Phase-segregated NiP x @FeP y O z core@shell nanoparticles: ready-to-use nanocatalysts for electro- and photo-catalytic water oxidation through in situ activation by structural transformation and spontaneous ligand removal.

The high overpotential of the oxygen evolution reaction is a critical issue to be overcome to realize efficient overall water splitting and enable hydrogen generation powered by sunlight. Homogeneous and stable nanoparticles (NPs) dispersed in solvents are useful as both electrocatalysts and cocatalysts of photocatalysts for the electro- and photo-catalytic oxygen evolution reaction, respectively, through their adsorption on various electrode substrates. Here, phase-segregated NiP x @FeP y O z core@shell NPs are selectively synthesized by the reaction of Fe(CO)5 with amorphous NiP x seed-NPs. The NiP x @FeP y O z NPs on conductive substrates exhibit higher electrocatalytic activity in the oxygen evolution reaction than those of other metal phosphide-based catalysts. The NiP x @FeP y O z NPs can also be used as a cocatalyst of an anodic BiVO4 photocatalyst to boost the photocatalytic water oxidation reaction. The excellent catalytic activity and high stability of the NiP x @FeP y O z NPs without any post-treatments are derived from in situ activation through both the structural transformation of NiP x @FeP y O z into mixed hydroxide species, (Ni, Fe)O x H y , and the spontaneous removal of the insulating organic ligands from NPs to form a smooth and robust (Ni, Fe)O x H y /substrate heterointerface during the oxygen evolution reaction.

Introduction

Hydrogen evolution by efficient and sustainable electrolysis of water is desirable for mass-production of hydrogen as a clean energy source.1 The water splitting reaction consists of two half reactions: the oxygen evolution reaction (OER) and the hydrogen evolution reaction. The OER is considered to be the bottle-neck in the water splitting reaction because the OER typically requires a multistep four-electron process for O–O bond formation, which is kinetically slow. A high overpotential for the OER is required in water electrolysis, even with the use of rare and expensive metal catalysts, such as Ir and Ru.2 These factors limit the mass-production of water electrolysis devices. Currently, extensive explorations have been made into earth-abundant, low cost, and highly active materials, with a focus on non-noble transition-metal-based materials.3 In particular, transition metal phosphides have attracted much attention as highly efficient earth-abundant electrocatalysts for the OER.4

The most recently developed OER catalysts are micrometer-sized materials, in the form of powders,5 thin films,6 and microstructures grown on conductive substrates.7 Nano-sized OER catalysts have also been widely studied because they exhibit remarkable activities owing to their large specific surface areas.8 Well-dispersed OER catalyst nanoparticles (NPs) in solvents are considered to be particularly effective, because they can be used to modify various kinds of substrates, including conductive electrodes and photocatalyst semiconductors by simple deposition methods.8,9 Such catalyst NPs usually require a ligand removal process after deposition. Hence, OER catalyst NP “ink” systems, which do not require ligand removal processes, are attractive for developing both catalyst/electrode and cocatalyst/photocatalyst hybrid systems on a large scale.

In our investigations of such OER catalysts, we found that the phase-segregated NiP@FePO core@shell NPs are colloidally stable and efficient OER active transition metal phosphide-based catalysts. The NiP@FePO NPs could be adsorbed on various kinds of substrates by simple deposition methods. The resulting composites exhibited high and stable OER activity without the need for any post-treatments due to in situ activation of NiP@FePO NPs. NiP@FePO NP-loaded carbon substrates exhibited an OER overpotential of 0.25 V at 10 mA cm–2 in 0.1 M KOH. Adsorption of NiP@FePO NPs also greatly enhanced the photocatalytic activity and durability of BiVO4, suggesting that the NiP@FePO NPs can also be used to fabricate photocatalyst/cocatalyst hybrid systems on a large scale.

