Artificial photosynthesis that converts solar energy into storable chemical fuels including H₂ is one of the most promising ways to meet the energy demands of future generations [1, 2]. In this process, water oxidation reaction involving four electron transfer process and various high-energy intermediates is quite challenging from both thermodynamic and kinetics point of view [3, 4]. To achieve efficient water oxidation, many water oxidation catalysts (WOCs) consisting of earth-abundant transition metal elements have been investigated [5-8]. Iron is the second most abundant metal element in the earth's crust, which has redox properties. Many iron-based compounds such as iron enzymes or complexes have been reported to exhibit biological or biomimetic activities for oxygen activation [9]. Therefore, it is desirable to develop active and robust iron-based WOCs for solar energy conversion to chemical fuels.

A number of iron-based oxides have been reported as WOCs [10-13]. In order to fast screening iron-based oxides for water oxidation, much research has been devoted to correlating the catalytic activity with a specific descriptor or feature of the catalyst. For example, Parmon et al. [14] suggested that the surface hydroxyls with stronger acidities on iron-based WOCs could lead to higher water oxidation activity. Berlinguette et al. [15] found that amorphous Fe₂O₃ showed improved electrochemical water oxidation activity over crystalline Fe₂O₃. However, the influence of hydration state on the catalytic water oxidation activity of iron-based oxides has been neglected. Lyons et al. [16] proposed that the electrochemical water oxidation reaction at hydrated iron oxide surface involves the active participation of octahedrally coordinated anionic iron oxyhydroxide surfaquo complexes to form the porous hydrated layer. This result suggested that the hydration state may affect the catalytic water oxidation activity of iron-based oxides.

In this study, we found that the turnover frequency (TOF) for water oxidation on hydrated amorphous iron oxide (FeO_x-hydrated) with a broad size distribution ranging from 3 to 100 nm is an order of magnitude higher than that of dehydrated one (FeO_x-dehydrated). Based on this finding, we further prepared a hydrated amorphous iron oxide nanoparticles with small size (ca. 2.2 nm) that is extremely active towards water oxidation, with TOF value up to 9.3 s^‒1 in the photocatalytic Ru(bpy)₃²⁺-Na₂S₂O₈ system.

FeO_x-hydrated sample was prepared following the literature [17]. The dehydration of FeO_x-hydrated was carried out at room temperature in vaccum and this process could be confirmed by its weight loss from thermogravimetry (TG) curve in Fig. 1(a). A significant amount of water was removed at room temperature for FeO_x-hydrated while only a minor amount of water lost for FeO_x-dehydrated powder which was likely caused by adsorbed moisture. Fig. 1(b) shows the Raman spectra of FeO_x-hydrated and FeO_x-dehydrated samples. A Raman peak around 3400 cm^‒1 assigned to the water stretching mode can be clearly observed for the FeO_x-hydrated sample. This peak disappears for the FeO_x-dehydrated, confirming the loss of water after drying at room temperature. This indicates that the adsorbed water molecule is weakly bonded with the surface of amorphous FeO_x-hydrated [18]. In short, both the TG and Raman results confirm the dehydration process occurs at room temperature for FeO_x-hydrated sample.

Water oxidation activities of FeO_x-hydrated and FeO_x-dehydrated were investigated in photocatalytic Ru(bpy)₃²⁺-Na₂S₂O₈ system (Fig. S1, ESI†) [19]. No obvious oxygen evolution was detected when Ru(bpy)₃²⁺ and Na₂S₂O₈ were used under irradiation in the absence of catalyst, suggesting that the catalyst is essential for the water oxidation reaction (Fig. S2, ESI†). Interestingly, after the dehydration process, TOF value of water oxidation on FeO_x-dehydrated decreased by an order of magnitude (Fig. 1(c) and Fig. S3, ESI†). We noted that after dehydration process, TOF value of amorphous FeO_x-dehydrated is still several times higher than those of typical crystalline Fe-based oxides (Table S1, ESI†) [15].

XRD patterns in Fig. 1(d) show no distinct changes between FeO_x-hydrated and FeO_x-dehydrated samples. In addition, a Raman peak around 690 cm^‒1 and the corresponding overtone peak around 1378 cm^‒1 assigning to Fe-O vibration are observed for FeO_x-hydrated and FeO_x-dehydrated samples with almost the same peak position and intensity (Fig. 1(b)) [20]. Both XRD and Raman results confirm no bulk structural changes of iron oxide after the dehydration process. This is reasonable since the dehydration process is carried out at room temperature. To exclude the possiblity that different sizes between FeO_x-hydrated and FeO_x-dehydrated nanoparticles contributes to their different water oxidation activities, dynamic light scattering (DLS) was carried out which shows they have almost the same size distribution (Fig. S4). The catalysts recycled after water oxidation were analyzed using Raman to check for structural changes (Fig. S5). Comparison of Raman spectra of recycled FeO_x-hydrated and FeO_x-dehydrated revealed that both of them did not undergo any observable change after reaction. This indicates that they are true WOCs instead of acting as precursors and they are relatively stable in water oxidation conditions.

