Photoelectrochemistry by Design: Tailoring the Nanoscale Structure of Pt/NiO Composites Leads to Enhanced Photoelectrochemical Hydrogen Evolution Performance.

Photoelectrochemical hydrogen evolution is a promising avenue to store the energy of sunlight in the form of chemical bonds. The recent rapid development of new synthetic approaches enables the nanoscale engineering of semiconductor photoelectrodes, thus tailoring their physicochemical properties toward efficient H2 formation. In this work, we carried out the parallel optimization of the morphological features of the semiconductor light absorber (NiO) and the cocatalyst (Pt). While nanoporous NiO films were obtained by electrochemical anodization, the monodisperse Pt nanoparticles were synthesized using wet chemical methods. The Pt/NiO nanocomposites were characterized by XRD, XPS, SEM, ED, TEM, cyclic voltammetry, photovoltammetry, EIS, etc. The relative enhancement of the photocurrent was demonstrated as a function of the nanoparticle size and loading. For mass-specific surface activity the smallest nanoparticles (2.0 and 4.8 nm) showed the best performance. After deconvoluting the trivial geometrical effects (stemming from the variation of Pt particle size and thus the electroactive surface area), however, the intermediate particle sizes (4.8 and 7.2 nm) were found to be optimal. Under optimized conditions, a 20-fold increase in the photocurrent (and thus the H2 evolution rates) was observed for the nanostructured Pt/NiO composite, compared to the benchmark nanoparticulate NiO film.

Introduction

With rapidly increasing global renewable energy harvesting capacities there is a rising interest for energy storage solutions, to overcome the intermittent nature of the various renewable energy sources. Photoelectrochemical (PEC) H2 evolution reaction (HER) and CO2 conversion are both promising avenues to generate solar fuels and thus store the energy of sunlight in the form of chemical bonds.[1,2] One of such PEC procedures is the direct reduction of H+ ions at the interface of a p-type semiconductor (SC) photocathode and a liquid electrolyte. Alternatively, when irradiating an n-type SC, water oxidation occurs at the SC/electrolyte interface, while HER can proceed on the counterelectrode. A range of photocathodes has been screened during the past decades,[3−6] and reasonable solar to hydrogen conversion efficiencies have been achieved employing complex electrode assemblies.[7,8] At the same time, there are still several open questions to be answered before high activity, stability, and selectivity could be achieved at the same time.[9]

Recent studies indicated that the use of nanostructured photoelectrodes may help to enhance electron transfer rates to levels which make practical significance.[10,11] While the high specific area (i.e., roughness factor) is a key advantage in electrocatalysis, the picture is much murkier in the photoelectrocatalytic scenario, because nanostructuring of photoelectrodes has both benefits and drawbacks. For example, charge carrier recombination due to both surface traps and the necessity of hopping from one particle to another, certainly decreases the achievable solar to chemical conversion efficiencies. Interconnected nanostructures may help to circumvent this issue, by eliminating the need of interparticle charge transfer. Different examples can be cited, ranging from mesoporous TiO2 vs TiO2 nanotubes,[12,13] nanoparticulate and nanoporous WO3,[14,15] and so forth.

NiO has been increasingly studied as a photocathode material, recently. NiO may also have practical significance in p-type DSSCs, in an inverted configuration compared to its TiO2-based counterpart.[16−18] Interestingly, although the NiO is long known to be capable of both evolving H2 and reducing CO2 (based on its conduction band edge position),[19] its solar fuel generation ability has not been extensively studied. There are a few examples, where NiO was studied in conjunction with either a solid state or molecular cocatalyst, in HER and CO2 ER. For example, a NiO/Cu2O heterojunction was employed as both light absorber and H2-evolution catalyst.[20] In another study, NiO was decorated with a Ru-dye and Ni catalyst to achieve visible light driven PEC H2-evolution.[21] PEC CO2 conversion to CO was demonstrated in combination with a Ru(II)–Re(I) containing supramolecular catalyst.[22] With respect to morphological questions, the PEC behavior of a nanoparticulate NiO film was compared with compact NiO recently, with special focus on their characterization with surface photovoltage spectroscopy.[23] On the other hand, there is only one single report on the PEC properties of interconnected NiO nanostructures.[24] The versatility of NiO is reflected in the fact that it has been also studied as protective coating for Si photoanodes, in solar fuel generation processes.[25,26]

Combination of a SC light absorber with a metal electrocatalyst is a promising avenue to further enhance the solar to chemical conversion efficiency, because the light absorption and charge transfer is decoupled. Consequently, by intimately linking an electrocatalyst to the SC surface a Schottky-junction can be formed, which in turn facilitates the rapid charge carrier extraction from the SC to the metal cocatalyst.[27,28] PEC H2-evolution takes place in the next step, on the surface of the metal nanoparticle cocatalyst. While various cocatalysts have been examined,[29] Pt is still considered as the state-of-the-art electrocatalyst for the EC HER.