Results and discussion

NiP@FePO core@shell NPs were synthesized through the reaction of a-NiP NPs and Fe(CO)5 (see ESI† for details). The a-NiP seed-NPs were 11.3 ± 0.7 nm in size (Fig. 1a) and their X-ray diffraction (XRD) pattern exhibited only one broad peak at 45°, indicating the amorphous structure of the NPs (Fig. 1c).10 After heating the a-NiP NPs with Fe(CO)5 in a mixture of 1-octadecene, oleylamine, and tri-n-octylphosphine (TOP) at 270 °C for 1 h, the spherical a-NiP NPs were transformed into a unique anisotropic structure, in which spherical particles (9.6 ± 0.7 nm) were connected with rod-shaped particles (10.8 ± 3.0 nm × 6.3 ± 1.2 nm) as shown in the transmission electron microscope (TEM) image (Fig. 1b). An XRD pattern of the resulting NPs featured diffraction peaks assigned to a mixture of the major Ni2P and minor Ni12P5 phases (Fig. 1c). Although no peaks from the Fe compounds were observed, we confirmed the presence of Fe [Ni/Fe = 78/22 (mol mol–1)] by X-ray fluorescence (XRF) analysis. High-resolution TEM (HRTEM) observations showed that both the spherical and rod-shaped phases were mainly composed of the Ni2P phase (Fig. 1d), and the Ni12P5 phase was rarely observed in our measurements (only one among eighteen NPs, Fig. S1†). These HRTEM images were consistent with the XRD results. The HRTEM images also show the presence of an amorphous shell layer surrounding the NiP core. Scanning TEM-energy dispersive X-ray spectroscopy (STEM-EDS) mapping of a single NP revealed that the elements Ni and P were mainly located at the core, and the elements Fe, O, and P were located at the shell, and therefore we describe the resulting NPs as NiP@FePO NPs (Fig. 1e).

Fig. 1

TEM images of (a) a-NiP NPs and (b) NiP@FePO NPs. The inset shows a hexane dispersion of the NiP@FePO NPs stored for more than 6 months. (c) XRD patterns of a-NiP NPs and NiP@FePO NPs. (d) HRTEM image of NiP@FePO NPs. (e) STEM-EDS mapping images of a NiP@FePO NP. (f) Schematic of the formation mechanism of NiP@FePO NPs.

The structural evolution of the NiP@FePO NPs was monitored during synthesis. At 10 min, large NiP NPs with small spherical domains were observed (Fig. S2a†). As the reaction proceeded, these small domains grew larger. The XRD patterns indicate that a-NiP started to change into crystalline NiP (c-NiP) phases, including Ni2P and Ni12P5 at 30 min. The peak intensities increased until 60 min (Fig. S2b†). The Fe/Ni molar ratios of the NiP@FePO NPs increased as the reaction progressed, indicating that Fe atoms were gradually incorporated into a-NiP seed-NPs (Fig. S3†).

The effects of Fe(CO)5 on the transformation of the a-NiP NPs were also studied. Without Fe(CO)5, the a-NiP NPs crystallized in a spherical shape (Fig. S4†), indicating that the Fe atoms induced a partial transformation of the spherical a-NiP into a rod-shaped phase.

In an XRD pattern of the NiP@FePO NPs, the (111) peak slightly shifted from the position of the pure Ni2P phase owing to Fe incorporation into Ni2P.11 From the (111) peak position of NiP@FePO NPs at 40.97°, the Fe content in the core of NiP@FePO NPs was estimated to be ∼5 mol% (Fig. S5†).11 XRF analysis revealed the Ni : Fe molar ratio of the c-NiP cores to be 96 : 4, through selective etching of the FePO shells by H2SO4. These results agreed with the XRD results (Fig. S6†). As previously reported, Ni2–FeP NPs tend to form rods or wires, because the Ni2–FeP phase preferentially grows along the P phase preferentially grows along the 〈001〉 direction.001P phase preferentially grows along the 〈001〉 direction. direction.12 HRTEM images of the NiP@FePO NPs showed that the long axis of the rod domains also grew in the direction of the NPs showed that the long axis of the rod domains also grew in the direction of the 〈001〉 plane for Ni001 NPs showed that the long axis of the rod domains also grew in the direction of the 〈001〉 plane for Ni plane for Ni2P (Fig. 1d); thus, both incorporation of Fe into NiP and the crystallization contributed to the anisotropic growth of the NiP NPs.

