Efficient Photocathode of an Ultrathin Organic p-n Bilayer Comprising p-Type Zinc Phthalocyanine and n-Type Fullerene for Hydrogen Peroxide Production

Abstract

Hydrogen peroxide (H₂O₂) is a clean and environmentally friendly oxidant. At present, as an alternative to the conventional industrial procedure, namely, the anthraquinone method, a clean H₂O₂ production method is desired. The construction of an artificial photosynthetic system in which H₂O₂ can ideally be prepared from water and dioxygen (O₂) is a promising approach. In such a system, an organic p-n bilayer comprising zinc phthalocyanine (ZnPc, p-type) and fullerene (C₆₀, n-type) acts as a photocathode capable of O₂ reduction to H₂O₂, where loading gold (Au) onto the C₆₀ surface is necessary to achieve the corresponding reaction. However, the enhancement of the photocathodic activity of the organic p-n bilayer for H₂O₂ formation remains a critical issue. In this study, the effect of the thickness of an organo-bilayer (organo-photocathode) on photocathodic activity for H₂O₂ production was investigated. When both ZnPc and C₆₀ were thin (approximately 10 nm each in thickness), the photocathodic activity of the ZnPc/C₆₀ organo-photocathode was approximately 3.4 times greater than that of the thick ZnPc/C₆₀ bilayer (i.e., ZnPc = ca. 70 nm and C₆₀ = ca. 120 nm). The thin ZnPc/C₆₀ bilayer exhibited a built-in potential at the p-n interface, where efficient charge separation occurs, resulting in a high concentration of electrons available for O₂ reduction.

1. Introduction

Artificial photosynthesis reactions have been actively investigated to obtain fuels and chemicals such as molecular hydrogen [1,2,3,4,5], hydrogen peroxide (H₂O₂) [6,7,8,9], and some products from carbon dioxide (CO₂) [10,11,12,13]. H₂O₂ is a clean and environmentally friendly oxidizing agent, which is industrially produced using the anthraquinone method [14]. However, the industrial manufacturing method of H₂O₂ involves complex and multistep processes. In addition, significant energy consumption and organic waste emissions are major challenges limiting the commercialization of H₂O₂ [15]. Therefore, a clean and energy-saving H₂O₂ production method is urgently required to replace the conventional method.

H₂O₂ can be photocatalytically produced from water (H₂O) and oxygen (O₂) via the following redox reaction (Equation (1)):

H₂O + 1/2 O₂ ⟶ H₂O₂, ∆G° = +117 kJ·mol⁻¹

(1)

Such an uphill reaction, which corresponds to artificial photosynthesis, can be expected to proceed via photocatalysis. Many studies on photocatalytic H₂O₂ production have been conducted [6,7,8,9,16].

We have a track record of constructing an artificial photosynthesis system for water splitting featuring an organic p-n bilayer [17]. The water splitting system by Honda and Fujishima [18], comprising titanium dioxide (TiO₂) photoanode and Pt counter, is recognized to be a typical instance, but the overall splitting of water into O₂ and H₂ occurs only when a voltage bias is applied to the system. We reported the water splitting system composed of the TiO₂ photoanode and organo-photocathode, where bias-free water splitting was found to occur (see Scheme S1). In the system featuring a TiO₂ photoanode, our instance is an achievement that had not been discovered up to that time. In the Honda–Fujishima system, the oxidizing and reducing power generated at the TiO₂ photoanode is available to the water splitting into O₂ and H₂, while our system can utilize oxidizing and reducing power separately photogenerated at TiO₂ and organo-bilayer, respectively. In the latter, the electrons photogenerated at TiO₂ do not move to Pt counter for H₂ evolution (in the former), but to the holes remaining in the p-type layer of the organic p-n bilayer. Thus, it is summarized that, compared with the Honda–Fujishima system, the reason for bias-free water splitting in our system is kinetically favorable for the overall decomposition of water.

Based on this knowledge, we reported a photocatalytic system in which both a photocathode based on an organic p-n bilayer, comprising zinc phthalocyanine (ZnPc, p-type) and fullerene (C₆₀, n-type), and a photoanode based on co-catalyst-loaded bismuth vanadate (BiVO₄) were configured for H₂O₂ production (Scheme S1) [19]. In summary, the results of our previous study confirmed the production of H₂O₂ without the use of bias voltage in the photocatalytic system; however, the resulting efficiency was very low (i.e., <0.01% in maximum), and the rate-limiting step was the reduction of O₂ to H₂O₂ at the photocathode in which the gold (Au)-co-catalyst was loaded at the C₆₀ surface of the organo-bilayer (denoted as ZnPc/C₆₀-Au). The activation of the ZnPc/C₆₀-Au photocathode is a critical issue to improve our previous system.

