandgaps cover a wide range of energy and exhibit strong interaction with electromagnetic waves across the mid-infrared, near-infrared, visible, and near-ultraviolet frequencies, making them suitable for various applications such as thermal imaging (0.1–1.0 eV), photovoltaics/photocatalysis (1.2–2.2 eV), and blue LED techniques (2.5–2.9 eV).

	Fig. 9 Comparison of the bandgaps of different EPMs and potential applications. The horizontal bars displaying a range of bandgap values indicate that the bandgap can be adjusted by changing the number of layers. The bandgap of green-P is currently based on theoretical calculations. Bandgap values are taken from ref. 50, 130, 132, 139, and 143.

Black-P and violet-P exhibit intrinsic p-type conductivity with a positive Hall coefficient, indicating that conductivity is primarily driven by holes.^37,69 Based on Qiao's phonon-limited scattering model, the hole and electron mobility of black-P can be extended to 10000–26000 cm² V⁻¹ s⁻¹ and 1100–1140 cm² V⁻¹ s⁻¹, respectively, which are comparable to those of graphene and greater than those of 2D TMDs.⁶⁹ Black-P shows in-plane anisotropic carrier mobility with electron mobility along the x-axis being greater than that along the y-axis. Regarding the hole mobility, the preference for migration along the x or y-axis depends on the number of layers. In the few-layered black-P, the hole mobility along the y-axis is about half that along the x-axis. In a monolayer, the hole mobility along the y-axis is surprisingly 16–38 times greater than along the x-axis, that is, 10000–26000 cm² V⁻¹ s⁻¹versus 640–700 cm² V⁻¹ s⁻¹, indicating that the y-axis is the direction of enhanced hole conductivity.⁶⁹ Despite the carriers being relatively heavy (6.35m₀), the high mobility in a monolayer is due to the low deformation potential (0.15 ± 0.03 eV).⁶⁹ Several people have designed black-P based field-effect transistors (FETs) and examined the charge carrier mobility.^{45,46,109,157} Violet-P has been predicted to have hole mobilities between 3000–7000 cm² V⁻¹ s⁻¹ based on theoretical calculations.¹³⁰ Applying PBE without dispersion corrections, the effective masses of electrons and holes are found to be greater along the y-axis and the effective mass of electrons along the x-axis (0.69 ± 0.01m₀) is the smallest. The hole deformation potential along the y-axis is nearly 7 times smaller than the value along the x-axis, that is, 0.18 ± 0.05 eV versus 1.26 ± 0.06 eV, resulting in a significant hole mobility of ∼7000 cm² V⁻¹ s⁻¹ along the y-axis. However, when PBE with dispersion corrections (PBE + TS) is employed, the hole mobility is reduced to ∼3000 cm² V⁻¹ s⁻¹ as PBE + TS calculates a magnitude of the deformation potential that is twice that of PBE. Recently, Ricciardulli et al. have created a FET using exfoliated violet-P thin films and found that the hole mobility of the film is 2.25 cm² V⁻¹ s⁻¹ at room temperature.³⁷ Although charge transport in films may be impeded by inter-flake resistances and structural defects caused by the exfoliation process,¹⁵⁸ this work experimentally demonstrates the potential of violet-P as a p-type semiconductor.

Dominated by high electron mobilities, fibrous-P and green-P are potential materials for n-type devices.^50,132 In the fibrous-P monolayer, the effective masses of electrons (holes) along the x- and y-axis are calculated to be 0.41 (2.95) and 0.56 (0.44)m₀, respectively.¹³² The deformation potential (E₁) and elastic modulus (C_2D) are highly anisotropic in the plane. The calculated deformation potentials of electrons along the x- and y-axis are approximately 3.98 and 0.64 eV, respectively, with an anisotropy ratio greater than 6. C_{2D_y} is almost nine times larger than C_{2D_x}. This results in a significant electron mobility of ∼17700 cm² V⁻¹ s⁻¹ along the y-axis, which is comparable to that of other layered semiconductors like black phosphorene,⁶⁹ violet phosphorene,¹³⁰ and Al₂C monolayers.¹⁵⁹ However, the FET made from fibrous-P nanoribbons has a lower hole mobility (236.7 cm² V⁻¹ s⁻¹),¹⁶⁰ which is far less than expected. This may be due to the high contact resistance between the channel materials and electrodes. High mobility can be improved by modifying the contact electrodes or using high-k materials as the gate layer.^156,160 In green phosphorene, the effective masses of electrons (holes) along the x- and y-axis are predicted to be 0.274 (0.166) and 0.162 (1.333)m₀, respectively.⁵⁰ This suggests that the hole mobility is highly anisotropic, while the electron mobility is nearly isotropic. For a carrier concentration of 5 × 10¹² cm⁻², the electron mobility is determined to be around 200 cm² V⁻¹ s⁻¹ at 300 K, which is higher than the hole mobility (∼60 cm² V⁻¹ s⁻¹). This relatively high electron mobility is a result of lower scattering rates near the CBM.

Zhuang et al. have investigated the carrier mobility characteristics of 4 × 4 blue phosphorene on Au(111) theoretically.¹⁴⁴ The effective mass of holes is slightly greater along the x-axis than along the y-axis (2.231m₀versus 1.587m₀), whereas the effective mass of electrons along both axes is very close and small (∼0.130m₀). This small effective mass leads to high electron mobility, which can reach up to 11429 cm² V⁻¹ s⁻¹. The limited anisotropy in carrier mobilities of 4 × 4 blue phosphorene on Au(111) is related to the more symmetrical atomic structure, but even minor changes in the structure of layered materials can produce significant variations in the electronic properties. The strong in-layer strain caused by the blue-P–Au(111) interaction greatly influences the electron mobility.^144,161 Therefore, they have also calculated the carrier mobilities of 4 × 4 blue phosphorene in the absence of tensile strain.¹⁴⁴ The effective mass of electrons increases significantly in free-standing blue phosphorene, which greatly reduces the electron mobility by one or two orders of magnitude.

So far, studies on the carrier mobility of elemental phosphorus semiconductors, excluding black-P, have mainly been theoretical. In general, theoretical calculations for the room-temperature carrier mobilities (μ_2D) of layered materials use an acoustic phonon limited method and the following expression:^69,162

To address the requirement for high-pressure conditions in black-P crystal production, mercury flux or bismuth flux was introduced as supplementary catalysts.^205–207 In 1955, Krebs and his colleagues reported that under high-temperature conditions (370–410 °C), addition of 30 to 40 atomic percent of mercury would initiate the transition of white-P into black-P.²⁰⁵ The reaction was found to take a few days to complete, yielding Black-P crystals as the final product. However, the use of mercury or bismuth as catalysts is associated with several drawbacks, including mercury contamination, bismuth toxicity, prolonged reaction times, and limited yield, making the mercury or bismuth flux method far from ideal. Despite the limitations, this method of introducing additional catalysts seems to provide insights into the later-developed chemical vapor transport (CVT) method, which efficiently produces black-P crystals at moderate temperature and pressure through the addition of transfer agents. CVT refers to a chemical reaction that transforms a condensed phase into a gaseous phase in the presence of a transfer agent, followed by deposition elsewhere and eventual formation of a crystalline structure.²⁰⁸ In 2007, Nilges and colleagues introduced a pioneering CVT-based approach for the preparation of black-P crystals at low pressure in evacuated silica ampules.⁴² By heating red-P with a mixture of Au, Sn, and SnI₄ as transport agents, they were able to produce black-P crystals, where a sample batch was grown on the surface of the Au₃SnP₇/AuSn materials (Fig. 14a). One year later, this research group improved the previous reaction system by using the AuSn alloy instead of Au and Sn metals.⁴³ This change, along with the optimal temperature program, reduced the crystal growth time to 32.5 hours and increased the crystal size to 1.5 cm in diameter. However, the cost of synthesis remained high due to the use of noble metals. In 2014, they further refined the synthetic procedure by eliminating expensive Au and substituting it with Sn/SnI₄ as transfer agents, which significantly reduced costs and the number of impurities.⁴⁴ Building on these advances, Nilges et al. recently utilized DFT calculations to probe the formation mechanism of black phosphorus and phosphorene.²⁰⁹ They discovered that the gaseous species P₄ and SnI₂ play a crucial role in the gas-phase growth of black phosphorus and phosphorene, with SnI₂ directly inserting into the P–P bonds of P₄. This work not only contributes to the optimization of the gas-phase growth system for black-P crystals but also aids in understanding the principles of epitaxial growth of phosphorene on substrates, which will be discussed further below.

	Fig. 14 CVT-based methods for black-P bulk crystal preparation. (a) Photographs of the obtained black-P crystals. Reproduced with permission from ref. 42. Copyright 2007 American Chemical Society. (b) A comparison of the yield and purity of black-P synthesized by different methods, including high-pressure high temperature (HPHT) methods, sonochemistry, flux methods, and CVT methods. (c) Schemes of the LDT method and (d) SDT method to grow black-P crystals and their corresponding photographs of the resulting products. Reproduced with permission from ref. 210. Copyright 2020 Elsevier.

In contrast to Nilges' CVT growth approach,^42,43 which requires a temperature gradient for long-distance transport (LDT), Liu's group have devised an efficient short-distance transport (SDT) growth method that utilizes a uniform temperature.²¹⁰ As shown in Fig. 14c and d, a substantial number of black-P crystals formed during the SDT process, with 98% of red-P being converted to black-P, while only a limited number of black-P crystals formed during the LDT process. The comparison of the yield and purity of black-P synthesized by different methods including high-pressure high temperature (HPHT) methods, sonochemistry, flux methods, and CVT methods is shown in Fig. 14b, and the results indicate that this work has achieved the highest yield and purity of black-P crystals among the compared methods. Two key factors for obtaining high black-P production yields are the short-distance transport and uniform temperature.²¹⁰ As the temperature in the quartz ampoule rises, the reactants (red-P, Sn, and SnI₄) start vaporizing and fill the ampoule. This results in a reaction between gaseous phosphorus and Sn and SnI₄ gases, forming P–Sn–I complexes that accumulate at the bottom of the ampoule. These complexes act as nucleation sites for further crystal growth. The transport agent aids in the crystallization and growth of black-P during the gradual cooling process until all the gaseous phosphorus is transformed into black-P crystals. However, the Sn/SnI₄/red-P reaction system still faces several challenges, such as the time-consuming synthesis of SnI₄ and the requirement for precise temperature control. The amount of red-P loading, ampoule capacity, and Sn/SnI₄ proportion all play crucial roles in the synthesis of black-P crystals. Scaling up the production of black-P necessitates appropriate modification of the experimental conditions, as excessive red-P loading can cause ampoule breakage, while inadequate loading results in the production of only white-P or violet-P.⁹³ In addition, this method can introduce iodine- or tin-doped impurities during the crystallization process of black-P. However, the ability to dope black-P allows for the alteration of properties, thereby offering potential to broaden the range of applications.

Nano-etching of black-P nanoribbons is achieved by electron beam lithography, reactive ion etching, and scanning TEM nano-sculpting techniques. In 2015, Lee et al. reported the successful fabrication of black-P nanoribbons with desired width and length through the use of electron beam lithography and reactive ion etching.¹⁷⁶ The process involved placing an exfoliated black-P flake on a SiO₂ substrate, followed by spin-coating PMMA on the flake for electron beam lithography in either the armchair or zigzag direction. Reactive ion etching (90% SF₆ and 10% O₂) was then used to eliminate the exposed black-P stripes, and PMMA protective stripes were removed with acetone to obtain black-P nanoribbons (Fig. 17a). However, the produced nanoribbons had a thickness of ∼50 nm and width within the range of 500 nm, resulting in their electronic properties being similar to those of black-P in its bulk form. To produce atomic-thick black-P nanoribbons, a scanning TEM nano-sculpting technique has been reported to synthesize sub-10 nm wide black-P nanoribbons composed of a few layers.²⁵⁶ However, this method has low throughput, making it challenging to produce black-P nanoribbons in large quantities. Feng et al. have utilized reactive ion etching to prepare anisotropic black-P nanoribbon field-effect transistors (Fig. 17b).²⁵⁷ This process allows for precise positioning of the nanoribbon and control of its width (∼60 nm) and height (∼3 nm) (Fig. 17c). However, the reliance on exfoliated black-P flakes and need for crystallographic alignment still limit the potential of this etching technique in practice.

	Fig. 17 Synthesis and characterization of black-P nanoribbons. (a) Scheme of electron beam lithography and reactive ion etching process of black-P nanoribbons. Reproduced with permission from ref. 176. Copyright 2015 Nature Publishing Group. (b) AFM image of the black-P nanoribbon field-effect transistor along the armchair and zigzag directions. (c) An enlarged AFM image of the nanoribbons (top) and SEM image showing the nanoribbon width of 60 nm (bottom). Reproduced with permission from ref. 257. Copyright 2018 John Wiley and Sons. (d) Schematic diagram of phosphorene nanoribbon production by lithium intercalation and liquid exfoliation. (e) TEM image of phosphorene nanoribbons drop-cast from the liquid dispersion shown in the inset. Scale bar is 10 μm. (f) Scatter diagram depicting the relationship between the widths and lengths of 940 phosphorene nanoribbons (left); aspect ratios (top right), widths (middle right), and lengths (bottom right) of only those nanoribbons falling within a dashed rectangle located in the scatter diagram. (g) A combined TEM image of a phosphorene nanoribbon with an approximate length of 11 μm. The crystallographic direction in the zigzag direction, identified by SAED, is indicated by the use of red arrows. The scale bar is 1 μm. (h) SAED pattern corresponding to the blue box shown in (g). The scale bar is 5 nm−1. Reproduced with permission from ref. 258. Copyright 2019 Nature Publishing Group. (i) Scheme of CVT-growth of black-P columns. (j) Temperature profile of the CVT heating process. (k) Scheme of nanoribbon formation through ultrasonic treatment. Reproduced with permission from ref. 265. Copyright 2021 American Chemical Society.

In 2019, Watts and colleagues introduced a novel approach for synthesizing high-quality black-P nanoribbons on a large scale.²⁵⁸ Their method relied on the combination of lithium intercalation and liquid exfoliation and involved two steps.

Black-P bulk crystals are first immersed in an aprotic lithium-ion solution at −50 °C for 24 hours, resulting in the intercalation of lithium into black-P. After ammonia is removed, the intercalated black-P is transferred to a glovebox and added to a vial containing NMP. The resulting mixture is sonicated for 1 hour and then centrifuged at a low speed to obtain a black-P nanoribbon solution (Fig. 17d). These nanoribbons exhibit a predominantly monolayer thickness, widths ranging from 4–50 nm, lengths up to 75 μm, and aspect ratios of up to 1000 (Fig. 17e and f). The nanoribbons are highly ordered and align well along the zigzag crystallographic orientation (Fig. 17g and h). The responsible mechanism for the formation of black-P nanoribbons involves the difference in diffusion kinetics between the armchair and zigzag directions.^259,260 Specifically, the preferred diffusion of alkaline metal ions in the zigzag direction creates a strain between intercalated and non-intercalated regions. This strain eventually leads to the breakage of the longer armchair P–P bonds, resulting in the formation of stripes oriented in the zigzag direction. Using this method, McDonald et al. synthesized black-P nanoribbons and utilized the nanoribbons in photovoltaic devices.²⁶¹ A recent report demonstrated an alternative method for the synthesis of black-P nanoribbons, utilizing an electrochemical-based sodium intercalation process. The resulting nanoribbons have a constrained width of 10.3 ± 3.8 nm and a length of 250 ± 156 nm. The width of the nanoribbons obtained by this method is narrower compared to that of the nanoribbons prepared by Watts et al.²⁵⁸ Furthermore, this electrochemical-based method eliminates the requirement for cryogenic conditions (−50 °C), making it more cost-effective and readily scalable for the synthesis of black-P nanoribbons. Other electrochemical-based exfoliation approaches have also been developed to synthesize black-P nanoribbons. For example, Liu et al.²⁶² have synthesized zigzag-phosphorene nanobelts by an electrochemical method with oxygen-driven BF₄⁻ intercalations. Yu et al.²⁶³ have optimized the electrochemical exfoliation conditions and employed two different types of quaternary ammonium salt cations (TPA and THA) as intercalating agents in the preparation of zigzag black-P nanoribbons with a yield of over 80%.

The bottom–up CVT method has been recently reported as an alternative to the top–down synthesis methods to produce black-P nanoribbons. Wang's group developed a one-step CVT method for the synthesis of black-P nanoribbons under mild conditions.²⁶⁴ The process involves heating commercial red-P with iodine precursors at a rate of 2 K min⁻¹ in a sealed quartz ampoule to 873 K. The temperature is maintained for 2 hours, followed by cooling to 738 K after 490 minutes and keeping for another 2 hours before natural cooling to room temperature. Ultimately, the black-P nanoribbons are grown at the bottom of the ampoule. The nanoribbons are several tens of microns in length, ∼400 nm in width, and ∼20 nm in thickness. Macewicz et al. optimized the CVT synthesis method of producing black-P nanoribbons and introduced a two-step process involving CVT growth of black-P columns, followed by ultrasonic treatment and centrifugation.²⁶⁵ The process is initiated by placing red-P, Sn, and SnI₄ precursors at one end of a vacuum quartz ampoule with the silicon substrate being positioned on the other end (Fig. 17i). To maintain the CVT process, a temperature gradient of 50 °C is maintained between the precursor and silicon substrate during heating (Fig. 17j). After the reaction, a uniform layer of columnar black-P is formed on the substrate surface. The columnar black-P is then transferred to a beaker containing DMF and sonicated for 10 minutes (Fig. 17k), after which black-P nanoribbons are collected by centrifugation. The black-P nanoribbons have an aspect ratio in the range of a few hundred with the majority of the columns having a width of up to 1.5 μm and length of more than 500 μm.

Overall, nano-etching is capable of producing black-P nanoribbons with different crystallographic orientations, but achieving precise control over the thickness and width at the atomic level remains challenging. Liquid-based exfoliation methods, including ion intercalation and electrochemical exfoliation, can lead to the fabrication of high-quality black-P nanoribbons on a large scale, but the nanoribbons obtained are predominantly oriented along the zigzag direction. CVT synthesis offers a relatively straightforward experimental procedure, but effectively separating the nanoribbons from impurities during crystal growth is an obstacle. It is anticipated that further improvement of existing methods or emergence of new ones will hasten the incorporation of black-P nanoribbons into real-world applications.

	Fig. 18 Synthesis and characterization of black-P quantum dots. (a) Scheme of the two-step sonication method for BPQD preparation and modification. (b) TEM images of BPQDs. (c) Lateral sizes (right) and heights (left) of BPQDs. (d) BPQDs and PEG-modified BPQDs in H2O and PBS. Reproduced with permission from ref. 269. Copyright 2015 John Wiley and Sons. (e) Scheme of the HEBM process of BPQDs and modification. Reproduced with permission from ref. 278. Copyright 2020 American Chemical Society. (f) Scheme of the synthesis of BPQDs by using a household kitchen blender. Reproduced with permission from ref. 279. Copyright 2016 John Wiley and Sons.

Electrochemical exfoliation with bipolar electrodes for the synthesis of BPQDs has been developed. In 2016, Mayorga-Martinez and co-workers devised a simple electrochemical exfoliation-based approach to prepare black-P nanoparticles, yielding nanoparticles with a size ranging from 40 to 200 nm.²⁷⁶ Two years later, Tang et al. succeeded in preparing fluorinated BPQDs (F-BPQDs) using an electrochemical exfoliation and synchronous fluorination technique.²⁷⁷ This process involves using black-P bulk crystals as the working electrode in a sealed three-electrode electrochemical system, where 1-ethyl-3-methylimidazolium tetrafluoroborate ([EMIM][BF₄])/acetonitrile (MeCN) serves as the electrolyte.²⁷⁷ Application of a +8 V voltage against a Ag wire anode initiates a series of chemical reactions in solution, leading to gradual exfoliation of black-P and electrolysis of BF₄⁻ to generate BF₃ molecules and F⁻ anions. The resulting BF₃ molecules form donor–acceptor bonds with MeCN, while the F⁻ anions synchronously exfoliate BPQDs. The as-synthesized BPQDs exhibit robust environmental stability, allowing them to withstand persistent ambient exposure for up to 7 days due to the hindrance of charge transfer from the phosphorus atoms to external oxygen, induced by fluorine adatoms.²⁷⁷

Ball milling is a viable choice for the synthesis of BPQDs. Ren et al. have developed a liquid-phase high energy ball milling (HEBM) and centrifugal separation process to synthesize brown transparent BPQDs ethanol suspensions, as shown in Fig. 18e.²⁷⁸ First, the HEBM approach is used to prepare black-P powders from amorphous red-P which are then placed in a sealed stainless ball milling jar containing ethanol and milling balls. The milling procedure is carried out at 600 revolutions per minute for 6 hours. After centrifugation and dispersion, the BPQD solution is mixed with various polyhydric alcohols to enhance the stability in aqueous environments. The as-prepared BPQDs show stable dispersion in ethyl alcohol and ethylene glycol, with an average lateral size of 6.5 ± 3 nm and a thickness of 3.4 ± 2.6 nm. Interestingly, BPQDs can also be fabricated by utilizing a household kitchen blender to generate a high turbulent shear rate for layer-by-layer disassembly (Fig. 18f).²⁷⁹ Using DMSO as a stabilized solvent, black-P bulk crystals are processed for 40 minutes in a household kitchen blender set to 17000–21000 rpm. The high-shear turbulence gradually peels off and disintegrates the black-P, ultimately resulting in the formation of the BPQDs suspension.

Pulsed laser ablation with high peak power densities (usually >10¹³ W cm⁻²) and ultra-short periods (femto- and nanosecond) has been applied for the exfoliation of 2D materials and production of their low-dimensional structures.²⁶⁸ Some studies have demonstrated that this approach can be utilized to prepare BPQDs. Ge et al. have synthesized phosphorene quantum dots (PQDs) by pulsed laser ablation of a bulk black-P target in diethyl ether.²⁸⁰ Diethyl ether molecules are cleaved by laser ablation to passivate the surface and edges of the phosphorene domains. The PQDs exhibit strong photoluminescence in the blue–violet range, with an average lateral size of ∼7 nm. Ren and co-workers have fabricated a few-layered BPQDs using pulsed laser ablation of black-P bulk crystal in isopropyl ether.²⁸¹ A neodymium-doped yttrium aluminum garnet (Nd:YAG) pulsed laser with a repetition rate of 10 Hz and wavelength of 1064 nm is used in the experiment.²⁸¹ The ablation process lasts for about 30 minutes and the BPQDs exhibit a higher photoluminescence quantum yield (∼20.7%) than PQD (11.92%),²⁸⁰ showing potential in cell bioimaging.

