Enhanced Photoreduction Activity of Carbon Dioxide over Co3O4/CeO2 Catalysts under Visible Light Irradiation

A series of new two semiconductor catalysts, Co 3 O 4 /CeO 2 , were prepared by glycine-nitrate combustionmethod for photocatalytic reduction of carbon dioxide to produce methanol and ethanol under visible light (λ > 400 nm) irradiation. The catalysts were characterized by BET, UV-vis spectra, XRD, SEM, PL, and XPS and the results indicated that the catalyst with 5wt.% of Co 3 O 4 has the highest yield among all kinds of tests with the methanol yield of 1.52 μmol⋅g⋅h and the ethanol yield of 4.75μmol⋅g⋅h, which are about 2.34 and 1.71 times as large as those of CeO 2 . However, methanol and ethanol can hardly be detected for Co 3 O 4 under the same condition because of its too narrow band gap. The improvement of the photoreduction activity of Co 3 O 4 doped CeO 2 was caused by the separation of electron-hole pairs of Co 3 O 4 /CeO 2 and charge transfer betweenCo 3 O 4 andCeO 2 , mimicking the Z-scheme in photosynthesis.


Introduction
Carbon dioxide, one of the major greenhouse gases which is largely formed by fossil fuels consumption, can be conversed back to fuel with water or hydrogen by thermochemical or photocatalytic approach [1].Photocatalytic reduction of carbon dioxide to methanol or other organic fuels, which is a process to both reduce carbon dioxide emissions and resolve energy crisis, has been proved to be a prospective way to convert solar energy by using semiconductor catalysts, like TiO 2 [2][3][4], CdS [5], SiC [6], g-C 3 N 4 [7,8], and graphene oxide [9,10].Due to shortages of narrow band gap of single semiconductor catalyst, two-semiconductor ones such as Co 3 O 4 /CeO 2 are expected to be explored in order to improve photocatalytic properties and to widen the responsive wavelength range for visible light, thus enhancing the total efficiency of photocatalysis.
Among various semiconductor photocatalysts, CeO 2 has some properties similar to titania catalyst such as wide band gap, nontoxicity, and high stability [11].However, the main light absorption region of CeO 2 lies in the near ultraviolet and its low quantum efficiency and high recombination rate of electrons holes also constrain its development [12].Due to the special f and d electron orbital structure of cerium atom, some metal oxide semiconductors like TiO 2 [13], Bi 2 O 3 [14], and Cu 2 O [11] can be easily doped with pure CeO 2 to enhance visible light absorption and quantum efficiency.Cobalt-based catalysts have been recently reported as photocatalysts for water oxidation [15][16][17][18], hydrogen production [19], and organic pollutants degradation [20][21][22].Co is significantly less abundant than other transition metals like Fe, Mn, and Ni, but it is emerging as a potential metal for catalytic processes because of its light harvesting and electron mediating properties [19].Cobalt oxides, for example, CoO and Co 3 O 4 , are p-type semiconductors with interesting electronic and magnetic properties [23].The p-n heterojunction semiconductor structure combined with cobalt oxide and other n-type semiconductors such as BiVO 4 [23] or WO 3 [20] has been investigated as an effective way to improve the photocatalytic activities under visible light irradiation.In this case, CeO 2 as an n-type semiconductor with wide band gap may possibly bring out a similar heterojunction structure with Co 3 O 4 to enhance the efficiency of visible light absorption by shortening the band gap, in which such p-n International Journal of Photoenergy heterojunction structure can hardly work to reduce carbon dioxide according to the band theory because the conduction band potential of Co 3 O 4 is too positive to start the reduction.However, CeO 2 is also a good reducing agent because of its relatively more negative reductive potential (around −0.5 eV) and Co 3 O 4 doped CeO 2 with shorter band gap (2.07 eV) can act as a good cocatalyst, sensitizer, and oxidation agent of the composite with high positive potential (+2.44 eV) in the Z-scheme reaction mimicking photosynthesis.Recent studies on Z-scheme without mediators or so-called direct Z-scheme were carried out to explain charge separation at the interface between two semiconductors in a hybrid photocatalyst [24][25][26][27].Further application of Co 3 O 4 /CeO 2 photocatalytic reduction of carbon dioxide to methanol and ethanol in the Z-scheme model has not been reported, which could enhance photocatalytic activity.
In this paper, a series of Co 3 O 4 /CeO 2 catalysts were prepared by glycine-nitrate combustion method for photocatalytic reduction of carbon dioxide to methanol and ethanol under visible light ( > 400 nm) irradiation.The structures of catalysts were characterized by XRD, UV-vis absorption spectra, BET analysis, and SEM testing.And a series of measurements, such as XPS and PL, were carried out for the mechanism of improvement of the photoreduction activity of Co 3 O 4 doped CeO 2 .

