Operando Unraveling Photothermal-Promoted Dynamic Active Sites Generation in Spinel NiFe2O4 for Oxygen Evolution

The ability to develop highly active and low-cost electrocatalysts represents an important endeavor toward accelerating sluggish water-oxidation kinetics. Herein, we report, for the rst time, the implementation and unravelling of photothermal effect of spinel nanoparticles (NPs) on promoting dynamic active sites generation to markedly enhance their oxygen evolution reaction (OER) activity via an integrated operando Raman and density functional theory (DFT) study. Specically, NiFe2O4 (NFO) NPs are rst synthesized by capitalizing on amphiphilic star-like diblock copolymers as nanoreactors. Upon the NIR light irradiation, the photothermal heating of the NFO-based electrode progressively raises the temperature, accompanied by a marked decrease of overpotential. Accordingly, only an overpotential of 309 mV is required to yield a high current density of 100 mA cm-2, greatly lower than recently-reported earth-abundant electrocatalysts. More importantly, photothermal effect of NFO NPs not only signicantly reduces the activation energy necessitated for water splitting, but also facilitates surface reconstruction into high-active oxyhydroxides at lower potential (1.36 V) under OER conditions, as revealed by operando Raman spectra-electrochemistry. Moreover, the DFT calculation corroborates that these reconstructed (Ni,Fe)oxyhydroxides are electrocatalytically active sites as the kinetics barrier is largely reduced over pure NFO without surface reconstruction. Given the diversity of materials (metal oxides, suldes, phosphides, etc.) possessing the photo-to-thermal conversion, this effect may thus provide a unique and robust platform to boost highly-active surface species in nanomaterials for fundamental understanding of enhanced performance that may underpin future advances in electrocatalysis, photocatalysis, solar energy conversion and renewable energy production.


Introduction
Motivated by the need for accelerating sluggish reaction kinetics at the anode, [1,2] the focus of water electrolysis has been centered heavily on oxygen evolution reaction (OER) towards sustainable hydrogen fuel production. To date, there has been much effort in developing low-cost yet high-performance earthabundant transition-metal alternatives to commonly used noble metals for OER. Intriguingly, many Ni-, Co-, Fe-and Mn-based oxides experience dynamic surface-reconstruction process to form more active oxyhydroxides, which are recognized as true catalytically active species for OER in alkaline media. [3,4] Among various transition-metal-based OER catalysts, bimetal spinel-structured oxides in the form of AB 2 O 4 (A and B are different metal ions) have garnered much attention due to their rich compositions, electron con gurations and valence states. [5] Interestingly, inverse spinel NiFe 2 O 4 (NFO), in principle, exhibits enhanced catalytic activity toward OER because of the presence of multivalent elements (i.e., Ni 3+ /Ni 2+ and Fe 3+ /Fe 2+ ). [6] It is important to note that studies on facilitating the surface reconstruction of NFO to achieve high-performance OER are relatively few, and fundamental understanding as to what makes the derived OER catalysts perform well remains elusive.
Recently, introducing thermal energy to promote electrocatalytic conversion has attracted signi cant interest. [7] Clearly, the use of thermal energy would reduce the activation energy of water oxidation and thus accelerate the electrocatalytic kinetics, thereby leading to improved e ciency. Electrocatalysts with photothermal effect (referred to as photothermal electrocatalysts) enable in-situ heating due to photo-tothermal conversion under the illumination with visible or near infrared (NIR) light, thereby dispensing with the need for extra devices required to provide thermal energy. More importantly, in sharp contrast to common approaches where the entire solution is heated, the photothermal effect is localized on electrocatalysts themselves, [8] thus effectively enabling heat modulation to a de ned region (i.e., the working electrode). Despite recent impressive advances in transition metal oxides (e.g., Fe 3 O 4 and Co 3 O 4 ) as photothermal agents for cancer therapy, [8] their implementations for photothermal-assisted OER, in particular spinel oxides, are comparatively few and limited in scope. Moreover, it has been reported that Ni-and Co-based OER catalysts are prone to surface reconstruction into highly active oxyhydroxides. [9,10] Surprisingly, the photothermal effect on promoting surface reconstruction in spinel oxide catalysts has yet to be explored.