The effect of Pt size was studied in different electrocatalytic processes. For example, in oxygen reduction reaction (ORR) the mass specific activity was increased sharply in the range of 1.2–2.2 nm Pt particles and slightly decreased for even larger particle sizes (>2.5 nm).[30] In another study CO electrooxidation was studied on Pt nanoparticles ranging from 1 to 30 nm, and the importance of particle shape and specific adsorption was also highlighted.[31] Similarly, Pt has been heavily studied and utilized as (co)catalyst in both thermally- and photoactivated heterogeneous catalysis. The effect of Pt nanoparticle size (diameter of 1–10 nm) on their heterogeneous catalytic activity was studied for both gas and liquid phase reactions. For example, in the case of alcohol oxidation, larger particles had higher activity towards the formation of aldehydes, ketones, and carbon-dioxide.[32−34] For benzene and toluene hydrogenation, ∼ 2–3 nm sized Pt nanoparticles had outstanding activity compared to both smaller and larger nanoparticles.[35] In photocatalytic hydrogen evolution (using oxalic acid as sacrificial e– donor) over TiO2 supported Pt nanoparticles, 3 nm sized particles showed the best activity compared to both smaller and bigger particles.[36] Photodriven CO2 reduction was performed on Pt/TiO2 composites, containing photodeposited ultrasmall Pt nanoparticles (below 2 nm).[37] The particle size and density on the PEC performance was also studied for Pt/Si samples, involving relatively large Pt particles (40–300 nm) on flat Si electrodes.[38]

Taking all the above listed precedencies as a whole, it is clear that there is considerable scope for the nanoscale engineering of both the light absorber and the cocatalyst in metal/SC composite electrodes. In this paper we report on high PEC H2 generation rates on optimized Pt/NiO nanostructures. We studied the effect of the NiO nanostructure as well as the Pt cocatalyst size (and loading), where we found that there is an optimal particle size and loading. Under properly adjusted conditions, a 20-fold increase in the photocurrent (and thus the H2 evolution rates) was observed compared to the benchmark nanoparticulate NiO film. Finally, we verified that particle size of the cocatalyst in a PEC configuration is equally important as it was in the previously studied electrocatalytic and photocatalytic scenarios.[30−35]

Experimental Section

Materials

For the anodization experiments Ni foils (99.9+%, 0.125 mm thick), NH4F (≥99%), and ethylene glycol (≥99%) were all purchased from Sigma-Aldrich, and KOH (≥86% in water) was obtained from Fluka. For cleaning the metal foils ethanol (Chromanorm) 2-propanol (Chromanorm), and acetone (Rectapur) were purchased from VWR. For the electrochemical measurements Na2SO4 (99%, Alfa Aesar) was used, along with N2 gas (99.995%, Messer, Budapest, Hungary). All solutions were prepared with ultrapure water (Milli-Q, ρ = 18.2 MΩ cm). All chemical reagents were used without further purification.

Anodization of the Ni Foil

Prior to anodization, the Ni foils were mechanically polished to mirror finish using silicon carbide sandpaper with successively finer roughness. To ensure the removal of any attached organic contaminants three subsequent steps (5 min) of ultrasonication in acetone, 2-propanol, and finally ultrapure water were performed. The anodization was carried out in a standard two-electrode setup, where the Ni foil acted as the anode and a Pt foil as the cathode. The Ni foil was pressed between two O-rings, leaving 2.27 cm2 exposed to the electrolyte, while the electric contact was located on the back of the sample. The anodization was carried out at different voltages using a programmable DC power supply (Voltcraft PSP 1803), while the procedure was monitored by a digital multimeter (Keithley 2000). Both instruments were controlled by custom-written LabVIEW software. Immediately after the anodization, the samples were carefully washed with ultrapure water and dried in N2 stream. To obtain the desired crystalline phase of NiO, the anodized samples were annealed for 4 h at 500 °C, with a heating ramp of 10 °C min–1 (Thermo Scientific Heraeus K114 Furnace).

Three different electrolyte compositions were used (the first two were adopted from the literature)[24,39] to perform the anodization:

Aqueous solution consisting of 0.5 M NH4F dissolved in 85 wt % H3PO4, with an anodization voltage of 6 V − 10 V, for 10 min.[39]

Organic media consisting of 0.5 wt % KOH, 5 wt % H2O, and 94,5 wt % ethylene glycol with an anodization voltage of 40 − 60 V and duration of 1 h.[24]

A combined electrolyte consisting of an ethylene glycol solution containing 0.15 M KOH, 0.1 M NH4F, and 3% v/v of H2O. During these syntheses the voltage was held at 30 V for 1 h.