In the present study, the photoelectrochemical reduction of O₂ to H₂O₂ with ZnPc/C₆₀-Au was investigated in terms of the thickness of the organo-bilayer employed (Scheme 1). In our previous study on H₂O₂ production [19], the thickness of the ZnPc/C₆₀ bilayer, which is recognized as optimal for reducing H⁺ to H₂ [17,20,21], was used without any modification (i.e., ZnPc = ca. 70 nm and C₆₀ = ca. 120 nm). In this study, the validity of this approach was examined, along with the novel optimization of the organo-bilayer used as the photocathode.

2. Experimental

An indium tin oxide (ITO)-coated glass substrate (AGC Inc; sheet resistance: 8 Ω·cm⁻²; transmittance: >85%; ITO thickness: 174 nm) was used as the electrode substrate. ZnPc (TCI) was purified via sublimation before use (sublimation conditions: temperature, 530 °C; pressure, ~2 × 10⁻² Pa) [19,22]. Pure C₆₀ (>99.5%, TCI) was used as received. The other reagents used in this study were of extra-pure grade.

The ZnPc/C₆₀ organo-bilayer was fabricated on an ITO-coated glass substrate via vapor deposition (ULVAC KIKO Inc., Saito, Miyazaki, Japan, pressure: <1.0 × 10⁻³ Pa; deposition speed: ~0.03 nm·s⁻¹) [17,21]. ZnPc was first coated on the ITO substrate, after which C₆₀ was placed on top of the ZnPc layer. Au was photoelectrochemically deposited onto the C₆₀ surface of the organo-bilayer under potentiostatic conditions, where a single-compartment cell was composed of an organo-bilayer-modified working electrode (effective area: 1 cm²), a platinum counter electrode, and an Ag/AgCl (immersed in a saturated KCl solution; denoted as Ag/AgCl (sat.)) reference electrode in an acid solution (pH = 2) containing 5.0 × 10⁻⁴ mol·dm⁻³ HAuCl₄·4H₂O under an aerobic atmosphere. A potential of −0.2 V (vs. Ag/AgCl (sat.)) was applied to the modified ITO substrate for Au deposition. Photoelectrodeposition was performed using a potentiostat (HA-301, Hokuto Denko, Meguro-ku, Tokyo, Japan) equipped with a function generator (HB-104, Hokuto Denko, Meguro-ku, Tokyo, Japan), a Coulomb meter (HF-201, Hokuto Denko, Meguro-ku, Tokyo, Japan), and an X–Y recorder under illumination. The amount of Au deposited was controlled by monitoring the amount of charge that passed during photoelectrodeposition (i.e., 0.04 C). The aforementioned electrochemical setup was used to measure voltammograms and photocurrents to obtain an action spectrum under potentiostatic conditions, but using a phosphoric acid solution (pH = 2) as the electrolyte. The light intensity was measured using a power meter (type 3A, Ophir Japan, Saitama, Japan). A halogen lamp was utilized for irradiating the photocathode with a typical intensity of approximately 100 mW·cm⁻². For the irradiation in the action spectrum measurements, the light source was combined with a monochromator (S-10, Soma Optics, Hinode, Tokyo, Japan) to generate monochromatic light. The electrolysis experiment for reducing O₂ to H₂O₂ was conducted using a twin-compartment cell separated by a salt bridge (Scheme S2). The salt bridge preparation procedure is described elsewhere [17]. Calculation for the incident photon-to-current conversion efficiency (IPCE) and Faradaic efficiency (FE) are described in the Supplementary Materials (SM).

The crystal system of Au was analyzed by X-ray diffraction (XRD; SmartLab 9kW, Rigaku, Akishima, Tokyo, Japan). The surface of ZnPc/C₆₀−Au was observed via scanning electron microscopy (SEM; JSM-7000F, JEOL, Akishima, Tokyo, Japan). The absorption spectra of the organo-bilayer, single layers, and the peroxo–titanium complex solution (vide infra) were measured using a spectrophotometer (Lambda 35, PerkinElmer, Shelton, CT, USA).

The thickness of the organic p-n bilayer was determined by visible-light spectroscopy. The absorption spectra of both ZnPc [23] and C₆₀ [24] obtained were identical to those previously reported, and their absorption coefficients indicated the thickness of the films used (Table S1). Since the additivity of the absorption coefficient should be preserved in the visible-light absorption spectrum of the bilayer, the two unknown parameters for each thickness were estimated by solving simultaneous equations for the absorbance at two different wavelengths (see SM for the simultaneous equations). Such estimation procedures for the thickness of organic p-n bilayers are described in the literature [22,25].