Using white-P as starting materials, early synthesis of red-P often involves the pyrochemical method, which requires temperatures over 250 °C in the absence of air.⁸⁷ Vvedenskii and Frost studied the reaction kinetics of thermal conversion and claimed that the conversion of melted white-P to red-P followed a first-order reaction in the temperature range of 176–373 °C.²⁸² However, the increasing trend of constants with a rise in the percentage conversion at temperature below 263 °C suggests an autocatalytic conversion. To validate this phenomenon, DeWitt and co-workers conducted more experiments on the reaction kinetics in 1946.²⁸² The reaction mixture initially comprises red-P particles suspended in white-P. As the reaction progresses and the percentage conversion reaches approximately 50%, the particles grow in number and size forming a semi-fluid before forming a solid product. The reaction in the temperature range of 250–350 °C strictly follows a first-order rate law, with an activation energy of 37800 calories per mole. In addition, it is noted that addition of iodine or sulfur has a significant accelerating effect on the thermal conversion process. However, this effect is only noticeable in the initial stages and rapidly diminishes, resulting in the conversion proceeding at a rate similar to that without accelerators.

In contrast to the pyrochemical method of red-P synthesis, electromagnetic radiation utilizes a gentler synthetic approach to convert white-P to red-P (Fig. 19a, route a).²⁸³ Linus Pauling had proposed that red-P can be formed from white-P by cleaving a single bond in each P₄ molecule, resulting in an expansion of the unit into a figure comprising two equilateral triangles with a shared base.¹⁰⁸ The two apical phosphorus atoms in each P₄ unit can then bond with similar phosphorus atoms in neighboring P₄ complexes, leading to the formation of long chains of bonded atoms and ultimately the formation of red-P. The formation of red-P from white-P is easier than that of black-P because breaking a single bond is sufficient to convert each P₄ unit into red-P, whereas the formation of black-P requires breaking of three out of the six bonds in each P₄ unit.¹⁰⁸ A. Pedler and G. Rathenau have published two seminal papers demonstrating how light triggers the conversion of white-P to red-P.^284,285 These studies reveal two mechanisms that drive the transformation through light activation. One mechanism involves the symmetry cleavage of the P₄ tetrahedron and subsequent rearrangement of P₂ into the final polymeric red-P.²⁸⁵ The other mechanism, which occurs in non-polar media, involves a radical process.²⁸⁶ The dispersed phase also influences the concentrations of P₄ and P₂. When white-P is heated to 1000–1500 °C in an inert atmosphere, the resulting gas is a mixture of P₄ and P₂ molecules, with their vapor pressure ultimately reaching equilibrium (Fig. 19a, route f),^283,287 but when the dispersed phase is a solution at room temperature, P₄ molecules are photolyzed to P₂.²⁸⁸ M. Kraft and V. Parini have described a photochemical reaction between white-P and alkyl iodides at 20 °C, yielding a small quantity of red-P containing organic molecules.²⁸⁹ D. Perner and A. Henglein have carried out a similar reaction of white-P with alkyl (or aryl) iodides in CCl₄ under γ-⁶⁰Co at temperatures ranging from 20 to 140 °C.²⁹⁰ The primary product is red-P (79%) at 25 °C, but the primary product is alkyl-phosphorylated derivatives (80%) at 130 °C. However, when white-P is subjected to visible light up to 100 °C, red-P becomes the primary product. There seems to be a more sophisticated mechanism behind the transformation of white-P into red-P through electromagnetic radiation, and a further study is necessary.

	Fig. 19 Synthesis and characterization of the red-P bulk form and nanosheets. (a) Scheme of the conversion of the main significant phosphorus allotropes through different routes. Reproduced with permission from ref. 283. Copyright 2014 John Wiley and Sons. (b) Images of samples before and after thermal decomposition. Reproduced with permission from ref. 292. Copyright 2018 Elsevier. (c) Scheme of the fabrication procedure of ultra-thin red-P. (d) and (e) AFM images of samples before and after sonication. Reproduced with permission from ref. 293. Copyright 2021 Springer. (f) Scheme of the fabrication procedure of red-P porous nanosheets. (g) Photographs of the red-P dispersion after ultrasonic treatment in different solvents and after standing for some days. Reproduced with permission from ref. 294. Copyright 2020 Elsevier. (h) Scheme of the synthesis of red-P nanosheets and their potential anticancer applications. (i) TEM image of the obtained red-P nanosheets. The scale bar is 200 nm. (j) AFM image of the as-synthesized red-P nanosheets. The scale bar is 500 nm. Reproduced with permission from ref. 298. Copyright 2021 John Wiley and Sons.

Ultrasound-assisted liquid exfoliation can be utilized to prepare highly dispersed red-P nanosheets. Water exfoliation is carried out to produce ultra-thin red-P from commercial red-P (Fig. 19c), with the resulting nanosheets being highly dispersed and having an average thickness of ∼1 nm (Fig. 19d–e).²⁹³ When a similar liquid exfoliation process is conducted using probe ultrasonication in the NMP solvent, red-P porous nanosheets are produced (Fig. 19f).²⁹⁴ This is because of the strong synergistic effects of NMP and ultrasonication facilitating layered exfoliation and partial surface fragmentation of bulk red-P, leading to the formation of nanosheets with porous topologies. In contrast, bulk red-P exfoliated in a water solution displays insufficient fragmentation and fails to form porous structures. To further assess the suitability of red-P nanosheets with dispersants and the role of NMP in ultrasonication, other solvents are used in exfoliating nanosheets (Fig. 19g).²⁹⁴ However, the nanosheets are found to be unstable in these solvents and visible deposition occurs to varying degrees after three days. On the other hand, the nanosheets can remain stable in the NMP solvent for over three months, thus confirming the superior stability of NMP in the red-P exfoliation systems. However, the specific interaction between bulk red-P and solvents of varying polarity requires further elucidation.

While red-P nanosheets prepared by liquid exfoliation have shown potential for a variety of applications, they suffer from issues such as sheet-edge corrosion and uncontrollable size. Fortunately, an alternative method using amine-induced wet chemical synthesis has made significant progress in the production of red-P nanosheets.^295–297 This method involves a nucleophilic attack on phosphorus sources, which triggers the breakage of the P–P bond and the formation of polyphosphoric ions (P_n⁻). Subsequent treatment with water and acid replaces external ions with electrophiles, resulting in the decomposition of P_n⁻ to form red-P nanosheets. Recently, Mo and colleagues have described a unique wet-chemical approach for the synthesis of red-P nanosheets through a cysteine-mediated redox reaction and demonstrated their potential for anticancer applications (Fig. 19h).²⁹⁸ In this method, phosphorus triiodide (PI₃) serves as the phosphorus source, while L-cysteine (L-Cys) acts as the reductant. The confinement material CATB and stabilizer PVP are also employed. Through a mild redox reaction between PI₃ and L-Cys, phosphorus atoms are gradually released and then assembled into nanosheets. The optimal concentration of CATB is critical in the process, since it suppresses red-P natural 3D growth, allowing for 2D growth instead. The weakened reduction, combined with the appropriate confinement material, enables a mild reaction equilibrium, resulting in the formation of high-quality red-P nanosheets. TEM and AFM reveal that the RPNSs are atomically flat with a thickness of around 1.7 nm (Fig. 19i and j). Moreover, this method enables doping of metal atoms in red-P nanosheets to regulate the basic properties and expand the applications.

Song et al. have developed a drug delivery platform by constructing a nanocomposite of bismuthene and RPQDs for synergistic photothermal/photodynamic/chemotherapy.²⁹⁹ The nanocomposite is synthesized through an electrostatic assembly of exfoliated bismuth nanosheets and RPQDs, followed by PEGylation and loading of the organic drug molecule DOX (Fig. 20a). RPQDs are prepared by dispersing 0.5g of red-P powder and 0.2g of triphosphate in water and subjecting them to sonication in an ice bath for 10 hours, followed by centrifugation and filtration. The resulting RPQDs are then PEGylated to make them more suitable for physiological conditions and the average diameters of RPQDs before and after PEGylation are around 2.5 and 55 nm, respectively. The combination of bismuthene and RPQDs exhibits synergistic effects achieving photothermal conversion efficiencies of 54.2% and 38.5% upon 1064 nm and 808 nm (1.5 W cm⁻²) laser irradiation, respectively.

	Fig. 20 Synthesis and characterization of RPQD-based materials. (a) Scheme of the fabrication procedure of Bi@RP-PEG-DOX. Reproduced with permission from ref. 299. Copyright 2022 John Wiley and Sons. (b) Scheme of the synthesis of Ni-RPQD. (c) XRD pattern of Ni-RPQD and RPQD. (d) TEM image of Ni-RPQDs. (e) Statistical analysis of Ni-RPQD diameters. (f) EPR spectra of Ni-RPQD and RPQD. Reproduced with permission from ref. 300. Copyright 2022 John Wiley and Sons.

Jia and colleagues have reported in situ growth of single-atom nickel on the surface of RPQDs using the ultrasonic method.³⁰⁰ In the process, commercial bulk red-P is first ground into fine particles, followed by hydrothermal treatment to remove the oxide layer from the surface. The purified red-P and NiCl₂·6H₂O are introduced to 100 mL of NMP and sonicated in an ice bath for 48 h to form the nanocomposite Ni-RPQD (Fig. 20b). TEM demonstrates successful synthesis of a homogeneous nanocomposite with a particle size of around 2.7 nm (Fig. 20c and d). The uniform distribution of single-atom Ni is attributed to the unique coordination of single atoms docked with P vacancies on the RPQD surface, which is supported by significant reduction of the electron paramagnetic resonance signal and high-angle shift in the XRD patterns (Fig. 20e and f). The Ni–P bond potentially serves as an electronic antenna to promote facile extraction and transfer of charges. Furthermore, incorporation of Ni single atoms reduces the bandgap and enhances light absorption to improve photocatalytic activity under visible light irradiation.³⁰⁰

Ruck's detailed experimental synthesis procedure for fibrous-P bulk crystals using the CVT approach has become a widely adopted practice. Researchers have utilized various iodine-containing transport agents in experiments to improve the short-distance transport reaction and obtain high-purity fibrous-P crystals. Smith et al. developed a two-step method for producing ultra-long and uniform fibrous-P nanowires.³⁰² They first prepared thin films of red-P on a SiO₂/Si wafer, and then heated a mixture of purified red-P (110 mg), SnI₄ (15 mg), and Sn powder (30 mg) in a sealed ampoule with the red-P-coated silicon wafer plate to 630 °C for 1 hour, followed by programmed cooling. The resulting fibrous-P nanowires have a diameter of 300–800 nm and a length of over 1 mm. However, the nanowires are found to be sensitive to air due to their larger specific surface area and the presence of other crystalline forms of red-P. Other iodine-containing transporters such as CuI and SbI₃ are also found to produce other phosphorus allotropes besides fibrous-P. To obtain single-phase fibrous-P crystals, Nilges’ group substituted the iodine-containing transport agents with chloride compounds and used CuCl₂ as an assisting agent.³⁰³ They designed a temperature-control program that involves heating the precursors to 550 °C within 8 hours, followed by holding the temperature for 15 hours, cooling to 275 °C within 75 hours, maintaining this temperature for 1 hour, and finally cooling to room temperature within 75 hours.³⁰³ This method leads to the production of fibrous-P crystals in high yield with reasonable quality. However, not all products synthesized by the CVT method have well-defined fibrous structures and some crystals instead show different morphologies such as bunches-like, nanowire, and urchin-shaped structures.^302–305 These diverse morphologies may be attributed to the complex phase transformation that occurs during the gas phase process and the dispersion interaction between each tubular unit.

Aside from the transport agent, the vapor transport process is highly influenced by the reaction conditions. Our group has developed a low-temperature CVT method that effectively produces high-quality fibrous-P crystals on a large scale (Fig. 21a).⁴⁷ In this process, amorphous red-P is transported in the presence of iodine from a high-temperature region (480 °C) to a low-temperature region (450 °C), where crystallization occurs within a day. Compared to previous reports in which the typical temperature is over 550 °C and annealing time lasts several days to weeks, our method greatly reduces the overall reaction temperature and annealing time. In a single preparation, we are able to obtain a total amount of 9.937 g crystals with a yield exceeding 99% (Fig. 21b and c). Mass production of fibrous-P bulk crystals allows for exfoliation of fibrous-P nanoribbons, which will be discussed in the next part. Yan's group has investigated the impact of growth kinetics on the crystallinity of fibrous-P in the CVT reaction and found that the kinetics of CVT is regulated by the rate-determining steps of gas convection and diffusion. To obtain high-quality crystalline fibrous-P, it is important to slow down gas convection and diffusion. To achieve this, there are several strategies including reducing the amount of red-P during the reaction, increasing the length of the ampoule, lowering the average reaction temperature, and designing a “neck” on the ampoule to reduce the cross-sectional area and slowing down the reaction rate (Fig. 22a).¹⁶⁰ These measures are effective in modifying the growth kinetics of the CVT reaction, ultimately leading to the production of high-quality and well-crystallized fibrous-P crystals. Recently, Zhang et al. have developed a low-pressure CVT approach for the production of centimeter-scale fibrous-P micropillar arrays without the use of transport agents.³⁰⁶ The approach utilizes directional phase transformation of amorphous red-P at a specific temperature and pressure to generate fibrous-P micropillar arrays by self-assembly. In contrast to traditional transport agent-assisted CVT methods, this approach avoids nucleation aggregation induced by transport agents, paving the way for the bottom–up synthesis of fibrous-P microstructures and nanostructures.

	Fig. 21 Low-temperature CVT method for the preparation of fibrous-P bulk crystals and nanoribbons. (a) Scheme of the low-temperature CVT method. (b) Photograph and weight of the fibrous-P bulk powder. (c) Photograph of the bulk powder under high magnification. (d) Photograph of the fibrous-P dispersion before and after exfoliation. (e) TEM image of fibrous-P bulk crystals and (f) nanoribbons. (g) AFM image of a representative fibrous-P nanoribbon. For the sake of consistency, it should be noted that the term “fibrous-P” is used throughout this review to refer to the “triclinic crystalline red phosphorus (cRP)” mentioned in the original text. Reproduced with permission from ref. 47. Copyright 2020 John Wiley and Sons.

	Fig. 22 Modification of the growth kinetics in the CVT reaction for the preparation of fibrous-P bulk crystals and nanoribbons. (a) Scheme of a “neck” on the ampoule in the CVT reaction. (b) Dispersion concentration of fibrous-P nanoribbons versus solvent surface tensions. (c) Lambert–Beer's plot at λ = 650 nm of fibrous-P nanoribbons in IPA. (d) Photograph of fibrous-P nanoribbon dispersions in different solvents. Reproduced with permission from ref. 160. Copyright 2021 American Chemical Society.

Solution-based deposition of elemental phosphorus is an alternative method to prepare low-cost and scalable fibrous-P thin films. Ban et al. have demonstrated a wet chemical process that utilizes soluble polyphosphides as precursors to generate fibrous-P through moderate heating without the need for external agents.³⁰⁷ Amorphous red-P is the source in the polyphosphide synthesis and reacts with potassium ethoxide to produce a potassium mixture of polyphosphide powder. The resulting polyphosphide powder is then refined by ethanol dispersion and insoluble precipitate removal to eliminate unreacted precursors and by-products. To fabricate fibrous-P thin films, the polyphosphide solution is spin-coated or drop-cast onto glass or Si/SiO₂ substrates followed by annealing for 30 minutes at 250 °C. The Raman signals of fibrous-P films formed on glass and Si/SiO₂ substrates are almost identical, indicating that fibrous-P formation is mainly due to the precursor itself rather than the precursor–substrate interaction. However, similar to the issues encountered during the preparation of black-P crystals by wet chemical methods,²⁵² fibrous-P produced by this method lacks obvious lattice structures and may contain mixed phosphorus phases. Hence, investigation of the interaction between phosphorus clusters and solvents is necessary to better understand the dynamic growth process of crystalline phosphorus in solutions.

Liquid-phase exfoliation is currently the primary method to fabricate fibrous-P nanoribbons due to its scalability and high-yield production. As aforementioned in the preceding section, our group has developed a low-temperature CVT method for mass production of high-quality fibrous-P bulk crystals.⁴⁷ After dispersing these crystal powders in absolute ethanol and performing probe sonication, the resulting supernatant changes from colorless to a uniform orange color (Fig. 21d), indicating the formation of a stable fibrous-P colloid solution. The fibrous-P nanoribbons obtained from this solution are compact, micro-pillar-like lumps, which are quite distinct from the micrometer-scale fiber structures of the bulk materials (Fig. 21e and f). The width of the nanoribbons is approximately 120 nm and lengths range from hundreds of nanometers to several micrometers. The majority of the fibrous-P nanoribbons have an aspect ratio greater than 10, with some having an aspect ratio close to 100. AFM reveals that the thickness of a representative fibrous-P nanoribbon is ∼8.5 nm (Fig. 21g), indicating that liquid-phase exfoliation can lead to the production of few-layered fibrous-P nanoribbons. Hu et al. have proposed the concept of fibrous phosphorene based on the feature that fibrous-P can be exfoliated.¹³¹ Using N-cyclohexyl-2-pyrrolidone (CHP) as the liquid-sonication solvent, they were able to prepare fibrous phosphorene along the [001] direction with a thickness of roughly 1.67 nm. Given that the [001] facet has an average spacing of 0.583 nm,¹³¹ the fibrous phosphorene should consist of three layers. NMP is also used to exfoliate bulk fibrous-P. In the process, single-layered phosphorene is observed, indicating that NMP may have a greater exfoliation capability than CHP. This study demonstrates that the choice of solvents plays a critical role in the exfoliation of Fibrous-P.

To investigate the impact of solvents on the efficiency of fibrous-P exfoliation, Yan's group selected eight solvents with different surface tensions, including isopropanol (IPA), DMSO, NMP, DMF, formamide, ethanol, deionized water, and acetone (Fig. 22d).¹⁶⁰ Each solvent shows a different exfoliation efficiency and there is a broad linear relationship between the dispersed concentration of nanoribbons and solvent surface tension (Fig. 22b). Solvents with higher surface tension tend to have stronger phosphorus–solvent adhesion, which stabilizes the nanoribbon surface for a higher exfoliation efficiency.³⁰⁸ However, IPA and ethanol, which have relatively low surface tension, also show efficient exfoliation due to the appropriate molecular interaction with fibrous-P interlayers.³⁰⁹ Although formamide has the highest exfoliation concentration, IPA is recommend as the primary solvent for fibrous-P exfoliation due to its high efficiency, low boiling point, ease of handling, and cost-effective features. The yield of exfoliated nanoribbons in IPA is 16%, and the dispersion extinction coefficient is measured to be ∼2087 L g⁻¹ m⁻¹ under 650 nm incident light (Fig. 22c).¹⁶⁰ Liu et al. have used the Hansen solubility parameters (HSPs) to quantify the effects of the solvents on the exfoliation efficiency of fibrous-P nanoribbons.³¹⁰ The HSP approach considers three key parameters: δD (intermolecular dispersion force), δH (intermolecular hydrogen bond), and δP (intermolecular polar force).³¹¹ The desired exfoliating solvent should have HSPs that closely match those of fibrous-P. After comparing 19 common solvents, acetone and 1,3-dimethyl-3,4,5,6-tetrahydro-2(1H)-pyrimidinone (DMPU) were found to be more suitable for fibrous-P exfoliation. Moreover, the HSPs of a solvent can be modified by varying the solvent ratios. The water/acetone mixture is chosen as the solvent because it covers the HSPs of Fibrous-P, and both water and acetone are easy to remove without leaving a residual solution. The optimal solvent mixture for fibrous-P exfoliation is determined to be 10% water and 90% acetone, which yields 40% nanoribbons with a thickness of less than 10 nm and 70% with a thickness of less than 20 nm. These results demonstrate the potential of HSP modeling in identifying suitable solvents for exfoliating 2D materials and highlight the importance of solvent selection in the exfoliation process.

In 2019, Yu's group developed a thermal evaporation method that utilized liquid bismuth nanodroplets to produce large quantities of single-crystal violet-P microbelts measuring up to 100 × 10 × 0.3 μm³ (Fig. 24a).³¹⁴ The growth mechanism of microbelts is shown in Fig. 24c. The process involves the sealing of appropriate amounts of red-P and Bi powders in vacuum quartz capsules and heating to 550 °C, inducing the sublimation of the red-P precursor while Bi melts. At this temperature, the molecular motion of the phosphorus vapor becomes highly turbulent producing quick disturbances within the capsule to impinge into the Bi droplets and stretch then into nanodroplets. As the temperature drops to 300 °C, the sublimed phosphorus progressively adsorbs onto the surface of the Bi nanodroplets and nucleates, forming seed crystals that eventually grow into nanorods.³¹⁵ Due to the high surface energy of the nanodroplets, they tend to consolidate, causing the nanorods to cohere and form a thicker crystal that gradually expands into a microbelt. The Bi nanodroplets may depart during growth and after cooling to ambient temperature, most of them aggregate, with only a few droplets condensing singly on the microbelts.^316,317 The XRD pattern of violet-P microbelts reveals a high-intensity peak at 15.66 degrees from the (013) facet. This peak is markedly distinct from the corresponding peak in the standard violet-P bulk crystal with low intensity. The difference in the peak intensity suggests the specific orientation of facets within the crystal.³¹⁴ Additionally, high-resolution TEM shows a well-resolved interference fringe spacing of 5.67 Å along the crystal's edge (Fig. 24b), which is in good agreement with the interplane distance between the (013) planes of violet-P.³¹⁴ These findings provide evidence of successful synthesis of high-quality violet-P microbelts with the exposed active (013) facets.

	Fig. 24 Synthesis and characterization of the violet-P bulk form and phosphorene. (a) SEM image and (b) high-resolution TEM patterns of violet-P microbelts. (c) Scheme of the growth mechanism of microbelts initiated by liquid Bi through thermal vaporization. Reproduced with permission from ref. 314. Copyright 2019 Elsevier. (d) Scheme of the fabrication procedure of violet-P crystals through an improved CVT method. Reproduced with permission from ref. 319. Copyright 2020 American Chemical Society. (e) TEM images and atomic model of the violet-P/InN/InP layered structure. Reproduced with permission from ref. 320. Copyright 2015 American Chemical Society. (f) AFM images of violet phosphorene from mechanical and (g) liquid exfoliation. (h) High-resolution TEM image of a violet-P nanobelt. (i) SAED pattern corresponding to (h) with the zone axis of [001]. Reproduced with permission from ref. 48. Copyright 2020 John Wiley and Sons.

Different from direct thermal evaporation, CVT typically involves separate heating zones, which allow for the movement of gaseous intermediates in defined directions.³¹⁸ Zhang's group has made contributions to the synthesis of high-quality violet-P bulk crystals using the CVT method. In 2020, they obtained single-crystal violet-P and conducted a detailed analysis of its structure.⁴⁸ This achievement is of great significance, as it represents the first experimental determination of the lattice structure of violet-P and provides valuable information about the properties and potential applications of violet-P. However, the violet-P crystals obtained in this study contain a mixture of black-P with a yield of only about 36% after separation. To address this issue and improve the production yield, the CVT process has been optimized by using 10 mg of Sn and 18 mg of SnI₄ as the transport agents and 470 mg of amorphous red-P as the source materials in a vacuum quartz tube.³¹⁹ As shown in Fig. 24d, the tube is heated slowly in a three-zone muffle furnace for 8 hours to reach 600 °C at the sample end (source zone) and 580 °C at the other end (reaction zone).³¹⁹ After 5 hours, the tube is cooled for 10 hours to 550 °C in the source zone and 530 °C in the reaction zone. The tube is kept at these temperatures for 30 hours before slowly cooling to room temperature for about 100 hours. The violet-P crystals are produced with a high yield of ∼80%. The holding time of 550–530 °C and cooling time to room temperature are critical for violet-P crystal nucleation and growth.³¹⁹ A long holding time at 550–530 °C increases the nucleation of violet-P, with nucleation peaking at approximately 30 hours. Longer cooling time results in larger crystal formation with crystal size reaching its maximum at around 100 hours.³¹⁹ The ability to produce high-quality and high-yield violet-P bulk crystals serves as the foundation for the subsequent synthesis of nanostructured violet-P.

