Introduction

The worldwide human population is rising at a quick pace, which inherently demands a large amount of energy for consumption [1]. Current energy production largely depends on the utilization of fossil fuels. The excessive use of petroleum-based resources results in a continuous CO2 discharge [2, 3]. This constant release of CO2 is considered as one of the primary reasons for the variation in the climate parameters [4]. The environmental distress occurring due to increases in CO2 emission generated more interest in the conversion of CO2 into value-added products.

One of the possible options for CO2 utilization is to convert the captured CO2 into fuels. Solar thermochemical cycles driven based on the metal oxide (MO)-based redox reactions can split CO2 into CO (Fig. 1) [5, 6]. The solar CO produced can be combined with the solar H2 (produced via solar thermochemical splitting of water) for the manufacturing of the solar syngas, which can be further utilized in the catalytic Fischer–Tropsch process [7].

Figure 1
figure 1

Overall process of production of liquid fuels by using solar syngas generated via MO-based solar thermochemical H2O/CO2 splitting cycle

The MO-based solar-driven thermochemical conversion of CO2 is a two-phase process. In the first phase, the MO is reduced with the help of concentrated solar power, and in phase 2, the reduced MO is re-oxidized again via a CO2 splitting reaction. Zinc oxide [8, 9], tin oxide [10, 11], iron oxide [12, 13], CeO2 [14,15,16], doped ceria [17,18,19,20], ferrites [21,22,23,24], perovskites [25,26,27,28], and others [29,30,31,32] have been utilized for the solar thermochemical conversion of H2O and CO2. Among all these, the phase pure CeO2 is considered as one of the promising options due to its faster reaction kinetics and better thermal stability. Although the CeO2 is beneficial for the thermochemical conversion of H2O and CO2, a lower fuel production capacity is one of the major limitations associated with this MO.

Recently, Bhosale and Takalkar [33] reported that the doping of lanthanides such as La, Pr, Nd, Sm, Gd, Tb, Dy, and Er into CeO2 fluorite cubic crystal structure improved the thermal reduction (TR) capacity of the CeO2. It was also reported that although the TR capability of CeO2 was improved, only Ce0.9La0.1O2 was capable of producing higher quantities of CO than CeO2 via CO2 splitting reaction. In another investigation, Takalkar et al. [19] reported that the inclusion of Ag in the transition metal-doped ceria considerably improved the TR as well as CO2 splitting (CDS) capacity.

Based on the results reported in our previous investigations, in this study, the synthesis, characterization, and application of Ce0.9Ln0.05Ag0.05O2−δ materials (where, Ln = La, Pr, Nd, Sm, Gd, Tb, Dy, Er) for the thermochemical conversion of CO2 is reported. Synthesis of the Ce0.9Ln0.05Ag0.05O2−δ materials is achieved via a co-precipitation method. The derived materials are further tested for multiple thermochemical CDS cycles by utilizing a thermogravimetric analyzer (TGA) setup. The TR and CDS capacity of the Ce0.9Ln0.05Ag0.05O2−δ materials was estimated and compared with the previously studied phase pure CeO2 and lanthanide-doped ceria materials.

Experimental

Preparation and characterization of Ce0.9Ln0.05Ag0.05O2−δ

Nitrate-based precursors of ceria, silver, and all lanthanides were acquired from Sigma-Aldrich, USA. Aqueous 28% NH3OH as a precipitating agent was procured from the same supplier. Ultrapure deionized (DI) water (produced through Direct-Q system, Millipore, France) was utilized for the preparation of nitrate solution. An ultrapure grade Ar gas (purity 99.999%) and 50% CO2 + 50% Ar gas mixture are ordered from the Buzwair Scientific and Technical Gases, Doha, Qatar.