Preparation of Co
and Ce(NO 3 ) 3 ⋅6H 2 O were mixed in the appropriate molar ratios with a minimum volume of distilled water to form a transparent solution.Then the glycine solution was slowly dropped to the metal nitrate aqueous solution during stirring.The glycine to nitrate ratio was set to 0.3 as the literature described [28].The resulting solution was then kept stirring for an hour in order to gain transparent sol.Afterwards, the crucible was dried at 75 ∘ C in an oven for more than six hours until a transparent moisture-sensitive glassy material was obtained.This glassy material with fine mesh was heated in a muffle furnace up to around 180 ∘ C, and then it generated spontaneous combustion with large amounts of gases evolution and a foamy voluminous powder.Then the powder was calcined at 300 ∘ C or 400 ∘ C or 500 ∘ C or 600 ∘ C for an hour to burn off any carbonaceous residues in the catalyst.In addition, CeO 2 was prepared by the same method without adding Co(NO 3 ) 2 ⋅6H 2 O.

Catalyst
Characterization.Specific surface area, average pore diameter, total pore volume, and pore size distribution of the samples were determined from the adsorption and desorption isotherms of nitrogen at 77 K after outgassing procedure under vacuum at 250 ∘ C for 10 h, using a Quantachrome Autosorb-1 instrument.X-ray diffraction (XRD) analysis was performed with the PANalytical X'Pert diffractometer (X'Pert PRO MPD, PW3040/60) within the 2- ranging from 5 ∘ to 80 ∘ by a speed of 6 ∘ per minute with Cu-K ( = 0.154060 nm) radiation (40 kV, 40 mA).The crystallite size was calculated by the Scherrer equation from the XRD spectra.The surface morphology was observed by the field emission-scanning electron microscope (FE-SEM, S-4800).UV-vis spectroscopy in the 300 nm-1000 nm was measured with a Lambda 750 Diffuse Reflectance Spectroscopy.X-ray photoelectron spectroscopy (XPS) analysis was conducted on a thermo ESCALAB 250XI multifunctional imaging electron spectrometer (Thermo Fisher Scientific Inc.) equipped with an Al K radiation source.Fluorescence spectra were obtained using a photoluminescence (PL) spectrometer (Perkin Elmer, LS-55).The concentration of methanol and ethanol was detected by a gas chromatograph (GC-7890II), equipped a flame ionization detector and a stainless steel packed column (PorapakQ, 2 mm × 1 m).