Herein, we report, for the rst time, an integrated operando Raman and density functional theory plus Hubbard U (DFT + U) study to exercise and unveil the photo-to-thermal conversion of inverse spinel oxide nanoparticles (NPs) in promoting the generation of dynamic active sites via surface reconstruction into oxyhydroxides and thus greatly enhancing their OER activity. First, a series of amphiphilic star-like poly(acrylic acid)-block-poly(styrene-co-acrylonitrile) (denoted PAA-b-PSAN) diblock copolymers with wellde ned molecular weight (MW) and low polydispersity index (PDI) are exploited as nanoreactors to synthesize a set of PSAN-ligated NFO NPs with different sizes and PSAN chain lengths. The effects of the NFO NP sizes and the outer PSAN chain lengths on catalytic activity of NFO NPs are then scrutinized.
Interestingly, NFO NPs of largest size (~12 nm) ligated with shortest PSAN chains (MW = 7K) display the best OER reactivity on glassy carbon electrode in alkaline media as a result of a high fraction of exposed electrochemically active surface area and fastest electrocatalytic kinetics. Subsequently, the photothermal effect of PSAN-ligated NFO NPs is exploited to promote their surface reconstruction and thus boost OER. Signi cantly, the reaction kinetics of OER is found to be considerably improved due to the lowered activation energy E a , as modulated by the thermal energy originated from photothermal conversion of NFO. More importantly, operando Raman spectra-electrochemistry study is performed to unveil the mechanism of photothermal-assisted enhancement in OER reaction, revealing the emergence of electrocatalytically active γ-NiOOH at lower potential (1.36 V) during the surface reconstruction process with photothermal effect. Finally, the rst-principle calculations substantiate that the reconstructed surface (i.e., (Ni,Fe)oxyhydroxides) plays a pivotal role as active site for electrocatalytic reaction. As such, photothermal electrocatalysts (e.g., metal oxides, sul des, phosphides, etc.) may render signi cantly low overpotential and fast OER kinetics, representing an array of important materials that couple the localized heating with electrochemistry for effectively producing renewable energy production.
The as-prepared amphiphilic star-like PAA-b-PSAN diblock copolymers are then utilized as nanoreactors to direct the growth of NFO NPs ligated by the outer hydrophobic PSAN chains that are originally covalently connected to the inner hydrophilic PAA chains. The reaction is performed in the mixed solvent of diphenyl ether/benzyl alcohol (DPE/BA at a 9:1 ratio by volume) (see Experimental Section). First, the unimolecular micelles of amphiphilic star-like PAA-b-PSAN are dissolved in DPE with a shrunk PAA core as DPE is a good solvent for PSAN blocks yet poor solvent for PAA blocks. The NFO precursors (Ni(OAc) 2 and Fe(AcAc) 3 ) are added to the star-like PAA-b-PSAN DPE solution under stirring. In order to expand the inner PAA chains to allow more precursor loading as well as improve NP morphology, BA is added to swell the inner compartment occupied by PAA chains as it is a good solvent for PAA yet poor solvent for PSAN. It is notable that due to the strong coordination interaction between PAA blocks and metal moieties of precursors, as well as the polar a nity of precursors to BA, the precursors accumulate within the PAA compartment, thereby effectively yielding PSAN-ligated NFO NPs at 250 o C (lower left panel; Fig.   1). Due to the living polymerization characteristic of ATRP, polymers with well-de ned MWs and low PDI can be readily synthesized. Thus, the lengths of PAA and PSAN blocks in star-like PAA-b-PSAN can be controlled by varying the ATRP times of tBA (hydrolyzed from PtBA block into PAA block later) and SAN co-monomers, respectively, [14] which in turn dictate the size of formed NFO NPs templated