1 Introduction

Perovskite-type oxides (ABO3 or A2BO4) have received much attention because of their important physical properties such as ferro-, piezo-, pyroelectricity, magnetism and electro-optic effects [1,2,3,4,5]. In the ideal perovskite structure, A is either a lanthanide or an alkaline earth metal and B is a transition metal. Fine dispersion of “metal on oxide” sample can be obtained by reduction of the perovskite oxides, which benefits the catalytic reaction [6].

Many attentions have been paid on the CO2 hydrogenation to methanol, because that offers a way to the green-house gas mitigation and fossil fuel substitution [7, 8]. For thermodynamic limitation of this reaction, the CO2 conversion is low [9, 10]. Cu-based catalysts were researched mostly for the reaction, and the active sites contain one or combination of Cu0, Cu+ and Cuα+ [11,12,13,14,15,16]. The catalysts derived from perovskite-type precursors have good methanol selectivity (> 60%) but low CO2 conversion (< 15%) [15, 16].

The metallic sample can be obtained by gas (such as H2, CO, etc.) reduction or liquid (such as hydroboron, hydrazine hydrate, etc.) reduction. Generally, gas reduction is performed at 473 to 1027 K with 2000–20000 h−1 space velocity for 4–24 h. While liquid reduction with NaBH4 is normally carried out at low temperature (338 K) for less than 5 min in the liquid environment, thus reducing the agglomeration of metal species and producing samples with smaller metallic particles [17,18,19]. Meanwhile it can be seen that liquid reduction is low energy consumption. Liquid reduction is usually used for reducing noble metal and liquid sample.

Considering the advantages of liquid reduction and application of perovskite, the perovskite LaCuZn is chosen, which contains the essential composition for CO2 hydrogenation to methanol over Cu-based catalyst based on the references and our experience, and the effect of the NaBH4 liquid reduction over this sample is investigated. The perovskite LaCuZn sample (LCZ) and liquid reduced perovskite LaCuZn sample (l-LCZ) were prepared, characterized, and tested for the CO2 hydrogenation to methanol.

2 Experimental

2.1 Preparation of Samples

The perovskite LaCuZn sample (LCZ) was synthesized by co-precipitation method. A solution of La(NO3)3·nH2O, Cu(NO3)2·3H2O and Zn(NO3)2·6H2O with the mole ratio of La3+:Cu2+:Zn2+ = 1:1:0.5 (1.0 mol L−1, 100 mL) and a solution of NaOH and Na2CO3with the mole ratio of 2:1 (1.5 mol L−1) were added dropwise simultaneously using two peristaltic pumps to deionized water (100 mL) under stirring at 338 K and kept the pH value of 9.0 ± 0.2. The slurry was aged for 12 h followed by filtering and washing. The filter cake was dried at 353 K overnight and then calcined at 1073 K for 6 h. The LCZ sample was obtained.

The LCZ sample was ground and dissolved to deionized water (100 mL) to form a suspension. In order to avoid to hydrolyzation, a certain amount of NaBH4 (B/Cu = 5) was dissolved in NaOH solution (100 mL, 0.1 mol L−1) to prepare the reducing agent solution. The reducing agent solution was added dropwise to the LCZ sample suspension under N2 atmosphere with stirring at 338 K, and aged at this condition for 2 h. Then the slurry was filtrated and washed with deionized water for three times and ethanol for the fourth time. The filter cake was dried in vacuum at 353 K for 10 h. The sample was denoted as l-LCZ.

Sample LCZ which was reduced by H2 at 623 K for 6 h was named as LCZ-623.

2.2 Characterization Methods

Powder X-ray diffraction (XRD) was tested using a Panalytical X’Pert Pro X-ray diffractometer with CuKα radiation in the 2θ range of 5°–90°. The elemental composition was determined by inductively coupled plasma optical emission spectroscopy (ICP-OES) on a Thermo iCAP 6300 instrument. The specific surface area and pore volume were determined by N2 adsorption–desorption at liquid nitrogen temperature 77 K on a Micromeritics Tristar 3000 instrument. The isotherms were elaborated by the BET method for surface area calculation.