From these results, we propose the following mechanism for the formation of NiP@FePO NPs (Fig. 1f). Initially, Fe atoms become incorporated into the a-NiP NPs. When the a-NiP NPs partially crystallize into small Ni2–FeP domains, they grow along the P domains, they grow along the 〈001〉 direction to form rod structures. During rod growth, Ni atoms are supplied from spherical a-NiP001P domains, they grow along the 〈001〉 direction to form rod structures. During rod growth, Ni atoms are supplied from spherical a-NiP direction to form rod structures. During rod growth, Ni atoms are supplied from spherical a-NiP phases. When the a-NiP phases are completely crystallized, the Ni migration and structural transformations terminate. As a result, anisotropic spherical and rod-shaped NPs are formed. Finally, the FePO shells are generated by surface oxidation during the purification step in air.

The NiP@FePO NPs were stable in a hexane dispersion for more than half a year, and could be easily adsorbed on various kinds of substrates, including carbon powder, carbon paper, and FTO coated glass, by simple mixing or deposition methods (see ESI† for details). Their OER catalytic activities were examined without any post-treatments such as annealing or ligand exchange to remove the organic ligands. Cyclic voltammetry (CV) curves of the NiP@FePO NPs, Ni2P NPs, FeO NPs, and a mixture of Ni2P NPs and FeO NPs [Ni2P + FeO, Ni/Fe = 77/23 (mol mol–1)] loaded carbon powder in 0.1 M KOH (Fig. 2a, S7 and S8†) are shown in Fig. 2b. Interestingly, the simply mixed Ni2P + FeO NPs showed a lower overpotential than those of Ni2P and FeO NPs; however, the NiP@FePO NPs exhibited a further lower overpotential of 0.28 V at 10 mA cm–2 (0.36 V without iR compensation, Fig. S9†). NiP@FePO NPs with different Ni/Fe molar ratios (85/15 and 72/28) were synthesized by changing the reaction time and showed overpotentials of 0.29 and 0.30 V at 10 mA cm–2 (Fig. S10†). Thus, the NiP@FePO NPs with a Ni/Fe molar ratio of 78/22 were found to be the best OER catalyst in this work. The overpotential of 0.28 V is lower than those of many other previously reported metal phosphide-based OER catalysts (Table S1†).8a,13 More importantly, the amount of loaded NPs (0.02 mg cm–2) was much smaller than those of other catalysts, further confirming the excellent OER activity of the NiP@FePO NPs.8a,13 The Tafel slope of the NiP@FePO NPs, 43 mV dec–1, was smaller than those of Ni2P (44 mV dec–1), FeO (64 mV dec–1), and Ni2P + FeO (48 mV dec–1) NPs (Fig. 2c), and was also better than those of most of previously reported phosphide-based OER catalysts.8a,13 These results suggest favorable OER kinetics for the NiP@FePO NPs. We also checked the NP loading amount dependent OER activity of NiP@FePO NPs/carbon paper composites (Fig. S11†). The densely loaded electrode (0.5 mg cm–2) showed an overpotential of 0.25 V at 10 mA cm–2, which is better than those of other transition metal phosphide OER catalysts reported recently (Table S1†). Additionally, long-term chronoamperometry (CA) testing of NiP@FePO NP-loaded carbon powder and paper showed no major decrease of the current densities during the continuous OER for 10 h, indicating the high OER operational stability of the NiP@FePO NPs (Fig. 2d and S11c†).

