1 Introduction

Solar energy is being considered as one of the most promising alternative renewable energies, however, due to the intermittence of sunlight it must be stored in an appropriate manner. Artificial photosynthesis enables the possibility on photo-assisted water splitting with semiconductor materials for generating hydrogen as an energy carrier [1, 2]. Efficient photocatalytic water splitting requires highly stable and inexpensive photoelectrodes. Control of the semiconducting properties of photoelectrode materials is the primary concern in developing materials for solar water splitting, since they determine how much photoexcitation occurs in a semiconductor under solar illumination and the quantity of photoexcited carriers reaching the surface where water splitting takes place [3].

To absorb a major portion of the solar spectrum for solar water splitting, a semiconductor shall have a conduction band (CB) that is higher than the H2/H2O redox level and a valence band (VB) that is lower than the H2O/O2 redox level [4]. However, no single material has fulfilled all the required conditions. First result on photocatalytic water splitting using a TiO2 electrode was reported by Fujishima and Honda [5]. Along with TiO2 several metal oxides have been extensively studied, such as ZnO, WO3 and α-Fe2O3 since their bands are energetically favorable for such application. Nevertheless, these materials have large band gap, which implies low light absorption [6,7,8,9,10,11]. Recently, researchers put serious effort on addressing these issues by modifying semiconductors with noble metal loading, addition of sacrificial reagent and sensitization [6, 12, 13]. Momeni et al. [14,15,16,17,18,19,20,21,22,23] have made a valiant effort on introducing materials such as Cr, Fe, Cu, WO3, ZnO, Ag2S and CdS to TiO2 photoanode to enhance the photochemical activity of the later. The sensitized wide bandgap semiconductor with dyes or quantum dots (QD) such as CdS, CdSe, CdTe, PbS and PbSe, has demonstrated improvement of solar light absorption [3, 24,25,26,27,28,29,30,31,32,33]. Instead of bulk electrodes, the use of nanocrystalline semiconductors in photoelectrochemical cells (PEC) offer certain advantages such as larger surface area which is beneficial for the catalytic processes, flexibility in design and fabrication of materials, and short distance for charge transport due to quantum confinement effect [34,35,36]. Since the electron mobility of nanostructures such as nanotubes, nanowires and nanorods are better than nanospheres, application of such nanostructures can improve the photoelectrochemical response of a photoanode [37,38,39,40]. Photocorrosion occurs in semiconductors such as CdS and SiC, and hence, it is necessary to add sacrificial reagents (or hole scavengers) that react irreversibly with the photogenerated holes in the VB and can enhance the photocatalytic electron/hole separation, resulting in a higher quantum efficiency and lower charge recombination [6, 37, 41]. Furthermore, CdS–CdSe co-sensitized photoelectrodes have been reported in PECs as tandem or core/shell structures, mainly supported on TiO2 or ZnO nanotubes/nanowires using successive ionic layer adsorption and reaction (SILAR) deposition or chemical bath deposition (CBD) [38, 42,43,44,45,46]. This combination has the ability to produce hydrogen without the assistance of an external bias, placing it as a promising alternative photoelectrodes [38, 47]. However, SILAR or CBD methods have the disadvantage of taking longer deposition time and less control of the size and shape of the QDs compared to colloidal pre-synthesized QDs [48,49,50].

In this work, the photoactivity of sensitized TiO2 semiconductor with cadmium chalcogenides QD: TiO2/CdS/ZnS, TiO2/CdSe/ZnS and TiO2/CdTe/ZnS, as well as electrodes co-sensitized in the configuration TiO2/CdS-CdSe/ZnS and TiO2/CdS–CdTe/ZnS in hydrogen evolution was evaluated. CdS QDs were deposited by SILAR while colloidal CdSe and CdTe pre-synthesized QDs were deposited by electrophoresis. Electrochemical measurements were performed in a two-electrode configuration PEC. Moreover, the role of the CdS in the solar-to-hydrogen (STH) conversion efficiency of co-sensitized photoelectrodes was investigated.