by the inner PAA blocks and the length of the outer PSAN chains. The MWs of star-like PtBA homopolymers and PtBAb-PSAN and PAA-b-PASN (both inner PAA and outer PSAN blocks) diblock copolymers as well as the corresponding sizes of PSAN-ligated NFO NPs are summarized in Supplementary   Table 1), a decrease in sizes of the resulting NPs may be ascribed to the use of 9/1 volume ratio of DPE/BA in the reaction and the shrinking of the PAA blocks during the crystallization process of NPs. In addition to interrogate the size effect of PSAN-ligated NFO NPs on the electrocatalysis performance, the NPs with the same size yet different length of outer hydrophobic PSAN chains are synthesized (i.e., NFO-12-7K and NFO-12-21K in Supplementary Table 1; 12-nm NPs with 7K and 21K MW of PSAN chains, respectively). Interestingly, the TEM sample of PSAN-ligated NFO-12-7K exposed to RuO 4 vapor [15] that preferentially stains the PSAN chains, yielding a grey shell situated on the surface of dark NFO NPs ( Fig. 2e; also see the inset), signifying the PSAN chains are intimately and permanently ligated on the NP surface. accordingly the highest catalytic activity because of its largest exposed active surface area, as veri ed by the electrochemical active surface area (ECSA, Supplementary Fig. 7), which was measured based on the double-layer capacitance (C dl ) of samples. [17] As shown in Fig. 3b, the C dl of NFO-12-7K is 830 µF cm -2 , which is the highest among all PSAN-ligated NFO NPs electrocatalysts. Taken together, NFO-12-7K NPs expose a highest density of active sites to the electrolyte during OER.
The conductivity of the samples was measured by EIS ( Supplementary Fig. 8). The NFO-12-7K sample displayed a smallest charge transfer resistance (R ct , the radius of semicircle in the low-frequency range), which is indicative of a fastest electrocatalytic kinetics. [18] Additionally, the semicircular radii of the samples in the EIS plot progressively decreased with the increasing NP size ligated with shorter outer PSAN chains. This is due to the faster electron transfer rate because of the larger exposed active surface area (as substantiated by C dl results in Supplementary Fig. 7), thereby offering the greater accessibility of ions to the exposed surface area. Moreover, it has been reported that oxygen vacancies can offer the substantially enhanced electrical conductivity because the delocalized electrons around an oxygen vacancy can be easily excited to the conduction band, thus enhancing the conductivity of catalyst. [19] It is notable that the relative peak intensity of oxygen vacancies (531. suppressing the OER activity to some extent. As shown in Supplementary Fig. 10a, for NFO NPs ligated with nearly the same length of PSAN chains (8K for NFO-3-8K, and 7K for NFO-12-7K), NFO-3-8K with a smaller nanoparticle size (3.3 ± 1.7 nm) than NFO-12-7K (12.1 ± 2.5 nm) is further adequately covered by PSAN chains, resulting in less active surface area exposed. Similarly, with the same NFO NP size (12.0 ± 2.1 nm for NFO-12-21K, and 12.1 ± 2.5 nm for NFO-12-7K), the longer PSAN chains (21K in NFO-12-21K compared to 7K in NFO-12-7K) impart denser PSAN shell situated on the NFO NP surface and thus further effectively block the electrolyte penetration through the shell, leading to a decreased active surface area during the OER process. Consequently, it is not surprising when placing the PSAN-ligated NFO NPs in KOH, compared to NFO-12-7K NPs, NFO-3-8K NPs delivered a lowest OER performance due to the least accessible active surface area that prevents the effective electrolyte ions (OH -) transport (left panel; Supplementary Fig. 10b). Likewise, among all samples, NFO-12-7K least shielded by the PSAN shell would be most readily accessed by OHions (right panel; Supplementary Fig. 10b), thereby achieving the highest ECSA and lowest overpotential.