To date, violet phosphorene has been fabricated by both mechanical and liquid exfoliation of violet-P single crystals.^48,321–323 Zhang's group has demonstrated that violet phosphorene can be produced by immersing single-crystal violet-P (∼0.30 × 0.23 × 0.13 mm³) in 30 mL of ethanol and treating it by tip sonication at room temperature.⁴⁸ They also utilized Scotch tape to peel violet-P crystals into thin flakes several times and then transferred them to SiO₂/Si substrates. Fig. 24f and g show the AFM images of violet phosphorene prepared by both mechanical and liquid exfoliation methods. Mechanical exfoliation produces single-layered violet-P with a thickness of ∼1.9 nm, but most of the flakes are multi-layered. On the other hand, liquid phase exfoliation can more easily produce thin violet-P flakes (∼2.2 nm), but the edges of the flakes are irregular due to defects generated during liquid exfoliation. However, the internal lattice in the liquid-exfoliated flakes remains intact and high-resolution TEM shows that most flakes have two sets of crystal planes intersecting at an angle of 90° (Fig. 24h) consistent with the (110) and (10) lattice planes determined by single-crystal XRD.⁴⁸ The SAED pattern obtained from the flake with the zone axis [001] in Fig. 24i clearly demonstrates the high degree of crystallinity in the as-prepared phosphorene. Furthermore, EDS confirms the high stability of violet phosphorene by showing no detectable oxygen signal. These findings make a significant contribution to the development of high-quality violet phosphorene materials and potential applications in various fields.

The VPQDs synthesized in NMP emit a high fluorescence intensity. To rule out the possibility that the fluorescence originates from the NMP solvent, the excitation wavelength is set at 430 nm, as fluorescence from NMP disappears completely at excitation wavelengths longer than 430 nm. Upon 430 nm excitation, a robust fluorescence band at 522 nm is observed from the VPQDs.³²⁴ The VPQDs emit strong green light compared to conventional QDs such as BPQDs (450 nm) and MoS2 QDs (414 nm),^326,327 which emit light in the blue–violet range. Conventional QDs emitting blue–violet light penetrate biological tissues poorly due to the predominantly blue luminescence of biological tissues thus limiting in vivo optical imaging of deep tissues.³²⁸ However, the green fluorescence emitted by VPQDs may greatly mitigate this issue. The fluorescence quantum yield of VPQDs is determined to be 7.02% and the fluorescence properties have been demonstrated to be adjustable by varying the solvothermal temperature and time conditions, making them potential candidates for nanosensors in biology and medicine.^324,329

In addition to the solvothermal temperature and reaction time, Zhang's group demonstrated that the solvent used in the VPQD synthesis influences the fluorescence properties.³³⁰ They used six different solvents, including DMF, NMP, H₂O, 1,3-dimethyl-2-imidazolidinone (DMI), isopropanol (IPA), and ethanol (EA), to prepare VPQDs based on the solvothermal method (Fig. 25a).³³⁰ The quantum dots prepared in the six solvents had the characteristic lattice structure of violet-P with similar lateral dimensions (2.5–4.0 nm) and thickness (1.5–2.0 nm). However, FTIR reveals that the VPQDs have different oxygen-containing functional groups in different solvents (Fig. 25b), which can stabilize the edge-dangling phosphorus bonds of VPQDs during synthesis. This modification affects the bandgap size of VPQDs, and thus their fluorescence properties. By functionalizing the edge-dangling phosphorus bonds with different reaction solvents, fluorescence from VPQDs can be adjusted from 400 nm (3.10 eV) to 536 nm (2.31 eV) to span the violet to green spectrum.³³⁰ The VPQDs prepared in protic solvents (IPA, H₂O, and EA) exhibit shorter emission wavelengths (<430 nm) than those prepared in aprotic solvents (DMF, NMP, and DMI), suggesting larger bandgaps for VPQDs prepared in protic solvents (Fig. 25c and d).³³⁰ Theoretical calculations show that different functional groups can alter the HOMO and LUMO positions of VPQDs. In protic solvents, hydroxyl groups are ionized, leading to the formation of P–O bonds between the O atoms of solvent molecules and the edge-dangling phosphorus bonds. This results in significant changes in the HOMO–LUMO bandgaps. Functionalizing the edge-dangling phosphorus bonds with different reaction solvents to adjust the bandgap properties is a simpler and more feasible method than traditional doping and particle size control methods, thereby expanding the potential optoelectronic applications of the VPQDs.

	Fig. 25 Synthesis and characterization of VPQDs in various solvents. (a) Scheme of the synthesis of VPQDs with tunable bandgaps by using different solvents. (b) Fourier transform infrared spectroscopy analysis of VPQDs in different solvents. (c) Fluorescence emission spectra of VPQDs in protic solvents and (d) in aprotic solvents. The excitation spectrum is represented by the dotted line, and the emission spectrum is represented by the solid line. Reproduced with permission from ref. 330 Copyright 2022 American Chemical Society.

Photocatalysis is a process in which the photoreaction is accelerated in the presence of a semiconductor photocatalyst.³⁶⁹ The semiconductor absorbs solar photons and converts the reactant into the final product without depletion.³⁷⁰ The overall mechanism of photocatalysis involves three steps. First, solar photons excite electrons in the semiconductor, allowing them to move from the valence band to the conduction band, generating charge-negative electron (e⁻) and charge-positive hole (h⁺) pairs. Second, the photoexcited electrons and holes move to the semiconductor surface, where they act as redox agents. Third, the transfer of electrons leads to the formation of superoxide radicals (˙O₂⁻) or initiates reduction events with a corresponding reduction potential. Hole transfer, on the other hand, generates hydroxyl radicals (˙OH) or initiates oxidation events with an appropriate oxidation potential. The two critical steps in a photocatalytic process are light absorption and charge carrier excitation, and the application of this process in energy conversion is largely dependent on the light radiation intensity and the semiconductor's bandgap. To achieve photocatalytic hydrogen generation, the semiconductor bandgap should be greater than 1.23 eV with a suitable conduction band position for the H⁺/H₂ potential. Several common semiconductors such as TiO₂ and MoS₂^371,372 have been reported for photocatalytic water splitting to generate hydrogen. However, as these metal oxides consist of rare elements with relatively high manufacturing costs, their widespread industrial use is restricted. Moreover, these semiconductors have wider bandgaps (>3.2 eV), resulting in the primary absorption of ultraviolet light (<400 nm), which corresponds to only ∼4% of the solar spectrum, leaving a significant portion of the solar spectrum unused. Consequently, these disadvantages significantly limit the practical applications of conventional semiconductors and necessitate the development of innovative and cost-effective photocatalysts for hydrogen production.

EPMs are promising materials in the mono-elemental family, with the potential to serve as metal-free photocatalysts for hydrogen generation. First, phosphorus is one of the most earth-abundant elements on the earth,^373,374 and so the production cost of phosphorus-based catalysts is relatively low. Second, EPMs have narrow bandgaps (<2.7 eV), making them suitable for absorbing solar photons primarily in the visible light region and even down to the near-infrared region, with their conduction band minimum position being more negative than the water reduction potential. Third, EPMs have several layered allotropes with tunable bandgaps, allowing the selection of different allotropes in bulk or few-layered nanosheet forms for specific photocatalytic situations. Apart from black-P, amorphous red-P, fibrous-P, violet-P, and blue-P have all demonstrated their viability as candidates for photocatalytic hydrogen generation, both theoretically and experimentally.^{21,23,375–377}Table 2 summarizes the progress made in EPMs and EPM-based systems for photocatalytic hydrogen generation.

Allotropes	Materials	Methodology	Light source	H2 generation rate	Ref.
Black phosphorus	BP-BM	Ball milling	300 W Xe lamp with a 420 nm cut-off filter	512 μmol h−1 g−1	213
	BP nanosheets and nanoparticles	Liquid phase exfoliation	300 W Xe lamp with a 420 nm cut-off filter	64 μmol from the two cycles fornanosheets/45 μmol h−1 g−1 for nanoparticles	381
	BPQDs/g-C3N4	Sonication approach	300 W Xenon lamp with a 420 nm cut-off filter	271 μmol h−1 g−1	382
	Ni2P/BP	Solvothermal process	150 W Xe arc lamp l	406.08 μmol h−1 g−1	383
	BP-10000/MoS2	Solvothermal method	300 W Xenon lamp with a UV cutoff filter (λ > 420 nm)	1286 μmol h−1 g−1	384
	BP/CN	High-energy ball milling	Blue LED lamp (λ = 440–445 nm)	786 μmol h−1 g−1	385
	FPS/CS	Mechanical mixing	300 W Xenon arc lamp with a UV-cutoff filter (λ ≥ 420 nm)	11192 μmol h−1 g−1	386
	BP/PCN-HKM	Ultrasonication and wet-chemical self-assembly	300 W Xenon lamp with an optical filter (λ > 400 nm)	7380 μmol h−1 g−1	387
	BP/MBWO	Sonication	300 W Xenon lamp	21042 μmol g−1	388
Amorphous red phosphorus	URP	Liquid exfoliation	300 W Xe lamp with a UV-cut 420 nm filter	180 μmol h−1 g−1	293
	RP PNS	Solvothermal treatment-coupled ultrasonic exfoliation	300 W Xe lamp	1169.4 μmol h−1 g−1	294
	Red-P/g-C3N4	Solid state annealing	300 W Xenon lamp with a UV cut-off filter (λ > 420 nm)	310 μmol h−1 g−1	389
	Hierarchical P/YPO4 hollow microspheres	Hydrothermal method	300 W Xenon lamp with a 400 nm cut-off filter	65 μmol h−1 g−1	390
	g-C3N4/Red-P hybrid nanosheets	Sono-chemical treatment	300 W Xenon lamp with a 420 nm cut-off filter	1691 μmol h−1 g−1	391
	Co2P/RP	Hydrothermal method	300 W Xe lamp irradiation	5992.5 μmol h−1 g−1	392
Fibrous phosphorus	Micro-fibrous P/SiO2 and smashed-fibrous P	CVD and ultrasonication	300 W Xenon lamp	633 and 684 μmol h−1 g−1, respectively	393
	Red P/g-C3N4	One-step CVD	300 W Xenon lamp equipped with L40 cutoff filter	2565 μmol h−1 g−1	394
	Fibrous phosphorene	Ultrasonication	300 W Xenon lamp	1021 μmol h−1 g−1	131
Violet phosphorus	Violet phosphorus	CVT	300 W Xenon lamp	553 μmol h−1 g−1	395
	CRPMR	CVT	Solar simulator-300 W Xenon lamp	324.2 μmol h−1 g−1	396
	PCN@HP	CVD	300 W Xe lamp with an AM 1.5G filter as simulated sunlight (λ > 300 nm) or a 420 nm cutoff filter as visible light (λ > 420 nm)	33.2 and 17.5 μmol h−1 g−1 for simulated sunlight and visible light, respectively	397

Although theoretical studies suggest that pristine bulk EPMs may serve as photocatalysts for water splitting,^75,378,379 their inadequate catalytic efficiency limits the competitiveness compared to conventional photocatalysts. This is due to their narrow bandgaps, limited catalytic sites, and elevated recombination of photogenerated electron–hole pairs. As discussed in the electronic property section, exfoliation of EPM bulk crystals into few-layered nanosheets or phosphorene leads to a positive shift in the valence band and a negative shift in the conduction band, resulting in a significant increase in the bandgap. This promotes the efficient generation of strongly oxidative holes, preventing undesirable photoinduced charge pair recombination. In addition, the exfoliation process provides a large surface area and increases the molecule absorption space and active sites. As a result, low-dimensional EPMs are becoming increasingly popular for photocatalytic hydrogen generation. Zhu et al. conducted a systematic study of the visible-light photocatalytic efficiency of black-P bulk powder and nanosheets and found that black-P nanosheets exhibit an 18-fold greater hydrogen generation activity compared to the bulk form ref. 205. It is noteworthy that anhydrous LiOH powder is added during the synthetic process, and the photocatalytic hydrogen generation rate of black-P nanosheets in the presence of LiOH is 512 μmol h⁻¹ g⁻¹, which is greater than that of bulk black-P (28 μmol h⁻¹ g⁻¹) and well-known metal-free photocatalyst g-C₃N₄ (107 μmol h⁻¹ g⁻¹).³⁸⁰ After illumination for 9 hours with 420 nm homochromatic light, the apparent quantum yield of hydrogen generation is calculated to be approximately 0.47%. However, another study also using black-P nanosheets for photocatalytic hydrogen generation showed a lower generation rate.^213,381 This discrepancy may arise from the different fabrication methods (liquid exfoliation and ball milling, respectively) and addition of LiOH, which terminates the reactive black-P nanosheet edges and provides a stable structure for photocatalytic reaction.

Amorphous red-P boasting a lower manufacturing cost and well-established production technique compared to black-P has attracted a considerable interest due to its great photocatalytic characteristics. Chen and co-workers investigated the photocatalytic hydrogen generation efficiency of ultra-thin amorphous red-P nanosheets produced by liquid exfoliation.²⁹³ The nanosheets exhibited a good hydrogen generation activity of 180 μmol h⁻¹ g⁻¹, which is 2.88 times higher than that of bulk red-P. The apparent quantum efficiency of amorphous red-P nanosheets is approximately 0.28%. Cycling stability measurements show that the amorphous red-P nanosheets retain 89.1% of their original catalytic activity after three irradiation cycles, indicative of great chemical stability. The primary advantage is the low cost of the raw materials and environmentally friendly preparation process, boding well for commercializing. Zhang et al.²⁹⁴ used a probe ultrasonication technique to synthesize porous red-P nanosheets with a significantly higher photocatalytic hydrogen generation rate of 1169.4 μmol h⁻¹ g⁻¹. The BET-specific surface area of the porous nanosheets reach 84.37 m² g⁻¹ and this vast surface area significantly increases the number of reactive sites and molecule adsorption areas, leading to higher hydrogen production efficiency. Furthermore, other red-P derivatives such as fibrous phosphorene (1021 μmol h⁻¹ g⁻¹),¹³¹ and violet-P microbelts (553.4 μmol h⁻¹ g⁻¹)³⁹⁶ have also delivered excellent performance in photocatalytic hydrogen generation.

However, low-dimensional EPMs are not invincible, and they also face some challenges in photocatalytic hydrogen generation. Experimental findings show that the quantum efficiency of two-dimensional EPMs is relatively low due to the poor stability of pristine phosphorene. The electron-rich nature of the phosphorene surface and the presence of multiple unpaired lone pairs of electrons leads to the recombination of layers, preventing the materials from maintaining a long-term stability in the photocatalytic process. To overcome these issues, three potential strategies can be applied to improve the photocatalytic efficiency. First of all, chemical modification of phosphorene with hydroxyl (–OH) or pseudo-halogen (–CN and –OCN) functional groups can passivate the phosphorene surface and improve the durability.^213,398 Second, phosphorene can be utilized as a matrix to load photocatalytically active electron mediators (Ag/Au nanoparticles)^399,400 to produce co-catalytic effects in photocatalysis. Third, metal/nonmetal doping and heterostructure construction are powerful methods for regulating the electronic structure and mechanical strength of phosphorene.^401,402

Heterostructures play an important role in promoting the spatial separation of electrons and holes and the photocatalytic activity of materials like black-P. Zhu et al. have designed a binary nanohybrid composed of black-P nanoflakes and graphitic carbon nitride nanosheets to investigate the photocatalytic hydrogen performance in the presence of methanol as a hole quencher.⁴⁰³ At the interface between black-P and graphitic carbon nitride, a type-I heterojunction is formed. Under light irradiation above 420 nm, graphitic carbon nitride is the main excited materials generating electrons and holes in the conduction band and valence band, respectively. Black-P serves as an electron acceptor and accepts electrons from the conduction band of adjacent graphitic carbon nitride, while the transferred holes in black-P are quenched by methanol (Fig. 30a). This process inhibits recombination of electron–hole pairs in graphitic carbon nitride. Electrons and holes in black-P recombine rapidly under irradiation (lifetime of less than 1 ps) to prevent black-P from functioning as a photocatalyst. However, this type-I heterojunction can produce hydrogen under light irradiation above 780 nm. Time-resolved diffuse reflectance spectroscopy indicates that the P–N coordination bond at the interface between Black-P and graphitic carbon nitride can act as a trap site for electrons and captures electrons in the conduction band of black-P to generate hydrogen. This strong interfacial interaction and effective charge transfer between black-P and graphitic carbon nitride improves the photocatalytic efficiency by inhibiting the recombination of photogenerated charges in black-P. This demonstrates that the interfacial P–N coordination bond plays a critical role in photocatalytic hydrogen generation in both visible and near-infrared regions. Zhu et al. constructed 2D black-P/BiVO₄ heterostructures as a novel Z-scheme photocatalyst for overall water splitting.⁴⁰⁴ Under >420 nm light irradiation, the photocatalyst generates H₂ and O₂ yields of approximately 160 and 102 μmol h⁻¹ g⁻¹, respectively, without the need for any sacrificial agents or external bias. The staggered band positions of black-P and BiVO₄ facilitate separation of charges, leading to the reduction and oxidation of water on black-P and BiVO₄, respectively. In addition to binary heterostructures, researchers have constructed ternary nanostructures for photocatalytic hydrogen generation. Black-P–Au/La₂Ti₂O₇ shows high efficiency as a photocatalyst over a broad spectrum of wavelengths spanning the ultraviolet to near-infrared regime.⁴⁰⁵ The optimal hydrogen generation rates of black-P–Au/La₂Ti₂O₇ are 0.74 and 0.30 mmol h⁻¹ g⁻¹ at wavelengths longer than 420 nm and 780 nm, respectively. This work highlights that the strong surface plasmon resonance absorption of gold nanoparticles in the hybrid nanostructure can transform incident photon energy into plasma energy and transmit electrons to the semiconductor to induce charge separation (Fig. 30b), thus improving the photocatalytic performance. Furthermore, the strong optical absorption of black-P and plasma gold nanoparticles in the region from ultraviolet to near-infrared, along with the interfacial interaction of black-P and Au/La₂Ti₂O₇, enables effective electron transfer to La₂Ti₂O₇ by excited black-P and Au.

	Fig. 30 Applications of elemental phosphorus materials to photocatalytic hydrogen generation. (a) Proposed mechanism of photocatalytic hydrogen generation for the black-P/graphitic carbon nitride (CN) heterostructure under visible and near-infrared irradiation in the presence of methanol. Reproduced with permission from ref. 403. Copyright 2017 American Chemical Society. (b) Proposed mechanism of photocatalytic hydrogen generation for the black-P–Au/La2Ti2O7 (LTO) composite under visible (left) and near-infrared (right) irradiation in the presence of methanol. Reproduced with permission from ref. 405. Copyright 2017 John Wiley and Sons. (c) Band alignment of black-P/P3HT type-II heterojunctions. (d) Charge transfer at the heterojunction interface. Reproduced with permission from ref. 408. Copyright 2019 American Chemical Society. (e) Photoexcitation processes in black-P/carbon nitride heterojunctions. Reproduced with permission from ref. 410. Copyright 2021 John Wiley and Sons. (f) Proposed photocatalytic process of dendritic red-P (DRP). (g) Photocatalytic hydrogen evolution yield of DRP and red-P-based materials under visible light irradiation. (h) Photocatalytic stability of DRP. (i) Comparison of the visible-light photocatalytic hydrogen generation rate between DRP and other bare elemental/compound photocatalysts. Reproduced with permission from ref. 412. Copyright 2022 Elsevier.

The low-dimensional black-P materials exhibit powerful multibody effects resulting in strong exciton that dominates the photoexcitation process.^219,406,407 To enhance the photocatalytic effects of black-P, it is crucial to make reasonable use of these excitons. Xie's group has proposed a strategy to modulate exciton dissociation by heterojunction engineering,⁴⁰⁸ which combines black-P nanosheets with poly(3-hexylthiophene) (P3HT) leading to the construction of type-II heterojunctions (Fig. 30c). This strategy weakens the stability of excitons in black-P and promotes the dissociation and conversion to free carriers. The internal electric field generated at the heterojunction interface due to the difference in the band structure further promotes directional transfer of holes to P3HT and electrons to black-P (Fig. 30d), consequently increasing the output of the loaded carriers. Benefiting from this, the black-P/P3HT heterojunction shows efficient exciton dissociation and greatly improved photoresponsive properties of black-P such as a photocurrent switching ratio of approximately 18.3. Excitons can undergo resonant transfer between systems possessing compatible spin states and energy levels.⁴⁰⁹ In 2021, Xie's group proposed an exciton-mediated energy transfer strategy in heterojunction engineering for efficient infrared-light-driven photocatalysis.⁴¹⁰ They constructed a black-P/polymeric carbon nitride heterojunction in which black-P acted as an infrared light absorber and transferred its energy to activate polymeric carbon nitride through exciton interaction (Fig. 30e). The excitons in carbon nitride separated into free charge carriers at the heterojunction interface, where holes were further transferred into black-P and electrons were transferred into carbon nitride to achieve great photocatalytic efficiency under infrared light irradiation.