2 Experimental section

2.1 Electrodes preparation

2.1.1 Materials

WER2-0 Reflector and DSL 18NR-T TiO2 paste were obtained from DYESOL. Titanium (IV) isopropoxide (97%), acetylacetone (> 99%), trioctylphosphine (TOP 90%), trioctylphosphine oxide (TOPO 99%), cadmium oxide (CdO 99%), selenium powder (Se 99%), tellurium powder (Te 99%), cadmium acetate dihydrate (Cd(CH3COO)2·2H2O), zinc acetate dihydrate (Zn(CH3COO)2·2H2O), lead acetate dehydrate (Pb(CH3COO)2·2H2O), sodium hydroxide (NaOH), sodium sulfite (Na2SO3·98%) and toluene anhydrous (99%) were obtained from Sigma-Aldrich. 1-tetradecylphosphonic acid (TDPA 99%) was purchased from PCI Synthesis. Sulphur (S), sodium sulfide (Na2S·9H2O), absolute ethanol, and methanol were obtained from KARAL. Fluorine-doped tin oxide (FTO, TEC-15) was obtained from MTI.

2.1.2 TiO2 film preparation

Fluorine-doped tin oxide (FTO) glasses were cleaned with water, acetone and ethanol in an ultrasonic bath for 15 min before use. The electrodes consist of three different TiO2 layers stacked on top of the other. 1) A compact layer of about 150 nm deposited by spray pyrolysis, 2) An active layer of 8 µm composed of particles of 20 nm, and 3) a scattering layer of 7 µm composed of particles of 200 nm deposited by blading. Detailed description of the fabrication of electrodes is in previous work [51, 52].

2.1.3 Synthesis of colloidal CdSe and CdTe

CdSe QDs were synthesized by hot injection method based on the Peng protocol [53], with minor changes [54]. The QDs were dissolved in toluene prior to their use as a sensitizer. The synthesis of CdTe QDs was similar to the CdSe QDs, via a solution of TeTOP instead of SeTOP as tellurium precursor.

2.1.4 QDs deposition

CdS and ZnS were deposited into TiO2 electrodes by SILAR method. This was performed with 0.05 M Cd(CH3COO)2·2H2O dissolved in ethanol and 0.05 M Na2S in methanol:water solution (V:V = 1:1). Seven SILAR cycles were done for CdS QDs [52, 55]. ZnS was obtained using 0.1 M Zn(CH3COO)2·2H2O and 0.1 M Na2S both dissolved in water. The films were dipped in the solutions during two SILAR cycles [56]. ZnS was deposited taking into account that it reduces the recombination process of electrons in the TiO2-electrolyte interface by surface passivation [57, 58].

Since CdSe and CdTe were synthesized as colloidal QDs, these were deposited by electrophoresis. In this method, TiO2 electrodes were inserted face to face with a FTO in a colloidal QDs cuvette and kept at a distance of 0.2 cm. A constant voltage of 200 V was applied for 105 min to allow the deposition of QDs into the electrode [54]. The geometrical active area of the electrodes was 0.196 cm2. For the co-sensitized photoelectrodes, CdS QDs were deposited on TiO2 films by SILAR, followed by the CdSe or CdTe deposition by electrophoresis.

2.1.5 Photoelectrochemical cell assembling

The PEC consists of a two-electrode configuration using the TiO2/QDs/ZnS photoelectrode as working electrode and Pt wire as counter electrode. The electrolyte with sacrificial agent was 0.25 M Na2SO3 and 0.35 M Na2S. The hydrogen obtained under light irradiation was collected in a syringe locked on top.

2.2 Characterization

The scanning electron microscope (SEM), JEOL (JSM-7800F) was used to measure surface morphology of the samples. The UV–Vis absorption measurements were performed using an Agilent Technologies Spectrophotometer (Cary 5000) with an integrating sphere. The linear sweep voltammetry and chronoamperometry measurements were recorded with a Gamry reference 600 potentiostat/galvanostat under 100 mW cm−2 illumination intensity using an Oriel Sol 3A solar simulator. The light intensity was calibrated employing a NREL Si solar cell with a KG-2 filter. The energy levels were measured by Electrochemical Impedance Spectroscopy (EIS), as was reported in our previous work [54], with a three electrodes cell using FTO/QDs as working electrode, Ag/AgCl as reference electrode and Pt wire as counter electrode, while a solution of 1 M Na2S was the electrolyte.

