Abstract

In this work, we report for the first time the photocatalytic activity of Co3O4 nanostructures for the reduction of aqueous CO2 to methanol (MeOH). This could be considered a simple example of artificial photosynthesis. The photocatalysis experiments were developed under simulated solar light of 100 mW/cm2 and without using any sacrificial agent. To carry out this study, nanostructured mixed valence cobalt oxide (Co3O4) powders, with porous nanoparticle aggregates of different morphologies, have been prepared by two synthesis methods. The characterization of structural (PXRD, XPS, SEM, and TEM) and optical (UV-vis-NIR, Raman, and FT-IR) properties, magnetization curves, and surface area (BET) was accomplished.

1. Introduction

Global warming is the major issue that humanity has to face in the next 50 years. Some effects that could be devastating for humankind are the increase of global temperature, rise of sea level, extreme droughts, devastating floods, destructive storms, among others [15]. The reduction of gases that contribute to the Earth’s surface warming can solve some problems that we are experiencing right now as the weather changes and the global temperature is rising. A great number of authors claim that the main cause of global warming is CO2 [69]. Almost all the excess of this gas in the atmosphere is produced by the combustion of mineral fuels, like gasoline or coal. Many efforts have been done in the last years aiming to reduce CO2 emissions to the atmosphere: imposition of taxes for CO2 emissions, CO2 sequestration using different methods [1020], or the conversion of this gas using different techniques to obtain value added products, e.g., fuels [2133]. In order to reduce CO2 quantity in the atmosphere and produce fuels using this gas and water, different materials and methods have been applied: metal complexes [21], microalgae [28, 33], syngas procedure [24], polymers [31], electrochemical transformations [26, 32], photothermal and thermochemical conversions [23, 25], semiconductors [27, 30], and several others.

One of the most attractive ways of producing fuels from CO2 and water is through photocatalytic semiconductor-mediated solar energy conversion. This fascinating fundamental and technological problem has challenged the scientific community for many years. A great variety of semiconductors, sometimes combined with metal nanostructures, carbon-based materials, and/or organometallic molecules have been tested with the mentioned purpose, and important progresses have been reached [34]. However, typical semiconductors with an adequate band gap for solar energy absorption (Si, CdSe, CdTe, GaAs, GaP, PbS, SbS, among others) are unstable in contact with electrolytes, and those with higher stability (TiO2, ZnO, SnO, ZnS, inter alia) cannot absorb a huge component of solar radiation [35]. A group of semiconductors, such as CuO, Fe2O3, or BiVO4, display both a wide spectral range of solar light absorption as well as stability when in contact with aqueous solutions of electrolytes. This last group of semiconductors has been largely studied in photocatalytic water splitting and/or CO2 photoreduction [36]. Another interesting semiconductor is Co3O4; it is stable in aqueous suspensions and oxygen-rich atmospheres and thus can be prepared and easily manipulated under usual laboratory conditions. This material has unique optical properties because it shows several band gaps in different regions of the electromagnetic spectrum (0.7–4.4 eV) [3739]. Because it absorbs light in the UV, visible, and infrared regions, it can be a good candidate to carry out photocatalytic reactions when Co3O4 interacts with light. In addition, the preparation of different sizes and shapes of this material could potentially enhance the photocatalytic activity, since the optical properties may change and the area and surface defects of the material can increase with regard to its bulk counterpart. Despite the ad hoc properties of this material, only two studies have been reported regarding nanostructured Co3O4 as an active photocatalyst for CO2 reduction [40, 41]. Formic acid and formaldehyde were considered as major products for CO2 reduction assisted by randomly oriented Co3O4 nanocrystals [40], while formate was the only product using highly texturized {121} nanostructured Co3O4 [41]. However, Mendoza et al. [40] only used a small wavelength range in the visible spectrum (510 to 620 nm), neglecting the much wider absorption range (from UV to NIR) of this material. In the work of Huang et al. [41], the CO2 photocatalytic reduction was assisted by an external electric potential in a photoelectrochemical cell configuration, adding additional energy to that of the radiation. These reasons make it interesting to explore the possibilities that can be offered by an abundant, inexpensive, and stable material such as Co3O4 when it is only excited with solar light irradiation. Thus, here we report the properties of Co3O4. Herein, two different morphologies of this oxide were prepared based on synthesis procedures previously published. They were used as a heterogeneous catalyst for photochemical reactions with CO2 and H2O under artificial solar light irradiation, in order to evaluate their catalytic capabilities to generate hydrocarbons. In this work, the photocatalytic activity of Co3O4 (nanoparticles and porous structure) was evaluated for CO2 reduction in water and under visible light. The photocatalytic reactions were carried out in aqueous dispersions without the use of sacrificial agents, under normal conditions, i.e., room temperature and atmospheric pressure.

