Photoresponse of Visible Light Active CM-n-TiO₂, HM-n-TiO₂, CM-n-Fe₂O₃, and CM-p-WO₃ towards Water Splitting Reaction

Abstract

Photoresponses of visible light active carbon modified titanium oxide (CM-n-TiO₂), hydrogen modified titanium oxide (HM-n-TiO₂), carbon modified iron oxide (CM-n-Fe₂O₃), carbon modified tungsten oxide (CM-p-WO₃) towards water splitting reaction are reported in this article. Carbon and hydrogen in titanium oxide were found to be responsible for red shift from UV region to visible region which in turn enhanced the photoconversion efficiency by an order of magnitude for water splitting reaction. Photocurrent densities and photoconversion efficiencies of regular n-TiO₂ and CM-n-TiO₂ towards water splitting reaction under monochromatic light illumination from a xenon lamp and sunlight were compared and found in reasonable agreement. These oxides were characterized by photocurrent measurements, UV-Vis spectra, scanning electron microscopy (SEM), energy dispersive x-ray spectroscopy (EDS) and x-ray diffraction (XRD) studies and these results are also reported in this article.

1. Introduction

Sunlight is the unlimited source of clean and renewable energy if it could be efficiently utilized to split water to hydrogen and oxygen. Numerous studies focused on photoelectrochemical water splitting reaction [1–36]. The enhancement of efficiency of a solar hydrogen production process is of considerable importance because high efficiency translates into lower costs. It is, therefore, important to search for a stable and low-cost material for efficiently harvesting the solar energy. To use semiconductors as possible photoelectrodes, their viability depends on their ability to absorb enough sunlight as well as their stability against photocorrosion. The electronic structure of a semiconductor plays a key role in its photoactivity. Unlike a conductor, a semiconductor consists of a valence band (VB) and the conduction band CB). Energy difference between these two levels is said to be the bandgap (𝐸_𝑔) [37].

For efficient H₂ production using a visible-light-driven semiconductor (Figure 1), the bandgap should be less than 3.0 eV, but larger than 1.23 eV [38–41]. Moreover, the conduction band (CB) and valence band (VB) levels should satisfy the energy requirements set by the reduction and oxidation potentials for H₂O, respectively. For hydrogen production, the CB level should be more negative than hydrogen production level E_H2O/H2 while the VB should be more positive than water oxidation level E_H2O/O2 for efficient oxygen production from water by photocatalysis (Figure 1). Figure 2 shows the bandgap energies; CB and VB positions at pH 2 for different oxide materials that can be used as photoelectrodes. It shows that the CB levels of SrTiO₃ and KTaO₃ are above H⁺/H₂ but the VB levels of these oxides are below H₂O/O₂ redox energy level. This means that only the SrTiO₃ and KTaO₃ can photosplit water without external bias. The TiO₂ needs at least 0.3 volt of external bias to photosplit water. However, most of these oxides have very high bandgap energy (>3.1 eV) and hence cannot absorb the visible light of solar spectrum.

Figure 1 

Basic principle of overall water splitting on a cocatalyst-loaded semiconductor.

Figure 2 

Diagram showing bandgap energy of different oxide materials and relative energies in terms of vacuum level and normal hydrogen electrode level in electrolyte of pH = 2 [69].

1.1. Mechanism of Hydrogen Production from Water

Photoelectrolysis of water in a photoelectrochemical cell (PEC) involves several processes within photoelectrodes and at the photoelectrode/electrolyte interface, including the following.

Autoionization of Water Molecules
In an electrolyte solution, water molecules dissociates into H⁺ and OH⁻:4H₂O⟶4H⁺+4OH⁻. (1)

Light-Induced Excitation
When a semiconductor electrode is illuminated by a light source or directly by solar radiation that can provide the electrode with photonic energy (ℎ𝜈) greater than its bandgap energy, (E_g), then an electron from the valence band will be excited into the conduction band, and correspondently, a positively charged hole will be left in the valence band:photoanode+sunlight⟶4h⁺+4e⁻. (2)

Reduction of Hydrogen Ions
The electrons generated as a result of (2) are transferred over the external circuit to cathode, resulting in the reduction of hydrogen (H⁺) ions to gaseous hydrogen:4H⁺+4e⁻⟶2H₂(gas) (3) at cathode.

Oxidation of Hydroxyl Ions
The light-induced holes from (2), at the photoanode/electrolyte interface, oxidize the hydroxyl (OH⁻) ions to gaseous oxygen as 4OH⁻+4h⁺⟶O₂(gas)+2H₂O (4) at photoanode.
The overall reaction will bephotoanode+sunlight+H₂1O⟶2O₂(gas)+H₂(gas). (5)

1.2. Development of Photoelectrodes for Water Splitting

Among the photoelectrode materials, titanium dioxide (n-TiO₂) was found to be the most promising because of its low cost, chemical inertness, nontoxicity, and photostability. However, its wide bandgap (3.0–3.2 eV) limits its photoresponse in the ultraviolet region which is only a small fraction (~5%) of the sun energy compared to visible light from 400 nm to 750 nm (~49%). Hence, any shift in the optical response of n-TiO₂ from the UV to the visible spectral range will have a profound positive effect on the photocatalytic efficiency of the material. Several attempts were made to lower the bandgap of n-type titanium oxide (n-TiO₂) by transition metal dopants [42, 43] but no noticeable change in bandgap energy of n-TiO₂ was observed. The visible light absorption by the transition metal-doped n-TiO₂ was found to be mainly due to d-d transition of electrons in the transition metal dopants but not due to bandgap lowering. The transition metal dopants acted adversely on photocatalytic activity of n-TiO₂ because they acted as recombination centers for the photogenerated electron-hole pairs.

However, recent studies involved doping of n-TiO₂ by carbon [1, 15–24, 44–49], nitrogen [50, 51], sulphur [52], also fabricating the carbon-modified (CM)-n-TiO₂ in mesoporous [53, 54], and nanostructure [55, 56] forms. Photocatalytic activity of sulphur-doped n-TiO₂ was found to diminish under UV and visible light illuminations due to catalytic poisoning induced by oxidation of sulfur to SO₂ and SO²⁻₄. Photocatalytic activity of nitrogen-doped n-TiO₂ was reported to be much lower than carbon-doped n-TiO₂ [44].

It was found that carbon modification of n-TiO₂ photocatalyst synthesised by thermal oxidation of Ti metal sheet in a natural gas flame lowered the bandgap energy of n-TiO₂ to 2.32 eV and exhibited water splitting to hydrogen and oxygen with a photoconversion efficiency of 8.35% [1] under artificial light illumination from a Xenon lamp. This progress stimulated further investigation into carbon-modified n-TiO₂ (CM-n-TiO₂) as visible light active photocatalysts [43–49] and also as photoelectrodes [14–24, 55–59] for water-splitting reaction with enhanced photoconversion efficiency. Xu and Khan [16] also showed enhanced rate of water splitting at spray pyrolytically synthesised visible light active carbon-doped n-TiO₂ thin films. Enhanced photoresponse for water splitting at nanocrystalline carbon-doped n-TiO₂ thin films was also reported by Xu et al. [17]. Argon was found to be the most effective calcining ambient to enhance the carbon doping of n-TiO₂ films. The resultant carbon-modified n-TiO₂ thin film calcined in Ar at the optimum temperature of 700°C contained 3.8 at % carbon and showed higher photoresponse in both UV and visible regions compared to those calcined in air at the optimum temperature of 500°C that contained 2.3 at % carbon. Carbon content in these samples calcined in presence of oxygen in air reduced due to removal of some carbon in the forms of CO and CO₂. This wet process synthesis allowed molecular scale mixing of the carbon source with precursor of TiO₂ for improved doping homogeneity and was capable of tuning the carbon content in the film.