3 Results and discussion

Figure 1 shows the SEM images of the TiO2, TiO2/CdS, TiO2/CdSe and TiO2/CdTe films. The TiO2 film is shown as reference in Fig. 1A. We can observe that the TiO2 nanoparticles are completely covered with the CdS QDs deposited by SILAR (Fig. 1B). In the films deposited with CdSe and CdTe by electrophoresis it is possible to observe the TiO2 nanoparticles, suggesting a less saturated but uniform coverage, since the QDs are considerably smaller than the TiO2 nanoparticles top layer, thus allowing the QDs to penetrate deeper into the TiO2 layer (Fig. 1C, D).

Fig. 1
figure 1

SEM images of A TiO2, B TiO2/CdS, C TiO2/CdSe and D TiO2/CdTe

Full size image

The cross-sectional SEM image of photoelectrode and the EDS analysis of the QD-sensitized TiO2 are presented in Fig. 2. It is possible to observe the two TiO2 layers mentioned in the experimental section, active in the bottom and scattering in the top of the photoelectrode (Fig. 2A). The EDS distribution of the different QDs is shown in Fig. 2B. As shown in the Fig. 2B, the concentration profile of the CdS QDs is higher in the active layer than in the scattering layer. However, the difference in the distribution of the CdSe concentration between the two layers is not so high. The CdTe QDs showed a high concentration at the bottom of the photoelectrode (active layer), but decreases considerably over a short distance and remains uniform on the rest of the substrate. This difference in the concentration profile between CdS and CdSe (or CdTe) can be attributed to the sensitization technique used, since it is well known that SILAR produce high surface coverage and denser distribution of the QDs deposited, as demonstrated in Fig. 1B, while the ex situ deposition methods for colloidal QDs usually suffers from lower coverage of the QDs on the TiO2 surface [55, 59, 60].

Fig. 2
figure 2

Quantum-dot sensitized TiO2 photoelectrode cross-section A SEM image and B EDS distribution of QDs in the cross-section

Full size image

The UV–Vis absorption spectra of the TiO2 films with QDs semiconductors deposited are shown in Fig. 3. The TiO2/CdS film has a wide absorption, as far as 550 nm, which is in good agreement with previous reports for this kind of material [27, 39, 45]. TiO2/CdSe film, has a wider absorption spectrum (from 300 to 650 nm), with three absorption bands at 350 nm, 420 nm and 575 nm corresponding to the TiO2 absorption (350 nm) [61, 62] and to the CdSe QDs (420 and 575 nm) [29, 46, 63]. The TiO2/CdTe film has very weak visible absorption (Fig. 3A), that may be due to the degradation of CdTe QDs in the films. When CdS and CdSe are used together, the obtained absorption spectra is similar to that of TiO2/CdSe films but with a more pronounced peak at 450 nm, indicating the presence of CdS. In the case of TiO2/CdS/CdTe the absorption spectra is similar to that of CdS sensitized films, but with a long tail that reaches 800 nm (Fig. 3B).

Fig. 3
figure 3

UV–Vis absorption spectra of the photoelectrodes. A TiO2 sensitized with cadmium chalcogenide QDs semiconductors (a) CdS, (b) CdSe and (c) CdTe and B TiO2 co-sensitized with (d) CdS-CdSe and (e) CdS–CdTe QDs semiconductors

Full size image

The PEC is illustrated in Fig. 4. It contains the photoanode which absorbs the light and generates the electrons that are used to split the water from the electrolyte solution. The hydrogen is generated on the Pt surface and collected in the syringe container. The photoanode performance was characterized in a two-electrode configuration because the application of a potential bias versus the reference electrode excludes the second half-reaction at the counter electrode [3, 10].