2. Experimental

2.1. Preparation of Co3O4 Nanoparticles (Co3O4-NPs) Photocatalyst

The synthesis of these nanoparticles was accomplished using an analogous method to that reported by Xie et al. [42]. Briefly, 1.19 g of CoCl2⋅6H2O (Mallinckrodt Pharmaceuticals, 99.63%) was added to 30 mL of ethylene glycol (EG) (Analytyka, 99.1%), and the suspension was stirred and gradually heated to 80°C. Then, 100 mL of a 0.2 M solution of Na2CO3 (Reproquifin, 99% min.) in deionized water (15 MΩ cm) was added dropwise and the mixture was further aged for 1 h under argon gas (Praxair, grade 4.8, extra pure, 99.998% min.). The purple solid obtained was filtered, washed several times with deionized water until neutralization, and dried overnight at 100°C under vacuum. The obtained purple powder was characterized using powder X-ray diffraction (PXRD) and infrared (IR) spectroscopy. As shown in Figure S1 it corresponds to cobalt hydroxycarbonate [43]. Finally, the solid was calcined for 4 h at 450°C, obtaining a black powder.

2.2. Preparation of a Porous and Nanostructured Co3O4 (Co3O4-P-NS) Photocatalyst

This sort of cobalt (II, III) oxide was synthesized using the gel-combustion method, similarly to that previously reported by Makhlouf et al. [44]. In our case, the quantities used were 1.46 g of Co(NO3)2⋅6H2O (Fermont, 99.6%) and 2.5 g of glycine (Sigma-Aldrich, 99%). Both powders were dissolved in 100 mL of deionized water under vigorous stirring. 3 mL of concentrated HNO3 (J. T. Baker, 65.8%) was added when both powders were completely dissolved, and the heating of this reaction mixture started at 300°C. A purple gel was formed when almost all the water was evaporated, and then the temperature was increased up to 400°C. Finally, the combustion of the organic material took place and a black solid was obtained. The PXRD pattern is shown in Figure S2. After that, the solid was calcined for 4 h at 700°C in order to obtain the final material.

2.3. Characterization Techniques of the Photocatalysts

PXRD patterns were taken on a Bruker D2-Phaser diffractometer equipment using Cu Kα radiation (10 mA, 30 kV, ) and an integration time of 0.5 s. Raman spectra were recorded with an EZRaman-N (Enwave Optronics Inc.) Raman analyzer coupled to a Leica DM300 microscope with a 40x objective, using an excitation laser source of 532 (∼30 mW) or 905 (∼160 mW) nm wavelengths. The infrared spectra were collected on a PerkinElmer Spotlight 400 FT-IR-FIR spectrometer with a Universal ATR sampling accessory. UV-visible-NIR electronic absorption spectra (under a diffuse reflectance regime) were recorded on a Varian Cary 5E, and the Kubelka-Munk function was computed. The transmission electron microscopy (TEM) micrograph and the selected area electron diffraction (SAED) pattern were obtained using a JEOL JEM-2012 microscope operating at a voltage of 200 kV. The scanning electron microscopy (SEM) images were obtained using a field-emission scanning electron microscope, JEOL JSM-5900-LV, operating at 20 kV. This microscope is equipped with an Oxford-ISIS for elemental analysis using energy-dispersive spectroscopy (EDS). The magnetic hysteresis loops (M–H) were measured in a vibrating sample magnetometer (VSM-LDJ 9600, 30,000 Oe) at room temperature with a standard magnetization curve routine. BET measurements were carried out by N2 at −196°C in a Quantacore Nova 2000. Before the physisorption, the samples were degassed for 12 h at 150°C. XPS was performed in a spectrometer equipped with an Al Kα X-ray source of 1486.6 eV. The analyzer energy was 23.5 eV and the base pressure was about 4.5 × 10−8 Pa.