Importantly, Mohaputra et al. [18] synthesised CM-n-TiO₂ thin films by flame oxidation of Ti metal sheet and reported photoconversion efficiency of 8.5% for water splitting which is slightly higher than earlier results [1]. Carbon doping was considered to be responsible for the lowering of the bandgap of n-TiO₂ and consequent high photocatalytic activity under visible light illumination. Noworyta and Augustynski [19] observed shift in the spectral response (up to 425 nm) during water-splitting reaction for carbon-doped n-TiO₂ film electrodes formed in the flame of a burner fed with various gas mixtures. Shankar et al. [57] investigated the effect of flame annealing on the spectral photoresponse of titania nanotubes and demonstrated the enhancement of the visible light absorption and the photocurrent for water splitting due to annealing in the flame. However, in their study, the flame annealing was carried out at 1020°C which is much higher than the optimum temperature of 850°C observed earlier [1]. Such high temperature generally removes considerable amounts of carbon from the sample and thus reduces its photoresponse. Given the sensitivity of CM-n-TiO₂ to flame temperature, Shankar et al. [57] and Xu et al. [59] prepared highly ordered n-TiO₂ nanotube arrays by electrochemical anodization of Ti metal sheet followed by calcinations in an electric oven and further oxidation in natural gas flame at optimum temperature to incorporate carbon in it. Significant enhancement of the photoresponse of these carbon-modified (CM) n-TiO₂ nanotube films for water splitting was also reported [58, 59]. Hahn et al. [60] prepared carbon-doped self-organized TiO₂ nanotubes layers formed by electrochemical anodisation of Ti in an HF/Na₂HPO₄ electrolyte. The tubes were treated at 500°C under a mixed flux of N₂ and acetylene (C₂H₂). Nakano et al. [61] prepared TiO₂:C films by oxidative annealing of sputtered TiC films. They reported three bands having energies 0.86 eV, 1.30 eV, and 2.34 eV below the conduction band of carbon-doped n-TiO₂. They attributed the 0.86 eV level to the intrinsic nature of TiO₂, whereas the 1.30 eV and the 2.34 eV levels were newly introduced by C-doping and behaved as deep-level bands. In particular, the pronounced 2.34 eV band contributed to bandgap narrowing by mixing C 2p with the O 2p valence bands. Therefore, the 2.34 eV level played a significant role for the visible-light sensitivity in TiO₂:C. Ren et al. [62] synthesised a visible light active TiO₂ photocatalyst by carbon doping using glucose as carbon source. The preparation was performed by a hydrothermal method at temperature as low as 160°C. The carbon-doped n-TiO₂ showed absorption in the 400–450 nm range with a red shift in the bandgap transition. Sun et al. [63] reported that carbon-doping-made In₂O₃ red shifted from UV to visible range up to 500 nm and 40% contribution from visible light was observed for the photoelectrochemical splitting of water to hydrogen. Cardenas et al. [64] reported marked decrease in resistivity of carbon-doped antimony sulfide (Sb₂S₃) thin films to 102 Ω cm compared to 108 Ω cm for undoped samples, and electrical resistivity was also possible to tune by controlling the carbon content (wt %) in the sample. The bandgap energy the carbon-doped Sb₂S₃ thin films was found to decrease to 1.7 eV from 2.57 eV for undoped sample. This result indicates that carbon doping of Sb₂S₃ is not only capable of bandgap lowering but can also enhance the conductivity of the sample.

In an important theoretical study, Nie and Sohlberg [65] reported the lowering of the bandgap of n-TiO₂ to 2.32 eV (535 nm) due to carbon incorporation and predicted that the bandgap value of 1.58 eV may be possible to achieve by some complex carbon incorporation. Interestingly, it was found experimentally that enhanced carbon doping lowered the bandgap of n-TiO₂ up to 1.45 eV [16]. Wang and Lewis [66, 67] addressed theoretically the effects of carbon dopants concentration on the photoresponse of n-TiO₂ in the visible-light region. They found that the substitutional and interstitial carbon dopants incorporated into TiO₂ drastically affected the electronic structure of the material, thus improving its photoactivity. They predicted the low bandgap of 2.35 eV for carbon-doped TiO₂ as it was experimentally observed earlier [1]. The theoretical findings of Di Valentin et al. [68] revealed the presence of substitutional and interstitial carbon in carbon-modified (CM)-n-TiO₂ (Figure 3) which were found to be responsible for the lowering of its main bandgap as well as generating a mid-gap band.

(a)

(b)

(c)

(d)

Figure 3 

Partial geometry of the models for (a) one substitutional C atom to O (CS-O), (b) one substitutional C atom to Ti (CS-Ti), (c) one interstitial C atom (CI), and (d) one interstitial C atom nearby an oxygen vacancy (CI+VO) in the rutile supercell. The yellow spheres represent O atoms, the small brown spheres represent Ti atoms, and the black sphere presents the carbon impurity [68].

Among various oxide semiconductor photocatalysts, iron (III) oxide (Fe₂O₃) is a low-cost semiconductor with high stability. Iron (III) oxide has a bandgap of 2.0–2.3 eV; therefore, it can absorb solar radiation up to 620 nm, which covers ~37% of the photons of the solar spectrum of global AM 1.5 (1 sun) [70]. However, challenges are set by its quite low photoresponse due to its high resistivity and consequent recombination of photogenerated carriers. To minimise these limitations, it has been reported that dopants, either n-type [4, 7, 26, 27] or p-type [28–30], improved the photoresponse of iron (III) oxide (Fe₂O₃) towards water splitting to hydrogen. Thin films of n-Fe₂O₃ were extensively studied [4, 7, 25–36]. For example, n-Fe₂O₃ films were synthesised by a sputtering method [30], by pressing powders of Fe₂O₃ [32–34], and by spray pyrolysis method [4, 7, 28–30, 32]. Silicon-doped Fe₂O₃ films synthesised by ultrasonic spray pyrolysis showed 1.0 mA cm⁻² at +0.1 V versus AgCl [36] under solar-simulated light illumination at 1 sun, AM 1.5 (100 mW cm⁻²).

Tungsten trioxide (WO₃) is also of great interest and has been investigated extensively due to its promising physical and chemical properties [70–73] The WO₃ films are nontoxic and have relatively low price. Semiconducting tungsten oxide (WO₃) is an interesting candidate for photocataylsis because of its relatively low bandgap energy (E_g = 2.6 eV), [74–76] resulting in possible utilization of 12% of the solar spectrum [77]. Spectral response studies were carried out for n-type WO₃ by different workers [78, 79]. For example, n-WO₃ films were synthesized by conventional evaporation techniques [80, 81] and, more recently, with the sol-gel process [82, 83]. It is important to note that regular undoped tungsten oxide is an n-type semiconductor, but when it is carbon-doped or carbon-modified, it was found to be a p-type tungsten oxide. Carbon-modified p-type tungsten oxide (CM-p-WO₃) films were synthesized by thermal flame oxidation method at several flame temperatures for different lengths of time [84].

In this paper, we summarize our studies on thin-film photoelectrodes of the following(i)visible light active carbon-modified titanium oxides (CM-n-TiO₂);(ii)visible light active hydrogen modified n-type titanium oxide (HM-n-TiO₂) thin films;(iii)carbon-modified iron oxides (CM-n-Fe₂O₃) thin films;(iv)visible light active carbon-modified p-type tungsten oxides (CM-p-WO₃) thin film.