Dehydration at room temperature would result in the loss of free water and surface weakly bounded water molecules. Water oxidation is a multiple electron-transfer process which involves the coordination of the substrate H₂O molecule to the surface active sites, namely the surface bounded H₂O participating in forming O‒O bond [21]. FeO_x-hydrated surface can be considered as compounds with great affinity to the H₂O molecule [14]. We guess that during the dehydration process, the FeO_x-hydrated sample undergoes surface reconstructing and the obtained FeO_x-dehydrated shows much lower affinity to the substrate H₂O molecule, thus becoming unfavorable for catalytic water oxidation. Further investigations are needed to examine the surface structural changes of FeO_x-hydrated during the dehydration process as well as its interaction with the substrate H₂O molecule.

Inspired by the above results that FeO_x-hydrated exhibited much higher catalytic water oxidation activity than FeO_x-dehydrated, we tried to prepare hydrated amorphous FeO_x nanoparticles with small sizes possessing more active sites to achieve even higher water oxidation activity. The synthetic route involves using tris(hydroxymethyl)aminomethane (Tris) to inhibit the growth of FeO_x nanoparticles. Tris molecule is reported as good capping agent to prepare metal oxide nanoparticles with small sizes [22]. Hereafter, the as-prepared iron-based nanoparticles is denoted as FeO_x-Tris for simplicity.

Fig. 2(a) and (b) gives the representative TEM image and histogram of size distribution of FeO_x-Tris nanoparticles which shows they are well-dispersed with an average size of about 2.2 nm according to random statistics of 100 nanoparticles. The existence of large FeO_x-Tris nanoparticles can be excluded from the dynamic light scattering measurements (Fig. S6, ESI†). As a comparison, the size of FeO_x-hydrated sample prepared without using Tris as capping agent is much bigger and more broadly distributed, ranging from 3 to 100 nm (Fig. S7, ESI†). These results indicate that the Tris molecule plays a crucial role in limiting the growth of FeO_x-Tris. The selected area electron diffraction (SAED) pattern of FeO_x-Tris in the inserted Fig. 2(a) shows no diffraction spots or rings for FeO_x-Tris, indicating its amorphous feature.

Fig. 3(a) presents the oxygen evolution rates catalyzed by different concentrations of FeO_x-Tris catalysts. TOF value was calculated assuming that every iron atom is involved in catalytic water oxidation. An average TOF value up to 9.3 s^‒1 towards water oxidation can be obtained using FeO_x-Tris as WOC according to five times' parallel test results (Fig. S8, ESI†) in the photocatalytic Ru(bpy)₃²⁺-Na₂S₂O₈ system. This value is much higher than those of reported iron-based oxides WOCs as well as typical iron oxides such as γ-Fe₂O₃ and γ-FeOOH prepared according to reported methods [23] (Fig. S9, ESI†), as listed in Table S1. The TOF value of FeO_x-Tris is even comparable to that of the very active Fe-based polyoxometalate WOC [24].

The initial quantum efficiency of oxygen evolution on FeO_x-Tris nanoparticles is calculated to be 67% based on the number of evolved O₂ molecules and the irradiation spectrum of the halogen lamp (Fig. S10, ESI†). This further indicated the outstanding activity of hydrated FeO_x-Tris nanoparticles for water oxidation.

A general advantage of heterogeneous WOCs is their robustness compared with homogeneous WOCs. We immobilized the FeO_x-Tris nanoparticle on SiO₂ surface (denoted as "FeO_x/SiO₂") to investigate its stability, since the small FeO_x-Tris nanoparticles of only 2.2 nm is difficult to isolate from the post-solution after water oxidation test. SiO₂ is oxidation-stable as well as hydrolysis-stable which is suitable to act as support for FeO_x-Tris nanoparticle. FeO_x-Tris nanoparticles can be effectively mobilized on the surface of SiO₂ from the color changes of SiO₂ (Fig. S11, ESI†). About 98% of FeO_x-Tris nanoparticles are adsorbed on the SiO₂ surface as determined from the UV-vis spectra (Fig. S12, ESI†).

After each run of photocatalytic test, a new batch of Ru(bpy)₃²⁺, Na₂S₂O₈ and borate buffer in the same concentration as the original one were added to the recycled FeO_x catalyst to examine its stability. Fig. 3(b) shows the time course of oxygen evolution with FeO_x/SiO₂ catalyst for repetitive use. The oxygen evolution rates of FeO_x/SiO₂ catalyst decreased slightly after 2 cycles of water oxidation tests, which may be ascribed to the loss of FeO_x/SiO₂ during the recycling process. This result indicates that FeO_x-Tris is stable during water oxidation process as WOC. The high catalytic activity of FeO_x-Tris WOC combining with its robustness suggests its potential application in artificial photosynthesis devices.

In summary, dehydration process would greatly reduce the catalytic water oxidation activity of amorphous iron-based oxide. Based on this result, we prepared hydrated amorphous FeO_x nanoparticles with an average size down to 2.2 nm, which shows extremely high activity as WOC with a TOF value up to 9.3 s^‒1 in the photocatalytic Ru(bpy)₃²⁺-Na₂S₂O₈ system.

There are no conflicts to declare.