Preparation of Pt Nanoparticles

Pt nanoparticles with various sizes were synthesized as described in the Supporting Information.[34] Briefly, NaOH and polyvinylpyrrolidone (Mw = 29 000) were dissolved in ethylene glycol together with the platinum precursor (H2PtCl6·xH2O or platinum(II) acetylacetonate). Subsequently, the mixture was heated to 160–200 °C in an oil bath and held at that temperature between 10 min and 2 h under argon atmosphere. Pt nanoparticles in five different sizes were synthesized, with a diameter of 2.0 ± 0.4, 4.8 ± 0.7, 7.2 ± 0.8, 8.6 ± 1.3, and 12.3 ± 1.4 nm. The resulting Pt nanoparticles were precipitated with acetone, centrifuged and dispersed in ethanol. Finally, the nanoparticles were repeatedly washed with hexane, centrifuged, and redispersed in ethanol before use. The exact concentration of the Pt suspensions was determined by ICP-MS measurements (Agilent 7700x type ICP-MS spectrometer).

Drop-Casting of Pt on the NiO Films

First the NiO samples were cut into four pieces with equivalent geometric areas of 0.57 cm2 to minimize the possible error arising from the use of different samples. One piece was always kept as an internal reference and every normalization was carried out with respect to this. The suspension of different sized Pt nanoparticles was prepared in ethanol (sonicated for 60 min) and was then drop-casted on the surface of the NiO samples (preheated on a hot plate at 65 °C). Investigations were carried out with different Pt loadings (0.5–4.0 μg cm–2). To improve the adherence between the Pt nanoparticles and the NiO scaffold, and to remove the capping polyvinylpyrrolidone from the Pt surface, the as prepared Pt/NiO composites were annealed in air at 300 °C for 2 h with a heating ramp of 5 °C min–1.

Physical Characterization

The X-ray diffraction (XRD) patterns were recorded with a Rigaku MiniFlex II benchtop instrument with Cu Kα X-ray source (λ = 1.5406 Å), in the 20–80° range, with a 2° min–1 scan rate. Raman spectra were recorded with a Thermo Scientific DXR Raman microscope at an excitation wavelength of 532 nm, applying 10 mW laser power, and averaging 20 spectra with an exposition time of 6 s. Scanning electron microscopic (SEM) images were captured using a Hitachi S4700 FE-SEM instrument. Transmission electron microscopic (TEM) investigation and electron diffraction analysis were performed using a FEI Tecnai G2 20 X-Twin type instrument, operating at an acceleration voltage of 200 kV. The Pt containing ethanol based suspensions were drop-casted on carbon film coated copper grids. In the case of the NiO samples, a small part of the surface of the electrode materials was scratched into a tiny jar filled with ethanol before ultrasonication. X-ray photoelectron spectra were recorded with a SPECS instrument equipped with a PHOIBOS 150 MCD 9 hemispherical analyzer. The analyzer was operated in the fixed analyzer transmission (FAT) mode with 20 eV pass energy. The Al Kα radiation (hν = 1486.6 eV) of a dual anode X-ray gun was used as an excitation source. The gun was operated at 210 W power (14 kV, 15 mA). The binding energy scale was corrected by fixing the main C 1s component to 285.0 eV, corresponding to the adventitious carbon. For spectrum acquisition and evaluation both manufacturers (SpecsLab2) and commercial (CasaXPS, Origin) software packages were used.

Electrochemical Measurements

All electrochemical measurements were performed using an Autolab PGSTAT302N instrument equipped with an FRA32 M module. The measurements were carried out in a standard three electrode setup, where the studied Pt/NiO acted as working electrode, a platinum foil and Ag/AgCl (saturated NaCl) electrode were used as the counter and the reference electrodes, respectively. The electrolyte was purged with N2 gas for 20 min before each measurement.

Photoelectrochemical measurements were carried out in a sealed two compartment, quartz cell where 0.1 M Na2SO4 solution was used as the electrolyte. Linear sweep voltammograms were recorded from 0.0 to −1.0 V with a sweep rate of 2 mV s–1 under periodically chopped illumination (0.1 Hz). The light source was a 300 W Hg–Xe arc lamp (Hamamatsu L8251). The radiation source was placed 3 cm away from the working electrode surface. During the long-term PEC measurements, the potential was held at E = −0.8 V, the solution was stirred continuously, and the solution temperature was kept constant at 25 °C. In these experiments, the gas phase products were detected with a gas chromatograph (Shimadzu GC-2010 Plus equipped with a barrier ionization discharge (BID) detector).

For the electrochemical impedance spectroscopy (EIS) measurements a closed electrochemical cell was employed, where 1.0 M Na2SO4 solution was used as the electrolyte. Prior to each EIS measurement, five cyclic voltammetry (CV) scans were recorded in the range of −0.8 to 1.0 V with a 50 mV s–1 sweep rate to ensure proper wetting of the porous electrodes and to estimate the electroactive surface area of the Pt nanoparticles. All impedance spectra were recorded at two different potential values (E = 0.0 V and E = −0.6 V), in the 10 Hz to 0.1 MHz frequency range, using a sinusoidal excitation signal (10 mV RMS amplitude).