Fig. 4
figure 4

PEC used to generate hydrogen with TiO2 electrode sensitized with QDs semiconductors under irradiation of solar sunlight

Full size image

The J-V curves for the photoelectrodes are presented in Fig. 5A. The linear sweep voltammetry of different electrodes in the range of water decomposition shows that, at 0 V the photocurrent rises from 0.3 mA cm−2 obtained with TiO2/CdTe/ZnS to more than 2.3 mA cm−2 and 3.7 mA cm−2 obtained with TiO2/CdS/ZnS and TiO2/CdSe/ZnS, respectively. The photocurrent transient response measurements for the photoanodes with different QDs semiconductors at 0 V bias applied under intermittent illumination are shown in Fig. 5B. It can be observed that for each switch-on and switch-off operation, the resulting current of TiO2/CdTe/ZnS was small, 0.1 mA cm−2. The photocurrent corresponding to TiO2/CdS/ZnS was reached 1.8 mA cm−2 and this remained uniform. The highest photocurrent achieved was 2.7 mA cm−2 with TiO2/CdSe/ZnS, slightly decreasing after some cycles. The difference between the current obtained with CdS and CdSe could be related to the wider light absorption range of the CdSe (Fig. 3).

Fig. 5
figure 5

A–CJ–V curve of the TiO2 sensitized with one and two cadmium chalcogenide QDs semiconductors and B–D photocurrent response of the photoelectrode in the PEC at 0 V applied voltage

Full size image

The CdTe was used in the photoanodes to absorb light in the longer wavelength region of the solar spectrum. However, as observed in the photocurrent-transient response (Fig. 5B), the photoelectrodes with these QDs resulted in low current and instability with the sulfide electrolyte similar to that observed in previous studies [61, 64].

In Fig. 5C, the J–V curves of the photoanodes with CdS QDs combined with other QDs semiconductor showed that the current at 0 V increased in comparison to the electrodes with one cadmium chalcogenide QDs semiconductor. The observed current for TiO2/CdS/CdTe/ZnS (3 mA cm−2) was higher than that obtained for TiO2/CdS/ZnS and TiO2/CdTe/ZnS configurations. Similarly, the current (4.2 mA cm−2) obtained for TiO2/CdS/CdSe/ZnS configuration is 23% higher than that of either TiO2/CdS/ZnS or TiO2/CdSe/ZnS configurations. The photocurrent transient measurements confirm that the photocurrent generated has been improved (Fig. 5D) to 2.7 mA cm−2 with the configuration TiO2/CdS/CdTe/ZnS and 3.9 mA cm−2 with TiO2/CdS/CdSe/ZnS.

The value of the photocurrent obtained in the PEC cell is equivalent to the amount of H2 generated [9, 65]. Figure 6 shows the hydrogen generated with the different photoanodes under illumination at 0 V external bias voltage. Note that the electrode with TiO2/CdTe/ZnS has a very low rate of hydrogen generation (RH2), 0.06 nmol s−1 (see Table 1). It increased to 1.38 nmol s−1 with TiO2/CdS/ZnS. The RH2 of TiO2/CdSe/ZnS is as high as 2.11 nmol s−1. For the photoelectrodes with two kinds of cadmium chalcogenide QDs semiconductors, the RH2 increased to a maximum of 2.33 nmol s−1 for TiO2/CdS/CdSe/ZnS.