It is also notable that two control samples (i.e., NFO NPs synthesized by using star-like PAA homopolymers as template and no template, respectively) displayed poor OER activity because of the aggregation of NPs ( Supplementary Fig. 11-12), further substantiating the importance of the presence of PSAN hairs on the NFO NP surface. Moreover, the LSV curve of pure star-like PAA-b-PSAN templates signi ed the negligible effect of polymers (either PSAN or PAA) on OER performance ( Supplementary Fig.  12).
Photothermal-assisted oxygen evolution reaction. After PSAN-ligated NFO-12-7K NPs were identi ed as the most electrocatalytically active nanomaterials, we turned our attention to invoke photothermal effect of NFO-12-7K NPs in an attempt to reduce the activation energy E a of water splitting and promote the formation of electrocatalytically active oxyhydroxides (-OOH) [10] for boosting OER reactivity. To this end, NFO-12-7K NPs were electrophoretically deposited on nickel foam (NF), which rendered intimate interfacial contact between NFO electrocatalysts and the underlying current collector (NF), thus decreasing the contact resistant and favoring electron and mass transport (Experimental Section and Supplementary Fig. 13-14). It is notable that in addition to the resulting PSAN-ligated NFO-12-7K/NF (hereafter referred to as NFO/NF) electrode prepared by electrophoresis as noted above, the bare nickel foam (denoted bare NF) was also selected as control sample for comparison.
First, UV-vis study was performed to examine the absorption characteristics of the samples. Clearly, compared to bare NF, the absorption edge of NFO/NF red-shifted and extended beyond 850 nm ( Supplementary Fig. 15), suggesting that NFO/NF is capable of utilizing and converting more photons into heat under NIR irradiation. In this context, the spatial electric-eld distribution and temperature distribution within NFO NP were theoretically modeled by the nite element method (FEM) (Fig. 4a). For a freestanding NFO NP (d = 12 nm), electric-eld distribution at an excitation wavelength of 808 nm is shown in upper left panel of Fig. 4a. This light ux concentration and localized electromagnetic eld enhancement would drive heat produced in NFO NP. The lower left panel of Fig. 4a illustrates the steadystate temperature distribution surrounding the NFO NP. Intriguingly, the close proximity of NFO NPs (with seven NFO NPs assembled together as an example) induces strong and respective coupling of electriceld distribution (central panel; Fig. 4a) as well as temperature distribution (right panel; Fig. 4a), clearly signifying that thermal energy is produced from photo-to-thermal conversion within NFO NPs.
To further elucidate photothermal responses of the electrodes, the temperature evolution of NFO/NF, bare NF electrodes, and 1.0 M KOH electrolyte under a NIR light irradiation (808 nm, 2.5 W cm -2 ) was recorded by infrared imaging (Supplementary Fig. 16 and Fig. 4b). We note that direct irradiation of KOH electrolyte by NIR light showed an almost unchanged temperature (increase of less than 1 o C), which can be considered negligible for electrocatalysis. The temperatures of both NFO/NF and bare NF increased with the illumination time, reaching approximately 45 o C and 35 o C in 6 min, respectively. A 10 o C higher temperature than that of bare NF clearly suggests that NFO/NF is a potential photothermal electrode ( Supplementary Fig. 16).
The stability of catalysts is a prerequisite for their long-term practical application. Prior to investigating the photothermal-assisted OER activity, the stability of NFO/NF was examined by a 12-h chronoamperometric test (Supplementary Fig. 17). The chronoamperometric curve shows the current density increased from ~16.56 to ~18.69 mA cm -2 (at 1.50 V vs. RHE), which can be attributed to an activation process taken place in the rst 2 h. [20] After that, it remained at ~18.69 mA cm -2 for 10 h, demonstrating an outstanding stability. The stability of this integrated electrode (i.e., NFO/NF) was further revealed by structural characterization of the used catalyst. SEM imaging on NFO/NF sample after electrochemical stability measurement showed no noticeable changes on the morphology of NFO/NF ( Supplementary Fig. 18a-b), con rming that the NFO NPs were stable on the NF electrode even in alkaline OER conditions.