The strategy has also been widely used for other EPMs, and Yu's group has made contributions in the preparation of crystalline red-P composites to promote sustainable application of low-cost and effective materials in photocatalysis. In 2016, they used different synthetic methods to prepare two kinds of fibrous-P-based materials, namely, micro-fibrous-P/SiO₂ composites and smashed-fibrous-P crystals.⁴¹¹ These materials show stable and high hydrogen release rates of 633 μmol h⁻¹ g⁻¹ and 684 μmol h⁻¹ g⁻¹, respectively. The high photocatalytic efficiency is due to the high conduction band minimum of fibrous-P (−0.9 eV vs. NHE) and the favorable full contact of the small-size microstructure with water. One year later, they developed a crystalline red-P/g-C₃N₄ composite with a record-high hydrogen generation rate of 2565 μmol h⁻¹ g⁻¹.³⁹⁴ The apparent quantum yield of the red-P/g-C₃N₄ composite is 13.8% at 420 nm, which is about 3.5 times that of pristine g-C₃N₄ (3.9%).³⁹⁴ The great catalytic performance of the composite is attributed to the formation of chemical bonds between red-P (fibrous phase phosphorus) nanoparticles and g-C₃N₄, which reduces the number of structural defects in g-C₃N₄. This, in turn, inhibits the inactive charge capture and prolongs the lifetime of the active charge in g-C₃N₄, resulting in improved photocatalytic activity. Recently, Yu's group⁴¹² has developed a bismuth-catalyzed CVD method to prepare violet-fibrous hetero-phase dendritic structures from amorphous red-P for photocatalytic hydrogen generation (Fig. 30f). The hetero-phase dendritic catalyst shows stable and efficient photocatalytic activity such as a hydrogen generation rate of 1280 μmol h⁻¹ g⁻¹, which is about 16 times greater than that of amorphous red-P and exceeds those of many elemental and bare compound photocatalysts (Fig. 30g–i). The alignment of band structures between two crystalline phases and intimate heterojunction binding at the stem-branch intersection contributes to the construction of a well-organized structure that reduces the interfacial resistance, promotes charge transfer, and inhibits charge trapping and recombination. Other phosphorus-based composites have also been explored for photocatalytic hydrogen generation, for example, red-P/YPO₄,³⁹⁰ fibrous-P/TiO₂,⁴¹³ violet-P/polymeric C₃N₄,⁴¹⁴ blue-P/Mg(OH)₂,⁴¹⁵ blue-P/g-C₃N₄,⁴¹⁶ and green-P/MoSe₂.⁷⁵

Photocatalytic CO₂ conversion is more challenging than photocatalytic H₂ generation. CO₂ is a non-polar linear molecule with a symmetric structure, and the CO bond energy is 804.4 kJ mol⁻¹ at 298 K, indicating that a large amount of energy is required to break the bond.^423,424Eqn (1)–(5) show the electrode equations and potentials (V vs. NHE) for CO₂ reduction to free radicals and typically produced compounds.^425,426 Although CO₂ can form a free radical by accepting one electron directly as shown in eqn (1), the reduction potential of −1.90 V is substantially negative, indicating a higher positive Gibbs free energy required, which is thermodynamically unfavorable. As a result, photocatalytic CO₂ reduction generally occurs via a proton-coupled multielectron transfer process.⁴²⁵ The typical reduction products are methane, methanol, carbon monoxide, and formic acid (eqn (2)–(5)). In addition to producing photoexcited electrons and delivering them to adsorbed carbon dioxide molecules for activation and reduction, the photocatalyst also plays a crucial role in stabilizing the corresponding intermediates in their respective reduction pathways, thus increasing the reaction selectivity.⁴²⁷

	(1)

	(2)

	(3)

	(4)

	(5)

In 2020, Zhu et al. reported a hydroxyl-modified black-P monolayer showing outstanding photocatalytic efficiency, and it was capable of converting CO₂ into CO with a generation rate of 112.6 μmol h⁻¹ g⁻¹.⁴²⁸ This value is four times that of bulk black-P, and comparable to that of 2D g-C₃N₄. Competing reactions leading to the production of by-products such as H₂ are constantly in progress. However, hydroxyl-modified black-P exhibits a better reaction selectivity to CO (90.8%) than pristine black-P. This can be attributed to the electron-donating character of a hydroxyl group (Fig. 31a), resulting in the formation of modified black-P with a higher bandgap (∼2.2 eV) than the unmodified one (2.0 eV). This elevated bandgap makes black-P more accessible to the CO₂/CO redox potential. In addition, hydroxyl-modified black-P has outstanding stability due to surface passivation, and its degradation rate is 8.8 times slower than the pristine black-P monolayer during light cycling measurements. This shows that an appropriate choice of electron-acquiring/electron-donating functional groups influences the electron distribution of the materials. Reasonable control over the reaction conditions can lead to the formation of elemental homojunction comprising mixed crystalline phases of red-P and black-P. Band alignment of red-P and black-P allows for the formation of a Z-scheme structure and introduction of another suitable semiconductor into the system results in the formation of a ternary dual Z-scheme system. Chai's group designed a dual Z-scheme homo-heterojunction of red-P/black-P@WO₃,⁴²⁹ with great photocatalytic efficiency under visible light irradiation by converting CO₂ into CH₄ with a yield of 6.21 μmol g⁻¹ within 6 hours, whereas the original components, red-P, WO₃, and red-P/black-P, do not produce any CH₄. In the dual Z-scheme system, red-P acts as a key bridge connecting black-P and WO₃, forming a unified and synergistic ternary homo-heterojunction composite with the appropriate bandgap energy for CO₂ reduction (Fig. 31b). Specifically, red-P has a higher conduction band position than black-P and WO₃,⁴³⁰ indicating its more negative reduction potential. Under visible light irradiation, black-P, red-P, and WO₃ generate charges in their respective conduction bands and valence bands. The photoinduced electrons in the conduction bands of black-P and WO₃ follow the Z-scheme vectorial transfer pathway to reach the valence band of red-P, where they recombine with holes in red-P. This effective separation and directional transport of the photoinduced electrons and holes in the ternary system contribute to the continuous progress of the photocatalytic reduction reaction.

	Fig. 31 Applications of elemental phosphorus materials to photocatalytic carbon dioxide conversion. (a) Charge difference of O2 adsorbed on the black-P monolayer (M-BP) (left) and the hydroxyl-modified black-P monolayer (M-BP-OH). Reproduced with permission from ref. 428. Copyright 2020 Elsevier. (b) Scheme of the dual Z-scheme homo-heterojunction of red-P/black-P@WO3 for photocatalytic CO2 reduction. Reproduced with permission from ref. 429. Copyright 2022 American Chemical Society. (c) Photocatalytic CO2 reduction rates of single-atom Au/red-P (Au1/RP). (d) Comparison of TOF for photocatalytic CO2 reduction of single-atom Au/red-P and other photocatalysts. (e) CO2 adsorption curves of red-P and single-atom Au/red-P. (f) Proposed mechanism of single-atom Au/red-P for C–C coupling. Reproduced with permission from ref. 432. Copyright 2022 American Chemical Society. (g) Band structure alignment of the red-P/Bi2MoO6 heterostructure following the type-II mechanism. (h) The product formation rate of red-P/Bi2MoO6 in different cycles. (i) Proposed mechanism of photocatalytic ethanol production for red-P/Bi2MoO6. Reproduced with permission from ref. 434. Copyright 2022 Royal Society of Chemistry.

Generating C₁₊ compounds in the photocatalytic reduction of CO₂ is typically more challenging than producing C₁ compounds (CO, CH₄, and CH₃OH). This process involves multiple electron and proton transfer reaction pathways, resulting in unsatisfactory catalytic efficiency and selectivity. Nonetheless, several C₂ compounds such as ethanol are crucial industrial chemicals and potential energy storage materials, and it is imperative to produce them in a cost-effective and environmental-friendly manner. Photocatalytic reduction of CO₂ to C₂ compounds involves a C–C coupling process, which requires overcoming a larger energy barrier during the reaction.⁴³¹ However, EPMs exhibit great potential for photocatalytic C₂ compound production based on the above discussion. From the physical perspective, EPMs are semiconducting materials with relatively narrow bandgaps, making them efficient for light harvesting in the infrared-visible light range, and they possess suitable conduction band edges for CO₂ photoreduction. From the chemical point of view, electron-rich EPMs can adsorb CO₂ on the surface through Lewis acid–base interactions, interact with O in the molecule, promote cleavage of C–O bonds, and generate large C₁ intermediates.^432,433 These physical and chemical properties have recently been exploited by some research groups.

In 2022, Li's group designed a single-atom Au loaded red-P photocatalyst for efficient ethane production.⁴³² The optimized single-atom Au/red-P catalyst exhibits a high ethane yield of 1.32 μmol g⁻¹ h⁻¹ without the need of any sacrificial agents. In contrast, the yields of CO and methane are only 0.02 and 0.03 g⁻¹ h⁻¹, respectively, indicating excellent selectivity (96%) for ethane production (Fig. 31c). The turnover frequency (TOF) of the catalyst is also noteworthy, reaching as high as 7.39 h⁻¹, which exceeds those of many C₁ compound conversion photocatalysts (Fig. 31d). Various characterization techniques have been employed to demonstrate that CO₂ preferentially adsorbs on the surface of red-P (Fig. 31e) and provides active sites near each single-atom Au. This proximity effectively reduces the transfer distance of photoinduced electrons and C₁ intermediates. The presence of single-atom Au aids the desorption and movement of the C₁ intermediate at the phosphorus reactive site, followed by C–C coupling at another phosphorus site (Fig. 31f).⁴³² Das et al. have developed a type-II red-P/Bi₂MoO₆ heterostructure photocatalyst capable of converting CO₂ into ethanol under visible light irradiation and water conditions (Fig. 31g).⁴³⁴ The composite shows an excellent ethanol conversion rate of 51.8 mmol g⁻¹ h⁻¹, whereas pristine red-P shows no detectable ethanol production as it mainly produces methane and formic acid after light irradiation for 10 hours. To expand the applicability of the catalyst, researchers have investigated the sunlight-directed photoreduction performance of the optimized red-P/Bi₂MoO₆ composite. The results reveal an ethanol yield of 34.1 mmol g⁻¹ h⁻¹, which is 90.91% of that under Xe lamp irradiation. To evaluate the catalytic stability, the optimized composite undergoes five consecutive photocatalytic cycles and a continuous 30 hour catalytic test. The catalyst shows great recyclability and stability in CO₂ photoreduction and no noticeable change in the photoactivity (Fig. 31h). In the catalytic process, transformation of the C₁ intermediate is crucial for ethanol production. It has been found that using methane as a reactant does not lead to ethanol formation. However, a 300 mM methanol solution produces a high yield of 996.3 mmol h⁻¹ g⁻¹ of ethanol, indicating that methanol plays a crucial role as an intermediate in ethanol formation. A new ethanol formation mechanism has been proposed for the red-P/Bi₂MoO₆ heterostructure based on the kinetic isotope effects and in situ diffuse reflectance infrared Fourier transform spectroscopy, showing that one OCH₃* on the Bi atom and another OCH₃* on the Mo atom interact with the adjacent OH* groups on red-P and promote C–C coupling and C–H bond cleavage in OCH₃.⁴³⁴ This leads to the formation of a C₂H₄O₂* intermediate with protons adsorbed on O centers (Bi) and C centers (Mo–O), ultimately resulting in ethanol production (Fig. 31i). The red-P in the composite provides strong visible light harvesting capability and abundant active sites, forms a heterojunction interface with Bi₂MoO₆ nanoparticles, prevents recombination of photoinduced charges, and improves the charge transfer efficiency of the system.

The electrocatalytic HER mechanism is a multi-step reaction that typically takes place on the surface of the catalytic electrode layer.⁴³⁸ It involves three main steps (adsorption, reduction, and desorption) and requires either hydronium ions in acidic electrolytes or water molecules in basic media (Fig. 32a).⁴³⁹ The initial step in this process is the Volmer reaction (eqn (6) and (7)), in which a proton adsorbs at the active sites of the electrocatalyst and accepts an electron from the external circuit to form an adsorbed hydrogen atom (H*). Subsequently, a hydrogen molecule is generated through the Heyrovsky (eqn (8) and (9)) or Tafel (eqn (10)) routes, or both. In the Heyrovsky route, another proton migrates to the electrode surface to capture a second electron, which then combines with the adjacent H* to produce a hydrogen molecule. In contrast, in the Tafel route, two neighbouring H* interact on the electrode surface to generate H₂. Finally, hydrogen molecules are released from the surface of the catalytic electrode. The electrocatalytic HER can be written as follows:^435,440

	Fig. 32 Applications of elemental phosphorus materials to electrocatalytic hydrogen evolution. (a) Scheme of the hydrogen evolution mechanism on the surface of a catalyst in acidic (left) and basic (right) media. Reproduced with permission from ref. 439. Copyright 2019 Elsevier. (b) Corresponding Tafel plots for Ni2P@black-P (BP) and Pt/C. (c) Time-dependent current density curves of Ni2P@black-P under an overpotential of 107 mV. The inset shows the TEM image and SAED pattern after continuous sweeps for 24 hours. Reproduced with permission from ref. 444. Copyright 2016 John Wiley and Sons. (d) Voltammetry of different materials for the electrocatalytic HER at a scan rate of 5 mV s−1 in 1 M KOH. (e) Tafel plots derived from (d). Reproduced with permission from ref. 446. Copyright 2018 Royal Society of Chemistry. (f) Voltammetry of porous red-P nanoparticles (P-RPNPs) and commercial red-P at a scan rate of 5 mV s−1. (g) Corresponding Tafel plots for porous red-P nanoparticles and red-P. Reproduced with permission from ref. 448. Copyright 2020 Royal Society of Chemistry. (h) Corresponding overpotentials at a current density of 10 mA cm−2 (η10), (i) Tafel plots, and (j) TOF plots of four electrocatalysts. (k) The η10vs. Tafel slope of SAP-Mo2C-CS compared to the reported TMC-based catalysts. Reproduced with permission from ref. 454. Copyright 2020 John Wiley and Sons.

(1) Electrochemical hydrogen adsorption (Volmer reaction)

(2) Electrochemical desorption (Heyrovsky reaction)

(3) Chemical desorption (Tafel reaction)

Based on the analysis above, EPMs exhibit electron-rich properties, abundant active sites, and reasonable stability, boding well for the HER. Black-P has been extensively studied as an electrocatalyst or as a conductive scaffold with a large surface area. Luo et al.⁴⁴⁴ utilized a simple solvothermal approach for preparing a 0D-2D Ni₂P@black-P heterostructure and demonstrated its excellent electrocatalytic activity for the HER. In the synthesis process, black-P nanosheets are used as precursors to react with nickel chloride, resulting in the formation of Ni₂P nanocrystals that are monodispersed and incorporated in the black-P nanosheets to create a co-catalytic system. The electrocatalytic performance of Ni₂P@black-P was evaluated using a conventional three-electrode setup with a 0.5 M sulfuric acid solution. The overpotential of Ni₂P@black-P is only ∼107 mV at 10 mA cm⁻², which is substantially lower than that of commercial Ni₂P particles (∼311 mV) and pristine black-P nanosheets (600 mV). As shown in Fig. 32b, Ni₂P@black-P shows a small Tafel slope (38.6 mV dec⁻¹), which is comparable to that of a commercial Pt electrocatalyst (20 wt% Pt/Vulcan XC-72, 30 mV dec⁻¹), and the entire electrocatalytic process follows the Volmer–Tafel pathway.^444,445 Furthermore, the Ni₂P@black-P electrode maintains a stable current density at an overpotential of 107 mV and a well-defined crystalline morphology even after continuous scanning for 24 hours (Fig. 32c). The high performance of Ni₂P@black-P is related to the well-designed heterostructure which efficiently prevents Ni₂P nanoparticles from agglomeration and black-P nanosheets from restacking, thereby increasing the surface area and abundance of active sites. Additionally, the presence of Ni₂P tethered to the black-P surface modulates the charge carrier concentration and optimizes the HER kinetic rate. The electrocatalytic HER performance of NH₂-functionalized black-P nanosheets in alkaline electrolytes is also outstanding as exemplified by an overpotential of 290 mV at 10 mA cm⁻² (Fig. 32d).⁴⁴⁶ The Tafel slope of NH₂-black-P is 63 mV dec⁻¹ (Fig. 32e) indicative of the Volmer–Heyrovsky mechanism. Moreover, NH₂-black-P exhibits good stability in the 12 hour chronoamperometric test, and the polarization curve remains nearly unchanged after 1000 consecutive cycles. Li et al.⁴⁴⁷ have recently employed a modified CVT approach to synthesize black-P bulk crystals cost effectively, which are peeled off into nanostructured sheets for use as an HER electrocatalyst. The improved electrocatalytic activity of black-P nanosheets derived from low-cost bulk crystals can be attributed to the presence of more edge defects than in high-cost black-P, abundant exposed active sites, and promoted charge transport. Compared to the traditionally high price of black-P bulk crystals (about $600 per gram) made from ultrapure red-P,⁹³ there is potential to reduce the manufacturing costs of black-P and in fact, cost reduction by hundreds of times will promote widespread use of black-P-based electrocatalysts in industrial production.

Porous red-P nanoparticles have been reported as a potential metal-free electrocatalyst for the HER.⁴⁴⁸ Uniform porous red-P nanoparticles have been prepared by a simple one-pot injection method involving decomposition of the precursor P_xI_y and adding oleic acid for shape control. To evaluate the electrocatalytic performance, the nanoparticles are dissolved in toluene as ink and applied to enable red-P to stick to the surface of the working electrode. As shown in Fig. 32f, the porous red-P nanoparticles show an overpotential of 218 mV at a current density of 10 mA cm⁻², which is substantially lower than that of commercial red-P (523 mV), indicating that the porous morphology improves the HER electrocatalytic efficiency. At this current density, the overpotentials of most metal-free electrocatalytic materials such as CNTs⁴⁴⁹ and N-S co-doped graphene⁴⁵⁰ are in the range of 200–425 mV,⁴⁵¹ and porous red-P nanoparticles have excellent properties in this range, which are even comparable to metal catalysts such as MoS₂ nanosheets with an overpotential of 205 mV.⁴⁵² The Tafel slope is 156 mV dec⁻¹ (Fig. 32g), suggesting that the overall reaction rate is controlled by diffusion.^441,448 The long-term stability of porous red-P nanoparticles is confirmed in an acidic medium after 45 hours at a static potential of −0.5 V (vs. RHE) and the cathode current density remains constant even after 6000 cyclic voltammetry cycles from −0.6 to 0.2 V.

Single-atom catalysts have garnered significant attention in electrocatalysis due to their exceptional atom utilization efficiency and notable reaction selectivity.⁴⁵³ However, the majority of single-atom catalysts employ metal elements and the synthesis of non-metal single-atom catalysts is uncommon. Nevertheless, the electron-rich active sites of non-metal catalysts exhibit reasonable interactions with hydronium ions in acidic media, indicating their great potential for the electrocatalytic process. To maximize the utilization efficiency of phosphorus in the electrocatalytic process, Fu et al. prepared a highly dispersed single-atom phosphorus embedded on a single-crystal Mo₂C nanosheet array supported by a carbon sheet (SAP-Mo₂C-CS) and demonstrated the robust performance as an electrocatalyst for the HER.⁴⁵⁴ As shown in Fig. 32h, SAP-Mo₂C-CS shows an overpotential of 36 mV at a current density of 10 mA cm⁻², which is comparable to that of the commercial Pt/C electrode (15 mV), and significantly lower than that of Mo₂C on carbon sheets (143 mV) and pristine carbon sheets (316 mV). The Tafel slope of SAP-Mo₂C-CS (38.1 mV dec⁻¹) indicates that the HER process mainly follows the Volmer–Tafel mechanism (Fig. 32i) and hence, recombination of the adsorbed hydrogen species determines the overall reaction rate. To elucidate the intrinsic catalytic activity of SAP-Mo₂C-CS, researchers have calculated the turnover frequency (TOF) of each active site (Fig. 32j). At an overpotential of 0.05 V, the TOF of SAP-Mo₂C-CS is 0.23 H₂s⁻¹, which is similar to the value of Pt/C (0.39 H₂s⁻¹) and considerably higher than that of Mo₂C on carbon sheets, confirming that the phosphorus single atoms are the primary active sites in SAP-Mo₂C-CS. Notably, SAP-Mo₂C-CS exhibits higher catalytic activity in acidic electrolytes than many transition metal carbides (Fig. 32k) and also maintains the stable HER activity during the chronopotentiometry test for 11 hours. To probe the mechanism for the high-performance activity of SAP-Mo₂C-CS, researchers have conducted DFT calculations to simulate the entire catalytic process. The results indicate that the presence of anchored phosphorus single atoms on Mo₂C results in chemical coupling and accompanying electron redistribution for the unique HER kinetics. This leads to a free energy of hydrogen adsorption (ΔG_H*) that is almost zero and hydrogen binding strength that is sufficient to cover the surface, but weak enough to facilitate desorption of H₂ molecules for the exceptional HER activity.

Regarding the mechanism, owing to the complexity of the basic reaction involving multi-electron transfer, a comprehensive and unified mechanism for OER electrocatalysis is yet to be established. Several studies have proposed alternative processes for the OER under different conditions. Most of the reported studies on OER electrocatalysis have been conducted under alkaline conditions, and the mechanism of OER electrocatalysis in the alkaline medium can be described as follows:^457,458

Significant efforts have been dedicated to investigating the OER electrocatalytic activity of EPMs. In 2016, Jiang et al.⁴⁶⁰ evaluated the OER performance of black-P materials and observed that even black-P bulk forms supported on Ti foil or carbon nanotubes required only 1.6 V (vs. RHE) to achieve a current density of 10 mA cm⁻², which is comparable to that of well-known commercial RuO₂ (1.59 V)⁴⁶¹ and IrO₂ (1.57 V)⁴⁶² electrocatalysts. This work demonstrates the potential of EPMs for OER electrocatalysis and encourages further research into the electrocatalytic efficiency of low-dimensional black-P and other phosphorus allotropes. Zhang's group systematically investigated the performance of few-layered black-P nanosheets in the electrocatalytic OER process.⁴⁶³ The 2D structure of nanosheets has more active sites for higher electrocatalytic efficiency than the bulk counterpart, even allowing it to compete with some heterostructure materials.^460,464,465 To examine the electrocatalytic performance of black-P nanosheets, a three-electrode setup with a gradient concentration of potassium hydroxide (KOH) electrolyte is used and surprisingly, when the electrolyte concentration goes up, the electrocatalytic capability of black-P nanosheets improves approximately linearly. Under optimal conditions, i.e., in 1 M KOH solution, black-P nanosheets exhibit an enhanced current density of 2.7 mA cm⁻² at 1.6 V potential and a reduced onset potential (∼1.45 V). The Tafel slopes of black-P nanosheets in 1, 0.2, 0.1, and 0.05 M KOH are 88, 138, 218, and 307 mV dec⁻¹,⁴⁶³ respectively, demonstrating that the OH⁻ concentration influences the electrocatalytic performance (Fig. 33a and b). The high-concentration hydroxide solution promotes electron transport between the electrolyte and electrode and also makes the nanosheets more stable. Fig. 33c shows the current density fluctuation of black-P nanosheets around 6.2 mA cm⁻² in 1 M KOH. It is higher than the current densities for lower concentrations of KOH over time. After 10000 s, the black-P nanosheets in 1 M, 0.2 M, and 0.1 M KOH show no decay in the current density, while a slight drop in the current density is observed from 0.05 M KOH. Furthermore, the polarization curves of black-P nanosheets before and after the reaction remain almost unchanged, except in 0.05 M KOH (Fig. 33d), indicating that few-layered Black-P electrocatalysts exhibit great long-term stability within a certain range of alkaline electrolyte concentrations. DFT-based theoretical calculations have shown that the oxidation level of pristine EPMs influences the electrocatalytic activity of the OER.⁴⁶⁶ Due to the lone electron pair feature of phosphorus atoms, the catalytic site shows excessive adsorption to the reaction intermediates (MOH, MO and MOOH), blocking the subsequent reactions or impeding O₂ desorption. Changes in the local oxidation degree of phosphorene can affect the adsorption energy of O*. When , the O* adsorption strength decreases rapidly with increasing . Therefore, partially oxidizing the phosphorus surfaces may alleviate the adsorption strength and increase the catalytic efficiency.