Incident photon-to-electron conversion efficiency (IPCE) measurements were performed on a Newport Quantum Efficiency Measurement System (QEPVSI-B) in a single-compartment, three electrode quartz electrochemical cell. The wavelength range was 250–400 nm (Δλ = 10 nm step size). The IPCE profiles were recorded in a 0.2 M Na2SO4 solution at E = −0.8 V bias potential.

Results and Discussion

Anodization of Ni Foils

As described in the Experimental Section, anodization of Ni foils was carried out in three different complexing agent containing media. The first method employed F– ions, dissolved in concentrated H3PO4 media.[39] Different anodization voltages were used, and by increasing the applied voltage, the gradual increase of the steady-state current was observed (Figure S1). This trend is related to the rate of the dissolution of the metal-oxide layer,[24] and also affects the pore size of the obtained porous electrodes.[40] The samples prepared at 6 V showed a compact morphology, whereas the presence of nanopores was observed at higher anodization voltages (see SEM images in Figure S2). Unfortunately, this method led to F and P-impurities in the structure (see further discussion and EDX data in Figure S3A), which deleteriously affected the PEC response of the prepared electrodes (Figure S3B). To prevent the incorporation of the phosphate impurity, our attention shifted to a method employing a glycerol based media with KOH as the complexing agent.[24] In this case, however, a high etching rate of the formed NiO layer was observed (hindering morphology control), which eventually leads to the complete dissolution of the Ni foils. To combat both the phosphate inclusion and the fast dissolution of the formed NiO, a complex media was employed. The optimized synthesis media consisted of ethylene glycol as the solvent, which contained 0.15 M KOH, 0.1 M NH4F, and 3% v/v of H2O.

The use of organic solvent necessitated higher anodization voltages compared to the aqueous synthesis (because of higher solution resistance). As the anodization voltage was increased, faster NiO etching was observed (Figure A). Another typical feature of the anodization curves was the appearance of a distinct current density maximum.[41,42] This can be attributed to the increasing electrode surface area caused by the extensive pore formation. When the two competing reactions (oxide formation and dissolution) reach a steady state level, however, near-constant current densities stabilized at the later stages of anodization.

Figure 1

(A) The effect of applied voltage on the anodization traces recorded in 0.15 M KOH, 0.1 M NH4F, and 3% v/v of H2O containing ethylene glycol. (B) SEM image of the heat treated (T = 500 °C, 1 h) NiO sample obtained at U = 30 V, 1 h. (C) Pore size distribution of a typical NiO sample (U = 30 V, 1 h) annealed at T = 500 °C for 1 h.

SEM images were recorded to probe the morphology of the NiO films obtained at different anodization voltages. The nanoporous morphology was witnessed for all samples, but the pore size was found to be dependent from the anodization voltage (Figure B and Figure S5). On the high magnification SEM image, the interconnected nanoporous structure can be observed (Figure B). The size of the obtained pores fall into the range of 15 ± 5 nm (see pore size distribution in Figure C), while the grain size was slightly larger (∼20 nm, see also TEM images later in Figure ). These morphological features show close resemblance to the porous NiO electrodes prepared in fluoride containing electrolytes (first method). We note here that in the case of the glycerol-based alkaline method the formation of macroporous structures were reported.[24] In all further experiments the samples prepared at U = 30 V for 1 h anodization time were used.

Figure 7

(A) Typical TEM images of a heat treated porous NiO and (B) a Pt/NiO sample containing 7.2 nm average diameter Pt nanoparticles with a surface concentration of 5 μg/cm2. (C, D) Electron diffraction patterns of the respective materials shown in parts A and B.

Electrochemical anodization usually results in the formation of amorphous oxides materials, thus a subsequent heat treatment step is often necessary. To investigate the effect of heat treatment (300–600 °C) on the crystallinity of the samples, XRD measurements were carried out (Figure A). XRD profiles of the pristine NiO and a commercial NiO powder were also included as reference. The pristine anodized sample exhibited no reflections attributable to NiO, only sharp reflections of the Ni foil underneath the amorphous layer were observed (marked with an asterisk). By employing a heat treatment above 500 °C, the development of the most intense reflections of NiO (JCPDS #47–1049) was observed. The average size of the crystalline domains was estimated using the Scherrer equation for the most intensive (111) diffraction. A d = 22 nm value was obtained, very similar to the grain size obtained from the SEM images, which in turn confirms that the samples are fairly crystalline.

Figure 2

(A) XRD patterns recorded for the anodized NiO samples (U = 30 V, 1 h) heat treated at different temperatures for 1 h and the reference NiO powder. (B) Raman spectra recorded for the nanoporous NiO samples (U = 30 V, 1 h) heat treated at different temperatures and the reference NiO powder.

