1 Introduction

Solar light driven photoelectrochemical (PEC) water splitting to produce hydrogen is a promising technology to meet future energy requirements of the world [1]. Continued research pursuits of nearly five decades have yielded substantial progress both in understanding the phenomenon and evolving the technological knowhow that has risen hopes for large-scale application of this technology in near future for renewable hydrogen generation [2, 3]. Key component of PEC cell device is the photoactive semiconductor working electrode (WE) that serves the twin functions of light-absorption and separation of photogenerated charge carriers for their participation in water splitting electrochemical redox reactions [4]. Investigations on design and fabrication of photoelectrode, using various possible semiconductors and their modifications, have yielded significant gain in energy conversion efficiency [5,6,7,8]. Yet, to attain the stage of commercial-viability and cost-competitiveness; achieving high efficiency through specific use of only earth-abundant low-cost materials is desired, instead of expensive, complicated and often unstable materials/material systems. From the long list of potential semiconductors for above application, few old work-horses, viz. oxides of iron, copper, titanium and zinc continue to interest researchers by being low-cost, earth-abundant and nontoxic [4, 9,10,11].

CuO is one of the visible-light absorbing semiconductors that absorbs nearly 42% of solar light. With conduction band edge placed at more negative potential, compared to water reduction potential for hydrogen evolution, and band gap energy being larger than the energy needed (~ 1.23 eV) to dissociate water molecule, CuO emerges as a strong material-candidate for PEC water splitting [12]. However, stability of CuO is a concern, as it photocorrodes easily in PEC cell [12,13,14]. ZnO, in contrast, is a reasonably stable ultraviolet-light absorbing wide band gap material, exhibiting ~ 4% solar light absorption [15]. Like other wide band gap materials, viz. TiO2 and WO3, ZnO is another promising material for above application, mainly due to its ability to generate the potential needed to propel water splitting redox reactions [16, 17].

Recent thrust on synthesis of three-dimensional nano-hetero-structures (3D-NHS) of metal oxides has opened up new technological opportunities for their use as efficient photoelectrode in PEC water splitting. Hence, bi-layered nano-hetero-structured thin films, fabricated carefully by choosing appropriate combination of metal oxide semiconductors, have stimulated much interest. Advantages envisaged with such nano-hetero-structures are enhanced absorption of light, increased electrical conductivity, diminished recombinations and increased surface area in contact with electrolyte [18,19,20]. Few such systems investigated, viz. ZnO/Cu2O [21, 22], Fe2O3/ZnFe2O4 [23], ZnO/TiO2 nanotubes [24], reduced graphene oxide/Cu2O/Cu [25], copper decorated WO3–TiO2 nanotubes [26] and CuWO4/Cu2O [27] have indeed shown enhanced PEC activity and the encouraging trend deserves further exploration with other combinations of prospective semiconductors.

Bilayered heterojunctions of the oxides of copper and zinc have been used in PEC water splitting by many researchers, yielding varied photoresponse. Shaislamov et al. reported insignificant increase in photoresponse on using highly stable CuO/ZnO nanorod-naobranch photocathode [28]. In contrast, Zhang et al. reported significant gain in PEC response with ZnO/CuO bilayered films [29]. Kwon et al. obtained enhanced optical and PEC response on using ZnO/CuO/Ag and ZnO/CuO/Au hierarchical nanostructures [30]. Kargar et al. grew three dimensional ZnO/CuO heterojunction branched nanowires and reported good correlation between electrode current, size and density [31]. Wu et al. also reported significantly improved PEC performance on using three dimensional nanostructure of CuO with ZnO [32]. Tsege et al. fabricated CuO/ZnO nanowire based photocathode, where again combined effect of charge separation and light absorption increased the PEC response [33].

This report is in continuation of authors’ earlier works on the use of nanostructured doped/undoped thin films of CuO and ZnO, reporting maximum photocurrent (1.12 mA cm−2 at 0.75 V bias vs. SCE) with Cu-incorporated ZnO films [34, 35]. Reporting further increment in photocurrent generation, this study deals with facile synthesis and characterization of bi-layered nano-hetero-structured thin films, CuO/ZnO (BNHS) and CuO/Cu-incorporated ZnO (Cu-BNHS), for possible use as photoelectrode in PEC water splitting under artificial illumination. Synergistic effect of the two oxides in uniquely evolved nano-hetero-junction thin film photoelectrodes could generate efficient absorption of light alongwith smoothened flow of photo-induced charge carriers. Cu-BNHS films, with 3% Cu-incorporation in ZnO yielded the highest photocurrent (2.98 mA cm−2, at 0.85 V bias versus SCE), with nearly 142 and 8 times increment against monolayered pritstine thin films of CuO and ZnO, respectively.