	Fig. 33 Application of few-layered black-P nanosheets to electrocatalytic oxygen evolution. (a) Polarization curves of black-P nanosheets in KOH electrolyte with varied concentrations. The inset shows their corresponding Tafel plots. (b) Calculated Tafel slope, the current density at 1.6 V, and potential at 10 mV cm−2 in relation to the concentration of OH−. (c) Electrocatalytic OER stability of black-P nanosheets in KOH electrolyte with varied concentrations. (d) Polarization curves of black-P nanosheets in corresponding KOH electrolytes before and after 10000 seconds. Reproduced with permission from ref. 463. Copyright 2017 John Wiley and Sons.

Similar to HER electrocatalysis, element doping,⁴⁶⁷ combination with other materials,⁴⁶⁸ and designing catalyst scaffolding⁴⁶⁹ are strategies to enhance the OER electrocatalytic activity of EPMs. Inspired by a novel photocatalyst of red-P–black-P heterostructure created by Shen and colleagues,⁴⁷⁰ Xu's group synthesized a series of composites of red-P–black-P chemically bonded with expanded graphite (RP-BP/EG) or carbon nanotubes (RP-BP/CNTs) using a high-energy ball milling technique and compared the OER electrocatalytic activities.^471,472 The primary factors influencing the catalytic performance of the composites are the ball-milling time and relative mass ratio of the carbon-based materials. The P–C bond cannot be formed if the ball-milling time is too short but if it is too long, the heterogeneous structure collapses, and the catalyst surface is damaged, hindering O* adsorption. Excess or insufficient EG or CNTs can make it difficult for the active sites to form or be buried. Under optimum circumstances, the overpotential of RP-BP/EG reaches 328 mV vs. RHE at 10 mA cm⁻². Although this value is similar to that of some nitrogen-doped carbon materials and surface-oxidized multiwall carbon nanotubes,^471,473,474 the overpotential of RP-BP/CNTs is only 263 mV at the same current density.⁴⁷² The Tafel slope of RP-BP/EG is smaller than that of RP-BP/CNTs, indicating that the former has a faster electron transfer rate or the two systems have different rate-determining steps. However, the strength of the P–C bond varies due to the diversity of carbon nanostructures thus altering the stability of heterostructures, and the nanostructures also impact the catalytic area and density of active sites. Further investigations are required to determine the influence of composite nanostructures on OER electrocatalytic efficiency.

EPMs have shown promise in both the HER and OER, and researchers have attempted to develop these materials as bifunctional electrocatalysts for overall water splitting. In 2018, our group synthesized monodispersed in-plane black-P/Co₂P heterostructures by a simple solvothermal method in which Co₂P nanocrystals were selectively grown on black-P edge defects.⁴⁷⁵ Co₂P in the heterostructure not only occupies the black-P defects to improve the stability, but also provides more effective electrocatalytic active sites. Moreover, the bulging morphology alleviates agglomeration of black-P nanosheets and increases the exposure of active sites and gas effusion. The electrocatalytic activities and stability for the HER and OER are improved compared to pristine black-P and they are even comparable to those of commercial electrocatalysts (Fig. 34a–d). Given the outstanding performance of black-P/Co₂P in the HER and OER, a dual-electrode system has been constructed with black-P/Co₂P as both the cathode and anode to evaluate the overall water-splitting activity in 1.0 M KOH electrolyte (Fig. 34e).⁴⁷⁵ The system achieves a current density of 10 mA cm⁻² at a voltage of 1.92 V (Fig. 34f), demonstrating the great potential of this electrocatalyst for overall water splitting. Similarly, Li et al. prepared a bifunctional HER/OER catalyst (CoFeO@Black-P) by growing amorphous multi-transition metal (cobalt and iron) oxides on 2D black-P nanosheets.⁴⁷⁶ The water electrolysis cell with CoFeO@Black-P electrodes reaches a current density of 10 mA cm⁻² at a voltage of 1.58 V in 1 M KOH, and maintains 93% of the initial current density after the 24 hour of operation (Fig. 34g and h). As shown in Fig. 34i, metastable CoFe oxides can react with the phosphorus source provided by black-P to in situ form CoFe phosphide as the active sites in HERs. On the other hand, the rich oxygen vacancies in CoFe oxides catalyze the OER through lattice oxygen oxidation mechanism. Other effective bifunctional electrocatalysts including BPQDs/MXene nanohybrids,⁴⁷⁷ black-P@N-doped graphene,⁴⁷⁸ and black-P@red-P-g-C₃N₄,⁴⁷⁹ have also been reported for overall water splitting.

	Fig. 34 Applications of elemental phosphorus materials to electrocatalytic overall water splitting. (a) HER polarization curves of pristine black-P (BP), BP/Co2P, and Pt/C in 0.5 M H2SO4 and 1.0 M KOH. (b) Corresponding Tafel plots for BP, BP/Co2P, and Pt/C in (a). (c) OER polarization curves of BP, BP/Co2P, and RuO2 in 1.0 M KOH. (d) Corresponding Tafel plots for BP, BP/Co2P, and RuO2 in (c). (e) Scheme of electrocatalytic overall water splitting for BP/Co2P. (f) Polarization curve of BP/Co2P for overall water splitting in 1.0 M KOH. Reproduced with permission from ref. 475. Copyright 2018 John Wiley and Sons. (g) Polarization curves of CoFeO@black-P and Pt/C-RuO2 for overall water splitting. (h) Stability test of CoFeO@black-P. The inset shows bubble formation from both electrodes during electrolysis. (i) Proposed mechanism of CoFeO@black-P for electrocatalytic overall water splitting. Reproduced with permission from ref. 476. Copyright 2020 John Wiley and Sons.

Understanding the mechanism of the NRR process is critical to designing an ideal NRR catalyst. Generally, nitrogen reduction on a heterogeneous catalyst surface follows either an associative or dissociative pathway, where the primary difference between the two is the time required to break the NN triple bond (Fig. 35a).^484,489 In the associative process, a nitrogen molecule initially adsorbs on the catalyst surface and is then hydrogenated. Following hydrogenation, the NN triple bond breaks, leading to the formation of ammonia molecules. Since the nitrogen molecule contains two nitrogen atoms, hydrogenation occurs through two distinct routes: distal and alternating hydrogenation. In the distal route, hydrogenation commences on the N atom farthest from the catalyst surface, and one NH₃ molecule is formed from this N atom. The hydrogenation process then proceeds to the N atom closest to the catalyst surface, generating the second NH₃ molecule. In contrast, in the alternating route, hydrogenation occurs alternately between the two N atoms, resulting in a continuous hydrogenation cycle where the first NH₃ molecule is released immediately followed by the second NH₃ molecule. On the other hand, in the dissociative process, the NN triple bond breaks before hydrogenation, leaving two adsorbed N atoms on the catalyst surface. These two atoms are then hydrogenated individually before forming NH₃ molecules. It is worth mentioning that the specific NRR mechanism (associative or dissociative) followed by different NRR catalysts is mostly driven by their atomic and electronic structures,⁴⁹⁰ which must be verified by theoretical calculations and experiments.

	Fig. 35 Applications of elemental phosphorus materials to electrocatalytic and photocatalytic nitrogen reduction. (a) Scheme of possible mechanisms of the NRR process on the catalyst surface. Reproduced with permission from ref. 489. Copyright 2019 Elsevier. (b) Faradaic efficiency and (c) NH3 yield rate of few-layered black-P nanosheets at different potentials. (d) Scheme of the possible reaction mechanisms (solid line: low-energy pathway and dotted line: unfavourable pathway) for the electrocatalytic NRR at the zigzag edge of few-layered black-P nanosheets under a N2 atmosphere. Reproduced with permission from ref. 494. Copyright 2019 John Wiley and Sons. (e) Scheme of the preparation of MnO2 nanosheets, BPQDs, and the vdW assembly of BPQDs onto MnO2 nanosheets for NRR electrocatalysis. Reproduced with permission from ref. 497. Copyright 2020 John Wiley and Sons. (f) Free energy diagrams of the NRR process on W-N3@blue-P along the distal mechanism. Reproduced with permission from ref. 498. Copyright 2020 Elsevier. (g) Photoelectrochemical NRR mechanisms of black-P under light irradiation and external electric field. Reproduced with permission from ref. 502. Copyright 2020 John Wiley and Sons.

Developing high-performance NRR catalysts is a challenging task. The cleavage of the NN triple bond requires a significant amount of energy (941 kJ mol⁻¹),⁴⁹¹ and throughout the catalytic process, the HER always competes with the NRR at the catalytically active sites,^487,488,492 leading to a catalytic selectivity issue. Therefore, a state-of-the-art NRR catalyst must not only fulfil the conventional requirements for a high density of active sites and material stability, but also exhibit exceptional catalytic selectivity. This means that the catalyst surface binding strength of nitrogen must be greater than that of hydrogen to favour electron transfer along the NRR pathway. Due to the similar valence electronic structures of phosphorus and nitrogen (3s²3p³ and 2s²2p³, respectively),^21,493 nitrogen has a stronger affinity for the surface of phosphorus-based materials, giving EPMs a unique advantage as NRR catalysts.

In 2019, Wang's group demonstrated that few-layered black-P nanosheets perform well as NRR electrocatalysts under ambient conditions.⁴⁹⁴ To determine the electrocatalytic activity, they selected 0.01 M HCl as the acidic electrolyte in a standard three-electrode setup, and a few-layered black-P suspension was dropped onto carbon fibers as the working electrode. As shown in Fig. 35b and c, the catalytic electrode achieved the highest Faradaic efficiency of 5.07% at −0.6 V, with a maximum NH₃ production rate of 31.37 μg h⁻¹ mg_cat.⁻¹ at −0.7 V. The few-layered black-P electrocatalyst outperforms its bulk counterpart (0.13% Faradaic efficiency, 0.036 μg h⁻¹ mg_cat.⁻¹), as well as various metal-free catalysts (e.g., N-doped porous carbon with a Faradaic efficiency of 1.42% and 23.8 μg h⁻¹ mg_cat.⁻¹) and metal-based catalysts (e.g., Au–TiO₂ sub-nanoclusters with a Faradaic efficiency of 8.11%and 21.4 μg h⁻¹ mg_cat.⁻¹).^495,496 Moreover, the UV/vis absorption spectra and chromogenic reactions indicate that no obvious side-product (N₂H₄) is generated throughout the electrocatalytic process, confirming the high catalytic selectivity of few-layered black-P. Based on DFT calculations for the molecular orbitals of black-P, the NRR catalytic process follows an alternating hydrogenation pathway in an association mechanism (Fig. 35d).⁴⁹⁴ This study not only demonstrates the feasibility of using black-P materials for the electrocatalytic NRR under ambient conditions, but also serves as a model for investigating the NRR mechanisms of EPMs. Zhu's group has reported a straightforward vdW assembly approach for loading BPQDs onto wrinkled MnO₂ nanosheets (Fig. 35e), which leads to enhanced NRR electrocatalytic activity compared to pristine materials.⁴⁹⁷ The maximum NH₃ generation rate of BPQDs/MnO₂ is 25.3 μg h⁻¹ mg_cat.⁻¹ and its Faradaic efficiency is 6.7% at an applied potential of −0.5 V (vs. RHE). The outstanding electrocatalytic performance can be ascribed to the strong synergistic effects of BPQDs/MnO₂.⁴⁹⁷ The wrinkled MnO₂ nanosheets act as a flexible substrate to prevent severe BPQD aggregation during the catalytic process. The edges and defect sites of MnO₂ nanosheets exhibit reasonable electrocatalytic activity. Moreover, BPQDs with an abundance of active sites serve as the primary NRR catalyst and physical spacer to prevent restacking of MnO₂ nanosheets and maintain the structural stability of the composite.

Other EPMs have also shown potential for NRR electrocatalysis. Our group has conducted a series of DFT calculations and found that crystalline red-P (fibrous phase) nanoribbons have potential electrochemical nitrogen fixation activity. On the surface of crystalline red-P nanoribbons, the NRR process can be carried out through the proton-coupled electron transfer pathway, with the associative mechanism being preferable, followed by distal hydrogenation. The initial step of adsorbed N₂ hydrogenation to NNH* is proposed to be the rate-determining step due to the largest energy input required. Experimentally, we have carried out the electrocatalytic test in a gastight H-type electrolytic cell with 0.1 M Na₂SO₄ electrolyte and placed crystalline red-P nanoribbons onto a nickel foam as the catalytic electrode. The catalytic electrode delivers efficient and durable NRR performance, at an optimal NH₃ production rate of 15.4 μg h⁻¹ mg_cat.⁻¹ at −0.4 V and a faradaic efficiency of 9.4% at −0.2 V. Compared to the few-layered black-P electrocatalyst developed by Wang's group, crystalline red-P nanoribbons exhibit a lower optimal voltage but a higher Faradaic efficiency due to the increased charge transfer and exposed-edge active sites of the quasi-1D structure.⁴⁷ However, the ammonia production rate of crystalline red-P nanoribbons is still limited and the development of red-P-based materials for NRR electrocatalysis still has a long way to go.

Compared to the deeper armchair ridges of black-P, the hexagonal lattice structure of blue-P with zigzag ridges is relatively isotropic, making it an ideal candidate for multi-electron transfer. Recently, Zhang et al. have demonstrated that a kinetically stable W-N₃ center supported by blue-P exhibited great NRR catalytic activity with a limit potential of −0.02 V based on DFT calculations.⁴⁹⁸ The W atom plays a crucial role in weakening the NN triple bond by accepting lone electron pairs from N₂ and donating electrons to the anti-bonding orbitals of N₂,⁴⁹⁹ enabling N₂ molecules to readily adsorb on the composite surface. The Gibbs free energy variations of the intermediates suggest that the associative distal mechanism is the most favourable in this system (Fig. 35f). Additionally, the high selectivity of the NRR can be achieved due to the high free energy barriers for the HER (0.83 eV) and the ultralow free energy barriers for the NRR (0.02 eV). This work provides theoretical guidance for the development of blue-P-based NRR electrocatalysts.

The solar-powered photocatalytic NRR is a promising low-cost, green, and sustainable strategy for ammonia production. However, the high energy barrier of the six-electron reaction that activates and breaks the strong NN bond by photocatalysts, hinders the reaction kinetics, resulting in relatively low photocatalytic NRR efficiency. Thus, developing highly active photocatalysts is crucial for achieving an efficient photocatalytic NRR. Our group has synthesized edge-rich black-P nanoflakes using a simple chemical etching exfoliation method, which achieves an efficient photocatalytic NRR without using any cocatalysts.⁵⁰⁰ The abundant edges of the resulting nanoflakes offer numerous active sites for N₂ adsorption and complete activation of the NN bond to enhance the reduction of N₂. Moreover, the black-P nanoflakes exhibit good water dispersibility, enabling full contact with the reactants. As a result, edge-rich black-P shows a high catalytic nitrogen fixation rate of 2.37 mmol h⁻¹ g⁻¹ under visible light irradiation, which is higher than 0.9 mmol h⁻¹ g⁻¹ shown by conventional black-P nanoflakes. This work highlights the impact of black-P morphology on catalytic efficiency, and the development of black-P materials with nanostructured design for NRR electrocatalysis is a promising research direction. Alternatively, incorporating co-catalysts has demonstrated acceleration of charge transfer and improvement in photocatalytic nitrogen conversion to some extent. However, controlling the interface state and studying the charge transfer at the interface remain a challenge. To address this, Shen et al. have selectively deposited a Ni₂P layer at the edge of black-P nanosheets, forming a 2D–2D heterojunction with plentiful Ni–P bonds.⁵⁰¹ The resulting Ni₂P–BP heterojunction has improved photocatalytic NRR properties such as an NH₃ generation rate of 6.14 μmol h⁻¹ g⁻¹ under visible light irradiation, which is 1.56 times higher than that of the Ni–P bond-free Ni₂P nanoparticle loaded black-P photocatalyst. This indicates that the interfacial Ni–P bonds plays a critical role in the catalytic process by facilitating interfacial charge transfer and enabling photoexcited electrons on the conduction band of black-P to efficiently transfer to Ni₂P and enhance the separation efficiency of electron–hole pairs. Moreover, the Ni₂P layer chemically passivates the edges of the black-P nanosheets and improves the stability of the system during the electrocatalytic process.

Compared to the electrocatalytic process with continuous and stable external bias potentials, the photocatalytic NRR faces challenges of low conversion efficiency and unstable conversion processes. However, by combining the advantages of both photocatalysis and electrocatalysis, the N₂-to-NH₃ conversion efficiency can be further improved through photoelectrochemical reactions. With these principles in mind, our group has developed a black-P photoelectrochemical electrode and investigated its catalytic nitrogen conversion activity under normal environmental conditions.⁵⁰² The electrode shows remarkable catalytic activity with an ammonia yield rate of 102.4 μg h⁻¹ mg_cat.⁻¹ and a Faradaic efficiency of 23.3% at −0.4 V, outperforming most metal-free photocatalysts or electrocatalysts for the NRR.⁵⁰² Moreover, the black-P electrode exhibits great stability after 6 consecutive cycles. The exceptional NRR activity of the black-P electrode can be ascribed to the synergistic effect generated by photoexcitation and external bias (Fig. 35g), in which the photoexcitation boosts the average electron energy and charge exchange efficiency and facilitates the electrocatalytic process, while the external bias potentials improve the separation of photoexcited electron–hole pairs, promoting the photocatalytic process. Considering the excellent charge mobility of black-P and the strong visible-light absorption and chemical stability of red-P based materials such as amorphous red-P, violet-P, and fibrous-P, the future development of black-P/red-P homojunctions for photoelectrochemical ammonia production holds great promise.

This section has discussed the mechanism and performance of EPMs in photocatalytic/electrocatalytic water splitting, carbon dioxide reduction, and nitrogen fixation. It has underscored the need for a deeper understanding of reaction pathways and intermediate products in the catalytic processes involving different phosphorus materials. Attention should be given to the role of excitons in photocatalysis, the selectivity in carbon dioxide reduction, and how certain material structures lead to different products. For electrocatalysis, the establishment of standardized criteria to assess the catalytic activity of EPMs and evaluating their contributions in composite systems are areas warranting further exploration.

Solid-state storage technologies such as metal hydride techniques, physisorption, and chemical adsorption have garnered considerable attention as a potential alternative to conventional hydrogen storage technologies.⁵⁰⁷ The search for suitable solid-state storage materials is ongoing and promises to yield exciting possibilities.^508–510 According to the US Department of Energy (USDOE) targets for 2020,^511–513 high-efficiency hydrogen storage materials with a gravimetric density of 5.5 wt% should be employed with hydrogen reversibility occurring at temperatures ranging from −40 to 60 °C at a filling/releasing rate of 1.5 kg H₂ min⁻¹. This criterion requires that the energy associated with adsorbing each hydrogen molecule must be within an appropriate range to regulate the adsorption/desorption kinetics. To date, many publications have demonstrated that carbonaceous materials can be used for hydrogen storage.^514–516 However, pure carbon atoms have a low affinity for hydrogen, and therefore, metal doping is a viable strategy for enhancing hydrogen adsorption. Nonetheless, metal atoms are prone to clustering on the carbon matrix, resulting in low reversibility and an inability to meet the long-term hydrogen storage requirement.

In comparison to carbonaceous materials, metal atoms on phosphorene exhibit significantly lower tendencies for cluster formation. Although phosphorene is heavier than graphene, it is lighter than many other prominent 2D materials, including MXenes and TMDs, resulting in a high gravimetric density. In 2014, Li et al.⁵¹⁷ conducted the first study on Li-decorated monolayer black-P using theoretical calculations. Li-decorated black phosphorene exhibits a hydrogen storage capacity of 8.11 wt% and average adsorption energy of 0.18 eV per H₂, exceeding the hydrogen storage targets established by USDOE. Luo et al.⁵¹³ have demonstrated better hydrogen storage performance for Li-decorated single-layered blue-P. At a Li/P ratio of 1:1, the interaction between Li atoms and blue-P is sufficiently strong, maintaining the Li atom dispersion. Due to the space constraint surrounding Li atoms, double-layer adsorption occurred, and each atom can adsorb two molecules of H₂, leading to a hydrogen storage capacity of 9.52 wt%. This can be attributed to blue phosphorene having a flatter in-plane structure than black phosphorene, and the zigzag ridges aiding in the dispersion of Li atoms. Recently, Xu et al.⁵¹⁸ have used a solvothermal approach to prepare Co-doped black-P nanosheets and evaluated their efficacy in electrochemical hydrogen storage. The optimal hydrogen storage capacity of Co-doped black-P reaches 6200 mA h g⁻¹ (i.e., 18 wt% of composite weight), and the average capacity is ∼10.3 times that of pristine black-P nanosheets. The active catalytic sites of Co atoms that facilitate hydrogen transfer are primarily responsible for the electrochemical hydrogen storage ability of the composite. Additionally, Co doping reduces the bandgap and increases the conductivity to provide favourable conditions for electrochemical hydrogen storage. However, most of the research on hydrogen storage in phosphorene is still in the theoretical stage, and several studies based on DFT calculations have been conducted to investigate the hydrogen capture capability of phosphorene.^519–523

In LIBs, the specific capacities of anode and cathode materials primarily determine the energy capacity and charge/discharge rate. Commercially available LIBs often employ graphite anodes and lithium metal oxide cathodes, resulting in a theoretical energy capacity of approximately 400 W h kg⁻¹ and a practical capacity of around 200 W h kg⁻¹.⁵³⁰ However, this performance falls short of the increasing requirements for electrochemical energy storage devices, thus necessitating enhancements in the specific capacity of LIBs. Beyond graphite anodes, considerable efforts have been devoted to layered materials, which can be divided into four categories: 2D monoelemental materials (Xenes),⁵³¹ metal nitrides/carbides (MXenes),⁵³² transition metal dichalcogenides (TMDs),⁵³³ and transition metal oxides (TMOs).⁵³⁴ Low-dimensional EPMs have garnered significant interest as potential LIB anodes among the Xenes. For example, black-P possesses a low diffusion energy barrier for lithium ions (0.08 eV),⁵³⁵ enabling it to exhibit 10² and 10⁴ times faster lithium ion diffusion rates than MoS₂ and graphene, respectively, thereby suggesting a rapid charge/discharge rate. Moreover, the theoretical specific capacities of EPMs reach 2596 mA h g⁻¹ surpass those of most TMOs and TMDs.^536,537 These characteristics render EPMs a promising anode material for the next-generation LIBs and other electrochemical energy storage devices (Table 3).