The results of photocurrent density measurements on these photoelectrodes for water splitting to H₂ and O₂ and the maximum photoconversion efficiencies under both Xenon lamp and natural sunlight of global AM 1.5 illuminations are also provided. Synthesized thin-film photoelectrodes are characterized in terms of bandgap energies, X-ray diffraction pattern, scanning electron micrograms, and carbon contents to correlate with their photoresponse towards water-splitting reaction.

2. Experimental Details and Results

2.1. Visible Light Active Carbon-Modified Titanium Oxides (CM-n-TiO₂) Thin Films

Shaban and Khan [23] synthesized visible light active carbon-modified n-type titanium oxide (CM-n-TiO₂) thin films by flame oxidation of Ti metal sheets both flat and grooved and also by using combination of spray pyrolysis and flame oxidation methods. Also un-doped reference n-TiO₂ samples were synthesized in an electric oven for comparison. The flame was custom designed by Knight Co. The flame temperature was controlled by adjusting the flow rates of oxygen and natural gas.

2.1.1. Dependence of Photocurrent Density on Measured Electrode Potential for Titanium Oxides

Plots of photocurrent density (𝑗_𝑝, mA cm⁻²) as a function of measured potential, 𝐸_meas (V/SCE) for oven-made n-TiO₂ synthesized at 825°C for 16 min (sample 1); CM-n-TiO₂ synthesized by only thermal flame oxidation of Ti at 825°C for 16 min (sample 2); grooved CM-n-TiO₂ synthesized by only thermal flame oxidation of Ti (0.005 inch groove depth) at 820°C for 18 min (sample 3); CM-n-TiO₂ synthesized by spray pyrolysis of 0.175 M TiCl₄ in ethanol at substrate temperature of 460°C for 20 sec at carrier gas (oxygen) pressure of 20 psi followed by thermal flame oxidation at 825°C for 16 min (sample 4) are shown in Figure 4. Sample 1 shows the lowest photocurrent density of 1.6 mA cm⁻² measured at –0.4 V/SCE. The observed low photocurrent density for sample 1 was due to absence of carbon as well as lower thickness in such electric oven-made samples. Sample 4 generated the highest photoresponse compared to samples, 1, 2, and 3. Photocurrent density, measured at –0.7 V/SCE, increased significantly from 9.17 mA cm⁻² for sample 2 to 11.44 mA cm⁻² for sample 3. The surface grooving of Ti metal increased the effective surface area of oxide layer in sample 3 and consequently generated higher photocurrent density. A sharp increase in photocurrent density to 14.68 mA cm⁻² measured at –0.7 V/SCE for sample 4 can be attributed to increased thickness of the oxide layer synthesized by combination of spray pyrolysis and flame oxidation methods. This is because increased thickness allows absorption of more light to generate more electron-hole pairs (excitons).

Figure 4 

Photocurrent density, 𝑗𝑝 (measured under light intensity of 100 mW cm−2 from a 150 Watt Xenon lamp) as a function of measured potential, 𝐸meas (V/SCE), for oven-made n-TiO2 synthesized at 825°C for 16 min (sample 1); CM-n-TiO2 synthesized by only thermal flame oxidation of Ti at 825°C for 16 min (sample 2); grooved CM-n-TiO2 synthesized by only thermal flame oxidation of Ti (0.005 in groove depth) at 820°C for 18 min (sample 3); CM-n-TiO2 synthesized by spray pyrolysis of 0.175 M TiCl4 in ethanol at substrate temperature of 460°C for 20 sec at carrier gas (oxygen) pressure of 20 psi followed by thermal flame oxidation at 825°C for 16 min (sample 4); electrolyte solution of 5 M KOH was used. The electrode potentials at open circuit condition under illumination, 𝐸aoc, were found to be −0.977 V/SCE (sample 1), −0.940 V/SCE (sample 2), −0.942 V/SCE (sample 3), and −0.915 V/SCE (sample 4) [21–23].

2.1.2. Dependence of Photoconversion Efficiency on Applied Potential for CM-n-TiO₂ Films

The calculation of total percent photoconversion efficiency (%𝜀_photo(total)) of light energy under white light illumination to chemical energy in the presence of an external applied potential (𝐸_app) was carried out using the equation given earlier [1, 4] as%𝜀_photo()=𝑗total^𝑝𝐸^𝑜_rev−𝐸_app𝑃_𝑜×100, (6) where 𝑗_𝑝 is the photocurrent density (mA cm⁻²), 𝐸^𝑜_rev is the standard reversible potential (which is 1.23 V for water splitting reaction), 𝑃_𝑜 is the power density of incident light (mW cm⁻²), |𝐸_app| is the absolute value of the applied potential which is obtained as [1]𝐸_app=𝐸_meas−𝐸_aoc, (7) where 𝐸_meas (versus SCE) is the potential at which the photocurrent density was measured, 𝐸_aoc (versus SCE) is the electrode potential at open circuit condition in the same electrolyte solution under the same illumination intensity at which the photocurrent density was measured. The values of 𝐸_aoc (versus SCE) were found to be −0.847 V, −0.940 V, −0.942 V, −0.915 V for samples 1, 2, 3, and 4, respectively.

The corresponding maximum photoconversion efficiencies (see Figure 5) were found to be 1.2% at 0.447 V bias, 9.08% at 0.240 V bias, 11.31% at 0.242 V bias, and 14.04% at 0.215 V bias for samples 1, 2, 3, and 4, respectively, under illumination intensity of 100 mW cm⁻² from a 150 watt Xenon lamp.

Figure 5 

Dependence of photoconversion efficiency, %𝜀photo(total), on applied potential 𝐸app (V versus 𝐸aoc), for samples 1, 2, 3, and 4 [21–23].

2.1.3. Monochromatic Photocurrent Density-Wavelength Dependence for CM-n-TiO₂ Films

The plots of monochromatic photocurrent density, 𝑗_𝑝(𝜆), as a function of wavelength of light, 𝜆, for samples 1, 2, 3, and 4 are shown in Figure 6. The monochromatic photocurrent density, 𝑗_𝑝(𝜆), was measured at measured potentials corresponding to applied potential at which maximum %𝜀_photo(total) was observed (see Figure 5) under white light illumination. The values of 𝐸_app used were 0.477 V, 0.240 V 0.242 V, and 0.215 V for samples 1, 2, 3, and 4, respectively. Performing integration under the curve, total photocurrent densities of 1.25 mA cm⁻², 8.95 mA cm⁻², 11.30 mA cm⁻², and 13.58 mA cm⁻² were found for samples 1, 2, 3, and 4, respectively. These values are a little lower than the corresponding photocurrent density values under white light illumination from the Xenon lamp because of loss of some light passing through the monochromator.

Figure 6 

The monochromatic photocurrent density, 𝑗𝑝(𝜆), as a function of wavelength of light, 𝜆, for samples 1, 2, 3, and 4, respectively [21–23].

Note that the observed photocurrent in the visible region for the CM-TiO₂ samples cannot be due to artifact of a monochromator or due to second harmonic generation because if that would be the case the regular n-TiO₂ would have shown photocurrent in the visible region. Figure 6 clearly shows no photocurrent for regular n-TiO₂ in the wavelength region between 500 nm 750 nm. Recent combined sputtering and XPS studies showed that carbon incorporation was limited to few angstroms (15 ́Å to 20 ́Å) deep. Hence, the amounts of visible light absorption were low as demonstrated by low photocurrent in the visible region between 500 nm to 750 nm (see Figure 6).