2.4. Light Irradiation of CO2 Bubbled in Water in the Presence of a Photocatalyst

In a general procedure, 20 mL of deionized water in a Schlenk vessel were bubbled for 1 h with gaseous CO2 (Praxair, grade 3.0, 99.9%) in order to saturate water with CO2. Then, 4.15 × 10−2 mmol of a photocatalyst (Co3O4-NPs or Co3O4-P-NS) was suspended in the CO2 solution. A stirring bar was introduced, the glass vessel was sealed, and the irradiation was carried out with a 150 W arc xenon lamp (Osram XBO 150 W/CR OFR), which roughly simulates the solar spectrum. The lamp was located at a distance from the vessel so that 100 mW/cm2 of intensity was reached; a calibrated silicon photodiode (Thorlabs SM05PD2B) was used for that purpose. The experiments were accomplished at room temperature. After 6 hours of irradiation, the produced gases were collected in a Tedlar® bag and the resultant gaseous mixture was analyzed by gas chromatography (GC) in an Agilent 6890 series gas chromatograph with a 30-meter PerkinElmer PE-5 column ((5%-phenyl)-methylpolysiloxane). The temperature of the injector was 100°C, the oven was at 100°C, and the FID detector was at 150°C. Note that all the dispersions were prepared in deionized water. All the experiments were performed at least twice.

3. Results and Discussion

3.1. Characterization of Co3O4-NPs and Co3O4-P-NS

The powder diffraction pattern of the Co3O4-NP sample, displayed in Figure 1(a), agrees well with that of Co3O4 (PDF no. 42-1467). The synthesized nanoparticles have cubic symmetry and its space group is Fd-3m. In this material, the lattice is formed with mixed-valence cobalt: Co2+ (33%) and Co3+ (66%). It has a spinel structure where the atoms of Co2+ are in a tetrahedral coordination and the atoms of Co3+ are in an octahedral coordination with oxygen atoms. The average crystallite size of Co3O4-NPs estimated from the full width at half maximum of the (311) diffraction peak by applying the Scherrer equation is about 15 nm. The PXRD pattern of the prepared particles indicates the formation of the single spinel phase without any observable traces of cobalt hydroxide or cobalt monoxide. When Co3O4-P-NS was synthesized by combustion, the dark grey solid obtained was also analyzed by PXRD. A mixture of Co3O4, CoO, and Co was identified in the corresponding diffractogram (Figure S2). This is due to the presence of reducing and oxidizing species in the combustion reaction. The average crystallite size, obtained by the same method used for the Co3O4-NPs, was 21 nm. After the solid was calcined for 4 h at 700°C, as can be observed in Figure 1(b), the diffractogram only shows the Co3O4 cubic phase (PDF no. 42-1467) and its space group is Fd-3m. Due to the calcination process, the crystallite size increased to 45 nm. When the sample was annealed, the presence of Co and CoO was not detected, showing the formation of the single spinel phase. It is important to mention that the shape of the baseline in both diffractograms is due to the mathematical subtraction of the background, in order to remove the distortion produced by the fluorescence generated in the interaction of the Cu Kα radiation with the Co atoms.

The electronic absorption spectrum of Co3O4-NPs in the UV-visible-NIR region, using the diffuse reflectance mode displayed in Figure 2(a), shows different bands. Using Tauc’s plot, different band gap energies were calculated. In the present study, the best fit of versus photon energy was obtained for for Co3O4-NPs, which corresponds to direct band gap transitions. Then, the obtained gap energy values for Co3O4-NPs were 0.78, 1.55, and 2.19 eV (see Figure 2(b)). The transition at 0.78 eV was recently attributed to a direct dipole-dipole forbidden d-d transition between the tetrahedral site of Co2+ cations [38]. The gaps at 1.55 and 2.19 eV are direct allowed transitions. These transitions are assigned to a charge transfer O2−➔Co3+ (the former) and a transition O2−➔Co2+ (the latter) [4549]. The corresponding spectrum of Co3O4-P-NS in Figure 2(c) shows, as in the case of Co3O4-NPs, different direct band gap transitions. Using Tauc’s plot, the calculated band gap energies for Co3O4-P-NS (Figure 2(d)) were 1.96, 1.48, and 0.76 eV. The band gap values found for the Co3O4-NP sample are slightly higher than those of the Co3O4-P-NS. This fact could be due to a small quantum confinement effect in the Co3O4-NPs [50, 51]; this is in accordance with the smaller crystallite size observed in the PXRD analysis. In addition, the colors of the two powders are different (Figure S5): the nanoparticles are black and the porous material is dark grey.