Fig. 2

(a) TEM image of NiP@FePO NP-loaded carbon powder. (b) Cyclic voltammograms and (c) Tafel plots of NiP@FePO, FeO, Ni2P, and FeO + Ni2P NPs supported on carbon powder in 0.1 M KOH at 10 mV s–1. (d) CA curves of the NiP@FePO NPs (0.075 mg cm–2) loaded on carbon paper and carbon powder at an overpotential of 0.35 V and 0.30 V in 0.1 M KOH, respectively.

To understand the origin of the high OER activity of the NiP@FePO NPs, we performed XRF and X-ray photoelectron spectroscopy (XPS) on the NiP@FePO NP-loaded carbon paper before and after the OER (100 cycles of CV in 0.1 M KOH). The XRF results of the NiP@FePO NPs after the OER revealed a considerable decrease of the element P, while the Ni : Fe molar ratio was maintained. Thus, P was selectively eliminated during the OER (Fig. S12†). Core level XPS spectra of the NiP@FePO NPs before and after the OER are shown in Fig. S13.† Before the OER, the Ni 2p peak intensity was small because of coverage of the NiP cores by FePO shells (Fig. S13a†). After the OER, the Ni 2p peak at 857 eV clearly emerged, which was attributed to the Ni 2p3/2 peak of Ni oxide or hydroxide.14 This result also indicates that Ni2+ ions were exposed to the surface of catalysts during the OER. For the case of P, before the OER, P 2p peaks appeared at 133 and 130 eV corresponding to PO43– species in the FePO shells and P0 in the partially exposed NiP cores, respectively (Fig. S13b†).15 After the OER, these P 2p peaks completely disappeared owing to the elimination of P, which is consistent with the XRF results. The Fe 2p peak at 711 eV before the OER could be assigned to Fe oxide or phosphate in FePO shells. This peak markedly shifted to 714 eV, corresponding to Fe (oxy)hydroxide, after the OER (Fig. S13c†).16 In the case of O, before the OER, the O 1s peak at 531 eV, attributed to metal oxide or phosphate, shifted to 532 eV, which could be assigned to metal (oxy)hydroxide after the OER (Fig. S13d†).17 We conclude from the XPS results that the NiP@FePO NPs were transformed into (Ni, Fe)OH during the OER. After the transformation, the elements Ni and Fe were homogeneously distributed over the entire catalyst surface, and P was dissolved. XPS spectra measured after Ar bombardment also indicated this transformation occurred (see Fig. S13† for details). Fe-doped NiOH has been reported to have much higher activity than pure NiOH, because the Fe ions surrounded by the Ni ions behave as active centres for the OER.18 Because the active Ni species in the NiP@FePO NPs are covered with the FePO shell, the catalytic activity for the OER should be low, as shown in the case of the FeO NPs. However, the elimination of element P and the subsequent structural transformation of NiP@FePO into (Ni, Fe)OH create the OER active sites.

The formation of (Ni, Fe)OH was further supported by the CV results. The CV of NiP@FePO NPs showed a smaller redox peak area at 1.48 V vs. RHE than that of Ni2P (Fig. 2b). This result implies that Fe diffused into the NiOH, because the Fe cations doped into NiOH suppressed oxidation of Ni2+.19 Although, Ni2P + FeO NPs also showed a smaller redox peak area of Ni2+ than that of Ni2P, and the peak was larger than that of the NiP@FePO NPs. This suggests that the Fe diffusion was incomplete for the case of Ni2P + FeO NPs because the Ni2P and FeO NPs were spatially separated. Thus, the direct contact of Ni- and Fe-containing phases in NiP@FePO NPs was advantageous for fabricating homogeneously mixed metal compound catalysts.