The dispersion of Cu (DCu) and exposed Cu surface area (SCu) were determined by dissociative N2O adsorption and carried out on a Micromeritics AutoChem 2920 instrument. The catalyst (0.10 g) was first reduced in 5% H2/Ar mixture (40 mL min−1) for 2 h at 503 K (l-LCZ)/623 K (LCZ), and the amount of hydrogen consumption was denoted as A. Then, the reduced sample was cooled to 323 K and isothermally purged with Ar for 30 min, after which the sample was exposed to 10% N2O/90% N2 (85 mL min−1) for 1 h to ensure complete oxidation of metallic copper to Cu2O. The sample was then flushed with Ar and cooled to room temperature. Finally, a pulse of pure H2 was passed over the catalyst at 503 K. The surface Cu2O was reduced in the pulse of pure H2, and the amount of consumed H2 was denoted as B. The dispersion of Cu and exposed Cu surface area of the catalyst were calculated by the Eq. (1) [20] and Eq. (3) [21].

$${\text{D}}_{\text{Cu}} = \frac{{2{\text{B}}}}{\text{A}} \times 100\%$$
(1)
$${\text{n}}_{\text{Cu}} = 2{\text{B}} \left( {\text{mol}} \right)$$
(2)
$${\text{S}}_{\text{Cu}} = ({\text{n}}_{\text{Cu}} \times {\text{N}})/(1.46 \times 10^{19} \times {\text{W}})({\text{m}}^{2 } {\text{g}}^{ - 1} )$$
(3)

where DCu is the dispersion of Cu, nCu is the molar number of copper, SCu is the exposed copper surface area per gram catalyst, N is Avogadro’s constant (6.023 × 1023 atoms mol−1), W is the weight of the reduced catalyst, and 1.4 × 1019 is the number of copper atoms per square meter.

The morphology of the samples was investigated by scanning electron microscopy (SEM) on a JSM-7001F instrument. H2 temperature programmed reduction (H2-TPR) was performed in a quartz reactor. The samples (50 mg) were pretreatment with Ar at 423 K for 1 h, and then treated in a flow of 10 vol.% H2/Ar at a heating rate of 5 K min−1 up to 1073 K. TCD was used to monitor the hydrogen consumption. X-ray photoelectron spectroscopy (XPS) was carried out on an AXIS ULTRA DLD spectrometer with an Al anode (Al Kα = 1486.8 eV) source. The binding energies were calibrated using the C1s peak with Eb = 284.6 eV and the experimental error is within ± 0.1 eV.

2.3 Catalytic Evaluation of Samples

The samples were used for CO2 hydrogenation to methanol catalytic evaluation. There are three reactions described as follows:

$${\text{CO}}_{ 2} + 3 {\text{H}}_{ 2} \leftrightarrow {\text{CH}}_{ 3} {\text{OH}} + {\text{H}}_{ 2} {\text{O}}$$
(4)
$${\text{CO}}_{ 2} + {\text{H}}_{ 2} \leftrightarrow {\text{CO}} + {\text{H}}_{ 2} {\text{O}}$$
(5)
$${\text{CO}} + 2 {\text{H}}_{ 2} \leftrightarrow {\text{CH}}_{ 3} {\text{OH}}$$
(6)

The reaction was carried out in a continuous-flow fixed-bed reactor. The catalyst is 40–60 mesh. Generally, catalyst need be reduced by H2 under atmospheric pressure at 623 K for 6 h before activity test, while l-LCZ sample was directly tested without further H2 reduction process. The reaction condition was 5.0 MPa, n (H2): n (CO2) = 3:1, and gaseous hourly space velocity (GHSV) = 4000 h−1. The reaction ran at the temperature of 503 K, 523 K, 543 K and 563 K successively. The products went through a cold well, and divided into gas phase and liquid phase. The liquid phase was collected in a storage tank. The steady-state activity values were quoted as the average of three different analyses taken after 24 h on the stream operation. Gas products were quantitative analyzed on-line with a gas chromatograph equipped with a thermal conductivity detector (TCD, TDX-01 column) and a flame ionization detector (FID, Carbosieve column). Liquid products were determined off line by a Porapak Q column. The CO2 conversion and the carbon-based selectivity of CH3OH and CO were calculated by an internal normalization method. The space time yield (STY) of CH3OH which gave the amount of CH3OH produced per milliliter catalyst per hour was defined as:

$${\text{STY}}_{\text{CH3OH}} = \, \left( {{\text{W}} \times {\text{X}}_{\text{CH3OH}} } \right)/\left( {{\text{V}} \times {\text{t}}} \right) \, \left( {{\text{g mLcat}}^{ - 1} {\text{h}}^{ - 1} } \right)$$
(7)

where W is the total weight of liquid products (unit: g); XCH3OH is the mass fraction of CH3OH in the liquid products; V is the volume of catalyst (unit: mL); t is the reaction time (unit: h).

3 Results and discussion

Figure 1 is the XRD patterns of LCZ and l-LCZ catalysts. For LCZ catalyst, the main phase is La2CuO4 (JCPDS # 82-2142), which indicates that the typical perovskite structure is prepared by co-precipitation. CuO phase (JCPDS # 45-0937) also exists in the catalyst, meaning that part of Cu element is not anchored in the perovskite structure. Besides the above two phases, ZnO can be observed. After the treatment of liquid reduction by NaBH4 for the LCZ catalyst, i.e. l-LCZ catalyst, the phases are changed obviously. The peaks of La2CuO4 cannot be observed, which means that the perovskite structure is disappeared. Lanthanum species exists in the phase of La(OH)3 (JCPDS#36-1481) and copper species is transferred to Cu0. The peaks at 43.3º, 50.4º and 74.1º are attributed to the (111), (200) and (220) planes of Cu0 (JCPDS # 04-0836), respectively. The phase change of copper species from LCZ sample to l-LCZ sample indicates that the reduction is occurred for sample during the NaBH4 treatment. When La2CuO4 is reduced by H2 gas, La2CuO4 is also disappeared but transferred to La2O3, and Cu species exists in the phase of Cu0 [22, 23]. The difference phase of lanthanum species between liquid reduction and gas reduction maybe attribute to that there are a amount of water in the liquid reduction, which leads to La2O3 transferred to La(OH)3.

Fig. 1
figure 1

XRD patterns of LCZ and l-LCZ catalysts

Full size image

The elemental composition of the catalyst was tested by inductively coupled plasma optical emission spectroscopy (ICP-OES) and the results were listed in the Table 1. Compared LCZ catalyst, the content of element La, Cu and Zn all decreases a little but the value of La:Cu:Zn is similar for l-LCZ catalyst. It indicates that the metal elements are lost but the composition is not changed when the liquid reduction treatment is added during the catalyst preparation.

Table 1 Elemental composition of the prepared catalysts
Full size table

BET surface area (SBET) and pore volume (Vp) of the catalysts were listed in the Table 2. For perovskite material, the surface area is very low and usually lower than 10 m2 g−1 [6]. From the SBET data of the LCZ and l-LCZ catalysts, it can be seen that the surface area is improved five times, and pore volume is improved more obviously after the liquid reduction treatment. These improvements facilitate the distribution and exposure of the active sites. It also proves that the liquid reduction has effect on the perovskite: not only change the valance of the metal element, but also change the structure. Moreover, the exposed Cu surface area (SCu) and the Cu dispersion (DCu) measured by N2O adsorption technique were listed in the Table 2. Higher SCu and DCu are important for the activity of CO2 hydrogenation to methanol. SCu and DCu of l-LCZ are improved obviously compared with that of LCZ, which indicates that liquid reduction can make the copper species exposed more.

Table 2 Textural character of the prepared catalysts
Full size table

The morphology of LCZ and l-LCZ measured by SEM is showed in Fig. 2. For LCZ catalyst, which is the perovskite structure, the particles are smooth and the average size is 270 nm. For l-LCZ catalyst, which is treated by liquid reduction, the particles become rough, some chips appear, and the average size is 200 nm. The morphology can explain the reason why the surface area of LCZ catalyst is so low, and that of l-LCZ is improved. Moreover, some small particles presented for l-LCZ, which maybe play key role for CO2 hydrogenation because of these particles all located on the surface.