The synthesis of redox Ce0.9Ln0.05Ag0.05O2−δ materials was achieved via co-precipitation of the hydroxide method. As shown in Fig. 2, an aqueous mixture of selected metal precursors was prepared by dissolving them in 300 ml of deionized water. To this mixture, aqueous ammonium hydroxide (28%) was added to attain a pH approximately equal to 10. The mixture further stirred for 24 h with a maintained pH ~ 10. Once the stirring was stopped, the precipitate of Ce0.9Ln0.05Ag0.05O2−δ material was allowed to settle via gravity (mixture kept undisturbed overnight). The next morning, the supernatant liquid was decanted, and the obtained precipitate was washed several times by deionized water with the help of a vacuum filtration unit. The obtained filtered cake of Ce0.9Ln0.05Ag0.05O2−δ was dried (120 °C, 5 h), crushed, and annealed (Nabertherm Furnace) up to 1000 °C for 4 h in the presence of air. The obtained annealed powder was analyzed for the determination of the phase/elemental composition and morphology by using Panalytical XPert powder X-ray diffractometer (PXRD) and scanning electron microscope (SEM, Nova Nano 450, FEI) equipped with the electron diffraction spectroscopy (EDS).

Figure 2
figure 2

Synthesis of Ce0.9Ln0.05Ag0.05O2−δ materials via the co-precipitation method

CO2 splitting experiments

The Ce0.9Ln0.05Ag0.05O2−δ materials were experimentally tested in a TGA setup (Setaram Instrumentation, France), which is shown in Fig. 3. The experimental parameters utilized to perform the thermochemical cycles are given in Table 1. Approximately, 50 mg of the calcined Ce0.9Ln0.05Ag0.05O2−δ powder was charged in an alumina (100 µl) crucible, and then placed inside the heating furnace of the TGA. Before performing the TGA experiments, the residual air filling the hollow space near to the furnace was evacuated by applying a vacuum followed by sweeping by the inert Ar. Chilled water (generated by Julabo FC 1600T) was utilized to decrease the exiting gas stream temperature. Additional details related to the TGA setup and the experimental procedure are already described in our previous studies [23, 29]. Multiple TR and CDS steps were performed by considering the operating parameters given in Table 1.

Figure 3
figure 3

TGA experimental setup

Table 1 Experimental conditions used for the TGA experiments

The mass variations allied with the Ce0.9Ln0.05Ag0.05O2−δ materials during the TR and CDS steps were documented after subtracting the blank TGA experimental data from the actual TGA experimental data. The amount of O2 liberated (µmol/g) during the TR step and the quantity of CO produced during the CDS step are calculated by utilizing Eqs. (1) and (2).

$$ n_{{{\text{O}}_{2} }} = \frac{{\Delta m_{\text{loss}} }}{{\left( {M_{{{\text{O}}_{2} }} \times m_{{{\text{Ce}}_{0.9} {\text{Ln}}_{0.05} {\text{Ag}}_{0.05} {\text{O}}_{2{ - \delta }} }}} \right)}} $$
(1)
$$ n_{\text{CO}} = \frac{{\Delta m_{gain} }}{{\left( {M_{\text{O}} \times m_{{{\text{Ce}}_{ 0. 9} {\text{Ln}}_{ 0. 0 5} {\text{Ag}}_{ 0. 0 5} {\text{O}}_{{ 2 {{ - \delta }}}} }} } \right)}} $$
(2)

Results and discussion

After synthesizing the Ce0.9Ln0.05Ag0.05O2−δ materials, the next important step was to determine the phase composition of the derived materials. PXRD profiles of the calcined Ce0.9Ln0.05Ag0.05O2−δ materials are shown in Fig. 4a, b. The presented patterns indicate a cubic fluorite crystal structure of the Ce0.9Ln0.05Ag0.05O2−δ materials, similar to the one reported in the case of CeO2 [23]. The PXRD profiles shown in Fig. 4a further indicates the absence of the formation of any metal or metal oxide impurities. As shown in Fig. 4b, the Ce0.9Ln0.05Ag0.05O2−δ material peaks shift toward either lower or higher 2θ angle (based on the crystal ionic radii of the dopants). This shift in the peaks further confirmed the successful incorporation of the dopants inside the ceria fluorite crystal structure. The formation of nominally phase pure Ce0.9Ln0.05Ag0.05O2−δ materials was also assured via EDS analysis (results given in Table 2).