The photothermal-promoted electrocatalytic activity of NFO supported on NF electrode toward OER was then scrutinized using a photoelectrochemical cell containing an O 2 -saturated 1.0 M KOH solution irradiated with 808-nm NIR light (Experimental Section; Supplementary Fig. 19). The circulating water was used to maintain the electrolyte at room temperature. As shown in Fig. 4c, the OER activity of NFO/NF was found to be progressively boosted, evidenced by higher current density j across all potentials and gradually decreased overpotential η 20  For NFO/NF, when the light was switched off, the current density promptly decreased yet higher than the initial value, which can be ascribed to the residual heat originated from the photo-to-thermal conversion. Clearly, these results verify that the NFO/NF electrode carries excellent photothermal conversion attribute for boosting OER performance. Remarkably, among an array of earth-abundant OER catalysts reported in literature (Supplementary Table 2), the NFO/NF crafted in this study emerges as one of highestperformance electrodes with the lowest overpotential (309 mV) at high current density of 100 mA cm -2 and very small Tafel slope (43 mV dec -1 ) (Fig. 4e and Supplementary Table 2).
Activation energy (E a ). To further uncover the mechanism of photothermal effect on promoting the catalytic activity of NFO/NF during OER process, the activation energy E a at the zero overpotential (η = 0) was calculated, which is widely regarded as an important descriptor for evaluating the performance of electrocatalysis. [21] Derived from the LSV curves in Fig. 4c (for example, reading the current densities j at different temperatures for a given overpotential), the Arrhenius plots displayed the semilogarithmic dependences of j on the inverse temperature (see Experimental Section), [22] that is, a linear relationship at varied overpotentials (Fig. 4f). Accordingly, E a of NFO/NF at η = 0 was calculated to be 98.50 KJ mol -1 (Fig. 4g), much lower than 147.62 KJ mol -1 of bare NF (Supplementary Fig. 22b). This signi es that the photothermal heating induced by irradiating NFO NPs with NIR light can effectively lower the energy barrier for initiating OER and accelerate the sluggish OER reaction that undergo a four-electron transfer over a large thermodynamic potential (1.23 V). Meanwhile, the charge transfer resistance measurements ( Supplementary Fig. 23-24) revealed the enhanced electron transport due to photothermal effect, which in turn led to improved OER reaction kinetics.
Observation of dynamic active sites generation. Earlier studies have revealed the importance of the oxidation of Ni 2+ in Ni-based oxides for OER, where Ni 2+ is susceptible to be oxidized to Ni 3+ or a higher oxidation state. Such oxidation is believed to be a critical step in generating active oxyhydroxide sites, that is, γ-NiOOH, for OER. [23] To study the dynamic oxidation of Ni 2+ in NFO/NF electrode, the oxidation peaks between 1.32 V and 1.44 V in Fig. 4c were closely examined. The LSV curves of NFO/NF electrode exhibited an enhanced current density and a negatively shifted peak of Ni oxidation upon exposure to NIR light ( Supplementary Fig. 25). This electrochemical behavior suggests that the surface of catalysts may undergo a dynamic reconstruction into oxyhydroxides, evolving into a more catalytically active surface for OER. [10] Moreover, it is notable that the peak area ratio of Ni 3+ /Ni 2+ in Ni 2p increased from 0.58 (prior to photothermal-assisted OER reaction) to 1.06 (after photothermal-assisted OER reaction) ( Supplementary Fig. 26a). And the peak area ratio of Fe 3+ /Fe 2+ in Fe 2p raised from 1.45 (before photothermal-promoted OER) to 1.69 (after photothermal-promoted OER) ( Supplementary Fig. 26b). These ex-situ XPS results indicate that the increased high oxidation states of Ni and Fe were achieved via localized photothermal heating of NFO/NF during OER. Such surface reconstruction of NFO/NF electrode was also evidenced by ex-situ HRTEM where surface amorphous (Ni,Fe)oxyhydroxides were emerged ( Supplementary Fig. 27).