Upon chemical transformation, the morphology of the NiP@FePO NPs on the substrates changed to a film-like structure owing to fusion of the NiP@FePO NPs (Fig. 3a, b and S14†), which led to the drastic change of their absorption spectrum. The transmittance of the NiP@FePO NPs film on the FTO-coated glass became higher after 30 CV cycles in 0.1 M KOH owing to the formation of hydroxide species with a low absorption coefficient (Fig. 3c and S15†).20 Highly transparent catalysts in the visible region are particularly beneficial as cocatalysts for photocatalysts, because they do not obstruct incident light from reaching the photocatalysts. Furthermore, Fourier transform infrared (FT-IR) spectroscopy of the NiP@FePO NPs on FTO before and after CV revealed that the C–H stretching vibration peaks at 2849 and 2918 cm–1 disappeared after the CV scans, indicating that the organic ligands (oleylamine and TOP) were completely removed during CV (Fig. 3d). This spontaneous removal of insulating ligands is a major advantage of the NiP@FePO NPs as both a ready-to-use electrocatalyst and as a cocatalyst for photocatalysts, because post-treatment processes can be omitted to form NP/substrate heterointerfaces directly (Fig. 3e).

Fig. 3

SEM images of NiP@FePO NP coated FTO glass (a) before and (b) after 30 CV cycles in 0.1 M KOH. (c) Transmittance and (d) FT-IR spectra of NiP@FePO NP coated FTO glass (blue) before and (red) after 30 CV cycles in 0.1 M KOH. The inset in (c) shows the photograph of NiP@FePO NP coated FTO glass. (e) Schematic illustration of the transformation of NiP@FePO NPs into (Ni, Fe)OH during the OER.

To prove the versatile application of NiP@FePO NPs, we applied the NiP@FePO NPs as an OER cocatalyst with an anodic semiconductor photocatalyst to boost photocatalytic water oxidation (light source: 300 W Xe lamp with a 385 nm short-cut filter). A hexane solution of FeO, Ni2P, Ni2P + FeO, or NiP@FePO NPs was deposited on porous BiVO4 film electrodes21 and spin-dried, followed by washing with ethanol (Fig. S16†). Note that no post-treatment processes were performed in the following measurements. Linear sweep voltammetry (LSV) measurements with chopped light in 0.125 M K2B4O7, in Fig. 4a, showed that the loading of the NPs enhanced the photocurrents and that the NiP@FePO NP-loaded BiVO4 film electrode exhibited the largest photocurrent among the NPs used in this work. CA measurements with continuous light irradiation (@1.23 V vs. RHE) showed that the NiP@FePO NPs loaded on BiVO4 possessed the highest durability to continuous water photo-oxidation and maintained 92% of their photocurrent after 1000 s (Fig. 4b and S17†). However, the current densities of the bare, Ni2P, FeO, and Ni2P + FeO NP-loaded BiVO4 decreased to 30, 49, 43, and 61% of their initial current densities, respectively. Interestingly, we found that the photocurrent of the NiP@FePO NP-loaded BiVO4 was further enhanced after the CA measurement (Fig. 3c, d and S18e†), because the CA measurement promoted the transformation of the NiP@FePO NPs into (Ni, Fe)OH. We also confirmed the transformation of NiP@FePO NPs and enhanced OER activity in 0.125 M K2B4O7 (Fig. S19†). Conversely, the photocurrents of bare and other NP-loaded BiVO4 electrodes decreased after CA measurements (Fig. 3c, d and S18a–d†). This effect was likely caused by degradation of BiVO4 owing to accumulation of photogenerated holes in BiVO4.22 Namely, slow water oxidation kinetics at the surface of BiVO4 led to photo-corrosion under continuous light irradiation. By loading efficient OER cocatalysts onto the photoanode, holes were immediately consumed to oxidize water, preventing photo-corrosion and improving stability. These results also indicate the excellent catalytic OER activity of NiP@FePO NPs as a cocatalyst.