Photocatalytic Reaction
Testing.The photocatalytic reaction was performed in a continuous-flow reactor system as shown in Figure 1.A 300W Xe lamp located in the center of the quartz cool trap was the light source and the UV light wavelength below 400 nm was filtered by a 2.0 M sodium nitrite solution layer between light source and reacting liquid [4].Prior to experiment, sodium carbonate (3.18 g) and sodium sulfite (3.78 g) were dissolved in 300 mL deionized water.In this paper, sodium sulfite and sodium sulfide play a role as the sacrifice agent to scavenge photogenerated holes.Then the solution was injected into the reactor.Before irradiation, ultrapure carbon dioxide was bubbled through the solution in the reactor for at least 30 minutes to ensure that all dissolved oxygen was eliminated while adsorption and desorption of carbon dioxide in the solution and on the photocatalyst reached the equilibrium.300 mg of catalyst powder was then added into above solution in the reactor for a five-hour irradiation.The solution of about 1 mL in the reactor was withdrawn as a sample each hour.The UV-vis spectra of the catalysts are shown in Figure 3.The light absorption of all catalysts covers the visible region.The band gap of CeO 2 is estimated to be 2.93 eV from the plot of the (ℎ]) 2 versus photon energy (ℎ]) [4,6] as shown in the inset of Figure 3(a), which is 0.27 eV narrower than that of the commercial CeO 2 with the band gap of 3.2 eV.It has been reported that CeO 2 , as an n-type semiconductor, has band gap varying from 2.7 to 3.4 eV depending on the preparation method [30].The red shift of the CeO 2 sample might be attributed to trivalent ionic cerium compounds.Similarly, the band gap of Co 3 O 4 is estimated to be 2.0 eV, which is matched with the reported band gap (2.07 eV) [20].In contrast to CeO 2 , the Co In order to investigate the electronic environment and oxidation state of the CeO 2 sample, XPS analysis was performed.Eight peaks for Ce 3d of CeO 2 sample are shown in Figure 4(a), which are different from those of the reported CeO 2 with six peaks at binding energies 882.5, 888.7, 898.2, 900.7, 907.6, and 916.5 (±0.2) eV [31,32].Other authors have proved that the shape of Ce 3d peaks changes after normalizing the intensity of Ce 3d peaks for Ce 2 O 3 and CeO 2 [32] when Ce 2 O 3 is mixed or doped with CeO 2 .And the peaks can be divided into more than six peaks, which may be attributed to the presence of Ce 3+ [33].Among all of these ten peaks in Figure 4(a), V, V  , V  , ,   , and   are attributed to the presence of Ce(IV), and V 0 , V  ,  0 , and   are attributed to the presence of Ce(III) [34].Three peaks for O 1s of CeO 2 sample are shown in Figure 4(b).Two peaks at 529.7 and 531.3 eV are attributed to O-Ce(IV) bond and O-Ce(III) bond, respectively [34], which indicates the existence of both CeO 2 and Ce 2 O 3 [35].In this case, Ce 2 O 3 with much shorter band gap of 2.4 eV can shorten the band gap of CeO 2 and enhance the visible light absorption, which is in line with the UV-vis spectra.

Characterization of Samples with Different Cobalt Loading.
Table 1 displays the specific surface area, the pore volume, the total pore volume of pure CeO 2 and Co 3 O 4 , and the catalysts with Co 3 O 4 loading of 2 wt.%, 5 wt.%, and 8 wt.%.As the Co 3 O 4 loading increases from 0 wt.% to 100 wt.%, the surface areas decrease from 100.14 to 24.47 m 2 /g, and the total pore volume decreases as well.It can be attributed to the mutual filling of the pore structure during the forming of Co 3 O 4 and CeO 2 which can be proved by the pore diameter distribution in Figure 5.And pore size distribution of Co 3 O 4 /CeO 2 catalysts that smaller than 10 nm decreases sharply compared to that of Co 3 O 4 or CeO 2 and mainly is distributed in mesopore region from 10 to 50 nm.When       ) and better sacrifice reagent (SO 3 2− ) can greatly enhance the photocatalytic activities for methanol and ethanol generation as shown in Table 2.