Fig. 2
figure 2

SEM images of LCZ and l-LCZ

Full size image

Temperature programmed reduction (TPR) can reflect the reducibility and interactions between sample compositions. Figure 3 shows the H2-TPR patterns of LCZ and l-LCZ catalysts. Both samples have two main peaks, which mean there are two kinds of copper species: the low temperature peak can be attributed to easier reduced copper species and the high temperature peak can be attributed to harder reduced copper species, respectively. Compared to the reduction peak at 573 K of catalyst LCZ, the reduction peak of catalyst l-LCZ is moved to the lower temperature 473 K, which is also much lower than pure CuO (573 K). The amount of hydrogen consumed for catalyst l-LCZ is fewer than that for catalyst LCZ. Based on the above results, it can conclude that: (1) the copper species of the two catalysts is similar, which is to say that liquid reduction cannot change the copper species; (2) liquid reduction can make the copper species reduction easier, which means that it can decrease the interaction between copper species with other species [12]; (3) only a part of copper species of catalyst l-LCZ has been reduced to lower valence after NaBH4 treatment, which leads to the amount of hydrogen consumed only decrease but not disappear completely.

Fig. 3
figure 3

H2-TPR patterns of LCZ and l-LCZ

Full size image

l-LCZ sample still has H2 consumption in TPR process, connected with the results that only Cu0 can be observed in XRD test for it, and it can obtain other conclusions: (1) NaBH4 treatment can only reduce a part of copper species to lower valence copper species, and some copper species still can be reduced; (2) these unreduced copper species is dispersed highly and not aggregated.

Generally, the metal catalyst is obtained by the gas H2 reduction from their metal oxide. LCZ-623 sample which is the LCZ sample reduced by hydrogen at 623 K for 6 h is prepared for comparison. Figure 4 is the XPS results of LCZ, l-LCZ and LCZ-623. For LCZ catalyst, Cu 2p3/2, Cu 2p1/2 along with their satellite peaks are detected, which means that Cu species exist in Cu2+ in the perovskite structure [24]. After liquid reduction, i.e. for l-LCZ catalyst, the satellite peaks still exist which indicates that not all Cu2+ are reduced by NaBH4. The area ratio of Cu 2p and satellite peak is higher for l-LCZ than that for LCZ, illustrating that lower valence Cu appear in the l-LCZ catalyst. Besides, for l-LCZ catalyst, bifurcation occurs in Cu 2p3/2 peak obviously, which is induced by various valence Cu. Cu 2p3/2 peak can be divided into two peaks: a peak centered at about 932.3 eV assigns to Cu0 and/or Cu+ species and a peak around 934.5 eV attributes to Cu2+ species [25, 26]. The area of lower binding energy is bigger than that of higher binding energy, stating that most of Cu2+ is reduced to the lower Cu after liquid reduction. There are no satellite peaks and no bifurcation in Cu 2p spectra for LCZ-623, declaring that full Cu2+ can be reduced by H2 reduction.

Fig. 4
figure 4

Cu 2p, La 3d, Zn 2p and O 1s XPS spectra of LCZ, l-LCZ and LCZ-623

Full size image

The La 3d5/2 peak of the three catalysts all locates at around 834.9 eV which is close to the value of pure lanthana, indicating that element La is present in the trivalent form [27, 28]. For the three catalysts, the peaks of La 3d3/2 and La 3d5/2 are split which are caused by the transfer of electrons from O 2p to the unoccupied orbital of La 4f. The degree of split increases in the following: LCZ, LCZ-623, l-LCZ. It is suggested that the influence of electrons transfer on the unoccupied orbital of La 4f is stronger in l-LCZ, and the effect to element La of the two reduction method is different [29].

The difference of the XPS spectra of Zn 2p for the three catalysts is not obvious, and the binding energy is assigned to Zn–O binding [30].

