Photocapacitive CdS/WOx nanostructures for solar energy storage

Through a facile solvothermal procedure, a CdS/WOx nanocomposite has been synthesised which exhibits photocapacitive behaviour under white light illumination at a radiant flux density of 99.3 mW cm−2. Photoelectrochemical experiments were undertaken to examine the self-charging properties of the material and to develop an understanding of the underlying electronic band structure responsible for the phenomenon. By employing XPS, UPS and UV-Vis diffuse reflectance spectroscopy for further characterisation, the ability of the composite to generate current following the removal of incident light was related to the trapping of photoexcited electrons by the WOx component. The presence of WOx yielded an order of magnitude increase in the transient photocurrent response relative to CdS alone, an effect attributed to the suppression of electron-hole recombination in CdS due to hole transfer across the CdS/WOx interface. Moreover, current discharge from the material persisted for more than twenty minutes after final illumination, an order of magnitude improvement over many existing binary composites. As a seminal investigation into the photocapacitive characteristics of CdS/WOx composites, the work offers insight into how the constituent materials might be utilised as part of a future self-charging solar device.

. EDX measurements from the CdS (a) and CdS/WO x (b) samples (deposited on adhesive carbon tabs), recorded at an accelerating voltage of 20 kV and emission current 10 µA; both spectra suggest a Cd/S atomic ratio of approximately one, while the measurements from CdS/WO x additionally verify the presence of a WO x phase and organic contaminants. The source of the Al content in each spectrum is unclear, and the element was not identified in XPS measurements from either sample.  Table S1. Atomic percentage measurements obtained from EDX spectra of the CdS and CdS/WO x samples, with the measured elemental composition of the underlying adhesive carbon tab provided for reference.

Fitting of UPS spectra
The UPS measurements shown in Fig. 5c and d of the paper are instrumental in determining the band structure of CdS/WO x . More specifically, estimation of the secondary electron onset, E k,SEO , and valence band maximum, E k,VB , of both CdS and the composite allows calculation of the ionisation energy in each case, defined as the energy of the valence band edge relative to the vacuum level, E vac .
To estimate E k,SEO , a Matlab program is used to plot a linear fit through the points of steepest gradient on the onset curve, while a second line is similarly extrapolated through the data defining the baseline at lower values of kinetic energy; these fits are depicted as dashed lines in Fig. S2a and b for CdS and CdS/WO x , respectively. In both cases, E k,SEO is taken as the kinetic energy at the point of intersection between the two linear fits.
The energy of the valence band maximum is estimated in a similar fashion. Illustrated as dashed lines in Fig. S2c and d, which correspond to CdS and CdS/WO x , respectively, are linear fits through the points of steepest gradient close to the valence band edge and through the baseline at higher values of kinetic energy; as in the case of E k,SEO , the kinetic energy at the point of intersection between the two lines is assumed equal to E k,VB .

Estimation of incident light intensity
To calculate the radiant flux density, Φ e , incident on a system from knowledge of the luminous flux density, Φ ν , one must address the relationships between these two quantities and the measured light spectrum, Φ e,λ , where the suffix denotes a dependence on the photonic wavelength, λ. The form of the light spectrum may be measured at an arbitrary position relative to the light source, provided that there is no spectral variation between the selected position and the location of the experimental system. In the present case, saturation of the spectrometer precluded measurement of Φ e,λ at the system location, so the measurement was instead carried out at a greater distance from the LED source; the resulting spectrum is displayed as a function of wavelength in Fig. S3a.
In addition to determining the form of Φ e,λ , the value of Φ ν was also measured at the system location; these variables are related via the equation where C is a constant and η λ is the standard photopic luminous efficacy function, plotted in Fig. S3b, which characterises the sensitivity of the human eye and provides the basis for the definition of the lumen unit. Having already measured both Φ ν and Φ e,λ , the constant C may be estimated directly from (S1). Finally, Φ e is given by and may therefore be calculated from the estimate of C and the measured form of Φ e,λ . For the photoelectrochemical experiments in the present study, the measured Φ ν value of 325,250 ± 2000 lux yielded a Φ e estimate of 99.3 ± 0.6 mW cm -2 , equivalent to an intensity of 1.0 Suns. S3. Spectrum of the LED source measured using an Ocean Optics USB 2000+ spectrometer at an arbitrary position relative to the sample location (a), and the CIE standard photopic luminous efficacy function, η λ (b). Fig. S4. Transient photocurrent response measurements from Ta 3 N 5 (a), Ta 3 N 5 /WO x (b) and WO x (c) on FTO-coated glass in a three-electrode configuration with a platinum mesh counter-electrode and Ag/AgCl (3.0 M) reference; aqueous Na 2 SO 4 (0.5 M) was used as the electrolyte and a potential of 0 V was applied to the sample with respect to the reference electrode. Each sample was backside-illuminated by an LED source of power density 99.3 mW cm -2 . The Ta 3 N 5 sample exhibited a diminishing cathodic current upon turn-on of the LED source, while a near-negligible cathodic photocurrent was measured in the case of Ta 3 N 5 /WO x . When used alone, WO x demonstrated negligible photoactivity in the present experimental setup.