Effects of Calcination
Temperature on the Photocatalytic Activity.Calcination temperature also affects the surface structures of catalysts.The samples with 5 wt.% of Co 3 O 4 loading were calcined at 300 ∘ C, 400 ∘ C, and 500 ∘ C, respectively.As shown in Table 3, the specific surface area decreases with the increase of calcination temperature.And the pore diameter distribution becomes uneven when catalyst  6, which indicates that the particle was melted and the pore collapsed.
Figure 7 shows the SEM images of the 5.0 wt.% Co 3 O 4 /CeO 2 catalysts prepared by calcining the spontaneous combusted precursor at temperatures of 300 to 600 ∘ C when CeO 2 calcined at 400 ∘ C. The shape of CeO 2 particle becomes very irregular.However, more pores can be seen on the surface of the CeO 2 particles (Figures 7(i .It can explain why the CeO 2 has larger surface area than the other samples.The particle size of the Co 3 O 4 /CeO 2 catalysts, however, increased greatly with the calcination temperature higher than 400 ∘ C, as shown in Figures 7(c) and 7(d).However, the particles of the sample calcined at 300 ∘ C are more irregular and the pores on the surface are bigger than those of the sample calcined at 400 ∘ C, which is in good correspondence to BET analysis results.
In Figure 3, XRD peaks for Co 3 O 4 of Co 3 O 4 /CeO 2 catalysts are absent, however, the peaks appear when the calcination temperature is higher than 400 ∘ C as shown in Figure 8.And the CeO 2 crystallite size of catalysts was estimated by using the Scherrer equation, and the order is 500 (41.8 nm) > 300 (37.5 nm) > 400 (35.4 nm) as shown in Table 3.The diffraction peak at 48.6 ∘ corresponding to hexagonal Co-Ce-O phase (JCPDS file 65-5917) [36] becomes stronger with increase of calcination temperature, which suggests that the treatment by combustion method strengthens the Co-Ce interaction and the calcination temperature improves the crystallization structure of catalysts.Thus, the CeO 2 crystallite size is getting smaller under calcination temperature of 400 ∘ C than under 300 ∘ C.And the enlarged size of CeO 2 crystallite under 500 ∘ C is caused by sintering of particles which is in line with the SEM.
In order to determine the amount of Ce(III) in cerium oxides, the deconvolution method of the Ce3d features [34,37] has been used.The amounts of Ce(III) and Ce(IV) were calculated from the components of the Ce 3d core level using the following equations [37]: The calculated result shows that the amount of Ce(III) increases from 3.7% to 6.5%, the amount of Ce(IV) decreases from 32.0% to 25.9%, and the ratio of Ce(III) to Ce(IV) increases with the calcination temperature increase from 300 ∘ C to 500 ∘ C.These may be attributed to the reduction of carbon species which come from pyrolysis of residual glycine.are 2 and 1.5, respectively.The measured results (Table 4) suggest that the oxygen amount on the surface of particles meets stoichiometric ratio with Ce(IV) and Ce(III) for the CeO 2 sample; however, it is far from sufficient for the Codoped samples, which means that oxygen vacancies can be significantly formed on the surface of CeO 2 crystallite after cobalt is doped.Thus, the catalytic activity can be largely enhanced [38].Among the Co-doped samples, the percentages of cobalt on surface of particle are almost constant, which indicates that the amount of oxygen vacancies decreases with the increase of calcination temperature.It is attributed to reduction of Co species on the surface during the generation of hexagonal Co-Ce-O phase [36].Figure 9(a) shows the XPS spectra for Co 2p in 5 wt.%Co 3 O 4 /CeO 2 prepared by calcining the spontaneous combusted precursor at temperatures form 300 ∘ C to 500 ∘ C. The doublet peaks, corresponding to Co 2p 3/2 and a satellite peak, for all the samples, were found at 788.0 ± 0.3 and 782.6 eV in Co 2p scanning pattern, which indicates the generation of Co 3 O 4 on the surface of catalysts, which is in line with XRD analysis (Figure 8).Other researchers have reported that the satellite peak is absent in the diamagnetic Co 3+ compound [39,40]; however, it grows along with the increase of calcination temperature, which indicates the transformation of Co 3+ to Co 2+ species [41].Whereas the binding energies of Co 2p 3/2 and the satellite peak are slightly higher than the reported binding energies at ∼780.0 and ∼786.0 eV [41][42][43], respectively, this shift toward higher energies may be caused by the conjunction with Ce orbits.Binding energies of Co 3s and Ce 4d have been shown in Figure 9(b) to present the connection between Co and Ce orbits.Binding energy of Co 3s shifts toward higher energy from 110.9 (handbook [44]) to 111.4 or 111.1 eV, while binding energy of Ce 4d shifts toward lower energy from 110.0 (handbook [44]) to 107.7 or 107.5 eV, which suggests the formation of a kind of conjunction structure, and this structure becomes stable after calcining at temperature higher than 400 ∘ C. Figure 10 shows that catalysts calcined at 400 ∘ C have highest methanol and ethanol yield among all, which can be well explained by the BET and SEM analysis.The order of the photocatalytic activities arranged as the calcination temperature is 400 > 500 ≈ 600 > 300 > CeO 2 , which is in line with that of XPS quantitative analysis.