To further investigate the structural features of the formed NiO layers Raman measurements were carried out (Figure B). The reference NiO powder exhibits three distinguishable Raman active vibration modes, which correspond to one-phonon LO modes (at ∼513 cm–1), two-phonon TO + LO (at ∼851 cm–1) and 2LO (at ∼1048 cm–1) modes, respectively. These values are in close agreement with the reported values in the literature.[43] Furthermore, increasing heat treatment temperature, the gradual evolution of the LO mode was observed. While this vibrational mode is missing in the case of the unannealed (as is) sample, it is immediately observable even after a heat treatment at 300 °C. On the basis of the results from XRD and Raman spectroscopic measurements a heat treatment temperature of 500 °C was chosen for future studies. EDX measurements (Figure S4), however detected some residual carbon in the samples annealed for 1 h, thus the heat treatment step was extended to 4h.

Electrochemical Measurements

To evaluate the electrochemical behavior of the prepared nanoporous NiO electrodes cyclic voltammograms were recorded (Figure A). As reference measurements, NiO powder spray-coated on a glassy carbon electrode (Figure B) and a compact NiO layer (Figure C) (obtained by replacing the complexing agents in the anodization media with Na2SO4) were also studied. NiO usually exhibits two distinct redox peak pairs.[44] The first oxidation peak at E = 0.3 V is attributed to the oxidation of Ni(OH)2 to various NiO(OH) phases eq .[44] The exact reaction, however, involves different phases of NiO(OH) as described by the Bode model.[45] The second oxidation peak starting at E> 0.75 V is related to the oxidation of Ni(II) to Ni(III) eq .[23]In our case, only one broad anodic redox wave was seen centered at E ∼ 0.4 V, attributed to the redox transformation of Ni(OH)2 to various NiO(OH) phases. By determining the charge capacitance values, it was found that the nanoporous NiO sample surpasses both the spray-coated NiO nanopowder (3-fold increase) and the NiO compact layer (27-fold increase) in term of electroactivity. This enhanced behavior can be mainly attributed to the superior surface area (vs the compact layer) and the interconnected structure (vs nanoparticulate films) of the anodized NiO samples. These observations already projected that the nanoporous NiO films will be a viable platform for future PEC studies.

Figure 3

Cyclic voltammograms recorded in 0.2 M Na2SO4 with 50 mV s–1 sweep rate for (A) nanoporous NiO sample (prepared at U = 30 V, 1 h; heat treated at T = 500 °C, 4 h), (B) NiO powder on glassy carbon electrode, and (C) compact NiO layer (prepared at U = 30 V, no complexing agent, 1 h; heat treated at T = 500 °C, 4 h).

Photoelectrochemical Measurements

To evaluate the photoresponse of the anodized NiO electrodes linear sweep photovoltammetry measurements were carried out (Figure ). During illumination, the measured cathodic currents indicated a p-type semiconductor behavior of all studied samples. In these cases, no electron scavenger species was added, thus the photoresponse can be mainly ascribed to the photoreduction of water. Compared to the anodized and heat treated NiO layer (Figure A), both the spray-coated NiO nanopowder (Figure B) and the compact NiO layer (Figure C) showed inferior photoactivity (note the difference in the scale bar of the current density). In the latter two cases this may be the result of inadequate charge carrier extraction. As for the powdered sample, extensive charge carrier recombination may occur on the surfaces of the nanoparticles, which ultimately limits the photoresponse of this material. The case for the compact sample is somewhat different. Although the interfaces between particles are absent, the photogenerated carriers on the surface of the layer cannot be extracted through the bulk of the oxide (note the low conductivity of NiO). Thus, it can be concluded that the anodized sample exhibits superior PEC performance because of its interconnected nanoporous structure.

Figure 4

Linear sweep photovoltammograms recorded in 0.2 M Na2SO4 with 2 mV s–1 sweep rate for (A) nanoporous NiO sample (prepared at U = 30 V, 1 h; heat treated at T = 500 and 300 °C, 4 h), (B) NiO powder on glassy carbon electrode, and (C) compact NiO layer (prepared at U = 30 V, 1 h; heat treated at T = 500 °C, 4 h).

A further decisive parameter dictating the photoresponse of a material is its crystallinity. In Figure A the photoresponse of the anodized NiO samples heat treated at two different temperatures are compared. The sample annealed at 300 °C, which exhibited no crystalline features, showed negligible photocurrents, while the sample heat treated at 500 °C showed jph = −80 μA cm–2 at E = −0.7 V.

Pt Decoration of the NiO Electrodes

Pt nanoparticles with controlled size were synthesized in ethylene glycol by the polyol method using different Pt-based precursors and polyvinylpyrrolidone (PVP) as a capping agent (for more details see Supporting Information). The as-prepared Pt nanoparticles have narrow size distribution with an average diameter of 2.0 ± 0.4, 4.8 ± 0.7, 7.2 ± 0.8, 8.6 ± 1.3, and 12.3 ± 1.4 nm, respectively (Figure ). This allowed to cover the 1.5–16 nm size regime, which is typically employed in most catalytic studies. The nanoparticles mostly have spherical shape, however octahedron type particles can also be observed in samples with larger nanoparticles.

Figure 5

TEM images of the different sized Pt nanoparticles (A–E), together with the comparison of their size distributions (F).