2.1.4. Photoconversion Efficiency from Monochromatic Photocurrent Density for CM-n-TiO₂ Films

Alternatively, the percent total photoconversion efficiency (%𝜀_photo(total)) of light energy to chemical energy can be obtained by using the following equation [20, 21, 68, 70]:%𝜀_photo()=∫total^𝜆g𝜆min𝑗_𝑝(𝜆)𝐸^𝑜_rev−𝐸_app(𝜆)𝑑𝜆∫^∞_𝜆min𝑃_𝑜(𝜆)𝑑𝜆×100. (8) where 𝑗_𝑝(𝜆) is wavelength-dependent photocurrent density under monochromatic light illumination (mA cm⁻²nm⁻¹) and 𝑃_𝑜(𝜆) is the power density of incident wavelength-dependent monochromatic light (mW cm⁻²nm⁻¹). Note that the applied potential, 𝐸_app(𝜆), was also found to be wavelength-dependent and can be expressed as𝐸_app(𝜆)=𝐸_meas−𝐸_aoc(𝜆), (9) where 𝐸_aoc(𝜆) is the electrode potential in volts at open-circuit conditions under monochromatic light illumination. However, 𝜆_min = 250 nm for light from Xenon lamp and 300 nm for sunlight were used, and 𝜆_𝑔 = 1239.85/𝐸_𝑔 is the threshold wavelength corresponding to bandgap energy, 𝐸_𝑔, of the sample in electron volt unit.

Note that wavelength-dependent percent photoconversion efficiency, %𝜀_photo(𝜆) can be expressed as [20, 21],%𝜀_photo(𝜆)=𝑗_𝑝(𝜆)𝐸^o_rev−𝐸_app(𝜆)∫^𝜆max_𝜆min𝑃_o(𝜆)𝑑𝜆×100, (10) The plot of %𝜀_photo(𝜆) versus wavelength, 𝜆, using (10) is given in Figure 7 where total intensity in the denominator of (10) was taken as 100 mW cm⁻². Performing the integration (as given in (8)) from 𝜆_min = 250 nm (for Xenon lamp white light) to 𝜆_𝑔 nm or determining the area under the curve in Figure 7 obtained using (10), the total photoconversion efficiencies of 0.94%, 8.86%, 11.16%, and 13.79% were obtained for samples 1, 2, 3, and 4, respectively. Note that 𝜆_max represents the maximum wavelength available in sunlight which is 4200 nm. These values are little lower but comparable to 1.2%, 9.08%, 11.31%, and 14.04% (when calculated using (6) under white light illumination of intensity 100 mW cm⁻² from a 150 W Xenon lamp) for samples 1, 2, 3, and 4, respectively. These lower values can be attributed to loss of some light through the monochromator as well as the use of average value of 𝐸_app(𝜆) instead of using its experimentally determined value at each wavelength, 𝜆.

Figure 7 

The photoconversion efficiency, %𝜀photo(𝜆), as a function of wavelength of light, 𝜆, for samples 1, 2, 3, and 4 [21–23].

2.1.5. Bandgap Energy Determination for n-TiO₂ and CM-n-TiO₂ Films

The bandgap energy, 𝐸_𝑔, can be determined from the frequency of light, 𝜈, dependent quantum efficiency, 𝜂(𝜈), using the following equation [4, 5, 7]:𝜂(𝜈)ℎ𝜈=Aℎ𝜈−𝐸_𝑔^𝑛, (11) where 𝐴 is a constant, 𝑛 equals either 0.5 for allowed direct transition or 2 for allowed indirect transitions, ℎ𝜈 is the photon energy of frequency 𝜈. The wavelength, 𝜆, dependent quantum efficiency, 𝜂(𝜆), or incident photon conversion efficiency, IPCE(𝜆), can be expressed as [20, 21]𝜂(𝜆)(𝜆)=𝑗=IPCE^𝑝(𝜆)𝑒𝐼_𝑜(𝜆=()𝑗)1239.85Vnm_𝑝(𝜆)mAcm⁻²nm⁻¹𝜆(nm)𝑃_𝑜(𝜆)mWcm⁻²nm⁻¹, (12) where 𝑒 is the electronic charge and the incident photon flux, and the wavelength-dependent photon flux, 𝐼_𝑜(𝜆), (cm⁻² nm⁻¹ s⁻¹) was expressed as [20, 21]𝐼_𝑜(𝜆)=𝑃_𝑜(𝜆)𝜆𝐼ℎ𝑐_𝑜(𝜆)=𝑃_𝑜(𝜆)mWcm⁻²nm⁻¹𝜆(nm)[()𝑒],1239.85Vnm (13) where ℎ is the Planck’s constant and 𝑐 is the velocity of light.

Figure 8 shows [𝜂(𝜆)ℎ𝜈]^1/2 versus ℎ𝜈 plot for samples 1, 2, 3, and 4, respectively.From the intercepts of the straight lines the bandgaps were obtained. It shows two indirect bandgap energies for all CM-n-TiO₂ photoelectrodes. The first bandgap energy of 2.65 eV corresponds to lowering of original bandgap (samples 2, 3, and 4) and second bandgap with values of 1.6 eV and 1.4 eV can be attributed to mid-gap band generated by carbon doping as shown in the schematic diagram in Figure 9. Theoretical analysis [65, 66, 70] and experimental findings [20, 21] also showed multiple bandgaps in CM-n-TiO₂. Oven-made n-TiO₂ showed only one bandgap of 2.9 eV, and as a result, poor photoresponse was obtained. Note that the first bandgap energy values for all CM-n-TiO₂ samples were found equal. The slopes of bandgap plots for sample 4 were found to be the highest, which is in accordance with the highest observed photoresponse generated by this sample.

Figure 8 

The respective plots of [𝜂(𝜆)ℎ𝜈]1/2 versus ℎ𝜈 to determine the bandgap for samples 1, 2, 3, and 4 [21–23].

Figure 9 

A schematic diagram of valence, conduction, and mid-gap bands to show the bandgaps in CM-n-TiO2 films synthesized by thermal flame oxidation of Ti metal sheet [21–23].

2.1.6. Spectral Distribution of Light for Global AM 1.5 Natural Sunlight (1 Sun) and Xenon Lamp Light (100 mW cm⁻²)

It is important to compare the spectral distribution of light from both Xenon lamp and sunlight. Figure 10 compares the spectral distribution of these two lights. The integrated value of the intensity from 300 up to 800 nm for global sunlight AM 1.5 (1 sun = 100 mW cm⁻² measured by NREL [85]) was found to be 59.00 mW cm⁻² clearly indicating that the contribution from remaining light from 800 nm to 4100 nm present in global sunlight will be 41.00 mW cm⁻² [= (100–59) mW cm⁻²]. Hence, total Xenon lamp light comparable to sunlight (1sun) will be 98 mW cm⁻² [= (57 + 41) mW cm⁻²] where 57.00 mW cm⁻² is the integrated value of intensity from 200 nm to 800 nm for Xenon lamp light.

Figure 10 

The wavelength-dependent intensity of light from a 150 W Xenon lamp at intensity of 100 mW cm−2, measured by using silicon photon detector (UDT Sensors. Inc., Model 10DP/SB), compared to the wavelength-dependent intensity of natural global AM 1.5 sunlight in the wavelength range from 300 to 800 nm obtained from NREL website [85].

Note that the spectral distribution of Xenon lamp light begins below 250 nm compared to global AM 1.5 sunlight that begins at 300 nm. Furthermore, the spectral distribution of Xenon lamp light is higher in the UV region, but in the visible region, it is lower than that of sunlight. Hence, one will expect comparable photocurrent density and photoconversion efficiencies for CM-n-TiO₂ samples for water splitting under actual global sunlight illumination of AM 1.5 (1 sun) and the Xenon lamp illumination of intensity of 100 mW cm⁻². However, relatively higher values for photocurrent density and photoconversion efficiency under Xenon lamp illumination will be expected compared to those under sunlight for CM-TiO₂ due to higher spectral distribution of the former in the UV region than in the visible region (see Figure 10) and that the absorption coefficient of light for CM-TiO₂ in UV region is much higher compared to that in the visible region. Also, only 1.5 to 2.0 nm deep carbon modified region in CM-n-TiO₂ could not absorb enough light in the visible region.