A representative SEM micrograph of the Co3O4-NP sample is shown in Figure 3(a). Considering the crystallite size obtained for this sample (15 nm), the observed microstructures in Figure 3(a) must be composed of “nanoblocks.” In the SEM image of Co3O4-P-NS (Figure 3(b)), large nanostructured arrays (crystallite size about 45 nm) with different pore sizes are observed. In this sample, the size of the observed pores are in the order of microns; this can be due to the high volume of gases produced during the combustion reaction [44].

It is clearly revealed in the TEM micrograph of Figure 4 that the microstructures observed in Figure 3(a) are in fact Co3O4-NP porous agglomerates with a size in the order of submicrons. The corresponding pore sizes are in the range of 5–15 nm. These nanoparticles are agglomerated and have a rectangular shape and a mean size of 15.1 ± 3.6 nm (Figure S12). This value fairly matches the crystallite size found from the PXRD analysis.

For more details about the composition and structural characterization of the materials, see the supplementary materials.

3.2. Photocatalytic CO2 Reduction

After characterization, the materials were dispersed in CO2-saturated water and irradiated. The purpose of our work was to evaluate the catalytic activity of the catalysts under normal conditions of temperature, pressure, and quantity of CO2 determined by the solubility of the gas in water at atmospheric pressure. The reader can find elsewhere [52] more information about the different equilibria related to the solubility of CO2 in water at different pH values; thus, the starting pH of the reaction solution was between 4.5 and 5. It is important to note that no sacrificial agent was used in the reaction. The collected gases were analyzed with a gas-chromatograph with a FID detector. Figure 5(a) shows the sharp band of the gas obtained after irradiation of the reaction suspension containing either Co3O4-NPs or Co3O4-P-NS. These peaks matched well with those corresponding to the methanol standard. As can be seen, the gas obtained with the two materials is the same, since the signal appears at the same retention time as that of methanol. Though this material has been widely used for water splitting and water oxidation, this is the first time MeOH production is reported when Co3O4 is used as a photocatalyst in a system as described herein. It is noteworthy that the reaction product obtained here is different from those reported in the two previous works using Co3O4 as a photocatalyst, which at the same time are different from each other [40, 41]. This apparent discrepancy may be explained, since different excitation conditions were used in every report: (i) In the work of Mendoza et al. [40], the irradiation was carried out using a small window of the visible spectrum (510 to 620 nm); the light intensity in the reaction medium was not reported. (ii) In the report of Huang et al. [41], a Xe lamp was used to irradiate the reaction system at an intensity of 9 mW/cm2 and the photoelectrode was polarized at 0.9 V (vs. SCE) during the experiments. (iii) In our case, the intensity of the irradiation in the reaction system was 100 mW/cm2, coming from a Xe arc lamp typically used in solar simulators. These different excitation conditions, together with slightly different light absorption capabilities, cause different concentrations of photogenerated carriers, different Fermi level positions, and consequently, different potential distributions across the semiconductor/electrolyte interface [53]. Thus, different reactions could be expected. The exposed planes of the materials could make a difference too, with regard to the photocatalytic activity. It is well known that the plane (110) mainly contains Co3+ ions in an octahedral coordination that are good for the oxidation of, e.g., CO and alcohols [42, 54, 55]. Also, the planes (001) and (111) mainly contain Co2+ ions in a tetrahedral coordination that are almost inactive for the oxidation [5658]. It is worth saying that the plane (111) is the one that we found in the HR-TEM images. Other possible reasons, like different relative band positions, different electrolyte compositions affecting the redox equilibrium, and different Co3O4 crystalline textures affecting the corresponding surface activities, cannot be discarded.