2 Experimental

2.1 Synthesis

Pre-cleaned ITO (In:SnO2) glass sheets (dimension: 1 cm × 1 cm; sheet resistance: 8 Ω) were used as substrate to grow thin films of chosen oxide semiconductors [36]. Nearly 1/3rd area of the substrate (considered from a side edge) was masked during film growth to enable the use of this protected area later in establishing electrical contact for using films as working electrode (WE) in PEC studies. Synthesized films were stored in moisture-free environment at temperature 15–25 °C.

2.1.1 Synthesis of monolayered pristine CuO thin films

To grow monolayered CuO pristine films, equimolar solution of copper (II) acetate monohydrate (Cu(NO3)2·H2O) and dextrose (C6H12O6) was prepared in 2-propanol ((CH3)2CHOH; 20% V/V in water) and sprayed onto pre-heated substrate using a pneumatic nebulizer (nozzle diameter 0.1 mm; optimized air pressure 2 kg cm2) [37]. Films were prepared through 3 spray-cycles, each of 20 s, with in-between-cycle interval of 2 min to allow drying of the coated layer. Films obtained after 3 cycles of deposition were subjected to ageing at room temperature (32 ± 2.6 °C) for 24 h to complete the process of crystallization.

2.1.2 Synthesis of monolayered pristine ZnO thin films

To grow monolayered ZnO pristine films, a sol (0.5 M) of zinc acetate dihydrate (Zn(CH3COO)2·2H2O) was prepared in 2-propanol ((CH3)2CHOH) and mixed with 8% V/V diethanolamine under continuous stirring for 4 h at 33 °C. Prepared sol was spin-coated at 20,000 rpm, using a photoresist spinner (Milman), for 5 cycles of deposition. Each cycle was of 20 s, with in-between-cycle interval of 5 min to dry the coat at 60 °C. Finally obtained films were annealed in air at 500 °C for 1 h followed by slow cooling for 24 h [38].

2.1.3 Synthesis of bilayered CuO/ZnO and CuO/Cu–ZnO thin films

CuO/ZnO and CuO/Cu–ZnO bilayered thin films were synthesized by depositing an over-layer of ZnO through sol–gel method on previously obtained CuO films. Procedure followed was the same as enumerated above. In the synthesis of CuO/Cu–ZnO films; zinc acetate dihydrate sol, used in the deposition of ZnO overlayer, was spiked with varied amounts of copper acetate monohydrate to achieve 1, 2, 3 and 4% at. Cu incorporation.

2.2 Characterization

Prepared thin films were characterized for different physico-chemical characteristics relevant for their use as photoelectrode in PEC water splitting. Film thickness was measured using surface profilometer (Tecnor Alpha Step D-120) [39]. Density was estimated gravimetrically from weight/volume ratio by using the method described elsewhere [35, 38]. In-situ resistivity (r) of films was estimated from Eq. 1 and the current (I)–voltage (V) plots of PEC cell obtained using films as working electrode (WE) in conjunction with platinum counter electrode (CE) and saturated calomel reference electrode (SCE) [38]. Ohmic resistance (R) of the cell, operative under darkness and beyond breakpoint voltage, was determined from slope of the linear portion of I–V plots. Under the condition, resistance of cell circuit is approximated to thin film sample alone, as platinum counter electrode and electrolyte together offered only negligible resistance [40]. Here, R is the resistance, A the geometrical area of film surface, and t the thickness of film.

$$r = RA/t$$
(1)

Crystallographic characteristics were examined by XRD analysis using Bruker AXS D8 Advance (Germany) X-ray diffractometer, equipped with CuKα radiation source, graphite monochromator, and mirror at fixed incidence angle of 1–5°. Angular accuracy and angular resolution of the measurement were 0.001° and 0.01°, respectively. By maintaining step size of 0.02° per minute, diffraction angle (2θ) was varied from 30 to 60°. Debye–Scherrer’s computations, based on Eq. 2, yielded average crystallite size (s) [40]. Dislocation density (η), microstrain (γ) and lattice strain (ζ) were estimated from Eqs. 3–5 [40, 41]. Here, k is constant (0.9), λ the wavelength of X-ray used (1.5418 Å), B the full width at half maxima, θ the half diffraction angle of peak-centroid.

$$s = k \, \lambda /B{\text{Cos}}\theta$$
(2)
$$\eta = { 1}/s^{ 2}$$
(3)
$$\gamma = \, \left( {B{\text{Cos}}\theta } \right)/ 4$$
(4)
$$\zeta = \, \left( {B{\text{Cos}}\theta } \right)/ 4 {\text{ Sin}}\theta$$
(5)

Light absorption properties of films were investigated, in the wavelength range 350–800 nm, using double beam UV–visible spectrophotometer (UV-2450, Shimadzu, Japan) and bare ITO glass sheets as reference. Absorbance data and Eq. 6 were used to generate (αhv)2 versus hv Tauc plots and estimate optical band gap energy (Eg). Here, C is the constant, α the optical absorption coefficient, and hv energy per photon. Value of 1 and 4 is ascribed to integar n in the computation for direct and indirect band gap energy, respectively [42].