To investigate the morphology of the Pt/NiO electrodes SEM and TEM images were recorded. SEM image of a Pt/NiO electrode (7.2 nm average Pt diameter) is compared to the corresponding bare NiO in Figure . Even at high magnifications, the 7.2 nm Pt nanoparticles are too small to be observed directly, however there are some distinguishable changes, which indicate the presence of Pt on the NiO surface. First, the surface is smoother before drop-casting of Pt, than afterward. Furthermore, upon drop-casting, the pore structure also altered: the diameter of the pores decreased and some were even clogged. The pore size distribution was also determined before and after decoration with Pt (Figure C). Most importantly, the fraction of the large pores decreased notably, due to the incorporation of Pt nanoparticles. The amount of the small pores increased in parallel as expected. These observations together indicate that a high amount of Pt nanoparticles accumulated in the pores, close to the top of the NiO layer.

Figure 6

SEM images of (A) an anodized NiO film, (B) a Pt/NiO composite (7.2 nm Pt), and (C) pore size distribution before and after anchoring Pt (determined from SEM images in parts A and B).

TEM images of the debris of porous NiO films were scrutinized, and the presence of a mesoporous structure was witnessed. The average diameter of the joined building blocks were around 20 nm (Figure A). Note that this value is in perfect agreement with the previously shown XRD and SEM data. As for the Pt/NiO sample (containing 7.2 nm Pt), individual Pt nanoparticles can be spotted on the image (Figure B). The electron diffraction study of the nanoporous nickel oxide reveals NiO(111), Ni(200), NiO(200), NiO(220), NiO(311), and NiO(222) crystallite planes characteristic for face-centered cubic (fcc) nickel(II) oxide (Figure C). For Pt-decorated NiO, crystallite planes of Pt(111), Pt(200), and Pt(311) showed the presence of metallic face-centered cubic (fcc) platinum (Figure D).

HR-TEM images were also recorded for a composite sample with higher Pt loading (10 μg cm–2, Figure and Figure S6). The most important message of this TEM images is that nanosized Pt particles can be spotted anchored to the interconnected porous structure of NiO. In addition, clear lattice fringes corresponding to the (111) facets of Pt as well as the (111) facets of NiO were observed (Figure ).

Figure 8

High-resolution TEM images of 7.2 nm Pt nanoparticles anchored onto the surface of nanoporous NiO.

XPS Measurements

To evaluate the surface chemical characteristics of the NiO electrodes and the drop-casted Pt nanoparticles, XPS measurements were carried out. NiO-based materials exhibit complex surface chemistry because of nonstoichiometry and hydration under ambient conditions.[46] The presence of different Ni–oxide/hydroxide species can make the quantitative evaluation of the XP spectra problematic and should be only attempted in the presence of high quality standards.[46−48] In this vein, we focused on the qualitative evaluation of the surface characteristics. In Figure regions corresponding to Ni 3p (A), Ni 2p (B), and O 1s (C) are shown for a NiO sample decorated with 8.6 nm average diameter Pt nanoparticles. Because of the extensive overlap between the Ni 3p and the Pt 4f signals, deconvolution of the Ni 3p components was not attempted. Judging from the position of the asymmetric Pt 4f7/2 component (71.2 eV), together with the Pt 4d5/2 peak positions (Figure S8) however, the zero valence of the drop-casted Pt can be confirmed in all cases (Figure A).[49,50] Notably, a small amount of surface oxide (PtO) was only detected in the case of the smallest Pt nanoparticles (Figure S8). The main Ni 2p3/2 line was modeled with three components: (i) NiO (853.7 eV) and (ii) Ni(OH)2 (855.2 eV) and NiOOH (856.4 eV), which are consistent with literature data.[46] The addition of NiOOH into the model was dictated by the chemical nature of the samples, because its formation is characteristic of electrochemically grown NiO layers.[24,51] In the O 1s region (Figure C) similar chemical entities were distinguished in accordance with previous literature data.[46] When fitting the spectrum the contribution from oxygens of NiO (529.2 eV), Ni(OH)2 (530.7 eV), NiOOH (532.1 eV), and physisorbed H2O (532.9 eV) were all taken into account.

Figure 9

(A) Pt 4f/Ni 3p XP spectrum, (B) Ni 2p3/2 XP spectrum, and (C) and O 1s XP spectrum of the NiO sample decorated with 8.6 nm Pt nanoparticles. Resolved curves are added only to guide the eye to changes in spectral shape and do not represent sufficient component analysis.

