1. Introduction
To make ammonia a hydrogen carrier, its dehydrogenation reaction to produce CO
x-free hydrogen is highly anticipated [
1]. Such ammonia decomposition is mildly endothermic, so it requires that temperatures higher than 773 K be applied with the most active Ru-based catalysts to obtain an appropriate H
2 production. In this context, photocatalytic ammonia decomposition that proceed at room temperature has attracted tremendous interest [
2]. However, examples of photocatalytic decomposition of aqueous ammonia solutions evolving only H
2 and N
2 is limited and most of them can use not visible- but UV-light. For instance, Kudo and co-workers reported Ru-loaded ZnS showed the ammonia decomposition activity upon photoirradiation, of which wavelength was less than 350 nm [
3]. Although Fe-doped TiO
2 [
4] and MoS
2/N-doped graphene hybrids [
5] have been described as visible and/or near-IR responsive photocatalyst for ammonia decomposition as exceptional examples, controlling active wavelength ranges was still a not-solved-problem. In this context, a new light absorber that can introduce the photocatalytic systems to convert ammonia to H
2 and N
2 is highly required.
Meanwhile we have developed photocatalytic water splitting systems using carbon nanotubes (CNTs) [
6,
7,
8,
9,
10,
11] and dye-encapsulated carbon nanotubes [
12,
13,
14,
15] as light absorbers. Interestingly, using CNTs as a platform for making coaxial heterojunction, we fabricated coaxal nanohybrids, CNT/C
60 and dye@CNT/C
60, that exhibit efficient charge-separated-state generation followed by electron migration to co-catalyst to generate H
2 upon visible light irradiation. From the viewpoint of visible light absorbers, dye
1 [
13,
16] is quite interesting, because
1 was easily encapsulated into CNT by solution process to obtain
1@CNT, of which physical modification with amphiphilic C
60-dendron gave
1@CNT/C
60-dendron nanohybrid (
Figure 1). Endohedral dye
1 shows the outstanding stability against light irradiation and as well, photosensitizing ability for water splitting, with an apparent quantum yield (AQY) reaching nearly 10% under 510 nm light irradiation [
12]. Based on this background, we decided to investigate ammonia decomposition using dye-encapsulated CNT photocatalysts. In this paper, we demonstrate the decomposition of an aqueous ammonia solution into H
2 and N
2 using
1@CNT/C
60-dendron nanohybrid as a photocatalyst in the presence of RuCl
3 under visible light irradiation at room temperature. To the best of our knowledge, this is the first example of a dye-sensitized ammonia decomposition reaction.
2. Result and Discussions
To explore dye-sensitized ammonia decomposition, we synthesized
1@CNT/C
60-dendron nanohybrid with a dye/CNT/C
60 coaxial heterojunction using dye
1, single-walled carbon nanotubes (CNTs), and amphiphilic C
60-dendron, as previously reported (
Figure 1) [
12]. In a typical run,
1 and CNTs were sonicated in chloroform for 1 h to obtain dye-encapsulated CNT,
1@CNT. The obtained filtrate was repeatedly washed with chloroform to remove any dye remaining on the surface of CNTs. Afterward,
1@CNT and C
60-dendron in water was sonicated for 4 h and centrifuged for 30 min (3000 G) to obtain a water dispersion of
1@CNT/C
60-dendron, of which unadsorbed C
60-dendron in the dispersion was removed by dialysis.
Figure 2a shows photographs of the water dispersions of CNT/C
60-dendron and
1@CNT/C
60-dendron. It is evident that the solution turns a purplish color due to the absorption of the endohedral dye
1.
Figure 2b shows the absorption spectra of CNT/C
60-dendron,
1@CNT/C
60-dendron, and
1. Although the wide absorption range of CNTs from visible to near-infrared masks the absorption of dye
1, the absorption band of
1 is included in the 450 - 600 nm wavelength range.