$$\alpha hv \, = \, C(hv - E_{g} )^{n/2}$$
(6)

To explore surface morphology of films, images were recorded using Scanning Electron Microscope (SEM, Carl Zeiss SUPRA 40 VP) at 15 kV accelerating voltage and ~ 4 mm working distance. Cross-sectional SEM images were also recorded for the highest PEC photocurrent yielding bilayered thin films (sample CZC3) to observe the exact growth of bi-layered nano-hetero-structure. Size-distribution of particles in the prepared samples was tentatively obtained by analysing 80–100 randomly picked particles in the obtained SEM images through ImageJ software [40]. Energy-Dispersive X-Ray (EDX) analysis of films, using MIRA II LMH TESCAN EDX unit attached to SEM, revealed their elemental composition. Observations were recorded at 6–8 randomly chosen points, spread all through the film surface, to obtain the average elemental composition of sample at ± 4.8% standard deviation. To explore the surface topography, thin films were analysed using Nanosurf easyScan Atomic Force Microscope (AFM, Version 1.8, Switzerland). Images (with 256 × 256 data points) were generated by scanning 5 μm × 5 μm of film area at 20 µN of force. The data was also utilized to evaluate Root Mean Square (RMS) surface roughness values [40].

XPS analysis of the highest PEC photocurrent yielding bilayered films (sample CZC3) was attempted using high resolution XPS unit equipped with AES Module, Ar ion and C60 sputter guns (PHI 5000 Versa Probe II, FEI Inc.). Equipped with Mg–Kα radiation, the overall energy resolution of the instrument was 0.7 eV. To revalidate compositional and crystallographic properties, representative sample CZC3 was analysed using JEOL JEM-1400 Transmission Electron Microscope (TEM), featured with TEMCON software, 120 kV maximum accelerating voltage, ± 70º tilted computer controlled stage, LaB6/Tungsten filaments emitter, large area SDD detectors and compact foot print.

Mott–Schottky (MS) analysis was performed to estimate flat band band potential (Vfb) and charge carrier density (Nd). Impedance data of PEC cell, recorded under darkness through LCR meter (Agilent Technolgy, Model 4263B), enabled determinations on the variation of space charge capacitance (Csc) with applied voltage (V). Assuming negligible contribution of surface states towards the built-in-capacitance, Eqs. 7 and 8 were employed to genmerate MS plots [43].

$$1/C_{\text{sc}}^{ 2} = \, \left[ { 2\left( {\varepsilon_{0} \varepsilon_{\text{s}} qN_{\text{d}} } \right)} \right] \, \left[ {V - V_{{{\text{fb}} - }} - \, \left( {k_{\text{B}} T/q} \right)} \right]$$
(7)
$${\text{S }} = { 2}/\left( {\varepsilon_{0} \varepsilon_{\text{s}} qN_{\text{d}} } \right)$$
(8)

Here, q is the electronic charge, εs and ε0 the permittivity of semiconductor electrode and free space (8.85 × 10−12 F m−1 for vacuum), respectively, T the absolute temperature, q the electronic charge, S the slope of MS plot and kB the Boltzmann’s constant. In this measurement, aqueous solution of NaOH (0.1 M, pH 13) was used as electrolyte. The capacitance at working electrode–electrolyte interface was measured under 1 kHz signal frequency and lowest possible (~ 5 mV) AC voltage amplitude that was also comparable with open circuit voltage (Voc). Values of Nd and Vfb were obtained, respectively, from the slope of MS plot and its intercept on the potential axis [43]. To explore charge transafer kinetics, films were studied by potentiostatic electrochemical impedance spectroscopy, using electrochemical work station (EWS, Model: PP211, CIMPS-pcs, Zahner, Germany). Nyquist plots, representing a plot of real and imaginary impedance, were obtained. Charge transfer resistance, operative in the system, is expressed from the diameter of Nyquist plot arc, with smaller radii arc/semi-circle indicating least charge transfer resistance [44].

2.3 PEC studies

Ohmic electrical contact was established with synthesized films to allow their use as working electrode (WE) in PEC studies [45]. PEC cell was constituted by taking sample film as WE, platinum mesh as CE and saturated calomel electrode (SCE) as reference—all dipped in 200 cm3 of 0.1 M aqueous solution of NaOH (pH ≈ 13). With 0.1 M NaOH solution filling the PEC cell, SCE may not be the best choice of reference electrode, as it may lead to the diffusion of hydroxide ions into the electrode filling solution and the formation of metal hydroxy/oxide species, which would cause shift in the electrode potential. In the present work, the likely problem with SCE, was ameliorated by using double junction that effectively slowed the hydroxide diffusion for the duration of PEC measurement. Using electrochemical work station (Model: PP211, CIMPS-pcs, Zahner, Germany), I-V plots of PEC cell were recorded both under illumination and darkness. A light source (Ozone-free Xenon Arc Lamp, 150 W, illumination intensity 130 mW cm−2, Oriel, USA) was used to shine the working electrode. To eliminate the possibility of any interference, electrolyte solution was cleansed from dissolved gases by purging with N2 gas for 15 min before each PEC measurement. Further, PEC cell was surrounded by a water jacket to prevent IR heating and maintain the temperature of the cell constant, throughout the measurement. Applied voltage (versus SCE) was varied from − 1.0 to 1.0 V with scan rate of 20 mV s−1. Based on photocurrent values (Iph= Idarkness − Iillumination) and Eq. 9, Applied Bias Photon-to-Current Efficiency (ABPE) for the PEC cell was estimated that provided implicit indications on overall performance and suitability of synthesized samples in PEC water splitting.