Figure 4
figure 4

PXRD patterns of the Ce0.9Ln0.05Ag0.05O2−δ materials

Table 2 Abbreviations, chemical composition, crystallite size, and cell parameter of each Ce0.9Ln0.05Ag0.05O2−δ material

In order to determine the crystallite size, a widely used Scherrer equation and the PXRD data associated with the Ce0.9Ln0.05Ag0.05O2−δ materials were utilized. Table 2 indicates that the variation in the crystallite size of the Ce0.9Ln0.05Ag0.05O2−δ materials does not follow any specific trend. As per the obtained results, the Ce0.9Ln0.05Ag0.05O2−δ materials can be arranged as follows based on their average crystallite size: La5Ag5Ce > Dy5Ag5Ce > Sm5Ag5Ce > Pr5Ag5Ce > Tb5Ag5Ce > Gd5Ag5Ce > Er5Ag5Ce > Nd5Ag5Ce.

After estimating the composition and crystallite size of each Ce0.9Ln0.05Ag0.05O2−δ material, the morphology was scrutinized by performing the SEM analysis. The SEM images obtained looks very similar to each other and indicate the formation of roughly spherical particles of Ce0.9Ln0.05Ag0.05O2−δ. The microscopic SEM analysis further confirmed that the particles were agglomerated. It was also observed that the average particle size was very close to the crystallite sizes given in Table 2. The representative SEM images of Pr5Ag5Ce, Tb5Ag5Ce, Gd5Ag5Ce, and Er5Ag5Ce are shown in Fig. 5.

Figure 5
figure 5

SEM images of Pr5Ag5Ce, Tb5Ag5Ce, Gd5Ag5Ce, and Er5Ag5Ce

The redox performance of the Ce0.9Ln0.05Ag0.05O2−δ materials is estimated by performing the thermochemical CDS experiments using the TGA setup. Ce0.9Ln0.05Ag0.05O2−δ materials thermally reduced at 1400 °C (10 K/min) for 60 min in the presence of the inert Ar (100 ml/min) and then re-oxidized at 1000 °C by using a feed gas mixture containing 50% CO2 + 50% Ar (100 ml/min). As an example, Fig. 6 represents a TGA profile of La5Ag5Ce material obtained during the first CDS cycle. As shown in Fig. 6, during the TR step, the mass of the La5Ag5Ce material reduced by 0.511 mg, and during the CDS step, the weight of the La5Ag5Ce material increased by 0.126 mg. These mass variations further converted into respective redox performances in terms of \( n_{{{\text{O}}_{2} }} \) released (320.2 µmol/g) and \( n_{\text{CO}} \) produced (157.8 µmol/g) by using Eqs. (1) and (2).

Figure 6
figure 6

Exemplified TGA profile of La5Ag5Ce material during the first CDS cycle

The mass variations associated with the Ce0.9Ln0.05Ag0.05O2−δ materials recorded during the first cycle are shown in Fig. 7a, b. The \( n_{{{\text{O}}_{2} }} \) released and \( n_{\text{CO}} \) produced by each Ce0.9Ln0.05Ag0.05O2−δ material was computed based on the obtained TGA profiles and given in Table 3. The data given in Table 3 show that the Pr5Ag5Ce was capable of releasing a higher amount of O2 at 1400 °C than the other Ce0.9Ln0.05Ag0.05O2−δ materials. Likewise, the CO production aptitude of La5Ag5Ce was the uppermost when compared with the remaining Ce0.9Ln0.05Ag0.05O2−δ materials. The numbers listed in the \( n_{\text{CO}} /n_{{{\text{O}}_{2} }} \) ratio column shows that the re-oxidation ability of the Nd5Ag5Ce was better than the other Ce0.9Ln0.05Ag0.05O2−δ materials.