Allotropes	Materials	Applications	Methodology	Capacity	Capacity retention rate (%) (cycle number, rate or current density)	Ref.
Black phosphorus	BP-Graphite	LIB	High energy mechanical milling	Initial discharge capacity of 2786 mA h g−1 at 0.2C	80 (100, 0.2C)	549
	Sandwich-structure G-BPGO	LIB	Vacuum-filtration	Initial discharge specific capacity of 2587 mA h g−1 at 0.1 A g−1	1401 mA h g−1 (200, 0.1 A g−1)	550
	Bi–P/C	LIB	Ball-milling method	High-rate discharge capacities of 1788.2 mA h g−1 at 13 A g−1	86.3 (300, 7.8 A g−1)	551
	BP/NiCo MOF	LIB	Solution reaction route	Reversible capacity of 853 mA h g−1 at 0.5 A g−1	91.3 (1000, 5 A g−1)	552
	BP/G/CNTs	LIB	Ball-milling method	Initial reversible capacity of 1375 mA h g−1 at 0.15 A g−1	79 (1000, 1 A g−1)	553
	(BP-G)/PANI	LIB	Sonication and ball milling	Reversible discharge capacity of 1520 mA h g−1 at 0.26 A g−1	440 mA h g−1 (2000, 13 A g−1)	554
	PCNF/S/BPQD	LSB	Sonication and wet chemistry	Rate capacity of 784 mA h g−1 at 4C	∼90 (200, 0.1C)	555
	BP-coated separator	LSB	Vacuum-filtration	Specific capacity of 725 mA h g−1 at 1.8 A g−1	86 (100, 0.4 A g−1)	556
	BPQD/δ-MnO2	LAB	Hydrothermal route	Discharge capacity of 8463 mA h g−1 at 0.1 A g−1	1000 mA h g−1 (182, 0.4 A g−1)	557
	Phosphorene-coated Li metal	LAB	Ultrasonication and spin coating	Constant discharge capacity 1000 mA h g−1 at 0.25 A g−1	100 (50, 0.25 A g−1)	558
	BP–CNT	LIB and SIB	Surface oxidation-assisted chemical bonding procedure	LIB: Initial discharge capacity of ∼1922 mA h g−1 at 0.2C	LIB: 87.5 (400, 0.2C)	559
				SIB: Initial discharge capacity of ∼2073 mA h g−1 at 0.2C	SIB: 75.3 (200, 0.2C)
	BPQD/TNS	LIB and SIB	Interfacial assembly strategy	LIB: Reversible capacity of 828 mA h g−1 at 0.1 A g−1	LIB: over 100% (2400, 1 A g−1)	560
				SIB: Initial discharge capacity of ≈723 mA h g−1 at 0.2 A g−1	SIB: nearly 100% (1000, 1 A g−1)
	4-RBP	SIB	4-NBD modification and solvothermal reaction	Initial discharge specific capacity of 2500 mA h g−1 at 0.1 A g−1	1472 mA h g−1 (50, 0.1 A g−1)	561
	Layered BP/graphene	SIB	Flash-heat-treatment and straightforward pressure synthesis	Specific charge capacity of 1377.6 and 720.8 mA h g−1 at 1 and 40 A g−1, respectively	1250 mA h g−1 (500, 1 A g−1)	562
					640 mA h g−1 (500, 40 A g−1)
	RP@BP/3DNG	SIB	Solvothermal strategy	Reversible capacity of 1440.2 and 521.3 mA h g−1 at 0.05 and 10 A g−1, respectively	89.3 (1200, 10.0 A g−1)	563
	Few-layer black phosphorene	SIB	Electrochemical exfoliation	Activated capacity of 1968 mA h g−1 at 0.1 A g−1	60.5 (50, 0.1 A g−1)	564
	Phosphorene–graphene	SIB	Self-assembly	Specific capacity of 2440 mA h g−1 at 0.05 A g−1	84 (100, 8 A g−1)	565
	Phosphorene/MXene	SIB	Self-assembly	Reversible capacity of 535 mA h g−1 at 0.1 A g−1	87 (1000, 1 A g−1)	566
	MoS2/BP	SIB	Hydrothermal method	Reversible capacity of 435.5 mA h g−1 at 1.0 A g−1 after 150 cycles	263.2 mA h g−1 (1000, 10 A g−1)	567
	BP–C	PIB	Ball-milling method	Initial specific capacity of 617 mA h g−1 at 0.05 A g−1	∼30 mA h g−1 (50, 0.05 A g−1)	568
	(P-CNTs)/PANI	PIB	Ball-milling and in situ solution polymerization	Reversible capacity of 642.7 mA h g−1 at 50 mA g−1	94 (500, 0.5 A g−1)	569
	Restacked BP nanoflakes	SC	Liquid exfoliation	Specific capacitance of 13.75 F cm−3 at 0.01 V s−1	71.8 (30000, 0.5 V s−1)	570
	BP sponges	SC	Electrochemical technique	Specific capacitance of 80 F g−1 at 0.01 V s−1	80 (15000, 0.1 V s−1)	571
	BP/CNTs	SC	High-heat treatment and microfluidic-spinning technique	Specific volumetric capacitance of 308.7 F cm−3 at 0.1 A cm−3	90.2 (10000, 0.4 A cm−3)	572
	E-BP/ZIF-67	SC	Droplet microfluidic	Specific capacitance of 506 F cm−3 at 500 mA cm−3	89.3 (12000, 2 A cm−3)	573
Amorphous red phosphorus	RP/A-TiO2	LIB	In situ hydrolyzation	Rate capacity of 202 mA h g−1 at 1 A g−1	89.3 (100, 0.1 A g−1)	574
	P/C	LIB	Vaporization-adsorption method	Specific capacity of 2173 mA h g−1 at 1C	90 (500, 0.2C)	29
	RPN/rGF	LIB	Evaporation–condensation method	Rate capacity of 1073 mA h g−1 at 3C	78.4 (300, 0.3C)	575
	RP/TiN/CNT	LIB	High-energy ball milling method	Reversible capacity of 868.7 mA h g−1 at 2.3 A g−1	96.9 (3rd ∼ 850th, 2.3 A g−1)	576
	HPNs	LIB and SIB	Wet solvothermal method	Specific capacity of 1285.7 and 1364.7 mA h g−1 at 0.2C for LIBs and SIBs, respectively	LIB: 1048.4 mA h g−1 (600, 1C)	577
					SIB: 969.8 mA h g−1 (600, 1C)
	P@CMK-3	LIB and SIB	Vaporization–condensation–conversion process	LIB: Specific capacity of 2185 mA h g−1 at 1.2C	LIB: ∼80 (1000, 1.2C)	578
				SIB: Specific capacity of 2331 mA h g−1 at 0.6C	SIB: 1600 mA h g−1 (140, 1C)
	BP/RP	LIB and SC	Sono-chemical process	LIB: Initial discharge capacity of 2449 mA h g−1 at 0.05 A g−1	LIB: 491 mA h g−1 (100, 0.05 A g−1)	579
				SC: Specific capacitance of ≈60.1 F g−1 at 0.5 A g−1	SC: 83.3 (2000, 1 A g−1)
	RPEN@CF	LSB	Laser-assisted exfoliation	Specific capacity of 782.3 mA h g−1 at 3C	769.5 mA h g−1 (500, 1C)	580
	P@C	SIB	Shear emulsifying and electrospinning	Reversible capacity of 1308 mA h g−1 at 0.2 A g−1	81 (1000, 2 A g−1)	581
	P-CNT	SIB	Ball-milling method	Specific capacity of 2134 mA h g−1 at 0.26 A g−1	∼91 (100, 0.52 A g−1)	582
	P@RGO	SIB	Physical vapor deposition	Specific capacity of 1165.4 mA h g−1 at 159.4 mA g−1	87.9 (300, 1.5939 A g−1)	583
	Pred@CNF	SIB	Vaporization–condensation method	Specific capacities of ∼1850 mA h g−1 over 500 cycles at 0.1 A g−1	>1000 mA h g−1 (5000, 1 A g−1)	584
	NPRP@RGO	SIB	Redox reaction and boiling process	Specific capacity of 1848.67 mA h g−1 at 0.2563 A g−1 after 150 cycles	775.3 mA h g−1 (1500, 5.12 A g−1)	585
	RP@Ni–P	SIB	Electroless deposition and chemical dealloying	Reversible specific capacity of 1256.2 mA h g−1 at 0.26 A g−1	409.1 mA h g−1 (2000, 5 A g−1)	586
	P@N-MPC	SIB	Vaporization-condensation method	Specific capacity of ≈600 mA h g−1 at 0.15 A g−1	∼450 mA h g−1 (1000, 1 A g−1)	587
	P2@N-SGCNT	SIB and PIB	High-temperature infiltration	SIB: Reversible capacity of 2480 mA h g−1 at 0.1 A g−1	SIB: 1936 mA h g−1 (2000, 1 A g−1)	588
				PIB: Reversible capacity of 762 mA h g−1 at 0.05 A g−1	PIB: 319 mA h g−1 (1000, 2 A g−1)
	P50@ZRCod-0.025	PIB	Solvothermal synthesis	Reversible capacity of 595.8 mA h g−1 at 50 m A g−1	150.7 mA h g−1 (400, 2.5 A g−1)	589
	Red P@N-PHCNFs	PIB	Vaporization–condensation	Reversible capacity of 650 mA h g−1 at 0.1 A g−1	84 (200, 1 A g−1)	590
	H-P@NCNS/NCNT	PIB	Magnetic field assisted methodology	Reversible charge capacity of ≈527 mA h g−1 at 2.0 A g−1	≈85% (500, 2 A g−1)	591
	R-BP/SPC	SC	Sonochemical process	Specific capacitance of 364.5 F g−1 at 500 mA g−1	89 (10000, 2 A g−1)	592
	P-rGO	SC	Potentiodyna-mic deposition	Specific capacitance of 334.2 F g−1 at 1 A g−1	99.2 (10000, 1 A g−1)	593
Fibrous phosphorus	Fibrous phosphorus	LIB	Iodine-assisted CVD	Initial discharge capacity of 1783 mA h g−1 at 0.1 A g−1	75 (10, 0.1 A g−1)	305
	FP-C	LIB	CVT	Reversible specific capacity of 1621 mA h g−1 at 0.2 A g−1	742.4 mA h g−1 (700, 2 A g−1)	594
	MWCNT@f-RP@BP hybrid	LIB	CVD	Initial discharge capacity of ∼1333 mA h g−1 at 0.5 A g−1	560 mA h g−1 (250, 0.5 A g−1)	595
Violet phosphorus	Hittorf's phosphorus	LIB	Iodine-assisted CVD	Initial discharge capacity of 2113.8 mA h g−1 at 0.1 A g−1	558 mA h g−1 (40, 0.1 A g−1)	305
	Hittorf's violet phosphorene	LIB and SIB	DFT calculations	Adsorption energies of 1.65 eV for Li and 1.32 eV for Na (theoretical)	—	596
Blue phosphorus	Single-layer blue phosphorus	LIB	DFT calculations	Charge capacity of 865 mA h g−1 (theoretical)	—	597
	Double-layer blue phosphorus	LIB	DFT calculations	Charge capacity of 649 mA h g−1 (theoretical)	—	597
	BlueP/graphene heterostructure	LIB	DFT calculations	Specific capacity of 626 mA h g−1 (theoretical)	—	598
	BlueP/MS2 heterostructures (M = Nb, Ta)	LIB	DFT calculations	Specific capacities of 528.257 mA h g−1 for BlueP/NbS2 and 392.154 mA h g−1 for BlueP/TaS2 (theoretical)	—	599
	Blue phosphorene	LSB	DFT calculations	Adsorption energies in the range of −0.86 to 2.45 eV for lithium polysulfides	—	600
	2D-N-doped blue phosphorus	LAB	DFT calculations	Storage capacity of >314.22 mA h g−1 (theoretical)	—	601
	h-BN/blue-P heterostructure	LIB and SIB	DFT calculations	Specific capacities of 801 mA h g−1 for Li and 541 mA h g−1 for Na (theoretical)	—	602
	C3N/blue-P heterostructure	LIB and SIB	DFT calculations	Specific capacities of 333 mA h g−1 for Li and 658 mA h g−1 for Na (theoretical)	—	603
	Monolayer blue phosphorene	SIB and PIB	DFT calculations	Specific capacities of 865 mA h g−1 in both cases (theoretical)	—	540
Green phosphorus	Single-layer GP	LIB	DFT calculations	Specific capacity of 432.7 mA h g−1 (theoretical)	—	74
	2D green phosphorene	SC	DFT calculations	Specific capacitance of ≈90 F g−1 (theoretical)	—	604

The working mechanism of LIBs is characterized by two primary phases: lithiation and delithiation.^538,539 During the charging process, energy is stored as chemical energy and an external current drives lithium ions to the anode, where they are intercalated into the phosphorus interlayer, forming a new phase of Li₃P. During the charging process, the thermodynamically favourable delithiation continues. Li₃P progressively decomposes, and the intercalated Li atoms contribute their valence electrons to the external circuit, resulting in the formation of Li⁺. Eventually, the contributed electrons congregate at the cathode, dragging Li⁺ across the electrolyte and accumulating in the cathode materials. Given the significant volume changes that occur during each lithiation/delithiation process, maintaining the anode structure is critical to achieving long-cycle performance in LIBs.

Experimental studies have been conducted on black-P, amorphous red-P, violet-P, and fibrous-P as anode materials for LIBs, with blue-P and green-P LIBs being theoretically proven to be viable.^74,540 In the past few years, black-P has garnered significant interest, and the mechanism of the lithiation/delithiation process in black-P has been more explicitly defined. The structure of black-P is highly anisotropic, and lithium ions prefer diffusing along the shallower zigzag ridges rather than the armchair ridges with a relatively higher energy barrier of 0.68 eV.⁵³⁵ Moreover, a semiconductor-to-metallic transition occurs during the Li⁺ intercalation process,⁵⁴¹ leading to increased electronic conductivity of the black-P anode. Parker et al.⁵⁴² were the first to report the black-P-based anodes for LIBs and found that bulk black-P exhibited a reversible rated charge capacity of 1279 mA h g⁻¹. While this value exceeds that of commercial graphite (372 mA h g⁻¹), it only reaches half of the theoretical capacity, and the initial cycle efficiency remains relatively low (57%). In 2019, Chen et al.³⁰⁵ synthesized fibrous-P and violet-P bulk crystals through a CVD method and demonstrated that these two EPMs have the potential to be LIB anodes as well. The first discharge capacity of the fibrous-P anode is 1783 mA h g⁻¹ and that of the violet-P anode is 2113.8 mA h g⁻¹, with Coulombic efficiencies of 81.04% and 75.76%, respectively. However, fibrous-P and violet-P anodes display irreversible capacities after several dozen cycles, limiting their long-term capacity retention. Generally, pristine bulk materials lack rational structural designs,⁵⁴³ and abrupt pulverization of the anode materials during the delithiation process leads to undulation of the lattice structure, resulting in a substantial volume expansion of up to 300%.⁵⁴⁴ This can lead to the loss of electrical contact within the active material, resulting in a rapid decay of the battery capacity.

Nanostructured design can enhance electrochemical reactions and improve active material utilization.^545,546 Among various nanostructured strategies, exfoliating the bulk phase into 2D nanosheets is a widely adopted method that can decrease electrode thickness, reduce electron and ion diffusion distances within the electrode layer, and ultimately enhance the volumetric energy density of LIBs. However, 2D phosphorene nanosheets with large specific surface areas suffer from restacking and detrimental parasitic reactions with the electrolyte. These reactions often occur at the edge of the phosphorene zigzag diffusion channel, resulting in the distortion of the surrounding phosphorus atoms.⁵⁴⁷ Alternatively, the development of 3D electrode architectures is another design option for enhancing performance. Zhu et al. have developed a green and template-free hydrothermal approach to synthesize honeycomb-like micron-sized red-P with a tunable porous architecture.⁵⁴⁸ The optimal porous red-P anode delivers exceptional lithium storage performance, achieving a substantial reversible capacity of 2587.4 mA h g⁻¹ at 50 mA g⁻¹ and a capacity retention of ∼81.9% at 500 mA g⁻¹ after 500 cycles. The results from in situ TEM and ex situ SEM measurements indicate that the porous structure buffers the volume variations throughout the lithiation/delithiation process, improving the long-cycle stability of the anode materials.

Currently, commercial LIBs offer limited power densities typically ranging from 100 to 300 W kg⁻¹ and require long charging time to ensure safe operation.^605,606 To enhance battery utilization efficiency in portable electronics and electric vehicles, it is crucial to develop high-energy-density LIBs that possess superior fast-charging capabilities. EPMs are promising candidates as anode materials due to their high theoretical specific capacity (2596 mA h g⁻¹) and moderate lithiation potential (0.7 V vs. Li⁺/Li). However, challenges such as limited conductivity, sluggish reaction kinetics, and unstable solid-electrolyte interphase (SEI) impede the high-rate capability of the phosphorus anode during the lithiation/delithiation process. Despite self-structure redesign efforts, there are still limitations, and new strategies are required to optimize the performance of EPMs as anode materials in fast-charging LIBs.⁵²⁶ Carbonaceous materials possess good electrical conductivity and material stability and their incorporation into the anode can effectively improve the fast-charging capabilities of EPMs. In carbon–phosphorus (P/C) composites, the presence of high-conductive carbon-rich materials amplifies lithium ion diffusion in the anode by reducing the polarization effect,⁶⁰⁷ and carbon-rich materials also act as an external buffer layer to control volume changes.⁶⁰⁸ Cui's group designed a red-P/C nanocomposite for fast-charging LIBs by an adsorption and surface cleaning approach.²⁹ The average thickness of red-P/C electrodes is 21.5 μm, featuring an industrial-grade areal capacity of around 3.5 mA h cm⁻² at 0.5 mA cm⁻². This thickness is considerably less than that of the commercial graphite (76.3 μm) and Li₄Ti₅O₁₂ (124.5 μm) electrodes at the same areal capacity (Fig. 36a). Consequently, the total volumetric capacity of the red-P/C electrode is 1628 mA h cm⁻³, far surpassing that of graphite (459 mA h cm⁻³) and Li₄Ti₅O₁₂ (281 mA h cm⁻³) electrodes (Fig. 36b). The red-P/C nanocomposite also exhibits an extended cycle life. After 500 discharge–charge cycles, the red-P/C electrode shows a specific capacity of 1625 mA h g⁻¹ and retention rate of around 90% at 0.86 mA cm⁻² (0.2C). Following long-cycling, the red-P/C electrode retains its morphological integrity and electrical connectivity without fractures or contact losses after long-cycling tests. A uniform and thin solid-electrolyte interphase layer (<50 nm) is formed on the surface of a red-P/C nanoparticle, while the interior nanoscale void spaces are well preserved to buffer the volume changes after 100 cycles.²⁹ These findings demonstrate that the red-P/C nanocomposite is structurally robust at both the electrode and particle levels, enabling stable battery cycling and high specific capacity at relatively high current densities.

	Fig. 36 Applications of elemental phosphorus materials to LIBs. (a) Cross-sectional SEM images of P/C, graphite, and Li4Ti5O12 anodes with the same areal capacity of 3.5 mA h cm−2 at 0.5 mA cm−2. (b) Comparison of capacities and volumetric capacities of P/C, graphite, and Li4Ti5O12 anodes with the same areal capacity of 3.5 mA h cm−2 at 0.5 mA cm−2. Reproduced with permission from ref. 29. Copyright 2019 Elsevier. (c) Scheme of the synthesis of Bi–P/C composites. (d) Long-cycling stability of Bi–P/C and P/C at 7.8 A g−1. (e) Corresponding voltage profiles of Bi–P/C at 7.8 A g−1. (f) Rate performance comparison between Bi–P/C and P/C anodes across a range of rates from 0.052 to 13 A g−1. Reproduced with permission from ref. 551. Copyright 2022 John Wiley and Sons.

Sun et al. have proposed that the formation of robust P–C bonds in P/C composites can minimize electrical contact loss between the active materials and the carbon conductor and improve the cyclic stability of P/C composites.⁵⁴⁹ They evaluated the impact of the carbon structure on stable P–C bond formation by comparing four common carbon materials: graphite, graphite oxide, carbon black, and fullerene. Their findings demonstrate that the black-P–graphite composite has the greatest proportion of P–C bonds, making it an optimal choice as a LIB anode material with extended cycle life and high-rate performance. At a constant current of 0.2C, the black-P–graphite composite displays an impressive initial discharge capacity of 2786 mA h g⁻¹ and excellent stability. Specifically, it maintains a remarkable capacity of 1849 mA h g⁻¹ after 100 cycles, whereas the capacity of black-P/graphite without robust P–C bonds declines rapidly after only 40 cycles. However, the fast-charging capabilities of P/C composites are limited by the high lithium diffusion barrier (0.34 eV) of carbon carriers and the heterogeneous interface.⁶⁰⁹ To address these issues, Sun et al. have recently incorporated electrochemically active bismuth nanoparticles into P/C composites by ball milling (Fig. 36c) to improve fast-charging and cycling.⁵⁵¹ In principle, Bi possesses a slightly higher initial lithiation potential than P, enabling it to act as a modest Li reservoir that traps Li during lithiation. Meanwhile, initial delithiation potential of Bi is slightly lower than that of P, allowing it to emit Li before P during delithiation. Moreover, the Bi anode exhibits a low Li⁺ diffusion barrier (0.08 eV) and excellent electron transfer ability. Its strong interaction with P promotes Li⁺ transport while reducing large transient stress and capturing dissolvable polyphosphide at elevated current densities. The optimized Bi–P/C anode delivers excellent electrochemical performance, with a substantial fast-charging capacity of 1755.7 mA h g⁻¹ at 7.8 A g⁻¹ (5.2C) and it maintains a capacity retention rate of 86.3% after 300 cycles, which is better than the 60.7% retention rate of the pristine P/C anode after 300 cycles (Fig. 36d).⁵⁵¹ Morphological studies reveal that the Bi–P/C anode maintains a smooth and dense surface with a volume expansion of ∼164.7% after 300 cycles, while the P/C anode has huge surface cracks with a larger volume expansion of ∼268.2%. Moreover, the discharge platform of the Bi–P/C anode is around 0.66 V at the 200th cycle (Fig. 36e), effectively inhibiting Li dendrite formation. The ΔE_p of the Bi–P/C anode (0.45 V) at the 200th cycle is considerably lower than that of P/C (0.91 V) at 7.8 A g⁻¹, suggesting that the Bi–P/C electrode provides ultrafast lithiation/delithiation reaction kinetics. As depicted in Fig. 36f, the Bi–P/C anode exhibits remarkably high-rate discharge capacities at rates ranging from 0.052 to 13 A g⁻¹, while the P/C anode shows inferior discharge capacities. Additionally, both Bi and LiBi are calculated to have high adsorption energies for LiP₇ and LiP₅, indicating their ability to capture and facilitate the conversion of soluble lithium polyphosphides, resulting in increased reversible capacity and improved cycle stability.