Fig. 4

(a) Photocurrent density curves before CA, (b) CA curves, and (c) photocurrent density curves after CA, and (d) photocurrent densities at 1.23 V vs. RHE before and after CA of bare and NP-loaded BiVO4 in 0.125 M K2B4O7 at 1.23 V vs. RHE (300 W Xe lamp with a <385 nm cut filter).

For practical use, the photoelectrochemical measurements of NiP@FePO NPs/BiVO4 were also conducted under simulated sunlight (Fig. S20†). By loading NPs, the photocurrent density of NiP@FePO NPs/BiVO4 at 1.23 V vs. RHE reached 2.3 mA cm–2, which is more than double that of bare BiVO4 (1.1 mA cm–2). Loading NiP@FePO NPs increased the surface charge transfer efficiency (ηsurface) from 40% to 73% at 1.23 V vs. RHE (Fig. S20b†). Especially, the ηsurface at lower potential, 0.6 V vs. RHE, was greatly improved from 9% to 58%, also proving the fast OER kinetics of NiP@FePO NP cocatalyst. A long-term stability test was also conducted for each electrode, and the highest durability was 62% photocurrent retention in 3 h at 1.23 V vs. RHE (Fig. S20d†).

Generally, a robust cocatalyst layer on BiVO4 drastically improves both activity and durability (Table S2†).21 On the other hand, partial coverage of BiVO4 with nanosized particles or molecules tends to show limited improvement of BiVO4 stability (Table S2†).22 Because, in our case, the NiP@FePO NPs partially attach to BiVO4, it seems to be insufficient to fully boost the activity of BiVO4. However, ηsurface = 58% at 0.6 V vs. RHE is relatively high and the best durability (62% in 3 h) is better than that of partially covered BiVO4, despite the use of an ultimately simple and fast method (completed within ∼10 s under ambient conditions). However, there is plenty of room for further improvement of the photocatalyst performance. In addition to the OER kinetics on the surface of the cocatalyst, the hole transfer from the photocatalyst to the cocatalyst should be considered. Tuning the band structure of NiFe(OH) must be effective and may be realized by incorporating a foreign metal into NiP@FePO NPs.23

Recent studies on efficient Ni–Fe oxide, hydroxide, or oxyhydroxide based OER electrocatalysts showed significantly small overpotentials less than 0.3 V at 10 mA cm–2.24 Most of these catalysts are in bulk form, such as micrometer scale NiFe layered double hydroxides,24a,b composites with carbon,24 and Ni foam.24 Such bulk electrocatalysts, however, are difficult to directly hybridize with semiconductor photocatalysts, because of the small interfacial contact area and the lack of a robust bond between electrocatalysts and photocatalysts by simple mixing. This would be the reason why excellent electrocatalyst/photocatalyst hybrid system combinations have been rarely reported. Our NiP@FePO NPs allow us to readily form a number of durable NPs/substrate heterointerfaces through an in situ activation and provide excellent OER activity with various kinds of conductive and semiconductive substrates. This feature is a considerable advantage of our catalytic NPs in the fabrication of large scale electro- and photo-catalyst systems.

Conclusions

In conclusion, we developed a selective synthesis of monodisperse, colloidally stable, and phase-segregated NiP@FePO core@shell NPs with high OER activity. Using our NiP@FePO NP ink, we loaded the NPs onto various conductive and semiconductor substrates and found excellent OER activity. We discovered that migration of Ni and Fe occurred between the phase separated NiP and FePO phases, which served as efficient OER active sites for the OER. This process also induced spontaneous removal of ligands and in situ formation of the NP/substrate heterointerfaces, which provided ready-to-use OER hybrid catalysts without the need for any post-treatments. We demonstrated that even phase-segregated structures could be transformed into homogeneous active phases, suggesting a new way to design efficient nanostructured catalysts.

Conflicts of interest

There are no conflicts to declare.

Click here for additional data file.