Scrutiny of photothermal-assisted electrocatalytic OER mechanism via operando Raman studies and DFT + U calculations. To gain insight into the photothermal-promoted electrocatalytic activities, operando Raman spectra electrochemical measurements of NFO/NF electrodes in 1.0 M KOH were performed. Compared to the pristine states of both bare NFO powders and NFO/NF (i.e., without immersion in KOH) ( Supplementary Fig. 28a), there was clear positive peaks shift of the NFO/NF electrode after being immersed in KOH under OCP (open circuit potential) condition, which is due to the decreased laser power attenuated the electrolyte. [24] For bare NFO powders and NFO/NF, ve peaks at 194, 313, 470, 541 and 679 cm -1 can be assigned to the corresponding Raman active bands of inverse spinel NiFe 2 O 4 (i.e., T 2g(1) , E g , T 2g(2) , T 2g (3) , and A 1g , respectively). [24] In the case of NFO/NF electrode under OCP, the Raman peak at 120 cm -1 can be ascribed to Ni(Fe) hydroxide species (Supplementary Fig. 28b), [25] which was spontaneously formed after immersing the Ni(Fe)-based electrode into aqueous alkaline solution. [26] Thus, the peak of T 2g(1) cannot be seen because of the partial overlapping.
It is important to note that the active modes between 400 and 700 cm -1 were potential dependent according to operando Raman measurements, indicating the electrode surface experienced a dynamic phase transformation in the potential range from OCP to 1.50 V, as clearly evidenced in Fig. 5a and Supplementary Fig. 29. As shown in Supplementary Fig. 29, for the NFO/NF electrode without being irradiated with NIR light (i.e., no photothermal effect), there were no shifts of the Raman peaks at 478, 563 and 692 cm -1 for potentials ≤ 1.39 V vs. RHE. It is noteworthy that as the potential increased to 1.41 V, the peaks at 474 and 554 cm -1 appeared, which are assigned to γ-NiOOH, suggesting the formation of Ni 3+ species on the surface of electrode. [27] On the other hand, it is interesting to note that for the photothermal-promoted OER (Fig. 5a), the NFO/NF electrode displayed identical spectroscopic and electrochemical behavior when the potential was ≤ 1.35 V. Likewise, the γ-NiOOH phase emerged at 1.36 V, a 50 mV lower than the conversion potential of the electrode without experiencing photothermal heating ( Supplementary Fig. 29). Moreover, much higher ratios of 474 cm -1 to 692 cm -1 (∆I 2 ′) and 554 cm -1 to 692 cm -1 (∆I 1 ′) were observed for NFO/NF with photothermal effect than those (∆I 2 and ∆I 1 ) without photothermal effect ( Supplementary Fig. 30), suggesting the surface reconstruction into oxyhydroxide was more thorough for NFO/NF with photothermal effect. It is noteworthy that earlier reports of NiFe-based catalysts revealed that Fe ions would also be oxidized and reconstructed into oxyhydroxides. [28] However, our operando Raman studies discussed above did not show the peaks of FeOOH species. This may be due to the relatively less amount of FeOOH produced during surface reconstruction, which is consistent with our ex-situ XPS results ( Supplementary Fig. 26).
To further understand the effects of reconstructed oxyhydroxides on OER activity, rst principles density functional theory plus Hubbard U (DFT + U) calculations based on the 4e-mechanism were employed to simulate the OER process on both the reconstructed (Ni,Fe)OOH and the original NFO structure models.