2.1.7. Photocurrent Density and Photoconversion Efficiency for Titanium Oxides from IPCE(𝜆) Using Illumination Intensity, 𝐼_𝑜(𝜆) of Natural Global AM 1.5 Sunlight

Utilizing the values of the wavelength-dependent incident photon conversion efficiency IPCE(𝜆) or quantum efficiency 𝜂(𝜆) of these samples, it is possible to determine the photocurrent density, 𝑗^sunlight_𝑝(𝜆), under natural global AM 1.5 sunlight illumination as [20, 21]𝑗^sunlight_𝑝(𝜆)=IPCE(𝜆)𝑒𝐼^sunlight_𝑜(𝜆), (14) where 𝐼^sunlight_𝑜(𝜆) is the wavelength-dependent photon flux of incident global AM 1.5 sunlight which can be obtained using (13) in which photon power density of sunlight 𝑃^sunlight_𝑜(𝜆) was used from the tabulated data given in National Renewable Energy Laboratory (NREL) website [85].

The plots of 𝑗^sunlight_𝑝(𝜆) [data obtained from (14)] versus the wavelength of light, 𝜆, for samples 1–4 are given in Figure 11. The integrated values of total current density under sunlight illumination were found to be 0.85 mA cm⁻², 5.89 mA cm⁻², 7.71 mA cm⁻², and 12.27 mA cm⁻² for sample 1, 2, 3, and 4, respectively.

Figure 11 

Wavelength-dependent photocurrent density under global sunlight illumination, 𝑗sunlight𝑝(𝜆)=[IPCE(𝜆)×e×globalsunlightintensity,𝐼sunlight𝑜(𝜆)], versus wavelength, 𝜆 of incident sunlight for samples 1, 2, 3, and 4 (14) [23].

The total photocurrent density, 𝑗^sunlight_𝑝(total), under sunlight illumination can be expressed as [20, 21]𝑗^sunlight_𝑝(total)=e^𝜆g_𝜆minIPCE(𝜆)𝐼_𝑜(𝜆)𝑑𝜆. (15) The total photoconversion efficiency under natural sunlight illumination can be expressed as [20, 21]%𝜀^sunlight_solar()=∫total^𝜆g𝜆min𝑗^sunlight_𝑝(𝜆)𝐸^𝑜_rev−𝐸_app(𝜆)𝑑𝜆∫_𝜆^𝜆max_min𝑃^sunlight_𝑜(𝜆)𝑑𝜆×100, (16) where the denominator was taken as 100 mW cm⁻² for actual natural global AM 1.5 sunlight (1 sun).

Hence, using (15) and the denominator in (16) as 100 mW cm⁻² for global AM 1.5 sunlight, one can express the simplified form of (16) as%𝜀^sunlight_solar(𝜆)=𝑗^sunlight_𝑝(𝜆)mAcm⁻²1.23−𝐸_appaveragevolt100mWcm⁻²×100,%𝜀^sunlight_solar()=𝑗total_sunlight𝑝(total)mAcm⁻²1.23−𝐸_appaveragevolt100mWcm⁻²×100, (17)(18) where 𝐸^𝑜_rev = 1.23 volt was used for water-splitting reaction and an average value of the absolute value of applied potential, |𝐸_app|_average, was used for all wavelength, 𝜆 values.

Photoconversion efficiencies under sunlight illumination were obtained using (16) or (18). Figure 12 shows the calculated wavelength-dependent percent photoconversion efficiencies under sunlight illumination, %𝜀^sunlight_solar(𝜆), as a function of wavelength of incident light 𝜆 using (17). Performing the integration under the curve from 300 nm to 𝜆_g, photoconversion efficiencies of 0.67%, 5.63%, 7.62, and 12.26% were found for samples 1, 2, 3, and 4, respectively. These results indicate that CM-n-TiO₂ thin films synthesized by combination of spray pyrolysis and natural gas flame oxidation (sample 4) can be expected to split water with ≥10% even under actual sunlight illumination. The photoconversion efficiency values for these 4 samples under different illumination conditions are summarized in Table 1.

Table 1 

Maximum photoconversion efficiencies (%) under Xenon lamp light and actual natural sunlight illumination conditions for reference electric oven-made (n-TiO₂), flame-made, and spray-flame-made CM-n-TiO₂ samples and their bandgap and mid-gap Energies [21–23].

Figure 12 

Wavelength-dependent photoconversion efficiency under sunlight illumination, %𝜀sunlightsolar(𝜆), versus wavelength, 𝜆 of incident light for samples 1, 2, 3, and 4 (17) [21–23].

2.1.8. UV-Vis Spectra of n-TiO₂ and CM-n-TiO₂ Films

The UV-Vis absorbance of n-TiO₂ and CM-n-TiO₂ photoelectrodes is shown in Figure 13. The absorption spectra of CM-n-TiO₂ samples 1, 2, 3, and 4 demonstrate a wide absorption in the UV and visible regions with a tail extending to near infrared region up to 800 nm. This indicates low bandgap energy values for these samples. The reference oven-made n-TiO₂ (sample 1) shows absorption only up to 415 nm which corresponds to bandgap energy of 3.0 eV. The lower bandgaps of CM-n-TiO₂ samples may be due to carbon incorporation during the flame oxidation of Ti metal sheet. Note that carbon modification also generates oxygen vacancy by partial reduction of TiO₂ and this intern may be partially responsible for visible light activity of carbon-modified titanium oxide (CM-n-TiO₂). The UV-Vis spectroscopic results show higher absorption in UV as well as in the visible regions for CM-n-TiO₂ films synthesized by combination of spray pyrolysis and flame oxidation (sample 3) compared to sample 2. Also, these UV-Vis spectra are consistent with the higher photocurrent densities at CM-n-TiO₂ thin-film electrodes prepared by the combination of two methods. Importantly, this observation conforms to the important hypothesis that the spray pyrolysis increased the thickness of the visible light active CM-n-TiO₂ films. Consequently, these films absorbed more light.

Figure 13 

UV-Vis Spectra of titanium oxides (samples 1–4) [21–23].

However, the UV-Vis spectra (see Figure 13) do not match well with the monochromatic photocurrent density-wavelength dependence (see Figure 6). The UV-Vis spectra represent the excitation of electron from lower energy state (to generate holes in this state) to higher energy state by absorption of light. However, to have observed photocurrent, photogenerated holes will have to (1) transport to the interface and (2) undergo transition across the interface to available (thermally distributed) hole acceptor states (e.g., OH⁻) in solution. These two extra processes are responsible not having UV-Vis spectra to match with the monochromatic photocurrent density-wavelength-dependent plots.

The electrons that are excited to conduction band from the shallow mid-gap states in CM-n-TiO₂ (see Figure 9) by light of around 500 nm reasonable amount of absorption were observed in the UV-Vis spectra but the holes generated in the mid gap band (see Figure 9) were lost due to very low availability of acceptor states of holes in solution (e.g., OH⁻ ion) and consequently generated negligible photocurrent density (see Figure 6). However, absorption in UV-Vis spectra around 500 nm is observed since such absorption does not need availability of acceptor state of holes in solution. Furthermore, the light around 500 nm (2.48 eV) having sub-bandgap energy cannot be absorbed by CM-n-TiO₂ to generate electron-hole pairs in its valence band. However, the light around 600 nm to 700 nm can excite electrons from the valence band to mid-gap band states of CM-n-TiO₂ (see Figure 9) and these electrons move to the counter electrode. The corresponding holes generated in the valence band undergo transition to available hole acceptor states in solution. Consequently, some photocurrents are observed in this wavelength region (see Figure 6). Furthermore, it should be noted that the carbon modification or doping was confined to surface region (2.0 nm deep) in CM-n-TiO₂, the absorption of light in the visible region was found low.