$${\text{ABPE }} = \, [I_{\text{ph}} \times \, ( 1. 2 3 { } - \left| V \right|]/P$$
(9)

Here, V is applied voltage, P the incident illumination power density (mW cm−2), and Iph the photocurrent density (mA cm−2).

Throughout the study, only AR grade reagents (purity > 99.9%) and double distilled de-ionized water (specific conductance < 10−6 mho cm−1) were used. Triplcate measurements yielded results with standard deviation < 6.4%.

3 Results and discussion

3.1 Sample synthesis

Terntative set of chemical reactions (I–IV) for the synthesis of CuO film (underlayer in case of BNHS and Cu-BNHS samples) suggest critical role of dextrose (reducing agent) in regulating reaction kinetics [46]. Temperature is another important variant. Below 300 °C, Cu2O (cuprous oxide) is formed, which is converted to CuO (cupric oxide) on annealing at temperature above 300 °C [46].

$${\text{Cu}}^{ 2+ } + {\text{ C}}_{ 6} {\text{H}}_{ 1 2} {\text{O}}_{ 6} \left( {\text{dextrose}} \right) \, + {\text{ H}}_{ 2} {\text{O}} \to {\text{Cu }} + {\text{ C}}_{ 6} {\text{H}}_{ 1 2} {\text{O}}_{ 7} \left( {\text{gluconic acid}} \right) \, + {\text{ 2H}}^{ + } \left( {\text{I}} \right)$$
$${\text{C}}_{ 6} {\text{H}}_{ 1 2} {\text{O}}_{ 7} + { 5}\raise.5ex\hbox{$\scriptstyle 1$}\kern-.1em/ \kern-.15em\lower.25ex\hbox{$\scriptstyle 2$} {\text{O}}_{ 2} \to 6 {\text{CO}}_{ 2} + {\text{ 6H}}_{ 2} {\text{O }}\left( {\text{II}} \right)$$
$${\text{Cu }} + {\text{ C}}_{ 3} {\text{H}}_{ 8} {\text{O }}\left( { 2 {\text{-propanol}}} \right) \, + {\text{ H}}_{ 2} {\text{O}} \to {\text{Cu}}_{ 2} {\text{O }}\left( {{\text{Below 3}}00\,^\circ {\text{C}}} \right) \, \left( {\text{III}} \right)$$
$${\text{Cu }} + {\text{ C}}_{ 3} {\text{H}}_{ 8} {\text{O }}\left( { 2 {\text{-propanol}}} \right) \, + {\text{ H}}_{ 2} {\text{O}} \to {\text{CuO }}\left( {{\text{Above 3}}00\,^\circ {\text{C}}} \right) \, \left( {\text{IV}} \right)$$

Instead of more commonly used metal alkoxides, which are expensive and involve complicated reaction routes, zinc acetate dihydrate (ZAD) was chosen as preferred precursor to grow ZnO films (overlayer in case of BNHS and Cu-BNHS samples), in this study. Easy hydrolysis of acetate group in ZAD facilitated formation of homogenous sol that remained stable for ~ 20 days. Only a mild heating was needed to induce decomposition of products into volatile compounds, which prevented uneven and premature drying and yielded crack-free films [40]. Use of (CH3)2CHOH as solvent eliminated the long duration refluxing at higher temperatures (needed with several other alcoholic solvents) and the reactions were possible at ambient temperatures (25–35 °C) [47]. DEA was used to maintain optimum pH, facilitate complexation and promote formation of nanostructured ZnO. In this case, three nucleophiles (CH3COO−, OH− and DEA) apparently competed for Zn2+ ions. Gradual forced hydrolysis of Zn-OCOCH3 or Zn-DEA, followed by progressive condensation, yielded colloids and precipitates. Sintering of the deposited layer and the associated evaporation of solvent led to the combination of colloidal polymers to yield organic molecules’ framework supported gel system. On further rise in temperature, metal ion-organic chain link is broken and gel-embedded metal oxide crystallites are evolved [48]. High temperature sintering helped to complete the process of crystallite growth and raise electrical conductivity by regulated induction of defects, especially ‘O’ vacancies in the lattice [49]. Tentative set of reactions (V–VIII), involved in the growth of ZnO film is presented below [34].