Electrochemical and Photoelectrochemical Characterization

To assess the dark electrochemical behavior of the Pt/NiO nanocomposites, cyclic voltammograms (CVs) were recorded (Figure A). Two main trends were immediately observed upon the addition of Pt: (i) a dark current developed in the range of E = −1.0 to −0.4 V (arising from direct water reduction on the Pt nanoparticles) and (ii) a change in the overall electroactivity of the electrodes. For pristine NiO, negligible dark current flow was observed at the cathodic end of the potential window (E = −1.0 to −0.4 V). When the surface of the NiO electrodes is decorated with Pt nanoparticles, however the gradual evolution of a dark current was witnessed. The onset potential of this process is located at E = −0.2 V regardless of the size of the Pt nanoparticles. This observation indicates that the size of the nanoparticles predominantly affects the kinetic aspects of the dark water reduction process. In the positive potential regime (E > 0.0 V) higher electrochemical activity was observed for the Pt loaded NiO samples compared to their pristine NiO counterpart. Although the electrochemical transformation of NiO and Pt overlap in this potential regime, the increased electroactivity can be attributed to the presence of Pt related redox peaks.[31,52] Among all the studied samples the one decorated with 2.0 nm Pt nanoparticles exhibited the highest electrochemical activity, and the relative enhancement decreased with the increasing Pt particle size. These data served as the basis of normalization of the PEC performance (see below), where the actual surface area of the cocatalyst may play a key role in the overall performance.

Figure 10

(A) Cyclic voltammograms of NiO and Pt/NiO composites 2 μg cm–2. (B) Linear photovoltammogram of Pt/NiO 2.0 nm, with different Pt loading in 0.2 M Na2SO4, saturated with N2 gas.

To characterize the PEC behavior of the Pt/NiO composites, linear sweep photovoltammetry measurements were carried out with a slow sweep rate (2 mV s–1) in 0.2 M Na2SO4. The illumination was periodically interrupted (0.1 Hz) to record the response of the electrodes under both dark and illuminated conditions. In almost all the cases (except for the largest Pt size), the cathodic photocurrents of the Pt-decorated NiO samples surpassed their pristine NiO counterparts. This superior behavior could be attributed to the more efficient electron–hole separation in the prepared nanocomposites.[23] By increasing the Pt loading, higher photocurrents were achieved until an optimal composition was reached. After this maximum, the optical shielding effect of the Pt nanoparticles on the surface resulted in a decrease in the photocurrents. For the largest particle, this shielding effect seemed to be deleterious even at moderate loadings. To strip the measured photocurrent enhancements from possible uncertainty arising from the use of different NiO electrodes, the photocurrents were normalized by the photocurrent measured for the respective pristine NiO electrodes (Figure S7). To find the optimal composition and particle size, these normalized photocurrent values were determined of each Pt decorated sample (Figure A, and Figure S7). It is apparent that the largest increase in the photocurrent value was achieved in the case of 2.0 and 4.8 nm sized Pt nanoparticles.

Figure 11

(A) Particle size dependence of the photocurrent enhancement for the Pt/NiO samples (2 μg cm–2 loading). The error bar represents measurements carried out for three different sample for each particle size. Lines connecting the measured data serve as guides for the eye only. (B) Nyquist plots of the electrochemical impedance spectroscopy data, recorded for the bare and different Pt-decorated NiO samples, at E = 0.0 V in the 10 Hz to 0.1 MHz frequency range, in 1.0 M Na2SO4 solution. The inset shows the magnified region of the semicircles.

The origin of this photocurrent enhancement, however, is complex and convoluted. First of all, the surface area difference of the various sized Pt nanoparticles can be reflected in these increased values. To exclude this factor and to shed light on possible other, more sophisticated size-effects, further normalization of the data was carried out. The extra redox peak observed for the Pt-decorated samples during CV measurements is a direct indicator of the electroactive surface area of the Pt nanoparticles. As the first step, the charge capacitance of the bare NiO was subtracted from that of the Pt-decorated samples. Subsequently, the normalization of the photocurrent enhancement with respect of the electrochemically active surface area of the Pt nanoparticles was carried out. When the trivial surface area effect was stripped from the data the maximum photocurrent increase was obtained for the samples decorated with 4.8 and 7.2 nm sized Pt nanoparticles.

This observation is indeed interesting, because in earlier studies the highest heterogeneous/electrocatalytic activity was typically observed for even smaller particles (mostly because of the larger number of active sites).[30−35] In our case, however, the compatibility of the NiO and Pt has to be also considered. As the average pore size of the optimized NiO samples was 15 ± 5 nm (see also pore size distribution in Figure C), the large sized Pt particles simply do not infiltrate into the deeper regions of the nanoporous NiO film, but rather remain on its top. Such architecture is obviously not favorable considering the low contact area between NiO and Pt, and the possible optical shielding effect. On the other hand, the smallest (2 nm) Pt particles are likely to penetrate to the bottom region of the anodized NiO layer. As the light penetration is limited in the deeper regions, the cocatalytic effect of these deeply embedded Pt particles cannot be fully harnessed. Finally, the intermediate sized (4.8 and 7.2 nm) Pt particles may partially penetrate, resulting in a beneficial Pt/NiO nanoarchitectures. This hypothesis is qualitatively supported by XPS data, where the surface concentration of the Pt nanoparticles was estimated. It was found that the high specific surface area of the small sized Pt nanoparticles is not reflected in the surface Pt/Ni ratio, most likely because it is compensated by their increased penetration into the NiO layer (where they are invisible for XPS).