Separation of Electron-Hole Pairs on Co
The separation efficiency of electron-hole pairs of the composite photocatalyst is always the key factor to enhance the photocatalytic activity [26], which is mainly resulting from the addiction of cocatalyst.Ce 3+ species formed on the surface of the composite might contribute to the separation of electron-hole pairs thus enhancing the photocatalytic activity [45] and further evidences will be discussed later.And doping Co 3 O 4 can contribute to enhancing the visible absorption and it plays a role as a kind of cocatalyst to increase methanol and ethanol yields of photocatalytic reaction.
Two widely used equations for band edge calculation are showed below: CB is the conduction band edge of a semiconductor at the point of zero charge. is the absolute electronegativity of the semiconductor, expressed as the geometric mean of the absolute electronegativity of the constituent atoms, which is defined as the arithmetic mean of the atomic electron affinity and the first ionization energy.  is the energy of free electrons on the hydrogen scale (∼4.5 eV).  is the band gap of the semiconductor [46,47].[20,30].In order to confirm that the enhancement of photocatalytic activity is due to charge transfer between Co 3 O 4 and CeO 2 , PL measurements were carried out.Figure 12 shows the spectra of CeO

Figure 2 Figure 1 :
Figure 1: Schematic view of reactor system for photocatalytic reduction of carbon dioxide.
3 O 4 /CeO 2 have strong absorption in the visible region ( > 400 nm) and the visible light absorbance increases along with the addition of Co 3 O 4 (Figure 3(c)).

Figure 4 :
Figure 4: XPS spectra of the Ce 3d region (a) and the O 1s region (b) for the CeO 2 sample.
to 5 wt.%, the methanol yield increased from 0.68 to 1.52 mol⋅g −1 ⋅h −1 and the ethanol yield increased from 2.77 to 4.75 mol⋅g −1 ⋅h −1 .Further increasing the Co 3 O 4 loading (>5 wt.%) leads to the decrease of the methanol and ethanol yields because of the significant decrease of the crystallization degree of CeO 2 shown in Figure 2. Photocatalytic activities of the sample with 5 wt.%Co 3 O 4 were tested in different solutions.It is obvious that solution containing 0.1 M Na 2 CO 3 and 0.1 M Na 2 SO 3 has the best performance among (0.1 M Na 2 CO 3 + 0.1 M Na 2 SO 3 ), (0.1 M NaOH + 0.1 M Na 2 SO 3 ), and (0.1 M Na 2 CO 3 + 0.1M Na 2 S), which indicates that extra carbon source (CO 3 2− ) and 7(j)) compared to those of the Co 3 O 4 /CeO 2 catalysts (Figures 7(a)-7(h))

e 2 Figure 11 :
Figure 11: The scheme of photocatalytic reduction of carbon dioxide with water to methanol and ethanol.