$${\text{Zn}}\left( {{\text{OCOCH}}_{ 3} } \right)_{ 2} \cdot 2 {\text{H}}_{ 2} {\text{O}} \to {\text{Zn}}\left( {{\text{OCOCH}}_{ 3} } \right)_{ 2} + {\text{ 2H}}_{ 2} {\text{O }}\left( {\text{V}} \right)$$
$${\text{Zn}}\left( {{\text{OCOCH}}_{ 3} } \right)_{ 2} + {\text{ 2ROH}} \to {\text{Zn}}\left( {\text{OR}} \right)_{ 2} + {\text{ 2CH}}_{ 3} {\text{COOH }}\left( {\text{VI}} \right)$$
$${\text{Zn}}\left( {\text{OR}} \right)_{ 2} + {\text{ 2H}}_{ 2} {\text{O}} \to {\text{Zn}}\left( {\text{OH}} \right)_{ 2} + {\text{ 2ROH }}\left( {\text{VII}} \right)$$
$${\text{Zn}}\left( {\text{OH}} \right)_{ 2} \to {\text{ZnO }} + {\text{ H}}_{ 2} {\text{O }}\left( {\text{VIII}} \right)$$

3.2 General physical characteristics

All the films prepared were porous, as evident from the observed bulk density being significantly less (pristine monolayered CuO films: 2.71, pristine monolayered ZnO films: 2.40, bilayered CuO/Cu–ZnO films: 2.07–2.27 g cm−3, compared to true density of pure ZnO (5.10–5.60 g cm−3) and pure CuO (6.30–6.49 g cm−3) (Table 1) [39, 50]. Further, in bilayered films, density decreased marginally on Cu incorporation in ZnO overlayer, which may be attributed to further opening-up of the ZnO lattice. In the I–V plots, appearance of photocurrent under anodic bias indicated n-type nature of samples, suggesting the presence of oxygen vacancies and/or metal interstitials, generated through non-stoichiometric growth of crystals. Observed in situ electrical resistivity were 20.25 × 103 kΩ cm (pristine monolayered CuO film), 7.86 × 103 kΩ cm (pristine monolayered ZnO film) and in the range 0.12–4.15 × 103 kΩ cm (CuO/Cu–ZnO bilayered thin films). Resistivity of bilayered films decreased significantly on Cu incorporation in ZnO overlayer (Table 1), albeit with random changes, as the amount of incorporated Cu was increased from 1 to 4%. With similar oxidation states and nearly same ionic radii of Cu and Zn ions, Cu can easily penetrate into ZnO crystal lattice. Fall in electrical resistivity in ZnO on Cu incorporation has been reported even earlier by different workers, who have attributed this observation to varied reasons, viz. (a) non-stoichiometric growth and creation of oxygen deficiency [51], (b) high carrier mobility [52], (c) smaller ionization potential of Cu in ZnO compared to pristine ZnO, causing increase in donor concentration [53] and (c) breakdown of dopant phase, which acts as recombination centers for carriers [54]. In the present study also, Cu incorporation in ZnO has led to significant rise in charge carrier density (Table 4), with the value apparently maximizing for sample CZC3, which also yielded the highest PEC performance. Thickness of film samples was 360 and 420 nm, respectively, for pristine monolayered films of CuO and ZnO and in the range 505–900 nm for bilayered films (CuO/ZnO and CuO/Cu–ZnO) (Table 1).

Table 1 Density (d), thickess (t) and electrical resistivity (r) of thin film samples prepared
Full size table

3.3 Crystal structure and microscopic characteristics

Polycrystalline nature of films is evident from observed XRD pattern (Fig. 1). Peaks at 30.7 and 51.3º are ascribable to ITO substrate. Monolayered pristine sample Z recorded peaks at 2θ angles 31.8, 34.5, 36.3, 47.7 and 56.7º, which correspond to the diffraction from planes (100), (002), (101), (102) and (110), respectively, of hexagonal wurtzite ZnO (JCPDS, Card 89-1397). Likewise, peaks at 2θ angles 35.5 and 38.7º with monolayered pristine sample C may be assigned to diffraction from (002) and (111) planes, respectively, of monoclinic tenorite CuO (JCPDS, Card 80-1917) [55]. XRD pattern of BNHS and Cu-BNHS films reveal the presence of both hexagonal wurtzite ZnO and tenorite CuO phases. No other phases of Cu (metallic or other oxides) were detected, which suggests that the copper added as impurity in ZnO overlayer is also largely converted into CuO, probably during the post-deposition annealing of films at 500 °C. Evidences from XPS and TEM analyses (discussed later) suggest the possible existence of CuO nanparticles dispersion in ZnO overlayer. Since, the amount of Cu incorporated in ZnO overlayer was within the solubility limit of Cu in ZnO, the diffusion of Cu in ZnO lattice, through creation of substitutional (CuZn) and/or interstitial (Cui) sites, cannot be ruled out [56]. Considering nearly similar ionic radii of Cu2+ (73 pm) and Zn2+ (74 pm) ions, this would not lead to any structural deformations or strain in the lattice. However, in this study, a low-angle shift is seen in the ZnO peak at 2θ angle 34.5º ZnO (Fig. 2), alongwith significant strain and dislocations in the lattice, particularly of BNHS and Cu-BNHS films (Table 2), which is indicative of mutual intrusion of hexagonal ZnO and monoclinic CuO lattices at the heterojunction. Compared to monolayered pristine ZnO films, the bilayered films recorded reduced crystallite size (s). Variations in crystallite size is reverse-correlated with changes in microstrain (γ), lattice strain (ζ) and dislocation density (η), suggesting that higher strain is detrimental to the growth of crystallites [57].