2.1.9. Scanning Electron Microscopy (SEM) of CM-n-TiO₂ Films

The surface characterization of a CM-n-TiO₂ film (sample 2) was performed by employing a scanning electron microscope (SEM) and the image is shown in Figure 14. The surface looks extremely rough and nanocrystalline nanowalls hence provided much higher surface area to absorb incident light, multiple absorption, and reflection to absorb more light and consequently generated enhanced photoresponse. Generation of multiple electron-hole pairs (excitons) by absorption of single high-energy photon in UV-region in nanocrystals (Ncs) was observed and also explained theoretically in terms of impact ionization [86–92]. Furthermore, the microscopic rough surface could temporarily trap microscopic oxygen gas bubbles on rough surface that may have acted as tiny lens and spherical mirrors that helped to concentrate the incident light [93] on the electrode surface and hence augmented the photoresponse for water splitting by an order of magnitude.

Figure 14 

SEM images for a CM-n-TiO2 (sample 2) [21–23].

2.1.10. X-Ray Diffraction (XRD) for Titanium Oxides

X-ray diffraction (XRD) measurements were carried out to characterize the structure of n-TiO₂ and CM-n-TiO₂ samples. Figure 15 shows the XRD patterns for oven-made n-TiO₂ (S1); CM-n-TiO₂ synthesized by only thermal flame oxidation (S2) and the CM-n-TiO₂ synthesized by spray pyrolysis and thermal flame oxidation (S3).

Figure 15 

X-ray diffraction (XRD) patterns for samples 1, 2 and 3, and identified as S1, S2, and S3, respectively [21–23].

The peaks of these XRD spectra of both CM-n-TiO₂ photoelectrodes (S2 and S3) are consistent with the rutile structure. The XRD results demonstrate that the crystal structure of oven-made n-TiO₂ (S1) is of mixture of anatase and rutile forms. It is important to note that some peaks of rutile structure are missing in both CM-n-TiO₂ samples (S2 and S3) compared to those in oven-made n-TiO₂ (S1). This indicates the marked influence of carbon modification on the structure of n-TiO₂.

2.1.11. Energy-Dispersive Spectroscopy (EDS) for CM-n-TiO₂ Films

Figure 16 shows the pattern of energy-dispersive spectroscopic (EDS) data for samples 1, 2, 3, and 4. The carbon contents increased significantly from 17.60 atom % for CM-n-TiO₂ synthesized by only thermal flame oxidation to 23.23 atom % for CM-n-TiO₂ synthesized by combination of spray pyrolysis and thermal flame oxidation (see Table 2 for detail).

Table 2 

Atomic percent of elements Ti, O, and C in n-TiO₂ and CM-n-TiO₂ films from EDS data.

(a)

(b)

(c)

(d)

Figure 16 

Pattern of energy-dispersive spectroscopy (EDS) of samples 1, 2, 3, and 4 [21–23].

2.2. Visible Light Active Hydrogen-Modified n-Type Titanium Oxide (HM-n-TiO₂) Thin Films

Frites and Khan [94] reported a new method for the synthesis of hydrogen-modified (HM)-n-TiO₂ thin films by thermal oxidation of Ti metal sheet (Alfa Co. 0.25 mm thick) to TiO₂ in an electric oven followed by incorporation of hydrogen in it by electrochemical generation of hydrogen gas on it under cathodic polarization at −1.6 V versus Pt in an alkaline medium at room temperature and also provided its photoelectrochemical behavior during water-splitting reactions.

2.2.1. Photocurrent Density-Potential Dependence for HM-n-TiO₂

The photoresponse of the HM-n- TiO₂ was evaluated by measuring the rate of water-splitting reaction to hydrogen and oxygen in terms of photocurrent density, 𝐽_𝑝. The optimized electric oven-made n-TiO₂ and HM-n-TiO₂ photoelectrodes showed photocurrent densities of 0.2 mA cm⁻² and 1.60 mA cm⁻², respectively, at a measured potential of −0.4 V versus Pt at illumination intensity of 100 mWcm⁻² from a 150 W Xenon lamp (Figure 17). This indicated an eightfold increase in photocurrent density for HM-n-TiO₂ compared to oven-made n-TiO₂ at the same measured electrode potential. The bandgap energy of HM-n-TiO₂ was found to be 2.7 eV compared to 2.82 eV for electric oven-made n-TiO₂ and a mid-gap band at 1.67 eV above the valence band was also observed. The HM-n-TiO₂ thin-film photoelectrodes were characterized using photocurrent density under monochromatic light illumination and UV-Vis spectral measurements.

Figure 17 

Photocurrent density, 𝐽𝑝, as a function of measured potential, 𝐸meas (V versus Pt) at light intensity of 100 mW/cm2 from a 150 W Xenon lamp for (1) oven-made n-TiO2 at 835°C for 5 min and (2) HM-n-TiO2 samples [94].

2.2.2. Monochromatic Photocurrent-Wavelength Dependence for HM-n-TiO₂

The monochromatic photocurrent density, 𝐽_𝑝(𝜆), was measured under wavelength, 𝜆, dependent monochromatic light illumination from a Xenon lamp by using a Spectra Physics monochromator (Model 77250) at a measured potential of −0.1 V versus Pt (see Figure 18). Performing integration under the curve in the wavelength range between 250 nm and 775 nm in Figure 18, a total photocurrent density of 2.2 mA/cm² was observed at HM-n-TiO₂ electrode. This value is little higher within the experimental error than the corresponding photocurrent density values of 2.1 mA/cm² for HM-n-TiO₂ samples under white light illumination, at the same measured potential of −0.1 V versus Pt (see Figure 17). The monochromatic photocurrent density versus wavelength plots shows a clear shift of the photosensitivity of the HM-n-TiO₂ to the visible region with a second peak around 630 nm compared to oven-made n-TiO₂ which shows only one pick in the UV region only (see Figure 18).

Figure 18 

The monochromatic photocurrent density, 𝐽𝑝(𝜆), as a function of wavelength of light, 𝜆 (nm), for HM-n-TiO2, and unreduced n-TiO2 under monochromatic light illumination from a Xenon lamp; using a Spectra Physics monochromator (Model 77250) with a 0.75 mm slit width. The applied potential was kept at −0.1 V/Pt for both electrodes [94].

2.2.3. Bandgap Energy Determination for HM-n-TiO₂

The bandgap of HM-n-TiO₂ becomes narrower from 3.0 eV for regular rutile TiO₂ to 2.7 eV (Figure 19). Also, a mid-gap band was created within the bandgap of n-TiO₂ due to hydrogen reduction; which is located 1.67 eV above the valence band for the hydrogen-modified titanium oxide.

Figure 19 

[𝜂(𝜆)ℎ𝜈]1/2 versus ℎ𝜈 plot for electric oven-made n-TiO2 and hydrogen-modified HM-n-TiO2 films [94].

2.3. Carbon-Modified Iron Oxides (CM-n-Fe₂O₃) Thin Films

Shaban and Khan [23] synthesized nanocrystalline carbon-modified iron (III) oxide (CM-n-Fe₂O₃) thin films by flame oxidation of Fe metal sheet at different temperatures (700–850°C) and oxidation time (3–15 min), using a custom-designed large area flame (Knight, model RN. 3.5 xa wc) under controlled oxygen and natural gas flows.