Fig. 6
figure 6

The evolution of the hydrogen production measured in the PEC with different QDs semiconductors and their combinations

Full size image
Table 1 Hydrogen conversion efficiency of the TiO2 photoanodes sensitized with different QDs semiconductors
Full size table

Moreover, the STH conversion efficiency for a PEC system at short circuit conditions could be estimated as the fraction of incident solar energy photoconverted into chemical energy as follows [1, 66, 67]:

$${\text{STH}}=~\frac{{{{\text{J}}_{{\text{sc}}}}{{\text{E}}^0}\eta_{{\text{F}}}}}{{{{\text{P}}_{\text{i}}}}}$$

where Jsc is the short-circuit current density, E° is the thermodynamic reaction potential taking into account the oxidation reaction caused by the sacrificial agent in the electrolyte [1]:

$$2{{\text{H}}_{\text{2}}}{\text{O}}+~{{\text{S}}^{2 - }}+{{\text{SO}}_{3}}^{{2 - }} \to ~{{\text{H}}_2}+2{\text{O}}{{\text{H}}^ - }+~{{\text{S}}_2}{{\text{O}}_{3}}^{{2 - }}\,\Delta {\text{G}}^\circ =44.3\,{\text{kJ}}\,{\text{mo}}{{\text{l}}^{ - 1}}$$

ηF is the faradaic efficiency of hydrogen generation in standard conditions (ηF = 1) and Pi is the incident solar irradiance. The STH of TiO2/CdTe/ZnS was small (∼ 0.07%), while with TiO2/CdS/ZnS and TiO2/CdSe/ZnS films, the STH is more than eight times higher (0.61 and 0.93% repectively). Nevertheless, when the CdSe and CdS QDs were deposited in the photoelectrode, the STH increased to 1.2% (a 23% relative increase compared to CdSe QDs alone). The combination of CdS and CdTe QDs also produce an increase of STH, reaching 0.79% (a relative increase of 29%).

Figure 7 presents the energy diagram of the QDs semiconductors used in this study. These were obtained by EIS measurements (Fig. 7A) and fitted with an equivalent circuit to obtain the Mott–Schottky plot (Fig. 7B) [54, 68,69,70]. It is worth mentioning that the EIS measurements were carried out in a range of 0–0.6 V bias for all samples, however we only included representative data of each QDs in this manuscript. The CB of all the QDs are above the potential necessary to generate hydrogen. For the hydrogen generation system, the effective voltage is the thermodynamic potential of the water splitting reaction, while the optimization criteria is to lower the band gap as much as possible to obtain as high photocurrent as possible without losing the necessary overpotential required to drive the reaction. The band gap corresponding to the QDs here analyzed were 2.45 eV for CdS, 1.83 eV for CdTe, 2.13 eV for CdSe and 3.2 eV for TiO2. The TiO2 photoelectrode has a wide band gap. Additionally, the overpotential given by the incorporation of QD and the band levels arrangement provide the necessary driving force to separate the photogenerated charge carriers, and the increase of light absorption helps to improve the generation of hydrogen.

Fig. 7
figure 7

A EIS spectra, B Mott-Schottky plot and C Energy diagram of QDs semiconductors (filled square CdS, filled circle CdSe and filled triangle CdTe) used in the photoanode for hydrogen generation

Full size image

The co-sensitization of the electrodes with CdS and CdSe or CdTe allows the synergistic effect for the improvement of light absorption, and thus the photocurrent enhancement [38]. Furthermore, the CdS QDs offer protection to the CdSe and CdTe QDs from corrosion or instability due to their compatibility with the sulfide electrolyte and provides suitable energy levels that allow the improvement of the injection of photon-generated electrons from the QDs to the TiO2, producing an increase in the photocurrent [24].

4 Conclusions

TiO2 thin films were sensitized with different QDs like CdS, CdSe, CdTe and its combinations in order to enhance its photo-absorption, and such films were analyzed in PEC. A comparative study among single-sensitized photoanodes reveals that CdSe QDs sensitization could increase the range of optical absorption and hence, results a higher photocurrent and a higher rate of hydrogen evolution (STH = 0.9). As compared to the single sensitized photoelectrodes, co-sensitized photoelectrodes showed higher photocurrent due to the combined effect of larger optical absorption and cascade effect of energy levels that enhance the transport of photogenerated electrons. Furthermore, the rate of generation of hydrogen was improved with co-sensitized systems TiO2/CdS/CdSe/ZnS (STH = 1.2). These results suggest that such co-sensitized TiO2 photoanodes are promising candidates for hydrogen generation systems.