Fig. 1
figure 1

XRD pattern of prepared thin films. Peaks correspond to: CuO (C), ZnO (Z), ITO substrate (*)

Full size image
Fig. 2
figure 2

Shift in (002) plane XRD peak of ZnO in BNHS and Cu-BNHS films

Full size image
Table 2 Average crystallite size (s), microstrain (γ), lattice strain (ζ), dislocation density (η), bandgap energy (Eg) and RMS surface roughness (Sa) values for films prepared
Full size table

3.4 Morphological and elemental analysis

SEM images of Cu-BNHS films (Fig. 3) indicate continuous growth and nearly homogeneous and uniform occurrence of hexagonal/polygonal shape particles along with some random and variable grain-agglomeration. Similar images were recorded for samples C, Z and CZ. Estimated particle size distribution from SEM images (Fig. 4) indicate mean particle size ranging 20–33 nm in Cu-BNHS films. Mean particle size estimated from SEM images is agreeable with the average crystallite size derived through Scherrer’s computations, based on XRD data. It is also evident that particle size decreases in Cu-BNHS films as the amount of incorporated Cu is increased from 1 to 3%. However, particle size increased again in sample CZC4. Cross-sectional SEM image of representative sample, CZC3, categorically presents the envisaged bilayered configuration with thickness of CuO underlyer and Cu–ZnO overlayer being, respectively, 408 and 552 nm (Fig. 5).

Fig. 3
figure 3

SEM images of Cu-BNHS films

Full size image
Fig. 4
figure 4

Particle size distribution in Cu-BNHS film, as estimated from SEM data

Full size image
Fig. 5
figure 5

Cross-sectional SEM image of sample CZC3

Full size image

Surface topography of films, investigated through AFM analysis (Fig. 6) also indicate the homogeneous and porous microstructure. In conformity with observations of different workers in case of ZnO thin films grown by sol–gel spin-coating; here also in BNHS and Cu-BNHS films, which are formed with an overlayer of ZnO, c-axis oriented preferred vertical growth of nanocrystallites is recorded. Root-mean-square (RMS) surface roughness of films was affected by Cu incorporation, with sample CZC3 recording the highest roughness. Films with 4% Cu incorporation are relatively smoother (Table 2).

Fig. 6
figure 6

AFM images of Cu-BNHS films

Full size image

EDX analysis indicated the elemental composition and stoichiometry in bilayered samples (Cu-BNHS films) (Fig. 7). Similar observations were recorded with samples C, Z and CZ. Results reveal the existence of Zn, Cu and O elements in expected atomic compositions in accordance with the formation of CuO and ZnO phases (Table 3). Presence of Si, Sn and In in the recorded pattern are attributable to contributions from substrate. The percentage of Cu reported here in Cu-BNHS films is the cumulative of the value arising from CuO underlayer and Cu–ZnO overlayer. Concentration of O being higher than expected for the formation of CuO and ZnO phases may be attributed to part inputs from underlying substrate (ITO) layer.

Fig. 7
figure 7

EDX pattern of Cu-BNHS films

Full size image
Table 3 Elemental composition of bilayered thin films observed through EDX analysis
Full size table

3.5 XPS analysis

XPS analysis of the best performing sample CZC3, that yielded highest PEC photocurrent (2.98 mA cm−2 at 0.85 V bias versus SCE), revealed surface chemical composition and oxidation states of different elements (Fig. 8). C 1 s peak (284.5 eV) was used as reference in binding energy calibrations. XPS survey spectra indicate occurrence of O, Zn, Cu, In Sn and Si. In the core level Zn 2p spectrum, the peaks were observed at binding energy 1020 and 1042 eV, which correspond to Zn 2p3/2 and Zn 2p1/2 respectively, and confirm the + 2 oxidation state of Zn, compatible with the formation of ZnO overlayer. Similarly, peaks observed at 933.6 and 950.8 eV in the core level Cu 2p spectrum, respectively, are assigned to Cu 2p3/2 and Cu 2p1/2, indicating the +2 oxidation state of Cu, attuned with the formation of CuO underlayer [42]. Collection of satellite features at 940–944 eV is also indicative of Cu2+ state. As Cu is seen here only in +2 oxidation state, it may be inferred that Cu incorporated in ZnO overlayer has been converted to CuO during the post-deposition sintering of films and possibly exists as CuO nanoparticles, dispersed throughout the ZnO overlayer. Cu2+ substituting Zn2+ ions in the lattice or interstitial positions also remains a possibility, but appears to be very low in occurrence as indicated by XRD analysis. The peak observed at 528.7 eV correspond to O 1s and is attributable to bridging oxygen, confirming the formation of ZnO and CuO phases.