For the products obtained in this work, the signals obtained were used by means of qualitative detection, not for quantitative purposes, due to the fact that very small amounts of them were obtained in these studies [59, 60]. Attempts to fully characterize MeOH in the remnant solution were unsuccessful due to the lower amounts of gas produced. The fact that methanol was detected as a gas and none was detected in the remnant solution is not surprising. This is probably due to the fact that the methanol produced in the reaction evaporated, since the reaction solution reaches 35°C upon irradiation and atmospheric pressure in our room conditions is 585 mm Hg. A possible general mechanism explaining the photoreduction of CO2 and the production of methanol can be seen in Figure 5(b). As it is well known, the reduction of CO2 to methanol needs protons and electrons. In our experimental conditions, there are two possible proton sources: (1) the oxidation of water with Co3O4 and the production of H+ [61] and (2) the dissociation equilibrium of H2CO3/HCO3 [62, 63]. For the initial pH values encountered in the solution reaction (4.5–5), the expected species are CO2(aq), H2CO3, and a small quantity of HCO3 in equilibrium with the carbonic acid. In the literature, it is reported that the concentration of CO2(aq) in a solution is approximately 500 times the concentration of H2CO3 in standard reaction conditions [63]. As mentioned above, the pH determines the species you can find and their concentration in the solution. That is, the concentration of H2CO3 in a solution is from 23 to 71 times greater than the concentration of HCO3; consequently, the concentration of CO2(aq) is from 11,500 to 35,500 times larger than the concentration of HCO3 (the calculations are developed in Appendix A of the supporting information). Considering the different carbon-containing species present in the reaction medium and their corresponding potential reduction (see Figure 6), we suggest that at least two pathways in the formation of methanol are possible. One path is where CO2(aq) is combined with protons and photogenerated electrons to produce methanol, and the other path is where carbonic acid is combined with protons and reduced to give MeOH. In terms of the concentration of the two species, the former pathway is favored because the concentration of CO2(aq) is 500 higher than that of the H2CO3 in solution; however, in thermodynamic terms, the second path is preferred because the reduction potential of the H2CO3/CH3OH pair is more positive than that of CO2/CH3OH (Figure 6). We do not have enough experimental evidences to establish which of these two possible mechanisms has a higher contribution.

For the Co3O4-NPs, the highest direct gap energy transition is 2.19 eV, while that of Co3O4-P-NS is 1.96 eV. In Figure 6, the redox potentials of different possible products are shown. Taking into account the band gap obtained for our materials, methanol production, and possible water oxidation [61], it is possible to assume that the conduction band could be more reductive than the redox potential of the CO2/CH3OH pair and the valence band might be more oxidizing than the oxidation potential of the water.

By preparing two different morphologies of Co3O4, it was expected to obtain different products, quantities, or both. Nevertheless, no significant differences were observed between the two morphologies regarding their photocatalytic behavior. It is important to mention that CO32− anions are not formed in the pH range of our experiments. At the end of the experiment, the pH of the remnant solution was 7, which means that the concentration of dissolved CO2 (or the H2CO3/HCO3 pair) significantly decreased in the solution.

Once the reaction was completed, the materials were collected with a centrifuge and dried in a vacuum oven at 80°C overnight. After that, the PXRD of the two materials were recorded again (Figure S16(a) and S16(b)). When the average size of a crystallite was calculated for Co3O4-NPs, it increased from 19 nm to 21 nm. For Co3O4-P-NS, the crystallite size increased from 45 nm to 51 nm. Considering that the smallest particles could be retained in the final reaction solution, the increase of the crystallite was not significant in the two cases. In addition, no signal of cobalt hydroxides or carbonates was observed, so the materials did not change in composition or in size by means of PXRD. No changes were observed in their Raman spectrum either (Figure S16(c) and S16(d)). These results indicate that no significant changes in the structure of the catalysts were produced after the first cycle. Moreover, the optical properties remained unaltered (not shown).

The reaction was carried out again under the same conditions as the experiment mentioned above, using the recovered catalytic materials. Even more interesting was to corroborate that those materials conserved their activity for the reduction of CO2. As can be seen in Figure 7, the chromatograms revealed that methanol was obtained again with the reutilized materials.

After the second cycle, we tried to recover the catalysts in order to characterize the structural and optical properties to evaluate possible changes, but we did not succeed. The solids remained stuck on the walls of the reactor.