Figure 7
figure 7

TGA profiles obtained for Ce0.9Ln0.05Ag0.05O2−δ materials during a first TR step and b first CDS step

Table 3 Redox performance of Ce0.9Ln0.05Ag0.05O2−δ materials during the first cycle

Interesting to note that the \( n_{\text{CO}} \) produced by each Ce0.9Ln0.05Ag0.05O2−δ material was considerably lower than the \( n_{{{\text{O}}_{2} }} \) released during the first cycle. The two probable reasons for these results are (a) poor re-oxidation ability of the Ce0.9Ln0.05Ag0.05O2−δ materials or (b) additional mass loss during the first TR reduction due to the release of volatile chemicals from the Ce0.9Ln0.05Ag0.05O2−δ materials (which remained unburnt during the calcination step). For further investigation of this matter, the Ce0.9Ln0.05Ag0.05O2−δ materials were tested for three cycles (by maintaining the same operating conditions utilized in the case of the first cycle). Figure 8 shows the variations in the mass of the Ce0.9Ln0.05Ag0.05O2−δ materials during the successive three thermochemical cycles. Besides, Fig. 9a and b compares the \( n_{{{\text{O}}_{2} }} \) released and \( n_{\text{CO}} \) produced by each Ce0.9Ln0.05Ag0.05O2−δ material from cycle 1 to cycle 3.

Figure 8
figure 8

TGA profiles obtained for Ce0.9Ln0.05Ag0.05O2−δ materials during three cycles

Figure 9
figure 9

a\( n_{{{\text{O}}_{2} }} \) released and b\( n_{\text{CO}} \) produced by Ce0.9Ln0.05Ag0.05O2−δ materials during three cycles

Figures 8 and 9 show that the \( n_{{{\text{O}}_{2} }} \) released by all the Ce0.9Ln0.05Ag0.05O2−δ materials during cycle 1 was considerably higher than cycle 2. For example, the \( n_{{{\text{O}}_{2} }} \) released by La5Ag5Ce, Nd5Ag5Ce, Gd5Ag5Ce, and Dy5Ag5Ce in cycle 2 was lower by 77.9%, 64.6%, 76.6%, and 80.0% as compared to cycle 1. The comparison between cycle 2 and cycle 3 shows a different story. The \( n_{{{\text{O}}_{2} }} \) released by all the Ce0.9Ln0.05Ag0.05O2−δ materials in cycle 3 was slightly less when compared to cycle 2. For instance, the \( n_{{{\text{O}}_{2} }} \) released by La5Ag5Ce, Nd5Ag5Ce, Gd5Ag5Ce, and Dy5Ag5Ce in cycle 3 was lower by 2.0%, 0.0%, 14.8%, and 1.8% than cycle 2. Based on the results given in Figs. 8 and 9, it can be concluded that the prime reason for the more substantial O2 evolution in cycle 1 was the additional loss in the mass of the Ce0.9Ln0.05Ag0.05O2−δ materials due to the release of volatile chemicals.

In the case of the CDS step, the \( n_{\text{CO}} \) produced by each Ce0.9Ln0.05Ag0.05O2−δ material first decreased in cycle 2 (compared to cycle 1) and remained approximately stable in cycle 3 (compared to cycle 2). For example, the \( n_{\text{CO}} \) produced by La5Ag5Ce, Nd5Ag5Ce, Gd5Ag5Ce, and Dy5Ag5Ce in cycle 2 was 11.1%, 20.0%, 7.4%, and 24.5% lower than cycle 1. In contrast, the %decrease in the \( n_{\text{CO}} \) produced by La5Ag5Ce, Nd5Ag5Ce, Gd5Ag5Ce, and Dy5Ag5Ce in cycle 3 was dropped to 1.1%, 0.5%, 0.9%, and 1.2% when compared with the cycle 2 data. The \( n_{\text{CO}} /n_{{{\text{O}}_{2} }} \) ratio of all the Ce0.9Ln0.05Ag0.05O2−δ materials increased significantly in cycle 2 when compared with cycle 1. For instance, the \( n_{\text{CO}} /n_{{{\text{O}}_{2} }} \) ratio of La5Ag5Ce, Nd5Ag5Ce, Gd5Ag5Ce, and Dy5Ag5Ce increased by 1.49, 1.01, 1.13, and 1.09 in cycle 2 when compared with cycle 1. The \( n_{\text{CO}} /n_{{{\text{O}}_{2} }} \) ratio of all the Ce0.9Ln0.05Ag0.05O2−δ materials in cycle 2 and cycle 3 were almost identical.