Quantification of H₂O₂ was performed using the so-called titanium sulfate method [26]: a 30% titanium(IV) sulfate solution (0.5 mL) was added to the electrolyte solution (2.5 mL) at the reduction site (see Scheme S2); when H₂O₂ was present in the solution, a peroxo–titanium complex was formed from a Ti⁴⁺ ion and an H₂O₂ molecule according to the following equation (Equation (2)):

Ti⁴⁺ + H₂O₂ + 2H₂O → H₂TiO₄ + 4H⁺

(2)

By measuring the absorption spectrum of the resulting solution, the amount of H₂O₂ formed was determined from the absorbance of the peroxo–titanium complex at 410 nm (molar absorption coefficient: 674 M⁻¹·cm⁻¹).

3. Results and Discussion

A typical X-ray diffraction pattern of Au particles is shown in Figure S1, in which Au is loaded on the C₆₀ surface of the ZnPc/C₆₀ organo-bilayer. The resulting pattern represents the formation of cubic Au, which is consistent with previous research [27]. The scanning electron microscopy image of the Au particles indicates the formation of fine Au particles in the order of <30 nm (Figure S2).

Voltammetric measurements were performed with regard to the thickness of the C₆₀ and ZnPc layers. Figure 1a–c show the cyclic voltammograms (CVs) of the ZnPc/C₆₀-Au photocathode under an O₂ atmosphere, where the C₆₀ layer thickness changed while the ZnPc layer thickness remained constant (i.e., ZnPc = ca. 70 nm). In the dark, almost no electrochemical response was confirmed. However, when ZnPc/C₆₀-Au was irradiated, photocathodic currents were generated because of the reduction of O₂; in addition, the photocurrents increased with thinning of the C₆₀ layer. Furthermore, the ZnPc thickness was varied while keeping the thickness of the thin C₆₀ layer (i.e., C₆₀ = 6~7 nm) constant. Similarly to the change with the C₆₀ thickness, the thinner the ZnPc layer, the more enhanced the photocurrent (Figure 1d,e). These voltammetric results demonstrate efficient photocurrent generation at ZnPc/C₆₀-Au when thin ZnPc and C₆₀ layers are used. By comparing the thinner layers (Figure 1e) with the thicker layers (Figure 1a), the photocurrent generated at −0.2 V (vs. Ag/AgCl (sat.)) was ca. 3~4 times higher with the former than with the latter. The photocathode shown in Figure 1e was subjected to repeated scans and demonstrated stable performance even after 10 scans (Figure 2).

Using thin organo-bilayer that showed the most efficient characteristics, some types of ZnPc/C₆₀-Au were prepared by changing the amount of charge passed during Au deposition. Those CVs for the reduction of O₂ are shown in Figure 3. It can be considered that the amount of Au deposited corresponds to the charge amount (see SM). CV characteristics were improved with increasing the deposited amount of Au; however, when the Au amount was too large, the CV characteristics deteriorated. Thus, the conditions of Figure 1e and Figure 2, where the thin organo-bilayers and constant amount of charge during Au deposition (i.e., 0.04 C) were employed, were confirmed to be optimum.

To analyze the results of Figure 3, SEM images were observed with three types of Au loaded on the organo-bilayer (Figure S3). Comparing Figure S3a with Figure S3b, the particle size appears to increase with increasing charge amount. Figure S3c displays that the number of particles increase while maintaining the particle size with Figure S3b. In Figure 3, the surface concentration of electrons reaching the C₆₀ surface should be constant because organo-bilayers of constant thickness were employed. As the number of Au particles increases, the number of active sites increases; however, the electron concentration per unit particle of Au decreases, and consequently, the activity of Au catalyst is diminished. Thus, it reveals that the deposition conditions of Figure S3b are optimum in the number and size of Au particles, efficiently highlighting the catalytic activity of Au.

Electrolysis for H₂O₂ production from O₂ was investigated with some types of organo-photocathodes. The typical data are summarized in

Table 1. Electrolysis data for H₂O₂ production by ZnPc/C₆₀-Au under illumination ^a.

	ZnPc/nm	C60/nm	Amount of H2O2/µmol
Entry 1	75	116	0.96
Entry 2	84	2.8	2.12
Entry 3	11	3.5	3.22

^a Applied potential: −0.1 V vs. Ag/AgCl (sat.); electrolyte solution: aqueous H₃PO₄ solution (pH = 2); reaction time: 3 h; light intensity: 70 mW·cm⁻². Irradiation was performed from the back of the ITO-coated face. The deposition of Au onto the organo-bilayer was performed in the same way for each photocathode (see Experimental).