2.3.1. Photocurrent Density-Potential Dependence for Iron Oxides Photoelectrodes

Figure 20 shows the dependence of photocurrent density (𝐽_𝑝, mA cm⁻²) for water splitting on measured electrode potentials (𝐸_meas, V/SCE) at light intensity of 100 mW cm⁻², for oven-made-n-Fe₂O₃ (sample A) and flame-made CM-n-Fe₂O₃ (sample B). Under illumination of light intensity of 100 mW cm⁻², and at measured potential of −0.1 V/SCE, photocurrent density increased from 2.8 mA cm⁻² for oven-made n-Fe₂O₃ to 7.83 mA cm⁻² for carbon-modified n-CM-n-Fe₂O₃ synthesized by thermal flame oxidation. The electrode potentials at open circuit condition under illumination of light, 𝐸_aoc, were found to be –0.498 V/SCE and −0.510 V/SCE for samples A and B, respectively.

Figure 20 

Photocurrent density (𝑗𝑝, mAcm−2) versus measured potential (𝐸meas, V/SCE) for oven-made n-Fe2O3 (sample A) and flame-made CM-n-Fe2O3 (sample B) under light intensity of 100 mW cm−2. Both samples were prepared at the same optimum conditions of 850°C for 15 minutes. 5.0 M KOH was used as an electrolyte solution [23].

2.3.2. Photoconversion Efficiency-Potential Dependence for Iron Oxides Photoelectrodes

Figure 21 shows the photoconversion efficiency, %𝜀_photo(total), as a function of applied potential, 𝐸_app (V versus 𝐸_meas), for samples A and B. Under illumination intensity of 100 m W cm⁻², the maximum photoconversion efficiency for water splitting improved significantly from 3.61% (at a minimal applied potential of +0.5 V versus 𝐸_aoc) for sample A 6.5% (at a minimal applied potential of +0.402 V versus 𝐸_aoc) for sample B.

Figure 21 

Photoconversion efficiency, %𝜀photo(total), at light intensity of 100 mWcm−2 versus applied potential, 𝐸app (V versus 𝐸meas) for oven-made n-Fe2O3 (samples A) and flame-made CM-n-Fe2O3 (sample B) [23].

2.3.3. Monochromatic Photocurrent-Wavelength Dependence for Iron Oxides

The dependence of monochromatic photocurrent density, 𝑗_𝑝(𝜆), versus the wavelength, 𝜆, of light for CM-n-Fe₂O₃ (sample B) is shown in Figure 22. The monochromatic photocurrent density, 𝑗_𝑝(𝜆), at measured potential of 0.0 V/SCE was measured under monochromatic light illumination from a 150 watt Xenon lamp by using a spectra physics monochromator (Model 77250). Performing integration under the curve, total photocurrent densities of 7.2 mA cm⁻² were found for CM-n-Fe₂O₃. This value is less than that under white light illumination (7.83 mA cm⁻²). This is due to loss of light through the monochromator. It is also due to steady-state measurements of photocurrents under monochromatic light illumination. Note that the photocurrent densities under white light illumination were measured with a scan rate of 50 mV/sec.

Figure 22 

The monochromatic photocurrent density, 𝑗𝑝(𝜆), as a function of wavelength of light, 𝜆, for CM-Fe2O3 [23].

2.3.4. Photoconversion Efficiency from Monochromatic Photocurrent Densities at Iron Oxide Electrodes

The plot of %𝜀_photo(𝜆) versus wavelength, 𝜆, using (10) for CM-n-Fe₂O₃ is given in Figure 23 where total intensity in the denominator of (10) was used as 100 mW cm⁻². Performing integration in (8) from 𝜆_min = 250 nm to 𝜆_g = 1240/Eg or determining the area under the curve in Figure 23 obtained using (10), a total photoconversion efficiency of 5.9% is also lower than that under white light illumination (6.5%) for the reasons mentioned in the above conditions.

Figure 23 

The photoconversion efficiency, %𝜀photo(𝜆), as a function of wavelength of light, 𝜆, for CM-Fe2O3 [23].

2.3.5. Bandgap Energy Determination for Iron Oxides

Figure 24 shows [𝜂(ℎ𝜈)]^1/2 versus ℎ𝜈 for samples A and B, respectively. From the intercept of these plots, the bandgaps were found to be 2.3 eV and 1.95 eV for samples A and B, respectively. These bandgap values are in agreement with those obtained from UV-Vis spectra (see Figure 25). Note that 𝜂(𝜆) is the wavelength, 𝜆, dependent quantum efficiency as expressed in (12).

Figure 24 

The respective plot of [𝜂(ℎ𝜈)]1/2 versus ℎ𝜈 to determine the bandgap energies of both oven-made n-Fe2O3 and flame-made CM-n-Fe2O3 films [23].

Figure 25 

UV-visible spectra for iron oxides [23].

2.3.6. UV-Vis Spectra for Iron Oxides

Figure 25 compares the UV-Vis absorbance of flame-made CM-n-Fe₂O₃ and oven-made n-Fe₂O₃ photoelectrodes having maximum photoconversion efficiencies of 6.5% and 3.61%, respectively. Both electrodes were synthesized at the same optimum conditions of 850°C for 15 min. The UV-Vis spectroscopic results show a significant increase in the absorbance of light in the UV as well as in the visible regions for flame-made CM-n-Fe₂O₃ film compared to that of the oven-made n-Fe₂O₃. The UV-Vis spectra are consistent with the higher photocurrent densities of flame-made CM-n-Fe₂O₃ thin-film electrodes. The absorption spectra demonstrate a wide absorption in the UV and visible regions with a tail extending up to 600 nm which corresponds to a bandgap energy of 1.95 eV for flame-made CM-n-Fe₂O₃ and up to 545 nm which corresponds to bandgap energy of 2.3 eV for oven-made n-Fe₂O₃. The low bandgap energy value for flame-made CM-n-Fe₂O₃ can be attributed to carbon incorporation during the flame oxidation process of Fe metal sheet.

2.3.7. X-Ray Diffraction (XRD) for Iron Oxides

XRD patterns (Figure 26) show that in addition to Fe₂O₃, Fe₃O₄ and FeO are also present in both oven-made and flame-made thin films. Analysis of XRD confirms carbon incorporation in flame-made iron (III) oxide film (sample B) by the presence the peaks for iron carbide compounds in the forms of FeC, Fe₂C, Fe₃C, and FeCO₃. None of these peaks for iron carbide compounds were observed in oven-made iron (III) oxide (sample A). It can be concluded that the flame oxidation process is responsible for carbon incorporation onto iron (III) oxide thin film.

Figure 26 

X-ray diffraction (XRD) patterns for flame-made CM-n-Fe2O3 (top plot) and the reference oven-made n-Fe2O3 (bottom plot) films. The detected probable formulas to comply with peaks are a = Fe2O3; b = Fe3O4; c = FeO; d = Fe; e = FeC; f = Fe2C; g = Fe3C; h = FeCO3 [23].

2.3.8. Energy-Dispersive Spectroscopy (EDS) for Iron Oxides

Table 3 shows the EDS results of oven-made n-Fe₂O₃ (sample A) and flame-made CM-n-Fe₂O₃ (sample B) photoelectrodes. The presence of carbon in CM-n-Fe₂O₃ photoelectrode and its absence in the oven-made n-Fe₂O₃ were observed. The corresponding EDS spectra of oven-made n-Fe₂O₃ and flame-made CM-n-Fe₂O₃ samples are shown in Figures 27(a) and 27(b), respectively. Note that the regular oven-made n-Fe₂O₃ did not show any carbon peak.