Fig. 8
figure 8

XPS spectra of sample CZC3

Full size image

3.6 Optical characteristics

Figures 9 and 10, depict, respectively, the optical absorption spectra and Tauc plots of films. Estimated values of band gap energy (Eg) are shown in Table 2. Sample C exhibits strong absorption in the visible range, unlike sample Z, which is primarily UV absorbing. Barring variations in magnitude, the absorption behaviour of bilayered films is almost akin to the pattern seen with sample Z. The increased absorption in bilayered films, especially at longer wavelengths, suggests some contribution arising from CuO underlayer as well. The absorption threshold at ~ 400 nm in pristine sample Z and bilayered films corresponds to the onset of fundamental absorption, O:2p → Zn:4s, of hexagonal wurtzite ZnO. Samples here also recorded some d–d transitions induced humps (wavy pattern), before the arrival of threshold [58]. With pristine sample C, absorption threshold is observed at λ ~ 750 nm. Besides the band gap, cyrstallinity and stoichiometry are other variables that might have influenced sample-to-sample minor variations in the values for Cu-BNHS samples.

Fig. 9
figure 9

Variation in absorption with wavelength in pristine and bilayered films

Full size image
Fig. 10
figure 10

(αhν)2 versus hν plots for pristine and bilayered films

Full size image

3.7 TEM analysis

TEM image (Fig. 11) of the best performing representative sample, CZC3, reveals supported morphology with continuous lattice fringes for particles of CuO and ZnO. By analysing fringe-width in different regions and interfaces, the estimated distance between adjacent crystal planes are 0.28 and 0.19 nm for (100) and (102) planes of ZnO and 0.25 nm for (111) plane of CuO, respectively. Non-occurrence of any other phase of Cu revalidates the claim that, Cu added as impurity in Cu-BNHS films is oxidised and occurs as CuO nanoparticles dispersed in ZnO overlayer. Synthesis protocol for samples, involving post-deposition annealing in air at 500 °C, support the above contention. Selected area electron diffraction (SAED) pattern is also confirming the presence of only ZnO and CuO phases and the polycrystalline nature of sample (Fig. 12).

Fig. 11
figure 11

TEM image of representative sample, CZC3

Full size image
Fig. 12
figure 12

Observed SAED pattern for sample CZC3

Full size image

3.8 PEC studies

Variations in photocurrent density (Iph), with applied voltage (versus SCE), were obtained from linear scan voltammogram (LSV) plots of films, recorded under darkness and illumination (Fig. 13). Significant gain in Iph (Fig. 14) was recorded with BNHS and Cu-BNHS films, against the pristine samples. Cu incorporation in ZnO overlayer had profound impact on PEC photoresponse, as sample CZC3 yielded Iph (2.98 mA cm−2 at 0.85 V versus SCE) (Table 4). Analogous to variations in Iph, bilayered films recorded significant rise in Isc (under illumination) also, against monolayered pristine films. Suggesting the onset of photocurrent much sooner (i.e. at lesser applied voltage); Voc was found more negative under illumination compared to its value under darkness with all the samples. Figure 14 also presents Iph vs. time (t) and voltage (E) vs. t plots for the best performing sample, CZC3, which are indicative of the films being stable as photoelectrode in PEC cell.

Fig. 13
figure 13

I–V plots for pristine and bilayered films, recorded under darkness and illumination

Full size image
Fig. 14
figure 14

Variations in Iph with applied potential (a); Plot of Iph vs. t (b) and E vs. t (c) for sample CZC3

Full size image
Table 4 Observed photocurrent density (Iph), Flat band potential (Vfb), Charge carrier density (Nd) and Applied bias photon to current conversion efficiency (ABPE)
Full size table

Mott–Schottky plots (Fig. 15) led to determination of Vfb and Nd values (Table 4). Bilayered films (BNHS and Cu-BNHS) recorded significant gain in Nd against monolayered films, which is attributable to creation of more native-defects. With n-type nature of films, negative shift observed in Vfb with BNHS and Cu-BNHS samples, in comparison to monolayered pristine sample Z, is expected to favour photocurrent generation and the same has been recorded. Variations in ABPE followed the similar trend, recording maximum efficiency of 1.4% with sample CZC3 (Table 4). From the radii of semi-circles formed in Nyquist plots of Cu-BNHS films, it can be inferred that resistance to charge transfer is lowest with sample CZC3 (Fig. 16), which is also the best performing sample in this study. Further, the obtained Nyquist plots have been best fitted to equivalent Randle circuit model. This has led to the estimated values of charge transfer resistance (Rct) and capacitance (C), which are, respectively, lowest and highest for sample CZC3, demonstrating a clear impact of above parameters on PEC photocurrent generation (Fig. 16).