The O1s spectra show different oxygen peaks and they can be attributed to the oxygen ions in the crystal lattice (e.g. O2−) at 528.5 eV, chemically adsorbed oxygen species (e.g. O−) on the oxygen vacancies or the defects at 531.4 eV, and physically adsorbed oxygen species (e.g. H2O, O2) on the surface at 533.1 eV, respectively. The latter two can be combined and assigned to adsorbed oxygen [31,32,33]. Compared to lattice oxygen peak (528.5 eV) of catalyst LCZ, lattice oxygen peaks of catalyst l-LCZ and catalyst LCZ-623 are not obvious, which means that the latter two samples are reduced. For the three samples, the main peak is at 531.4 eV, which can be attributed to that it is easy to form a lot of vacancies for the perovskite structure material. The presence of large quantities of adsorbed oxygen indicates that there are many vacancies and defects associated with adsorption and catalysis activity [34,35,36]. It is noticed that the physically adsorbed oxygen species (533.1 eV) is more obviously for catalyst l-LCZ than that for the other two catalysts. This implies that gas reduction and NaBH4 liquid reduction have different effects on LCZ sample, and the catalysis process may different for l-LCZ and LCZ-623.

LCZ catalyst is reduced by hydrogen at 623 K for 6 h before reaction, and l-LCZ catalyst is used directly for reaction. Figure 5 shows the activities of LCZ catalyst and l-LCZ catalyst at different reaction temperature (503–563 K). With the reaction temperature increase, for the two catalysts, CO2 conversion is increased, methanol selectivity is decreased, and methanol space time yield (STY) is increased. The catalytic activity of l-LCZ is higher obviously than that of LCZ. The good performance of l-LCZ may be attributed to the higher exposed Cu surface area (SCu) and the Cu dispersion (DCu). The turnover frequency (TOF), which represents the number of CO2 molecules hydrogenated in a unit site per second (s−1), is calculated from the exposed copper surface area for the l-LCZ catalyst and LCZ catalyst. For comparasion, Cu/ZnO/Al2O3 catalyst was prepared according to the Ref. [37] and its TOF was given. The TOF values of the l-LCZ catalyst and LCZ catalyst are higher than that of Cu/ZnO/Al2O3 catalyst and Cu/ZnO/ZrO2 system [38], indicating the better efficiency for copper atoms on perovskite-type catalysts. Furthermore, the l-LCZ catalyst and LCZ catalyst have been run for 288 h and no apparent activity decrease, so they are stable.

Fig. 5
figure 5

Catalytic activities of methanol synthesis over LCZ and l-LCZ. Reaction conditions: P = 5.0 MPa, n (H2): n (CO2) = 3:1, GHSV = 4000 h−1

Full size image

Figure 6 shows the XRD patterns of LCZ and l-LCZ catalysts after reaction. For LCZ catalyst, suffered from H2 reduction and reaction, the perovskite structure is disappeared, and the main phases change to La(OH)3, Cu0 and ZnO. La species may undergo the phase of La2O3 in the process, but there are water produced in the reduction and reaction, so it exists in the phase of La(OH)3 at last. The phases of l-LCZ catalyst after reaction are the same as that before reaction, i.e. La(OH)3, Cu0 and ZnO, but the strength is increased which means that there is growth for the grains. The main phases of the two catalysts are the same, but some other phases which cannot be well assigned to only exist in LCZ catalyst. Combining the results of Fig. 1, these indicate that the reaction processes are different for the samples which are treated by the liquid reduction and gas reduction respectively.

Fig. 6
figure 6

XRD patterns of LCZ and l-LCZ catalysts after reaction

Full size image

4 Conclusion

This paper investigated the influence of liquid reduction on perovskite material. La2CuO4 perovskite material (LCZ) was prepared. After treated by NaBH4 reduction (l-LCZ), the sample was part reduced, the particle size became small, BET surface area increased. Liquid reduction can decrease the interaction between copper species and other species, but cannot change the type of copper species. The XPS results indicate that the effect of liquid reduction and gas reduction on Cu, La and O is different. The sample which is treated by NaBH4 reduction can be used directly for CO2 hydrogenation to methanol, and the catalytic activity is higher than that over the catalyst LCZ which is reduced by H2 at 623 K for 6 h.

Based on the conclusion of this paper, it is suggested that liquid reduction can be used for the samples which were usually reduced by gas, especially for those with high reduction temperature.