. Supporting the voltammetric results, the amount of H₂O₂ produced increased with decreasing thickness of the organo-bilayer. The FE value for the formation of H₂O₂ was always >90% (see SM for the FE calculation), where the H₂O₂ amount in entry 3 was ca. 3.4 times larger than that in entry 1.

Action spectrum measurements were performed for the photocurrents generated at ZnPc/C₆₀-Au (Figure 4). Even when a thin ZnPc/C₆₀ bilayer was used as photocathode, photocurrents were generated across the entire visible-light region. By comparing the resulting action spectrum with the single-layer ZnPc and C₆₀ absorption spectra, the photocurrents originated from the absorption of both ZnPc and C₆₀: in other words, the photocurrent generation around at 460–490 nm, where the absorption of ZnPc is almost absent, can be induced by the sole absorption of C₆₀; photocurrents generated at wavelengths of >530 nm may have originated from ZnPc absorption.

To gain insight into the thinner ZnPc/C₆₀ organo-bilayer, rest potential was measured using single layers of ZnPc and C₆₀. The results are presented in

Table 2. Rest potential measured using single-layer ZnPc and C₆₀ ^a.

	ZnPc/nm	C60/nm	Rest Potential/V vs. Ag/AgCl (sat.)
Entry 1	-	142	–0.01
Entry 2	-	9.6	–0.06
Entry 3	86	-	0.44
Entry 4	9.6	-	0.43

^a Electrolyte solution: aqueous H₃PO₄ solution (pH = 2); light intensity: 100 mW·cm⁻². Single-layer ZnPc and C₆₀ were prepared on the ITO substrate, similar to the ZnPc/C₆₀ bilayer. This measurement was performed under an Ar atmosphere. Irradiation was performed from the back of the ITO−coated face.

. By comparing the rest potential values of n-type C₆₀ and p-type ZnPc, the latter exhibited a more positive potential than the former; furthermore, the magnitude of the potential difference between the ZnPc and C₆₀ layers was independent of their thicknesses (i.e., ca. 0.5 V). These results demonstrate that built-in potential can be formed at the p-n interface even with a thin organo-bilayer, supporting the finding that series of photophysical events occur similar to photovoltaic cells of organic p-n bilayers [28]. Therefore, the photoelectrochemical reduction of O₂ to H₂O₂ at ZnPc/C₆₀-Au can be explained as follows. First, visible-light absorption occurs in both the ZnPc and C₆₀ layers, and consequently, excitons are formed within both bilayers. The excitons attain at the p-n interface through excitation energy transfer, followed by the charge separation of excitons into electrons and holes at the hetero-interface. The photogenerated electrons move to the Au-coated C₆₀ surface, thus resulting in the generation of reducing power for O₂ reduction. The reducing power induced the reduction of H⁺ to H₂, as previously reported by our group [29]. The potential of O₂/H₂O₂ (+0.38 V vs. Ag/AgCl, at pH = 2) is more positive than that of H⁺/H₂ (−0.32 V vs., Ag/AgCl, at pH = 2); thus, the O₂ reduction occurring at the ZnPc/C₆₀ bilayer is thermodynamically reasonable. The kinetics of O₂ reduction should be proportional to the surface concentration of O₂ and the number of electrons available for H₂O₂ formation on the Au co-catalyst. Considering that the O₂ concentration was saturated under these experimental conditions, the photocathodic activity of ZnPc/C₆₀-Au for producing H₂O₂ can be associated with the concentration of electrons on the Au co–catalyst. When employing the thin bilayer as a photocathode, the transport distance of electrons from the p-n interface to Au can be shortened, thus effectively enhancing the electron concentration at Au. This may also relate to the reduced carrier recombination between electrons and holes.

4. Conclusions

This study investigated the effect of the thickness of a ZnPc/C₆₀ bilayer for the development of efficient photocathodes capable of H₂O₂ formation from O₂. The results demonstrate that a thin ZnPc/C₆₀ bilayer acts as a superior photocathode when Au loaded on the C₆₀ surface is employed as a co-catalyst. The thin bilayer reduced O₂ to H₂O₂ by approximately 3.4 times more than the thick bilayer used in our previous study. This result can be attributed to the high concentration of electrons available for H₂O₂ formation on the Au co-catalyst, which is achieved through a series of efficient photophysical processes that occur within the thin bilayer.