It is worth mentioning that the viability of these catalysts for long-term use is still under investigation.

4. Conclusions

We have an unprecedented report of methanol production using a photocatalytic system with a very simple architecture working under ordinary reaction conditions. Co3O4 nanostructures showed photocatalytic activity in the CO2 reduction process, and methanol is the only detected reaction product. It is possible to assert that this is a simple example of artificial photosynthesis. Co3O4 has been widely used for water splitting and water oxidation, but this is the first time MeOH production is reported when this material is used as a photocatalyst.

Co3O4-NPs and Co3O4-P-NS were satisfactorily synthesized by adding changes to two reported methods. No impurities were detected in the synthesized Co3O4-P-NS powder. A small quantum confinement was found for the two samples compared with the bulk material, being higher for Co3O4-NPs. No significant differences were observed between the two reported morphologies regarding the photocatalytic behavior.

Data Availability

The data used to support the findings of this study are included within the article and the supplementary information files.

Disclosure

Inti Zumeta-Dube’s current address is Centro de Investigación en Ciencia Aplicada y Tecnología Avanzada, Unidad Legaria, Instituto Politécnico Nacional, Calzada Legaria No. 694, Col. Irrigación, Del. Miguel Hidalgo, CP 11500, México D.F., Mexico.

Conflicts of Interest

The authors declare that there is no conflict of interest regarding this publication of this paper.

Acknowledgments

This research has received partial funding from the European Community’s Seventh Framework Programme (FP7-NMP-2010-EU-MEXICO) and CONACYT (BisNano), under Grant Agreement nos. 263878 and 125141, respectively. Inti Zumeta-Dube is grateful to CONACYT (project number 174247), ICyTDF, BisNano, and the project CM-SECITI/095/2017 “Planta piloto sustentable operada con energía solar para el tratamiento de agua pluvial de la CDMX” for the postdoctoral fellowship. Salvador Pocoví-Martínez is thankful for the postdoctoral fellowship from DGAPA-UNAM. David Diaz thanks DGAPA-UNAM (IN115212 and IN116214 PAPIIT projects) and CONACYT (Redes193850 and SEP-CB-132094 projects) for the financial support. The authors want to thank Dr. Edilso Reguera-Ruiz (CICATA-U Legaria, IPN), Dr. Aida Gutiérrez Alejandre (FQ-UNAM), and Dr. Tobías Noel Nava-Entzama (IMP) for the access to several instrumental facilities.

Supplementary Materials

Figure S1: (a) XRD pattern and (b) IR spectrum of CoCO3(OH)2. Figure S2: XRD pattern of the black solid after combustion. Figure S3: Raman spectra recorded using a wavelength excitation laser source of 532 nm for (a) Co3O4-NPs and (b) Co3O4-P-NS. Figure S4: IR spectra of (a) Co3O4-NPs and (b) Co3O4-P-NS. Figure S5: pictures of (a) Co3O4-NPs and (b) Co3O4-P-NS powders. Figure S6: EDS spectra of (a) Co3O4-NPs and (b) Co3O4-P-NS. Figure S7: SEM images of Co3O4-NPs. Figure S8: SEM images of Co3O4-NPs. Figure S9: SEM images of Co3O4-NPs. Figure S10: (a) HR-TEM, (b) FFT of the framed region, and (c) reciprocal lattice of 2 different zones of Co3O4-NPs. Figure S11: (a) SAED pattern and (b) unit cell representation of Co3O4-NPs. Figure S12: size distribution of Co3O4-NPs. Figure S13: magnetization curve of (a) Co3O4-NPs and (b) Co3O4-P-NS. Figure S14: XPS of Co3O4-NPs: (a) wide-range spectra and high-resolution spectra of (b) Co2p, (c) C1s, and (d) O1s. Figure S15: isotherm plot for adsorption-desorption of (a) Co3O4-NPs and (c) Co3O4-P-NS; BET surface area plot of (b) Co3O4-NPs and (d) Co3O4-P-NS. Figure S16: XRD patterns and Raman spectrum after photocatalytic reaction of (a, c) Co3O4-NPs and (b, d) Co3O4-P-NS. Appendix A: calculation of the different concentration relations of species in solution. (Supplementary Materials)