The results obtained during cycle 3 indicate that most of the Ce0.9Ln0.05Ag0.05O2−δ materials are moving toward achieving a stable redox reactivity. For attaining further confirmation about the stable redox reactivity, the Ce0.9Ln0.05Ag0.05O2−δ materials were tested for ten successive cycles. The TGA profiles associated with the ten cycles are shown in Fig. 10. It was already confirmed that the data obtained in cycle 1 is misleading, and hence the TGA analysis was more focused on the remaining nine cycles. The \( n_{{{\text{O}}_{2} }} \) released and \( n_{\text{CO}} \) produced by each Ce0.9Ln0.05Ag0.05O2−δ material from cycle 2 to cycle 10 are shown in Figs. 11 and 12.

Figure 10
figure 10

TGA profiles obtained during ten successive cycles performed using Ce0.9Ln0.05Ag0.05O2−δ materials

Figure 11
figure 11

\( n_{{{\text{O}}_{2} }} \) released by Ce0.9Ln0.05Ag0.05O2−δ materials during ten successive cycles

Figure 12
figure 12

\( n_{\text{CO}} \) produced by Ce0.9Ln0.05Ag0.05O2−δ materials during ten successive cycles

As per the data given in Fig. 11, the La5Ag5Ce, Pr5Ag5Ce, and Nd5Ag5Ce shows a stable release of O2 from cycle 2 to cycle 10. The Gd5Ag5Ce indicates redox stability in terms of constant O2 release from cycle 3 to cycle 10. For the rest of the Ce0.9Ln0.05Ag0.05O2−δ materials, i.e., Sm5Ag5Ce, Tb5Ag5Ce, Dy5Ag5Ce, and Er5Ag5Ce, a steady \( n_{{{\text{O}}_{2} }} \) evolution was realized after cycle 5 or cycle 6. In terms of average \( n_{{{\text{O}}_{2} }} \) released from cycle 2 to cycle 10, La5Ag5Ce (72.2 μmol of O2/g cycle) and Tb5Ag5Ce (60.2 μmol of O2/g cycle) were observed to be the best and worst among all the Ce0.9Ln0.05Ag0.05O2−δ materials.

As shown in Fig. 12, the La5Ag5Ce, Pr5Ag5Ce, Nd5Ag5Ce, Gd5Ag5Ce, and Dy5Ag5Ce showed a stable production of CO from cycle 2 to cycle 10. In contrast, a constant \( n_{\text{CO}} \) production in the case of Sm5Ag5Ce, Tb5Ag5Ce, and Er5Ag5Ce was noticed from cycle 6 to cycle 10. In terms of the average \( n_{\text{CO}} \) produced from cycle 2 to cycle 10, the Ce0.9Ln0.05Ag0.05O2−δ materials can be arranged as: La5Ag5Ce > Nd5Ag5Ce ~ Pr5Ag5Ce > Gd5Ag5Ce > Tb5Ag5Ce > Sm5Ag5Ce ~ Er5Ag5Ce > Dy5Ag5Ce. According to the data given in Fig. 13, the re-oxidation ability (average \( n_{\text{CO}} \) / \( n_{{{\text{O}}_{2} }} \) ratio) of the La5Ag5Ce and Tb5Ag5Ce was the highest as compared to the rest of the Ce0.9Ln0.05Ag0.05O2−δ materials. Based on \( n_{{{\text{O}}_{2} }} \) released and \( n_{\text{CO}} \) produced from cycle 2 to cycle 10, the La5Ag5Ce appears to be the most excellent candidate among all the Ce0.9Ln0.05Ag0.05O2−δ materials investigated in this study.