Introduction of a conductive polymer coating layer into P/C composites has been shown to be effective in stabilizing the solid–electrolyte interphase and preventing the formation of low-conductivity by-products during fast charging. Ji's group has developed a covalently bonded black-P/graphite anode coated with a thin polyaniline gel ((BP-G)/PANI), which has a high rate, high capacity, and robust cycling stability.⁵⁵⁴ The covalent bonding to graphite inhibits edge reconstruction of layered black-P and maintains the structure of phosphorus atoms surrounding the zigzag diffusion channel, while the conductive PANI facilitates charge transfer at the electrode–electrolyte interface (Fig. 37a). After 2000 cycles at current densities of 2.6, 5.2 and 13 A g⁻¹, the reversible capacities reach 910, 790 and 440 mA h g⁻¹, respectively (Fig. 37b). The high-rate performance of (BP-G)/PANI exceeds that of conventional carbon anodes and advanced silicon anodes and is comparable to that of several well-known high-rate materials (Fig. 37c). To probe the role of different components in (BP-G)/PANI composites, researchers have prepared several composite materials for comparison (Fig. 37d). The (BP-G)/PANI composite exhibits a reversible capacity of 1520 mA h g⁻¹ at 0.26 A g⁻¹, which is 10 times higher than that of BP-G without the PANI coating at the same rate. Furthermore, the discharge capacity of (BP-G)/PANI is ∼3 times that of (RP-G)/PANI, indicating the significant contribution of black-P to the capacity of (BP-G)/PANI. The synergistic effects of the composite enable excellent Li⁺ transport kinetics and high-rate and high-capacity lithium storage.

	Fig. 37 Engineered interface black-P composites designed for high-rate and high-capacity lithium storage. (a) Scheme of electrolyte-swollen (BP-G)/PANI that retains the solid–electrolyte interphase and pristine BP-G that suffers from the formation of less-conductive by-products during lithiation/delithiation. (b) Long-cycling tests of (BP-G)/PANI at current densities of 2.6, 5.2 and 13 A g−1. (c) Volumetric performance metrics of the (BP-G)/PANI composite in comparison to a range of high-rate materials. (d) Cycling performance comparison between the (BP-G)/PANI composite and various control samples. Reproduced with permission from ref. 554. Copyright 2020 Science.

Graphene is typically arranged in a layered honeycomb-like pattern, which enables ion insertion and extraction. However, the limited interlayer spacing in graphene is insufficient for sodium ions, significantly constraining the gravimetric capacity of graphene-based anode materials. In contrast, black-P exhibits a considerably larger interlayer spacing than graphene (3.08 Å vs. 1.86 Å), indicating its potential to accommodate more sodium ions per gram, resulting in a higher theoretical specific capacity (2596 mA h g⁻¹). The sodiation mechanism in black-P involves two processes: intercalation and alloying (Fig. 38a).⁵⁶⁵In situ TEM and ex situ XRD studies have revealed that intercalation occurs within interlayer spaces of black-P, where Na⁺ preferentially diffuses along the one-dimensional pore channels (zigzag ridges) due to the wide channel size. The channel along the zigzag direction has a sufficient width of 3.08 Å to permit Na⁺ (2.04 Å) diffusion, whereas the channel width along the armchair direction is only 1.16 Å and cannot accommodate Na⁺ diffusion. During the intercalation process, the host structure of black-P remains largely unaltered, resulting in a highly reversible cycle in the 0.54–1.5 V range. However, the specific capacity is limited to 150 mA h g⁻¹ at this point, with a spatial conformation of Na_0.17P. Further sodiation below 0.54 V proceeds to the alloying reaction stage, where the P–P bond is broken, leading to the formation of amorphous Na_xP. Eventually, as the potential reaches 0.1 V, the Na₃P phase is formed, and the alloying reaction is almost complete. However, this is accompanied by a significant volume expansion of over 500%. First-principles calculations also support this proposed mechanism. The diffusion energy barriers for sodium atoms along the zigzag, armchair, and interlayer directions are 0.18, 0.76, and 4.2 eV, respectively.⁶¹⁹ These values indicate that sodium atoms primarily follow one-dimensional pathways along the zigzag ridges in black-P. This diffusion behaviour is similar to that observed in nanotube materials along their pores, suggesting that black-P could potentially exhibit exceptional rate capacity when employed as an anode material for SIBs.^619,620

	Fig. 38 Sodiation process and electrochemical performance in black-P-based anodes. (a) Scheme of the mechanism of sodiation in black-P. (b) Scheme of sandwich-like phosphorene–graphene nanohybrids during sodiation/desodiation. (c) Cycling performance of phosphorene–graphene nanohybrids evaluated at different current densities. Reproduced with permission from ref. 565. Copyright 2015 Nature Publishing Group.

Huang et al. have electrochemically exfoliated black-P bulk crystals into few-layered phosphorene and directly utilized the obtained phosphorene as an anode material for SIBs.⁵⁶⁴ Electrochemical tests reveal that few-layered phosphorene exhibited an activation capacity of up to 1968 mA h g⁻¹ at 100 mA g⁻¹ and a capacity retention of 60.5% after 50 cycles. At a higher current density of 1500 mA g⁻¹, the retained capacity is 603.3 mA h g⁻¹ after 100 cycles, suggesting the great potential of phosphorene as an SIB anode material. Given that phosphorene is directly used in this work, the electrochemical performance can be further enhanced by doping and compositing strategies. As a representative example, Cui's group has prepared sandwich-like phosphorene–graphene nanohybrids using a self-assembly method (Fig. 38b).⁵⁶⁵ In this hybrid material, phosphorene nanosheets reduce the diffusion distance for both sodium ions and electrons to improve the rate performance. Simultaneously, the graphene nanosheets facilitate electron transport from phosphorene to the current collector and provide a buffer space for the anisotropic expansion of phosphorene during Na₃P alloy formation. These lead to exceptional electrochemical performance with a high specific capacity, rate capability, and capacity retention. Specifically, the phosphorene–graphene composite shows a specific capacity of 2440 mA h g⁻¹ at 0.02C (50 mA g⁻¹) and maintains a reversible capacity of 2080 mA h g⁻¹ after 100 cycles (with a decay of only 0.16% per cycle). Moreover, even at the significantly higher rate of 10C (10 A g⁻¹), a specific capacity of 645 mA h g⁻¹ can be achieved (Fig. 38c).

As one of the most prevalent phosphorus allotropes, cost-effective red-P possesses a high theoretical sodium capacity and a relatively low safe working potential (0.4 V for Na⁺/Na),⁶²¹ which promotes fast charging without the risk of dendrite formation. However, the fundamental mechanism behind its long-cycle capacity decay remains a subject of debate. It has been assumed that electrochemical fading of red-P-based anodes is similar to the decay process of silicon anodes in LIBs, wherein the anode undergoes significant volume expansion during long-term cycling, leading to severe particle pulverization and active material losses.^583,622,623 Nevertheless, Liu et al. have discovered the “liquefication” phenomenon of red-P in SIBs by in situ TEM and chemo-mechanical simulations.⁵⁸⁴ This finding suggests that red-P should be less susceptible to volume fluctuation and fracture compared to the silicon anodes in LIBs. Instead, the primary cause of capacity decay in red-P-based anodes stems from side reactions that occur during the formation of reactive sodium phosphides. To mitigate capacity decay, researchers have developed an effective encapsulation technique, that is, synthesis of red-P-impregnated carbon nanofiber composites (P_red@CNF), which has been confirmed to prevent undesirable side reactions in red-P anodes. They employed a dry-format electrochemical cell setup (Fig. 39a) for real-time observation of the sodiation process.⁵⁸⁴Fig. 39b illustrates the “liquid-like” material behaviour of red-P during sodiation, where red-P may produce low-yielding stress and exhibit low stiffness (i.e., softening) as sodiation continues, thus allowing red-P to flow plastically inside the CNF like a fluid.⁵⁸⁴ Model stimulations in the various states of sodiation (Fig. 39c) align well with the longitudinal flow of red-P observed by in situ TEM. The average longitudinal expansion of red-P along the CNF is ∼330% during the complete sodiation process (Fig. 39d and e). Another series of cropped time-lapse STEM images display two distorted and blended red-P segments during the sodiation expansion (Fig. 39f), resembling two droplets and directly confirming the sodiation induced red-P softening effect. In contrast to a free-standing red-P segment, the CNF-encapsulated red-P exhibits reduced tensile stresses near the segment surface and is mechanically constrained by the CNF shell (Fig. 39g and h) to prevent fracture formation in the red-P segments. The designed P_red@CNF anode demonstrates a specific capacity of 1850 mA h g⁻¹ at 0.1 A g⁻¹ after 500 cycles, and over 1000 mA h g⁻¹ after 5000 cycles at 1 A g⁻¹.

	Fig. 39 In situ TEM analysis and corresponding chemo-mechanical simulations of Pred@CNF composites. (a) Scheme of in situ TEM experimental setup of a dry-format electrochemical cell. (b) STEM image series captured in real-time showing red-P volume expansion during sodiation. The scale bar is 200 nm. (c) Simulated state of sodiation (SOS) corresponding to the morphological changes in red-P in the region labelled with a blue box in (b). (d) Energy dispersive spectroscopy mapping of phosphorus (green), sodium (yellow), and carbon (red) elements of the region labelled with a red box in (b). The scale bar is 200 nm. (e) The length change of four red-P segments marked in (b) over time during sodiation. (f) STEM image series captured in real-time showing two red-P segments merging during sodiation. The scale bar is 100 nm. (g) Simulated hoop stress distribution of Pred@CNF and (h) a free-standing red-P particle. Reproduced with permission from ref. 584. Copyright 2020 Nature Publishing Group.

Heterojunction construction can enhance the electrochemical stability and conductivity of phosphorus-based electrodes.^578,624,625 However, the incorporation of another component in these heterojunctions often results in a lower theoretical capacity (e.g., carbon materials with a theoretical capacity of 530 mA h g⁻¹),⁶²⁶ sacrificing a part of the total electrode capacity. To optimize the stability of phosphorus-based electrodes while preserving the theoretical capacity as much as possible, it is beneficial to develop monoelemental phosphorus heterojunctions. Recently, Ma et al. have fabricated RP@BP core–shell heterostructures on 3D nitrogen-doped graphene using a straightforward one-step solvothermal method.⁵⁶³ The resulting RP@BP core–shell heterogeneous structure features a crystalline black-P shell tightly enclosing an amorphous red-P core. The internal electric field at the heterojunction not only provides exceptional electron conductivity and an extremely low Na⁺ diffusion barrier, but also induces electron cloud transfer from black-P to red-P. This suppresses the reactivity of lone-pair electrons in black-P atoms, resulting in excellent air stability of the composite. Electrochemical tests reveal that the composite anode exhibits an impressive capacity of 1440.2 mA h g⁻¹ at 0.05 A g⁻¹, superior to that of both pristine red-P (1137.9 mA h g⁻¹) and black-P (713.3 mA h g⁻¹) supported on 3D nitrogen-doped graphene.⁵⁶³ The composite anode also delivers an exceptional rate performance with a reversible capacity of 521.3 mA h g⁻¹ achievable at 10.0 A g⁻¹. Moreover, the electrode displays great electrochemical and air stability with a retention rate of 89.3% after 1200 cycles at 10.0 A g⁻¹ and stable cycling over 500 cycles at 10.0 A g⁻¹ after exposure to air.

The synthesis of crystalline red-P phases, such as fibrous-P and Violet-P, remains a challenging endeavour with limited reports on their utilization for SIBs. Nevertheless, the crystalline phase offers several distinct advantages over its amorphous counterpart. First, the crystalline lattice structure enhances charge mobility and electrode conductivity.³³ Second, crystalline red-P generates polymeric chains,³⁵¹ which mitigate the occurrence of side reactions during the charge/discharge process. Third, the crystalline structural phase enables a more comprehensive understanding of the sodiation mechanism in red-P anodes. Yan et al.⁶²⁷ have synthesized a 3D structured nanocomposite by incorporating highly dispersed crystalline red-P (violet-P phase) nanorods within a reduced graphite oxide aerogel matrix. This composite anode has a remarkable initial specific capacity of 2427 mA h g⁻¹ with an initial Coulombic efficiency of 82% at 0.1 A g⁻¹ and capacity retention even at higher current densities. This investigation highlights the potential of crystalline red-P as promising anode materials for SIBs.

	Fig. 40 Applications of elemental phosphorus materials to PIBs. (a) Electrochemical voltage windows of LIBs, SIBs, and PIBs in carbonate electrolyte solutions. (b) Comparison of ionic radii and Stokes radii in propylene carbonate and water among Li+, Na+, and K+. Reproduced with permission from ref. 629. Copyright 2020 American Chemical Society. (c) Cycling performance of the red-P/3D carbon nanosheet framework nanocomposite at 100 mA g−1. Reproduced with permission from ref. 634. Copyright 2018 John Wiley and Sons. (d) Long-term cycling performance of red-P/3D sulphur and nitrogen co-doped carbon nanofiber electrodes at 2 A g−1. Reproduced with permission from ref. 637. Copyright 2022 John Wiley and Sons. (e) SEM images of the thickness of (P-CNTs)/PANI (top) and P-CNTs without PANI coating (bottom) in the initial and depotassiation states after 20, 50, and 100 cycles at 1000 mA g−1. Reproduced with permission from ref. 569. Copyright 2022 John Wiley and Sons.

However, the large ion radius of K⁺ leads to sluggish potassiation kinetics,⁶³¹ which poses a challenge in selecting suitable electrode materials. For example, silicon-based materials that exhibit a high capacity for LIBs are inactive in PIBs.⁶³² The commercial graphite anode for KIBs only achieves a capacity of 279 mA h g⁻¹.⁶³³ Despite various modification strategies being developed for graphite, the capacity improvement has remained limited due to the intrinsic properties of carbonous materials. In contrast, the theoretical capacities of EPMs upon K₄P₃ and KP formation are 1154 and 843 mA h g⁻¹, respectively.^590,634 To the best of our knowledge, these are the highest specific capacities among all elements for PIB anodes, indicating that phosphorus-based anodes have a higher upper limit for PIB development. However, the existence of potassium phosphide-rich polymorphisms (such as K₃P, KP, K₄P₃, and K₃P₁₁) makes it difficult to investigate the potassium storage mechanism of phosphorus.⁶³⁵ Currently, research on PIBs for phosphorus-based anodes mainly focuses on well-established EPMs including amorphous red-P and black-P, as well as their heterostructures with carbon or metal species.

Amorphous red-P has been investigated for PIBs due to its non-toxicity and cost-effectiveness for mass production. However, its intrinsic low conductivity (∼10⁻¹⁴ S cm⁻¹) and significant volume expansion during the potassiation/depotassiation make pure red-P unsuitable as electrodes. To address this issue, researchers have explored the incorporation of highly conductive carbon materials into the phosphorus anode to enhance its storage performance. In 2017, Zhang et al. measured the K⁺ storage capacity of red-P-based composites.⁶³⁶ They utilized red-P and carbon black as precursors to prepare the P/C anode via ball milling and electrochemical tests showed that this P/C anode exhibited an initial discharge capacity of 2171.7 mA h g⁻¹ at 50 mA g⁻¹, significantly surpassing the performance of the Sn₄P₃/C anode (588.7 mA h g⁻¹). However, the P/C anode displayed rapid capacity fading, experiencing a 91% reversible capacity decay by the 20th cycle and reaching near-zero capacity by the 50th cycle. This capacity fading arose from the disordered microstructure of the composite, which impeded interfacial charge transport. To address the significant structural changes during the potassiation/depotassiation process, Xiong et al. anchored red-P nanoparticles to a 3D porous carbon nanosheet framework.⁶³⁴ This approach yielded an initial charge capacity of 715.2 mA h g⁻¹ at 100 mA g⁻¹ through KP alloy formation, with a capacity retention of 427.4 mA h g⁻¹ after 40 cycles (Fig. 40c). Although this method demonstrates improved cycle stability compared to Zhang's work,⁶³⁶ there are limitations. Yu's group has further addressed the aggregation and pulverization issues of red-P particles by incorporating them into free-standing nitrogen-doped porous hollow carbon nanofibers, achieving a remarkable reversible capacity and exceptionally long cycle life (465 mA h g⁻¹ at 2 A g⁻¹ after 800 cycles).⁵⁹⁰ Compared to nitrogen, lower electronegativity of sulphur leads to stronger chemical interactions with phosphorus atoms, increasing the adsorption energy of phosphorus atoms on sulphur-doped carbon matrices. Feng and colleagues have demonstrated that encapsulating strain-relaxed red-P within 3D interconnected multi-channel sulphur and nitrogen co-doped carbon nanofibers can enhance potassium ion capture and promote electrochemical reaction kinetics.⁶³⁷ This composite structure not only buffers the stress and strain caused by significant volume expansion, but also utilizes the defect sites and high electronic conductivity generated by the synergistic effects of dual heteroatom doping. Impressively, the composite sustains a reversible capacity of 282 mA h g⁻¹ after 2000 continuous cycles even at a high current rate of 2 A g⁻¹ (Fig. 40d). In addition, employing metallic Bi as an electrochemically active coating has recently been shown to boost the electrical conductivity and cycle stability of red-P-based composites.⁶³⁸

The unique thermodynamically stable structure and high electronic conductivity (∼10² S cm⁻¹) of black-P make it a potential anode material for PIBs. However, pristine black-P is not practical for PIBs, and engineering is necessary to achieve reasonable electrochemical stability. Sultana et al.⁵⁶⁸ have synthesized a black-P-carbon nanocomposite by ball milling and compacted the nanocomposites into dense aggregates to improve packaging of the black-P active surface. The nanocomposite shows a first cycle capacity of 617 mA h g⁻¹ at 50 mA g⁻¹ in 0.75 M KPF₆ in ethylene carbonate/diethyl carbonate (1:1 by vol.), which is more than twice that of the graphite anode (270 mA h g⁻¹). However, the cycle stability of the nanocomposite is relatively limited. After 50 cycles at the same current rate, the anode capacity decays to around 30 mA h g⁻¹. The gradual degradation of black-P during the cycle is responsible for the destruction of the solid electrolyte interphase layer and subsequent effects on the heterogeneous charge transfer kinetics. To improve the cycle stability of black-P-based materials, Lu's group has recently proposed the use of a PANI coating layer to alleviate black phosphorene degradation and replacement of graphite with regular nanostructured carbon materials to improve electrode conductivity.⁵⁶⁹ They have assembled carbon nanotubes and black-P flakes by ball milling and performed in situ polymerization of PANI to obtain a phosphorene–carbon nanotubes/polyaniline ((P-CNTs)/PANI) composite. In 3 M KFSI/dimethoxyethane electrolyte, (P-CNTs)/PANI shows a maximum charge capacity of 417.6 mA h g⁻¹ at 500 mA g⁻¹ and maintains a reversible capacity of 392.5 mA h g⁻¹ and an impressive retention rate of 94% after 500 cycles. Morphological studies show that the (P-CNTs)/PANI electrode has good structural stability during the potassiation/depotassiation process (Fig. 40e). After 20, 50, and 100 cycles at 1000 mA g⁻¹, the thickness of the (P-CNTs)/PANI electrode shows a slight increase with the number of cycles, but the overall electrode structure remains largely unchanged. In contrast, the P-CNT electrode without the PANI coating undergoes significant alterations in thickness and there are noticeable cracks in the electrode material. These observations suggest that the PANI coating mitigates volume expansion of the composite and prevents contact loss with the current collector during the charge/discharge process. Zhang et al.⁶³⁹ have demonstrated that the incorporation of a localized high-concentration electrolyte enhances the ionic conductivity of the electrolyte and facilitates the formation of a robust solid electrolyte interphase for improved long-cycling stability of black-P/C electrodes. This finding emphasizes the role of the electrolyte in determining the electrochemical performance of KIBs and motivates further exploration to develop electrolytes that complement the characteristics of phosphorus-based electrodes.

The electrode is a critical component of SCs and the material selection and structural design impact the electrochemical and cycling properties.⁶⁴² EPMs comprise various phosphorus allotropes and corresponding nanostructures boding well for electrodes. EPMs have the following advantages as SC electrodes: (i) both black-P derivatives and red-P modifications feature nanoscale channels within their structures, allowing for shorter charge transfer paths and facilitating electrochemical reaction kinetics. (ii) Low-dimensional EPMs exhibit great mechanical flexibility and packing density,⁶⁴³ suggesting their potential use in fabricating sophisticated and flexible SCs. (iii) Low-dimensional EPMs possess large lateral dimensions and exposed active-site surfaces. In comparison with chemically inert carbon materials, EPMs are more likely to form composites with metal/non-metal species, enhancing electrochemical performance.

In 2016, Hao et al.⁵⁷⁰ investigated the electrochemical performance of liquid-exfoliated black-P nanoflakes in all-solid-state SCs. The fabricated flexible electrodes achieved a high stack volumetric capacitance of 13.75 F cm⁻³ (48.5 F g⁻¹) at a scan rate of 0.01 V s⁻¹ and a great capacitance retention of 71.8% at 0.5 V s⁻¹ over 30000 cycles (Fig. 41a and b). Although this charge storage performance is better than that of several graphene-based and multi-walled carbon nanotube electrodes,^646–648 the as-fabricated SCs still have some limitations, including a rapid volumetric capacitance reduction with increased charge/discharge rates, which is due to morphological degradation caused by long-term sonication. Yang et al.⁶⁴⁹ have described a modified electrochemical exfoliation approach for the production of high-quality few-layered black-P nanoflakes. Vacuum filtering of the black-P nanoflake dispersion yields free-standing thin films for flexible quasi-solid-state micro-SCs. The as-prepared micro-SCs show excellent steady electrochemical performance at a high scanning rate of even up to 100 V s⁻¹ (Fig. 41c) comparable to that of MXene-based quasi-solid-state micro-SCs at the same scanning rate.⁶⁴⁹ The micro-SCs have a remarkable energy density of 3.63 mW h cm⁻³ at a power density of 0.247 W cm⁻³, surpassing that of many 2D metal/non-metal materials (Fig. 41d). The device also exhibits an ultralong cycling lifetime of over 50000 cycles at 0.2 A cm⁻³ and remarkable flexibility by retaining more than 91.3% of capacity after 500 bending cycles (Fig. 41e). Using 2D nanosheets as the building blocks in the 3D architecture can enhance the electrochemical performance of electrode materials, particularly for LIBs. To expand this strategy to SC electrodes, our group has reported the synthesis of a semi-connected 3D black-P sponge in three minutes under ambient conditions using a facile and simple electrochemical approach.⁵⁷¹ As an electrode material for all-solid-state SCs, black-P sponge shows a specific capacitance of 80 F g⁻¹ at 10 mV s⁻¹, which is better than those of bulk black-P and black-P nanosheets. On the one hand, the high-quality ultra-thin nanosheets as a basic unit of the black-P sponge offer expanded active sites and lateral dimensions and remove the high energy barrier of individual nanosheets to promote the charge transfer kinetics. On the other hand, the pores and channels inside the sponge allow contact with the electrolyte for better ion diffusion.