Fig. 15
figure 15

Mott–Schottky plots for films prepared

Full size image
Fig. 16
figure 16

Nyquist plots and computed values of charge transfer resistance (Rct) and capacitance (C) for Cu-BNHS films

Full size image

In an attempt to explore the possible mechanistic pathway for the flow of photoinduced charge carriers in bilayered samples (BNHS and Cu-BNHS films), energy positions of conduction and valence band edges in ZnO and CuO were derived through computations based on Butler and Ginley theoretical approach, using Eqs. 10 and 11 [59, 60]. Here, X is geometrical mean of the absolute electronegativity of the constituent atoms in the semiconductor (obtained by computing the arithmetic mean of the first ionization energy and atomic electron affinity), Ec the energy of free electron in the hydrogen scale (~ 4.5 eV), ECB conduction band (CB) potential and EVB valence band (VB) potential [42, 61].

$$E_{\text{CB}} = X{-}E^{\text{c}} {-} \, 0. 5E_{\text{g}}$$
(10)
$$E_{\text{VB}} = E_{\text{CB}} + E_{\text{g}}$$
(11)

Thus, a tentative alignment of band edges in bilayered thin film samples before and after illumination has been drawn, along with redox energy levels for hydrogen and oxygen evolution through water splitting (Fig. 17). Considering n-type nature of both CuO and ZnO formed in the present work, the uniquely evolved n/n heterostructure enabled enhanced absorption of light and more efficient flow of photoinduced charge carriers and their participation in water splitting. Before the formation of junction CB of CuO lies above the CB of ZnO. Further, as the carrier concentration in CuO is also higher (Table 4), the conduction band and Fermi level energy separation in this case would be narrower. So, in bilayered films, as CuO is possessed with larger work function compared to ZnO, electrons would spontaneously flow from the Fermi level of CuO to the Fermi level of ZnO till the Fermi levels equalize. In this condition, the CB of CuO would get lower compared to CB of ZnO owing to smaller energy separation between the CB and Fermi level in CuO. So, under illumination, the photogenerated electrons in ZnO would easily be moved to the CB of CuO. However, a snag is on the transfer of photogenerated holes from VB of CuO underlayer to electrolyte, as the two are not in direct contact and the transfer through VB of ZnO is energetically unfavourable. However, overcoming this limitation, here exists a possible role of Cu incorporated in ZnO. The likely presence of incorporated copper as CuO nanoparticles, dispersed throughout in ZnO overlayer, as indicated from XRD, XPS and TEM analyses, seems to facilitate transfer of photogenerated holes, through hopping of electrons, from electrolyte to VB of CuO via CuO nanoparticles (Fig. 17). Additionally, electrode existing in equilibrium with electrolyte would generate surface band bending that might have also facilitated charge-separation, resulting in enhanced photoresponse in the present case.

Fig. 17
figure 17

Tentative mechanism for the movement of photogenerated charge carriers in bilayered films

Full size image

PEC current recorded with best performing sample CZC3 under chopped illumination (Fig. 18), highlights the effect of light in current generation. Besides the generation of high photocurrent at low overpotential, the long-term stability of photoelectrode is another crucial requirement for practical applicability of PEC water splitting system in renewable hydrogen generation. In this study, films were found quite stable on use as photoelectrode and the photocurrent was found nearly unaffected in repeat observations recorded even after 75 days. Moreover, continuous exposure of the photoelectrode to electrolyte for nearly 12.5 h also did not cause any major damage to the films. It was confirmed through insignificant changes in the XRD pattern and SEM images recorded with sample CZC3 before and after PEC measurements (Fig. 18).

Fig. 18
figure 18

Sample CZC3: Iph measured under chopped illumination (I); XRD pattern (II) and SEM images (III), recorded before (a) and after (b) PEC measurement

Full size image

4 Conclusions

The study leads to following conclusions: (a) Compared to pristine monolayered thin films (sample C and Z), BNHS and Cu-BNHS films, represented by bilayered nano-hetero-structure, are more efficient photoanode for PEC water splitting; (b) The conversion of Cu, added as impurity in ZnO overlayer, to CuO nanoparticles during post-deposition sintering of films could generate an important pathway for the transfer of photogenerated holes from VB of CuO underlayer to electrolyte, enabling their swift participation in oxygen evolution reaction; (c) Bilayered nano-hetero-structuring and Cu incorporation in ZnO overlayer in this case has also induced favourable changes in microstructure and optical characteristics of samples that further boosted generation and migration of light-induced charge carriers, resulting in the highest photocurrent (2.98 mA cm−2) with sample CZC3.