Figure 11 (
a) shows an assumption which was made to explain the separation of electron-hole pairs of Co 3 O 4 /CeO 2 catalysts.Photoelectron on the surface of Co 3 O 4 transfers to the CeO 2 to avoid recombining to holes, which may also partly trap in the holes on CeO 2 and tend to reduce the recombination of electronhole pairs on CeO 2 .It works like a Z-scheme reaction.However, if the separation of electron-hole pairs works as p-n heterojunction, no methanol and ethanol would be produced because the conduction band potential of Co 3 O 4 is too positive to reduce carbon dioxide (Figure 11(b)).

2 ,
Co 3 O 4 , 5 wt.%Co 3 O 4 /CeO 2 , and mixture of 5 wt.%Co 3 O 4 and CeO 2 while the 5 wt.%Co 3 O 4 /CeO 2 sample was prepared as a reference sample.The excitation wavelength was 425 nm.CeO 2 sample has a strong PL emission at about 630 nm, which indicates rapid recombination of electrons and holes.The other samples have a PL peak at the same position with a peak intensity order of CeO 2 > mixture > 5 wt.%Co 3 O 4 /CeO 2 ≈ Co 3 O 4 .The weak PL intensity of 5 wt.%Co 3 O 4 and CeO 2 mixture could be attributed to the decrease of CeO 2 content.Although the sample contains only 5 wt.% of Co 3 O 4 , it has similar PL

Table 1 :
Characteristic of Co 3 O 4 /CeO 2 catalysts with different mass fraction of Co 3 O 4 .
the Co 3 O 4 loading increased from 2 wt.% to 8 wt.%, there is no distinct difference in specific surface area, average pore diameter, and total pore volume, which indicates that addition of Co 3 O 4 in the Co 3 O 4 loading range has little effect on the corresponding structure of Co 3 O 4 /CeO 2 catalysts.The sample with 5 wt.%Co 3 O 4 loading has biggest specific surface area and uniform pore diameter distribution among all the tested Co 3 O 4 /CeO 2 catalysts.In addition, larger specific surface area, smaller pore volume, and total pore volume compared to those of commercial CeO 2 suggest that

Table 2 :
Mean methanol and ethanol yield on Co 3 O 4 /CeO 2 catalysts under visible light irradiation ( > 400 nm). 2 prepared by glycine-nitrate combustion method is a more potential carrier for catalysis.The average methanol and ethanol yields, denoted as the average production rate of methanol and ethanol for each hour over the photocatalysts, are shown in Table 2.The photocatalytic activity of CeO 2 can be well improved via loading Co 3 O 4 on CeO 2 by glycine-nitrate combustion 0.30 g catalysts reacted in 300 mL solution of (a) 0.1 M NaOH + 0.1 M Na 2 SO 3 , (b) 0.1 M Na 2 CO 3 + 0.1 M Na 2 S, and (c) 0.1 M Na 2 CO 3 + 0.1 M Na 2 SO 3 .A 300 W Xenon lamp ( > 400 nm) irradiated for 5 hours.CeO 3 O 4 catalyst under visible light irradiation becauseof its conduction band maximum being too positive to reduce carbon dioxide to methanol or ethanol.The sample with 5 wt.%Co 3 O 4 has the best performance to produce both methanol and ethanol, which can be attributed to its better specific surface area, the proper pore volume, and pore size (Table1).Increasing the Co 3 O 4 loading from 0
∘ C. Pore diameter bigger than 984 nm can be found from Figure The ratios of O to Ce atoms in CeO 2 and Ce 2 O 3 lattice values for Co 3 O 4 and CeO 2 are 5.93 and 5.53 eV, respectively.So the  CB of Co 3 O 4 and CeO 2 are calculated to be +0.43 and −0.44 eV, respectively, and the  VB of Co 3 O 4 and CeO 2 are estimated to be +2.43 and +2.49eV, respectively.These results are close to those of the reported references