Figure 13
figure 13

Average \( {\raise0.7ex\hbox{${n_{\text{CO}} }$} \!\mathord{\left/ {\vphantom {{n_{\text{CO}} } {n_{{{\text{O}}_{2} }} }}}\right.\kern-0pt} \!\lower0.7ex\hbox{${n_{{{\text{O}}_{2} }} }$}} \) ratio of the Ce0.9Ln0.05Ag0.05O2−δ materials

Table 4 reports the comparison of Ce0.9Ln0.05Ag0.05O2−δ materials with the CeO2 and Ce0.9Ln0.1O2 materials. The results given in Table 4 shows that all the Ce0.9Ln0.05Ag0.05O2−δ materials were capable of higher \( n_{{{\text{O}}_{2} }} \) release (except for Sm5Ag5Ce) and \( n_{\text{CO}} \) production than CeO2 and their corresponding Ce0.9Ln0.1O2 materials. For instance, the \( n_{{{\text{O}}_{2} }} \) released by La5Ag5Ce, Pr5Ag5Ce, Nd5Ag5Ce, Gd5Ag5Ce, Tb5Ag5Ce, Dy5Ag5Ce, and Er5Ag5Ce was higher by 21.8 μmol of O2/g cycle, 23.1 μmol of O2/g cycle, 23.0 μmol of O2/g cycle, 18.6 μmol of O2/g cycle, 0.20 μmol of O2/g cycle, 12.1 μmol of O2/g cycle, and 8.50 μmol of O2/g cycle as compared to Ce0.9La0.1O2, Ce0.9Pr0.1O2, Ce0.9Nd0.1O2, Ce0.9Gd0.1O2, Ce0.9Tb0.1O2, Ce0.9Dy0.1O2, and Ce0.9Er0.1O2, respectively. Similarly, the La5Ag5Ce, Pr5Ag5Ce, Nd5Ag5Ce, Sm5Ag5Ce, Gd5Ag5Ce, Tb5Ag5Ce, Dy5Ag5Ce, and Er5Ag5Ce produced 1.39, 1.44, 1.52, 1.32, 1.49, 1.43, 1.29, and 1.37 times higher CO when compared with the Ce0.9La0.1O2, Ce0.9Pr0.1O2, Ce0.9Nd0.1O2, Ce0.9Sm0.1O2, Ce0.9Gd0.1O2, Ce0.9Tb0.1O2, Ce0.9Dy0.1O2, and Ce0.9Er0.1O2, respectively. The overall results of this investigation indicate that the incorporation of Ag in the Ln-doped ceria was beneficial to improve the redox performance of Ce0.9Ln0.05Ag0.05O2−δ materials.

Table 4 Comparison between the pure CeO2−δ, Ce0.9Ln0.1O2−δ, and Ce0.9Ln0.05Ag0.05O2−δ [19,33]

Summary and conclusions

The PXRD and EDS analysis have confirmed the formation of nominally phase pure Ce0.9Ln0.05Ag0.05O2−δ materials via co-precipitation of the hydroxide method. The average crystallite size of the derived Ce0.9Ln0.05Ag0.05O2−δ materials was in the range of 32–64 nm. The SEM analysis further established a spherical nanostructured particle morphology of the Ce0.9Ln0.05Ag0.05O2−δ materials. The SEM analysis also indicates that the doping of the lanthanides and silver does not have any significant effect on the morphology of Ce0.9Ln0.05Ag0.05O2−δ materials. Based on the results associated with the TGA analysis, the redox reactivity of the Ce0.9Ln0.05Ag0.05O2−δ materials can be ranked as: Ce0.911La0.053Ag0.047O1.925 > Ce0.921Nd0.049Ag0.050O1.940 ~ Ce0.892Pr0.051Ag0.051O1.886 > Ce0.889Gd0.054Ag0.050O1.884 > Ce0.905Tb0.051Ag0.046O1.909 > Ce0.896Sm0.048Ag0.053O1.890 ~ Ce0.899Er0.050Ag0.055O1.900 > Ce0.910Dy0.053Ag0.048O1.923. The TGA results also indicate that the Ce0.911La0.053Ag0.047O1.925 and Ce0.905Tb0.051Ag0.046O1.909 possess the highest re-oxidation ability as compared to the rest of the Ce0.9Ln0.05Ag0.05O2−δ materials. The overall results of this investigation indicate that the inclusion of Ag into the Ln-doped ceria considerably improved the O2 releasing and CO production capability of the Ce0.9Ln0.05Ag0.05O2−δ materials as compared to the phase pure CeO2 and Ce0.9Ln0.1O2 materials. After estimating the CO production capacity, our research team is currently focused on the production of H2 by using the Ce0.9Ln0.05Ag0.05O2−δ materials.