As shown in Fig. 5b, Fig. 31 and Supplementary Note 5). [5] The proposed 4e-mechanism of OER and the optimized structures of the intermediates in the free-energy landscape of (Ni,Fe)OOH and NFO are presented in Fig. 5c and Supplementary Fig. 32, respectively. The highest reaction free energy barriers on (Ni,Fe)OOH and NFO surfaces are 2.05 eV and 2.20 eV, respectively, implying the corresponding overpotentials η of 0.82 V and 0.97 V. The lower Gibbs free energy on Ni atom sites of (Ni,Fe)OOH revealed a more favorable OER kinetics in reconstructed (Ni,Fe)OOH species. The calculation results highlight the pivotal role of (Ni,Fe)OOH species in improving catalytic reactivity.
It is worth noting that a suite of experiments discussed above demonstrated that the OER kinetics can be greatly facilitated by photothermal-enhanced surface reconstruction into highly active (Ni,Fe)OOH species, which in turn signi cantly improved intrinsic electrocatalytic activity. On the other hand, the formations of OH*, O* and OOH* display uphill energetics, signifying the reactions in (Ni,Fe)OOH system are endothermic at the applied voltage U = 0 V and U = 1.23 V (central panel; Fig. 5c). Taken together, the photo-to-thermal conversion would promote the oxygen evolution because of the assistance with localized heating.
Thus, a mechanism is proposed ( Supplementary Fig. 33) to rationalize the observed reduction in activation energy (Fig. 4g), enhanced active Ni 3+ species generation (Fig. 5a), and subsequently improved OER activity (Fig. 4c) of the NFO/NF electrode under 808 nm NIR irradiation. The photothermal effect casts the following two profound impacts on the OER catalysis. First, upon NIR irradiation, thermal energy, arose from photo-to-thermal conversion of NFO NPs, could reduce the activation energy of water splitting, thus promoting kinetics of electrocatalytic reactivity ( Supplementary Fig. 33a and Fig. 4g). Second, the judicious implementation of photothermal effect of NFO/NF electrode would also lower the energy barrier of Ni 2+ to Ni 3+ transformation (at a lowered potential (1.36 V) as opposed to 1.41 V without photothermal effect), and facilitate the surface reconstruction into more electrocatalytic active Ni 3+ oxyhydroxides species (Supplementary Fig. 33b, Fig. 5a and Fig. 5c), thereby boosting OER reactivity, as evidenced by LSV results in Fig. 4c. Moreover, given the semiconducting property of NFO NPs that can absorb and respond to NIR light ( Supplementary Fig. 15), the holes generated by NIR irradiation on the NFO NP surface may render the oxidation of inactive Ni 2+ into active Ni 3+ .

Discussion
In summary, we demonstrated the robustness of photothermal effect of inverse spinel oxides in promoting their surface reconstruction into oxyhydroxides (i.e., creating more dynamic active sites) for markedly enhanced OER catalysis, as unraveled by operando Raman spectra-electrochemistry study. First, spinel oxide NFO NPs of tunable sizes that are permanently ligated by polymers of different lengths (i.e., PSAN-ligated NFO NPs) are synthesized by capitalizing on amphiphilic star-like PAA-b-PSAN diblock copolymers as nanoreactors. Subsequently, investigation into the effects of the NP size and surface PSAN chain length reveals that NFO NPs of the largest size ligated with shortest PSAN chains manifest outstanding OER reactivity due to the presence of the highest fraction of accessible electrochemicallyactive surface area and the fastest reaction rate. Afterwards, photothermal effect is introduced to the NFO/NF electrode via NIR light, yielding a considerably reduced overpotential from 307 mV (25 °C; in the absence of photothermal heating) to 272 mV (45 °C; after photothermal heating) at 20 mA cm -2 and a small Tafel slope of 43 mV dec -1 . Such localized photothermal heating effectively lowers the charge transfer resistance and the activation energy for OER.