Figure 3 shows the time course of H
2 and N
2 production using
1@CNT/C
60-dendron as a photocatalyst. Upon visible light irradiation (λ > 422 nm),
1@CNT/C
60-dendron in an aqueous ammonia solution shows both H
2 and N
2 evolution activity, of which rates are 1.8 μmol/h and 0.67 μmol/h, respectively, in the presence of RuCl
3. The production ratio of H
2 to N
2 was close to 3:1, suggesting that the ammonia decomposition reaction (2NH
3 → N
2 + 3H
2) proceeded. The amounts of H
2 and N
2 increased linearly for 12 h, but after that a decline in activity was observed. Measuring the pH of the solution revealed that it decreased from pH 11 at the start of the reaction to pH 10 after the reaction stopped, 24 h, suggesting that the decrease in pH reduced the decomposition rate. In fact, we have observed the pH dependency of the activity where activity was decrease significantly below pH 11. We confirm that this reaction did not proceed in the dark. Hence, this process is photocatalytic reaction. Additionally, in the absence of RuCl
3, the ammonia decomposition reaction did not proceed. We have reported that RuCl
3 act as a hydrogen evolution cocatalyst in the water splitting system containing
1@CNT/C
60-dendron. Therefore, it is considered that the Ru complex acts as a hydrogen production cocatalyst in this reaction system as well.
To identify the light absorber in the photocatalytic system, we measured the action spectra (
Figure 4). Under monochromatic light irradiation (450, 510, 550, and 650 nm), the apparent quantum yield (AQY) values derived from the hydrogen production rates were 0.047% at 450 nm, 0.22% at 510 nm, 0.042% at 550 nm, and 0.0070% at 650 nm, respectively. The variation in activity with the wavelength of the irradiation light closely matches the absorption spectrum of endohedral dye
1. This indicates that the ammonia decomposition reaction is photosensitized by
1. Moreover, compared to the AQY of ZnS
2, 0.21% under 340-nm-light-irradiation,
1@CNT/C
60-dendron exhibits similar AQY (0.22%) under longer wavelength, 510-nm-light-irradiation.
In terms of a plausible mechanism of the dye-sensitized ammonia decomposition (NH
3 → 1/2N
2 + 3/2H
2), it has a ΔG° of +27 kJ/mol (0.28 eV), which corresponds to the energy in the infrared region (4.43 μm). Therefore, visible light absorption of endohedral dye
1 can provide sufficient energy for this reaction. The electrons generated by photoexcitation of
1 are utilized in the H
2 evolution via two-electron reduction of protons (2H
+ → H
2) at the co-catalyst, RuCl
3, and the holes are utilized in the N
2 evolution via six-electron oxidation of ammonia (2NH
3 → N
2 + 6H
+) at the CNT-surface.
Figure 5 shows the energy level diagram for this reaction. Although the CNTs are a chirality mixture, the energy levels of the HOMO and LUMO of a representative (13,8) tube are shown. The HOMO and LUMO of the dye
1 are shown as reported in the literature. The H
2 evolution through photoinduced electron transfer from
1 to C
60 and Ru(III) co-catalyst has already been reported in water splitting, demonstrating sufficient level of
1’s LUMO for ammonia decomposition reaction. The energy level of the photogenerated holes on
1 is +0.80 V, which is deeper than the energy level of the HOMO of the CNT (+0.35 V), suggesting that the holes migrate to the CNT and are consumed in the oxidation reaction of NH
3. Considering the standard electrode potential for the nitrogen production reaction via the oxidation of ammonia (N
2(g) + 6H
+ + 6e
– = 2NH
3 (aq)) is -0.092 V, the HOMO level of the CNT is sufficiently deep to enable nitrogen production via the oxidative decomposition of ammonia.
Figure 6 shows the time course of the ammonia decomposition reaction under simulated sunlight (AM1.5). After 26 hours, the amount of hydrogen produced reached 64.7 μmol, and the amount of nitrogen produced reached 15.5 μmol. The STH (solar-to-hydrogen) efficiency calculated from these results was 0.0011% for the ammonia decomposition reaction. The pH at the start of the measurement was 11.2, and after 26 hours, when hydrogen production stopped, the pH was 10.4. The result indicated that the ammonia decomposition reaction was stopped owing to the lowering pH of the solution. But the pH of the solution was adjusted to 11.2 by adding KOH solution and irradiation was resumed, hydrogen and nitrogen production resumed, where the ratio of N
2 and H
2 evolution rate, 0.90 and 0.30 μmol/h, respectively, was perfectly 3:1.