	Fig. 41 Applications of elemental phosphorus materials to SCs. (a) Stack capacitances of the liquid-exfoliated black-P based all-solid-state SC (LE-BP-ASSP) with PVA/H3PO4 gel electrolyte at different scan rates. (b) Cycle stability of LE-BP-ASSP at 0.5 V s−1 in alternating flat and bent configurations. The inset shows representative cyclic voltammetry curves and photographs of the device in both flat and bent configurations. Reproduced with permission from ref. 570. Copyright 2016 John Wiley and Sons. (c) Cyclic voltammetry curves of the as-prepared micro-SCs at various scan rates. (d) Energy densities and power densities of the micro-SCs compared to those of various previously reported SC devices. (e) The capacitance retention rate of different bending cycles for the micro-SCs. Reproduced with permission from ref. 649. Copyright 2019 American Chemical Society. (f) Microfluidic spinning technique for the assembly of non-woven fabrics, which can be tailored into various shapes. (g) Cycle stability of black-P-carbon nanotube based SCs at 0.4 A cm−3. The inset shows galvanostatic charge/discharge curves after 10000 cycles. CNTs refer to pristine carbon nanotubes. CNTs/BP refers to a physical mixture of carbon nanotubes and black-P. CNTs/BP–CNTs refer to the physical mixture of carbon nanotubes and carbon nanotubes chemical-bridged black-P, which achieved the best electrochemical performance among the three materials. Reproduced with permission from ref. 572. Copyright 2018 Nature Publishing Group.

Nanostructures can enhance the specific capacitances of pure black-P electrodes, but achieving capacitance values in the hundreds or thousands remains highly challenging. To fabricate flexible SCs with a high energy density, the development of black-P-based multicomponent materials is necessary. Chen's group has chemically engineered black-P with carbon nanotubes to produce heterostructure composites.⁵⁷² With the aid of the microfluidic spinning technique, the composites are assembled into non-woven fabrics (Fig. 41f) for high-performance SC electrodes. The optimal flexible SC shows an outstanding volumetric capacitance of 308.7 F cm⁻³ at 0.1 A cm⁻³, an energy density of 96.5 mW h cm⁻³, cycle stability with a 90.2% retention rate after 10000 cycles (Fig. 41g), and sustained bending durability. This enhancement stems from the open network structure of the composites featuring abundant well-defined micro/mesopores for enhanced conduction, rapid ion transport, and mechanical stability. Recently, Wu et al. have employed a droplet microfluidics method to fabricate an anisotropic black-P/ZIF-67 heterostructure composite and utilized 3D printing to construct fully integrated solid-state flexible SCs.⁵⁷³ The SC has a high volumetric energy density of 109.8 mW h cm⁻³ at a power density of 0.675 W cm⁻³, a capacitance of 506 F cm⁻³, and long-term stability for 12000 cycles. These research findings demonstrate the significant potential of SCs in next-generation wearable devices.

Researchers have also shown interests in red-P as an electrode material for SCs due to its environmental-friendless, cost-effectiveness, chemical stability, and high theoretical capacitance similar to that of black-P.⁶⁵⁰ However, as previously stated, the poor electrical conductivity of pristine red-P limits the charge storage effectiveness. To enhance the conductivity, red-P is often coupled with carbon materials such as graphene and carbon nanotubes through chemical bonding or vdW interactions.⁵⁹³ However, the main drawback of carbon materials is the low specific capacitance that impacts the energy storage efficiency. In this regard, black-P owns high conductivity and theoretical capacitance as well as compatibility with red-P. In 2017, Chen et al. reported the in situ synthesis of red-P/black-P composites using a facile sonication method and their application as SC electrodes.⁵⁷⁹ The cyclic voltammetry curves of the resulting electrode exhibited a rectangular-like shape, indicating the electric double-layer charge mechanism at the electrolyte/electrode interface.⁶⁵¹ At a current density of 0.5 A g⁻¹, the hybrid composite showed a specific capacitance of around 60.1 F g⁻¹ and 83.3% capacity retention for 2000 cycles. This work not only establishes the excellent energy storage properties of red-P/black-P hybrids, but also provides a straightforward and cost-effective preparation method. Inspired by Chen's work, Gopalakrishnan and colleagues incorporated sulfonated porous carbon materials into red-P/black-P hybrids to boost the electrochemical performance of elemental phosphorus hybrids.⁵⁹² As an SC electrode, the ternary composite has a high specific capacitance of 364.5 F g⁻¹ and maintains long-cycling stability retaining 89% of capacity after 10000 charge/discharge cycles. The improved capacitive properties are associated with the synergistic effects between elemental phosphorus hybrids and porous carbon materials. The highly stable carbon frameworks featuring phosphorus bonding and heteroatom doping provide numerous active sites for ionic interactions and improve the surface wettability for seamless electrode/electrolyte interactions. Other EPMs such as blue-P and green-P have recently also shown promising theoretical potential as electrodes in SCs.⁶⁰⁴

Few-layered black-P nanosheets offer several advantages such as large interlayer spacing (∼3.08 Å), high sulfophilicity, and high lithium-ion mobility, rendering them ideal encapsulating materials for sulphur cathodes. Zhou et al. have constructed a 2D heterostructure of black-P/graphene for encapsulating sulphur species to provide catalytic and chemisorptive benefits for electrolyte-lean Li–S batteries.⁶⁵⁸ The 2D heterostructure not only enhances the electrochemical stability of black-P nanosheets, but also facilitates electronic regulation between black-P and graphene. The electric field generated at the heterostructure interface increases the intrinsic conductivity of the materials and accelerates electron transfer in the redox conversion of polysulfides. Concurrently, it promotes chemisorption of polysulfides and reduces the migration barrier for lithium ions to catalyze the conversion of polysulfides. These features contribute to low capacity fading with an average decline as low as 0.021% per cycle after 600 cycles at 2C. Xu et al.⁵⁵⁵ have demonstrated that BPQDs have a higher lithium polysulfide adsorption capacity and more substantial Li₂S precipitation capacity than large black-P flakes (Fig. 42a and b). BPQDs show strong interactions with lithium polysulphide through P–S and P–Li bonds and the abundant active sites at the edges of BPQDs catalyze the conversion of lithium polysulphide. As a result, the porous carbon nanofibers/sulphur cathodes containing a small amount (2 wt% of the cathode weight) of BPQDs exhibit fast reaction kinetics, no polysulphide shuttling, and excellent cycling stability. The porous carbon nanofibers/sulphur cathode with BPQDs has an initial capacity of 1234 mA h g⁻¹ at 0.1C and after 200 cycles, it retains a capacity of 1072 mA h g⁻¹ (Fig. 42c) and shows a small capacity decay rate of 0.06% per cycle. In comparison, the porous carbon nanofibers/sulphur cathode without BPQDs has an initial capacity of only 1036 mA h g⁻¹ and a decay rate of 2% per cycle. The BPQD containing cathode also delivers greater rate performance than the BPQD-free cathode (Fig. 42d).

	Fig. 42 Applications of BPQDs to LSBs. (a) Potentiostatic discharge curves of Li2S8 tetraglyme solution on various substrates at 2.05 V. The dark/light colours denote the reduction of Li2S8/Li2S6 and the precipitation of Li2S, respectively. CF refers to carbon fibers. BP-4K and BP-8K refer to 4000 rpm and 8000 rpm centrifugated black-P products, respectively. (b) SEM images of the precipitation of Li2S on various substrates as indicated in (a). The scale bar is 5 μm. (c) Cyclic capacities and Coulombic efficiencies of porous carbon nanofibers/sulphur cathode containing BPQDs (PCNF/S/BPQD) and pristine PCNF/S at 0.1C for 200 cycles. (d) Cyclic capacities of PCNF/S/BPQD and PCNF/S at various rates. Reproduced with permission from ref. 555. Copyright 2018 Nature Publishing Group.

While both the component and structural design of the cathode materials discussed above inhibit polysulfide shuttling and enhance the cycle stability, they do not eliminate shuttling, particularly during extended cycling. Consequently, researchers have explored separator modification by introducing active materials to physically restrict polysulfides or chemically anchoring polysulfides.^656,659,660 The modified separators also decrease the impedance between the cathode and separator, enhance the wettability between the separator and electrolyte, and improve the electrochemical performance. Cui's group has prepared a uniform, dense black-P nanoflake coating on a Celgard commercial polypropylene separator by vacuum filtration deposition.⁵⁵⁶ The black-P coating intercepts and combines the polysulfides to prevent diffusion through the separator to the anode. The binding energy of the black-P coating to lithium polysulfide ranges from 1.32 to 2.82 eV (without vdW interaction) and 1.86 to 3.05 eV (with vdW interaction) and increases as the Li₂S_x molecular chain shortens (x decreasing from 8 to 1). This energy is higher than that of the graphene coating (0.21–0.79 eV). The capacity of the cells with a conventional separator decays to below 100 mA h g⁻¹ at 0.4 A g⁻¹ after 100 cycles, while the capacity retention of separators modified with classical graphene is around 66%. In contrast, the black-P modified separator displays an initial discharge capacity of 930 mA h g⁻¹ and retains a capacity of 800 mA h g⁻¹ after 100 cycles. In further high-rate cycling studies, the specific capacity of the black-P modified separator remains at 623 mA h g⁻¹ even at a high rate of 3.5 A g⁻¹.

In addition to black-P, red-P is used in separator modification. Red-P is a weak Lewis base agent that can undergo reasonable Lewis acid–base interactions with weak Lewis acids of lithium polysulphide, allowing it to chemically anchor the polysulphide. Concurrently, red-P can anchor polysulphides through sulphur chain interactions to inhibit shuttling and the interaction between red-P and polysulfide leads to the production of a high-conductivity by-product Li₃PO₄ which aids in Li⁺ transport in the separator and promotes the sulphur reaction kinetics. Wang et al. were the first to utilize red-P nanoparticles to modify conventional polypropylene separators, capturing and reusing lithium polysulphide through strong chemical interactions between red-P and polysulfides.⁶⁶¹ The LSBs based on red-P modified separators have excellent cycling properties. The unmodified polypropylene separators experienced substantial pore blockage due to polysulfide penetration with polysulphide decomposing, precipitating, and obstructing the pores. In contrast, the red-P modified separator maintained a highly porous surface to inhibit polysulfide penetration. The surface of lithium after cycling with the red-P modified separator was the smoothest, while the surface of lithium with the pristine polypropylene separators exhibited a large number of cracks and dendrites. This result suggests that red-P-based separators inhibit polysulphides and safeguard the lithium anode from corrosion. Consequently, the red-P modified separators achieved a high capacity retention of 82% during a long cycle test of 500 cycles at a current density of 1C, better than that of super P-coated separators (65%) and pristine polypropylene separators (43%). The red-P-based separators also had an excellent rate capability of 809 mA h g⁻¹ at a high rate of 2C. Moreover, incorporating an independent functional interlayer between the cathode and separator of LSBs can physically block and chemically absorb lithium polysulfide. An interlayer with exceptional conductivity can function as a secondary current collector to improve the cathode conductivity. Lee et al.⁶⁶² prepared a red-P nanoparticle-coated carbon nanotube film as an interlayer by pulsed laser ablation and direct spinning. The interlayer provides efficient Li⁺ and electron transport as well as strong chemical anchoring of lithium polysulfide. Consequently, the interlayer has a superior specific capacity of 782.3 mA h g⁻¹ at 3C, long-term cycle stability, and high Coulombic efficiency. At a rate of 1C, the specific capacity initially reaches 1050.2 mA h g⁻¹ and remains at 769.5 mA h g⁻¹ after 500 cycles, with a Coulomb efficiency exceeding 99%.

Lithium-air batteries (LABs), which operate through the reversible reaction of Li + O₂ ↔ Li₂O₂, have the potential to offer an impressive theoretical energy density of around 3500 W h kg⁻¹. However, the sluggish kinetics of the cathode leads to high overpotentials, substantial polarization, and poor reversibility. As discussed in Section 5.1.4, EPMs inherently offer several benefits for electrocatalytic oxygen evolution, rendering them promising candidates to enhance the reaction kinetics of LABs. Cheng et al.⁵⁵⁷ have designed a BPQD-modified δ-MnO₂ catalyst and prepared it on carbon cloth using a hydrothermal method as a binder-free cathode for LABs. The Li–O₂ cell with this BPQD modification shows a high capacity of 8463 mA h g⁻¹ at 100 mA g⁻¹ (Fig. 43a) which is greater than the capacity of 2965 mA h g⁻¹ without BPQD modification (Fig. 43b), demonstrating the high catalytic activity of BPQD/δ-MnO₂ in the oxygen reduction reaction (ORR). In the cyclic voltammetry test (Fig. 43c), the BPQD/δ-MnO₂ electrode exhibits a higher onset potential for the ORR (2.64 V) and a lower onset potential for the OER (3.04 V) than the δ-MnO₂ electrode without BPQD modification.⁵⁵⁷ Additionally, the peak OER current of the BPQD/δ-MnO₂ electrode increases significantly, suggesting that the introduction of BPQDs enhances the electrochemical reaction kinetics. Cycling stability tests (Fig. 43d–f) reveal that BPQD/δ-MnO₂ polarization gradually increases with the cycle number, whereas δ-MnO₂ polarization increases rapidly. These findings indicate that BPQDs enhance the ORR/OER catalytic activity, leading to superior discharge capacity and cycling stability for LABs. Moreover, researchers have identified black phosphorene and nitrogen-doped blue-P as highly effective catalysts for cathodic reactions through theoretical calculations.^601,663

	Fig. 43 Applications of elemental phosphorus materials to other energy storage applications. (a) Voltage curves of BPQD/δ-MnO2 and (b) δ-MnO2-catalyzed Li–O2 cells at 100 mA g−1. (c) Cyclic voltammetry curves of BPQD/δ-MnO2 and δ-MnO2-catalyzed Li–O2 cells at 0.1 mV s−1. (d) Cycling performance of Li–O2 cells with BPQD/δ-MnO2 and δ-MnO2 cathodes at 400 mA g−1 with a capacity limit of 1000 mA h g−1, and (e) voltage curves of BPQD/δ-MnO2 and (f) δ-MnO2-catalyzed Li–O2 cells. Reproduced with permission from ref. 557. Copyright 2019 Elsevier. (g) Scheme of an aluminium–phosphorus battery comprising a phosphorus cathode, aluminium anode, and CAM-IL electrolyte. (h) Scheme of the distribution of AlCl, Al2Cl, and PCl in an Al–P battery under a fully charged/discharged state. (i) Scheme of the reversible phosphorus-based five-electron transfer processes. Reproduced with permission from ref. 664. Copyright 2022 Elsevier.

Recent advances have shown that EPMs are being developed for new types of energy storage devices. Cai et al. have reported an innovative aluminium–phosphorus battery which incorporates red-P as the cathode, aluminium as the anode, and CAM-IL as the electrolyte (Fig. 43g).⁶⁶⁴ This battery boasts a high theoretical capacity of 4325 mA h g⁻¹ due to the reversible, phosphorus-based five-electron transfer reaction as shown in Fig. 43h and i. The researchers have assembled a prototype aluminium–phosphorus battery showing a capacity of 1512 mA h g⁻¹ and an energy density of 1176 W h kg⁻¹, which surpass those of many alkaline metal ion batteries. By employing modification strategies, the cutting-edge aluminium–phosphorus battery is anticipated to have large potential in low-cost, high-performance energy storage applications due to the multiple electron transfer reactions and abundant availability of both phosphorus and aluminium.

In summary, EPMs possess many intrinsic advantages including high theoretical capacity, resource abundance, and environmental compatibility and are promising candidates for various energy storage devices such as hydrogen storage systems, alkali metal batteries, and supercapacitors. As researchers delve deeper into various modification strategies including structural design and nanosizing, practical application challenges such as structural and electrochemical stability degradation stemming from the volume expansion of phosphorus allotrope materials must be addressed. The development of innovative energy storage devices with various phosphorus allotropes like aluminium–phosphorus batteries will offer more viable alternatives for energy storage systems. Arguably, the most prominent potential application of EPMs currently lies in the realm of energy storage. However, all these applications are still in their infancy and lack industrial-grade standardized testing. Therefore, further efforts are necessary to transition from research to commercial application.

Previous theoretical studies have demonstrated that EPMs exhibit ultra-high sensitivity and selectivity for NO₂ detection at room temperature. The strong interaction between the frontier orbitals of NO₂ molecules and p_z orbitals of phosphorus atoms leads to charge transfer, enabling NO₂ to act as an oxidant and remove electrons from EPMs.^671–673 Among the various allotropes of phosphorus, black phosphorene is currently the predominantly used sensing material. In 2015, Abbas et al.⁶⁷⁴ developed a multilayer black-P field-effect transistor for gas sensing (Fig. 44a) showing an NO₂ detection limit of 5 ppb (Fig. 44b). Cho et al.⁶⁷⁵ have investigated the sensing characteristics of black-P, graphene, and MoS₂ and found that black-P has an intrinsic sensing capacity that is 20 times greater than that of graphene and MoS₂ (Fig. 44c). Furthermore, only black-P shows selective response to NO₂ while being unresponsive to oxygen-functionalized molecules.⁶⁷⁵ A plausible explanation for the superior NO₂ detecting capability of black-P has been postulated. First, since NO₂ molecules are paramagnetic, their adsorption on black-P can provoke spin polarization, leading to the production of a spin-polarized current in the channel. Second, black-P has larger molecular adsorption energy than the other two materials rendering it more sensitive to NO₂ (Fig. 44d).

	Fig. 44 Applications of elemental phosphorus materials to pollutant sensing. (a) Scheme of a multilayer black-P field-effect transistor. (b) A plot of the time-dependent relative conductance change (ΔG/G0) is presented for a multilayer black-P sensor, which exhibits sensitivity to NO2 concentrations ranging from 5 ppb to 40 ppb. The inset displays an enlarged image of the response to a 5 ppb NO2 exposure, with marked time points denoting the activation and deactivation of the NO2 gas. Reproduced with permission from ref. 674. Copyright 2015 American Chemical Society. (c) Calculated molar response factor of the black-P sensor and comparison with that of MoS2 and graphene. (d) Minimum NO2 adsorption energy for monolayer and bilayer for black-P, MoS2, and graphene. Reproduced with permission from ref. 675. Copyright 2016 John Wiley and Sons. (e) A plot of the time-dependent relative conductance change (ΔG/G0) is presented for a black-P microribbon sensor under zero relative humidity, which exhibits sensitivity to NO2 concentrations ranging from 0.4 ppb to 100 ppm. (f) Black-P microribbon sensor response to NO2 and other common indoor air pollutants under dry room temperature conditions. (g) Five cycles of the on–off switching mode through the black-P microribbon sensor to 10 ppb NO2 under dry room temperature conditions. (h) Plots of the black-P microribbon sensor response to NO2 concentrations are presented under various relative humidity conditions of 0%, 26%, 60%, and 80% in air. The inset displays the response to concentrations of NO2 in the range of 0.4 ppb to 100 ppb under the same relative humidity conditions. Reproduced with permission from ref. 676. Copyright 2022 American Chemical Society. (i) Dynamic sensing response-recovery curves (black) measured for different NO2 concentrations ranging from 40 to 200 ppm. A fitting function based on an adsorption rate is used to fit the data, and the results are shown in red. (j) Red-P based sensor response to NO2 (200 ppm) and other gas molecules (1000 ppm). (k) Expansion of the amorphous red-P tetramer induced by NO2 adsorption. Reproduced with permission from ref. 30. Copyright 2018 American Chemical Society. (l) The adsorption of different molecules on the armchair and zigzag blue-P nanotubes with variances in electron density. Reproduced with permission from ref. 678. Copyright 2017 Royal Society of Chemistry.

To further improve the detection ability of NO₂, Sang's group has recently prepared a few-layered black-P microribbon-based gas sensor that achieves sub-ppb level detection under nitrogen and ambient conditions.⁶⁷⁶ As shown in Fig. 44e, the gas sensor shows response (ΔG/G₀) to different concentrations of NO₂ over time at room temperature and zero relative humidity. After exposing the sensor to 0.4 ppb of NO₂, the response is about 8.09%. The specific adsorption capacity of the black-P microribbon for NO₂ is assessed by comparing the response to 100 ppb NO₂ and other common indoor air pollutants under dry ambient conditions (Fig. 44f). When the sensor is exposed to 10 ppm NO₂ for five cycles (on–off mode) under dry conditions at room temperature (Fig. 44g), good repeatability is demonstrated and the selective detection limit is 0.4 ppb for NO₂ at room temperature and zero relative humidity. However, in practical applications, ambient humidity often complicates the calibration process and reduces the sensor accuracy. If the relative humidity goes up from 0 to 80%, the response of the black-P microribbon sensor to NO₂ is no longer evident in the range of 100 ppb to 20 ppm (Fig. 44h) because adsorbed water molecules capture electrons and degrade the sensitivity.⁶⁷⁷ Therefore, gas sensors made of black-P need to be surface passivated.

An alternative strategy is to select a relatively chemically stable allotrope of phosphorus. Zhu et al.³⁰ deposited a red-P thin film on an Au electrode-patterned SiO₂/Si substrate. The film is composed of micrometer-scale particles assembled from amorphous red-P nanoclusters with different sizes (10 or 100 nm). As shown in Fig. 44i, the sensor shows dynamic response-recovery curves for different concentrations of NO₂ (40 to 200 ppm) at room temperature in nitrogen. The sensor has high sensitivity (229%) to 200 ppm NO₂, but due to the poor conductivity of amorphous red-P, the bias stress and high noise levels (0.107%) reduce the detection limit.³⁰ The conductivity of red-P can be improved by modulating the internal structure. The selectivity of the sensor is evaluated in the presence of potentially interfering gases (Fig. 44j) and the response to NO₂ is significantly stronger, even though the concentration of the other gases is higher (1000 ppm) compared to NO₂ (200 ppm). This selectivity originates from the intrinsic physicochemical properties of elemental phosphorus. Amorphous red-P has several unique sensing features, including amphoteric sensing that allows it to respond negatively to both oxidizing and reducing gas species. In addition, unlike the adsorption mechanism of black-P,⁶⁷⁴ the adsorption process on amorphous red-P is dominated by synergistic Langmuir isotherm adsorption and P–P bond expansion (Fig. 44k).

DFT calculations show that other EPMs are also promising gas sensors. Montes et al. investigated the adsorption of inorganic gas molecules (CO, CO₂, NH₃, NO, and NO₂) on blue-P nanotubes (Fig. 44l).⁶⁷⁸ In comparison with black phosphorene, blue-P nanotubes exhibit better selectivity and sensitivity for N-containing gases. The current change before and after adsorption of NO₂ is 294% at a voltage of 0.2 V, which is more than 2.4 times higher than those of other materials such as phosphorene (122%),⁶⁷⁹ MoS₂ (57%),⁶⁸⁰ and PtSe₂ (12%).⁶⁸¹ Singsen et al. have observed that silicon-doped green phosphorene exhibits high selectivity towards carbonyl-containing volatile organic compounds such as acetone, propane, and formaldehyde because the pi and lone pairs of electrons in the carbonyl group interact with electron-deficient Si atoms in the doped green phosphorene to produce strong adsorption.⁶⁸² Dharani et al.⁶⁸³ and Bhuvaneswari et al.⁶⁸⁴ have studied the adsorption characteristics of hexanal, benzyl chloride, and chlorobenzene on violet phosphorene. Zhang et al.⁶⁸⁵ have conducted spin-polarized first-principles computations with vdW correction to investigate the properties of fibrous phosphorene in air pollution monitoring. Fibrous phosphorene with a lone pair of electrons in the p orbital shows strong attraction to electron-deficient substances, giving rise to high detection sensitivity for NO, NO₂, O₃, and SO₂. However, gas sensors made of blue, green, violet, and fibrous phosphorus are still in the theoretical stage.