Most importantly, the structures of active phase and the reaction mechanism are uncovered by combining operando Raman spectra electrochemical measurements and DFT + U calculations. Under applied anodic potentials, the surface of NFO NPs are found to transform into active (Ni,Fe)oxyhydroxides species. In contrast to applied potential of 1.41 V for non-photothermal-assisted OER, the surface of NFO NPs dynamically reconstructed into active γ-NiOOH phase occurs at lower applied potential of 1.36 V under OER condition with photothermal effect, as unveiled by operando Raman study, and the high OER activity of (Ni,Fe)OOH species is also supported by DFT calculations. Our study brings fundamental and practical insights into the effectiveness of in-situ heating enabled by photo-to-thermal conversion of photothermal nanomaterials in promoting electrocatalysis. operated at 100 kV) and high-resolution TEM (TECNAIG2 F30; operated at 300 kV). The crystalline structures of PSAN-ligated NFO NPs were evaluated by XRD (X'pert PRO, The Netherlands). X-ray photoelectron spectroscopy (XPS) was measured by the Thermo Scienti c K-Alpha XPS system. The UVvis absorption spectra of NFO/NF and bare NF were measured by Shimadzu UV-vis-2450 spectrometer.
Raman spectra for bare NFO powders and NFO/NF samples were collected using a Renishaw inVia Raman microscope with an excitation wavelength of 532 nm, a power of 10 mW, and an acquisition time  Fig. 19). The circulating water was used to make sure the temperature of electrolyte remain at room temperature. The temperature of the working electrode was monitored by an IR thermal camera (FLIR One Pro) for every 2 s until the electrode reached a steady temperature. Chronoamperometry was carried out at a xed potential to test the stability of catalyst.
The activation energy (E a ) for NFO/NF and bare NF were calculated from the slope of the modi ed Arrhenius plot: where j is current density, A is pre-exponential factor, E a is activation energy, R is gas constant (8.314 J mol -1 K -1 ) and T is Kelvin temperature. where x and t are the space vector and time, respectively. Thermal conductivity (k), density (ρ) and heat capacity (C p ) of NP were taken from the literature. [31] is the local temperature, and represents the thermal energy induced by photo-to-thermal conversion within NFO NPs.
First-principles Calculation. All the free energies were calculated using the generalized gradient approximation (GCA) and the Perdew-Burke-Ernzerhof functional for the exchange correlation to the density functional theory (DFT). The projected augmented wave method was used, as implemented in Cambridge serial total energy package (CASTEP) code. [32] It is well known that the localized nature of 3d electrons is di cult to be accurately described by GCA, thus the effective on-site Hubbard U eff correction on the 3d or 4d electrons for all the transition metals included in our calculation were also employed. [33] The U eff parameter was taken from the previous theoretical reports, that is, 4.2 eV and 6.4 eV for Fe and Ni, respectively. [34,35] A 3D slab model with periodic boundary conditions was used with an energy cutoff of 500 eV and an appropriate γ-point, 3 × 3 × 1 mes, was selected to ensure that the total energies converged within 5 meV per formula unit.
The OER based in an alkaline electrolyte undergoes the following four elementary steps: where * represents an active site on the catalyst surface; OH*, O*, and OOH* are the oxygen intermediates.
The computational hydrogen electrode model proposed by Norskov and co-workers was used to express the thermochemistry of the sub-reactions at any given pH and applied potential U. [36] The Gibbs free energies were calculated from total energies as: where i = 1, 2, 3, 4 corresponds to steps from Eq. (1) to Eq. (4).
The Gibbs free energy changes for each step were calculated using the following equations: where the zero point energy (ZPE) and entropy correction values (TS) are listed in the Supplementary Table 3 and 4. eU represents the free energy changes for one electron transfer, where U is electrode potential with respect to the standard hydrogen electrode. For pH ≠ 0, ∆ (pH) should be de ned as -k B T log(pH), k B is Boltzman constant (1.380649 × 10 -23 J/K) [36] , and k B T is 0.025692 eV (T = 298.15 K).
The theoretical overpotential (η OER ), which is independent of pH, was then given by Eq. (10): Data availability. The data that support the findings of this study are available from the corresponding author upon request.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. GaoLKetalNCSIasof08122020ver2.docx