Electrochemical impedance spectroscopy (EIS) measurements were carried out to characterize the electrical properties of the Pt decorated NiO electrodes. The Nyquist-plots of the samples are presented in Figure B. Qualitatively, one may notice that the impedance spectra exhibit a depressed semicircle at high frequencies. It is also apparent that all Pt containing samples possess lower charge transfer resistance (the decrease in the end point of the semicircle) compared to the bare NiO electrodes. Overall, the lowest charge transfer resistance was obtained for the sample with the smallest Pt size, but all Pt-decorated electrodes have much smaller Rct, compared to the bare NiO. This trend suggest that the photocurrent increase is rooted in the improved charge transfer at the electrode/electrolyte interface. Further details, including the fitting of EIS data and the employed equivalent circuit are shown in the Supporting Information (Figure S10 and Figure S11).

To further study the origin of the photocurrent enhancement, photoaction spectra were recorded. Figure S9 shows the photoaction spectra of the pristine and 2.0 nm sized Pt decorated NiO electrodes in 0.2 M Na2SO4 recorded at a constant bias potential of E = −0.8 V. The bandgap of the materials was estimated by fitting the cutoff region of the photoaction spectra with a straight line segment and extrapolating it to the wavelength axis. For both the pristine and Pt decorated samples a similar bandgap value of Ebg = 3.55 eV was found. This is in good agreement with the 3.5 eV value reported for NiO in the literature,[23] and our own optical data obtained via diffuse reflectance UV–vis spectroscopy (3.50 ± 0.1 eV, raw data not shown here). The decoration of the NiO electrodes with Pt nanoparticles leaves the bandgap of the oxide hybrids intact. Therefore, confirming our previous conclusion, the improvement in the PEC behavior of the Pt-decorated samples can be attributed to better charge separation rather than increased optical absorption.

Long-Term Photoelectrochemical Measurements

Long-term photoelectrolysis, combined with in situ gas phase analysis, was performed to monitor the formation of H2 gas (Figure ). At the initial stage of the photoelectrolysis some dark current was detected, most likely because of the reduction of some minority Ni(III) species in the sample (see also XPS data above). The measured photocurrents slowly decreased during the electrolysis, partly because of the leaching of Pt nanoparticles to the solution (as confirmed by comparing the XPS data before and after photoelectrolysis). After half an hour, there was a detectable amount of H2 in the gas phase. The H2 concentration increased linearly with the transferred charge. The Faradaic efficiency of this process was around 70%, most likely because of minor leaking of the electrochemical cell. Long-term measurements for other Pt/NiO electrodes (with different Pt sizes) were also conducted (not shown here), where a similar pattern was found. XPS investigations taken after the long-term measurements revealed that the chemical state of Pt and NiO was not altered during the process.

Figure 12

(A) Long-term photoelectrolysis on a Pt/NiO electrode (7.2 nm sized Pt) at E = −0.8 V potential in 0.2 M Na2SO4. (B) Amount of evolved H2 during the long-term measurement shown in part A.

Conclusions

In this study, we carried out the simultaneous optimization of nanoporous NiO and Pt nanoparticles to assemble Pt/NiO composite photoelectrodes with improved PEC performance. This system served as a model for photoelectrodes encompassing a semiconductor light absorber and a metal nanoparticle cocatalyst. As the first step a new anodization protocol was developed, which allowed the formation of pure NiO, with controlled morphological attributes. In the second step, Pt nanoparticles were synthesized with precise size control (five sizes between 2.0 and 12.3 nm average diameter) and narrow size distribution. Finally, the Pt nanoparticles were anchored on the NiO surface with different loadings. All the hybrid samples showed enhanced cathodic photocurrents compared to the bare NiO film, although to notably different extents. It was shown that the enhanced PEC activity is rooted in the lower charge carrier recombination rate in the case of the composite samples (i.e., Pt acted as a cocatalyst in the H2 evolution reaction). Comparing the relative enhancements in the photocurrents upon the addition of different sized Pt particles, it was found that the two smallest (2.0 and 4.8 nm) have the most beneficial effect. After peeling off the trivial surface area effect (performed by normalizing the photocurrent enhancement with the electrochemically active surface area of Pt) the two intermediate particle sizes (4.8 and 7.2 nm) were found to be significantly better than both the smaller and larger particles. The main outcome of this study is that the size of the cocatalyst exerts an important role in the overall performance, similarly to the electrocatalytic and photocatalytic scenarios.[30−35] The complete picture, however, is even more complicated because light absorption, charge carrier transport, and surface chemical reactions have to be considered at the same time. We have demonstrated that in the case of nanocomposite photoelectrodes (e.g., light absorber + cocatalyst) the nanoscale structure of the components has to be optimized simultaneously, to achieve reasonable PEC activity. The conclusions of this study may lead to new design concepts of photoelectrode assemblies, and such endeavors are in progress in our laboratories.