Table 3 

Atomic percent of elements Fe, O, and C in n-Fe₂O₃ and CM-n- Fe₂O₃ films from EDS data.

(a)

(b)

Figure 27 

Pattern of energy-dispersive spectroscopy (EDS) of samples (a) and (b) [23].

2.3.9. Scanning Electron Microscopic (SEM) Images of Iron Oxides

Morphological differences were observed from the scanning electron microscopic (SEM) images (Figure 28) between (a) oven-made n-Fe₂O₃ film and (b) flame-made CM-n-Fe₂O₃ film. The surface of oven-made n-Fe₂O₃ shows nanowires but CM-n-Fe₂O₃ surface of sample B does not show nanowires even at higher magnification (200 nm compared to 1 micro meter used for oven-made sample a). The nanowires on oven-made n-Fe₂O₃ films (Figure 28(a)) helped to minimize the recombination rate of photogenerated carriers and consequently generated much higher photoresponse compared to its nanocrystalline thin films synthesized by spray pyrolysis methods earlier (4). The CM-n-Fe₂O₃ films show only the nanocrystals (Figure 28(b)).

(a)

(b)

Figure 28 

SEM images of (a) oven-made-n Fe2O3 [23] and (b) flame-made CM-n-Fe2O3 films.

2.4. Visible Light Active Carbon-Modified p-Type Tungsten Oxides (CM)-p-WO₃ Thin Film

Shaban and Khan [84] synthesized carbon-modified p-type tungsten oxide (CM-p-WO₃) photoelectrodes by flame oxidation of tungsten metal sheets (Alfa Co. 0.25 mm thick) at several temperatures for different lengths of time.

2.4.1. Photocurrent Density-Potential Dependence for Tungsten Oxides Photoelectrodes

The photocatalytic activity of CM-p-WO₃ was evaluated by measuring the rate of water-splitting reaction to hydrogen and oxygen, which is proportional to photocurrent density, 𝑗_𝑝. Carbon-modified tungsten oxide photoelectrodes exhibited p-type photoresponse in acidic medium of 0.5 M H₂SO₄. The highest photocurrent density of 2.08 mA cm⁻² was observed under illumination intensity of 80 mW cm⁻² from a 150 Watt (Kratos Model LH 150/1) Xenon arc lamp at an optimal thermal oxidation temperature of 900°C for 15 min at applied potential of −0.4 V versus 𝐸_aoc (Figure 29).

Figure 29 

Photocurrent density (𝑗𝑝, mAcm−2) versus measured potential (𝐸meas, V/SCE) for CM-p-WO3 synthesized by thermal oxidation of tungsten metal sheets at different flame temperatures 850, 900, and 950°C. All these samples were synthesized at the same flame oxidation time of 15 minutes [84].

2.4.2. Photoconversion Efficiency-Potential Dependence for Tungsten Oxides Photoelectrodes

The optimized CM-p-WO₃ photoelectrodes were found to generate the highest photoconversion efficiency of 2.16% for photosplitting of water (Figure 30).

Figure 30 

Photoconversion efficiency (%𝜀photo(total)) versus measured potential (𝐸meas, V/SCE) for CM-p-WO3 synthesized by thermal oxidation of tungsten metal sheets at different flame temperatures 850, 900, and 950°C. All these samples were synthesized at the same flame oxidation time of 15 minutes [84].

2.4.3. Monochromatic Photocurrent Density-Wavelength Dependencefor CM-p-WO₃ Electrodes

The dependence of monochromatic photocurrent density, 𝑗_𝑝(𝜆), versus the wavelength, 𝜆, of light for CM-p-WO₃ thin-film electrodes, having photoconversion efficiency of 2.16% (synthesized by thermal oxidation of W metal at oxidation temperature of 900°C for 15 minutes), is shown in Figure 31.

Figure 31 

Monochromatic photocurrent density, 𝑗𝑝(𝜆), versus the wavelength, 𝜆, of light for CM-p-WO3 thin-film electrodes synthesized by thermal oxidation of W metal at temperature of 900°C for 15 minutes; electrolyte solution of 0.5 M H2SO4; under monochromatic light illumination of Xenon lamp; using a Spectra Physics monochrometer (Model 77250) with a 0.75 mm slit width, and Silicon detector (UDT Sensors. Inc. Model 10DP/SB) [84].

Significant visible light absorption activity by carbon-modified p-WO₃ can be noticed, where a maximum monochromatic photocurrent density value of 3.25 μA cm⁻² at 425 nm was observed.

2.4.4. Dependence of Percent Incident Photon Conversion Efficiency (IPCE%) on Wavelength of Light, 𝜆

Maximum IPCE% of 42.7% at wavelength of 260 nm can be observed (Figure 32). However, in the visible region, the CM-p-WO₃ shows incident photon conversion efficiency percent (IPCE%) in the range between 0.9% at 500 nm and 9.9% at 400 nm. These activities in the visible region can be attributed to carbon modification of WO₃ during the flame oxidation process of tungsten metal.

Figure 32 

Incident photon conversion efficiency percent (IPEC %) as a function of wavelength, 𝜆, of light for CM-p-WO3 synthesized by thermal oxidation of W metal at temperature of 900°C for 15 minutes; electrolyte solution of 0.5 M H2SO4; under monochromatic light illumination of Xenon lamp; using a Spectra Physics monochrometer (Model 77250) with a 0.75 mm slit width, and Silicon detector (UDT Sensors. Inc. Model 10DP/SB) [84].

2.4.5. Bandgap Energy

Figure 33 shows the plot for [𝜂(ℎ𝜈)]^1/2 versus ℎ𝜈 for thermally prepared CM-p-WO₃ at oxidation temperature of 900°C for 15 min. From the intercept of the straight line, the bandgap is obtained. It can be observed that the bandgap energy became lower from 2.6 eV for regular n-WO₃ (to 2.42 eV for of CM-p-WO₃). This low bandgap energy can be attributed to carbon incorporation during the flame oxidation process of tungsten metal.

Figure 33 

Plot of [𝜂(ℎ𝜈)]1/2 versus ℎ𝜈 to determine the bandgap of CM-p-WO3 thin-film electrodes synthesized by thermal oxidation of W metal at temperature of 900°C for 15 minutes; electrolyte solution of 0.5 M H2SO4; under monochromatic light illumination of Xenon lamp; using a Spectra Physics monochrometer (Model 77250) with a 0.75 mm slit width, and Silicon detector (UDT Sensors. Inc., Model 10DP/SB) [84].

2.4.6. UV-Vis Spectra on CM-p-WO₃

The UV-Vis absorption spectra show (Figure 34) a significant improvement in the absorbance of light in UV and visible regions for flame-made CM-p-WO₃ films compared to oven-made n-WO₃. The absorption spectra of flame-made p-WO₃ show higher absorption in the visible region with a tail extending to 512 nm which corresponds to bandgap energy value of 2.42 eV.

Figure 34 

UV-Visible spectra of flame-made-CM-p-WO3 and oven-made n-WO3 photoelectrodes synthesized by heating at temperature of 900°C for 15 minutes [84].

The threshold of absorption for oven-made-n-WO₃ is at 475 nm (2.6 eV) which is in agreement with a bandgap value of 2.6 eV for n-WO₃ [74, 75]. The low bandgap energy value of 2.42 eV for CM-p-WO₃ films can be attributed to the carbon incorporation during the flame oxidation process of tungsten metal sheet. This low bandgap energy values for CM-p-WO₃ determined by UV-Vis spectra (Figure 34) agree well with the values obtained from the monochromatic photocurrent densities and the corresponding quantum efficiencies (Figure 33).