2. Results and Discussions
Several milling techniques were used for the evaluation of their effect on the textural and redox properties of the milled CZ materials. Most of the experiments were conducted with commercial CZ20, with a composition of 20%CeO
2–70%ZrO
2–5%La
2O
3–5%Pr
6O
11. A typical particle size distribution curve of the unmilled CZ material is presented in
Figure 1a. Unmilled CZ20 has particles in a wide range from 5 microns up to 100 microns with a d
50 of ~28 microns. According to SEM, the CZ consisted of aggregates with rounded or irregular shapes. These aggregates appeared dense; however, they had a complex hierarchical structure. Primary particles (crystallites) with a 10–20 nm size were stuck together and formed first-order aggregates with a d
50 < 0.5 μm. Further assembly resulted in second- and higher-order aggregates with larger sizes (
Figure 2a).
The first technique used was wet–ball jar milling. Despite relatively slow jar rotation, this process enabled the production nano-slurries in which the CZ particles could have a d
50 of less than 0.20 microns. Jar milling efficiently destroyed large, high-order aggregates with the formation of rather soft and homogeneous by size first-order aggregates (
Figure 1b and
Figure 2b). The size and shape of such powders allows us to conclude that the jar mill energy is not sufficient for producing nanomaterials consisting of individual crystallites. The process drawbacks are a very long milling time and re-agglomeration of the nanomaterials during drying.
Pore size distribution curves of the unmilled CZ20 and CZ20 milled to d
50 = 0.2 microns are presented in
Figure 3. There was no change in the pore diameter despite severe material deagglomeration. This suggests that first-order aggregates are the main structural blocks in the CZ powders. They bear the main features of the material’s mesoporosity and are responsible for the overall thermal stability of the material.
The specific surface areas and pore volume of the values for fresh, aged-milled, and unmilled materials are summarized in
Table 1. The data show that jar milling resulted in a decrease in the S
a and V
pore of the milled materials. However, despite severe deagglomeration, the decreases in S
a and V
pore were not drastic. The nano-CZ retained ~70–75% of its surface area and pore volume after slurry drying. The retention of a high specific surface area and porosity is extremely important during the catalyst substrate coating.
Further evaluation of the milling effect on the CZ properties included phase stability and redox properties. The XRD powder patterns presented in
Figure 4 show that the unmilled CZ20 had a tetragonal crystalline structure and that jar milling did not change the CZ20 crystal phase.
The redox activity of CZ materials was studied using the TPR-H
2 technique. TPR-H
2 allows to determine the available oxygen storage capacity (OSC) and, based on the T
max position, provides an indication of the CZ oxygen mobility. The OSC depends mainly on the composition and increases linearly with the increase in ceria in CZ up to ~30–35%CeO
2, and then it remains practically unchanged at 1.0–1.1 mM H
2/g for materials with higher CeO
2 content [
18]. The oxygen mobility in CZ depends on the following factors, including:
The Ce/Zr molar ratio, the amount of trivalent RE dopants defining the concentration of oxygen defects in the crystal lattice, the type and size of the RE dopants influencing the lattice parameters and channel dimensions, the crystallite size determining the pathway that oxygen needs to take to reach the surface from the bulk, and the chaotic or ordered orientation of domains in the crystallites affecting the activation energy for grain–boundary oxygen diffusion.
Typically, CZ materials with a higher CeO
2 content have faster oxygen mobility than zirconia-rich CZ. The T
max position was in the range of 400 to 600 °C (after aging at 1000–1100 °C). The TPR-H
2 profiles for the reference and jar-milled CZ20 (d
50 = ~0.2 microns) are presented in
Figure 5. The data shows that jar milling did not improve or deteriorate the redox properties of CZ. The oxygen storage capacity in both materials after aging was close to the theoretical value (~0.7 mMol H
2/g), and the T
max, which characterizes how facile the oxygen mobility is, was in the range of 470–485 °C for both.
Summarizing, it is possible to conclude that jar milling is a powerful tool for making CZ nano-slurries that retain the main features of the starting material after drying, namely, the phase stability, mesoporosity, high surface area, and pore volume, as well as the unchanged redox properties.
The second tested milling technique was Eiger milling, which is widely used in industry. This is also a form of wet ball milling, but differing from jar milling in that it employs a variable agitator speed (from hundreds of rotations per minute (RPM) to several thousands of RPM), resulting in the CZ deagglomeration to nano-size via attrition, shear, and direct impact being much faster. Another difference is that the collision energy of the milling media with CZ particles was significantly higher, which created local temperature spikes (quasi-hydrothermal environment). The efficiency of Eiger milling depends on several factors, including the bead size, agitator speed, duration, and CZ composition. The data in
Table 1 show a significant effect of milling on the surface area and pore volume of fresh and aged CZ. The deterioration of S
a and V
pore was much stronger in comparison to that of jar milling. The pore volume was affected the most severely, as it decreased by more than 50%. However, the pore diameter of the milled CZ remained practically unchanged, which supports the suggestion that first-order aggregates are the main structural blocks in CZ powders. Eiger milling also did not affect the CZ phase stability.
However, Eiger milling had a significant impact on the CZ redox properties. The TPR-H
2 profiles of unmilled CZ20 and CZ20 Eiger-milled to different depths are presented in
Figure 6. Eiger milling did not affect the available oxygen storage capacity of CZ20, but it enhanced the oxygen mobility in CZ. There was a significant shift of the T
max to lower temperatures, which depended on the milling depth: T
max = 415 °C for CZ20 with d
50 = 1.35 microns and T
max = 350 °C for CZ20 with d
50 = 0.54 microns. This is clear evidence of the presence of mechanochemical activation of the CZ redox properties.
The last milling technique tested was steam jet milling. In this case, the grinding energy was provided by super-heated steam jets, with a temperature of ~280 °C and a pressure of ~10 bar. The use of super-heated steam in the milling process eliminated the possibility of CZ contamination with elements from the milling media (as in jar or Eiger milling), making the milling process fast and resulting in non-aggregated nano-size powders. In the steam jet milling experiments, we used two CZ compositions: CZ20 and CZ40 (40%CeO2–50%ZrO2–5%La2O3–5%Pr6O11). By changing the milling duration, two samples with different milling depths for each CZ composition were made and characterized similarly to the materials prepared via jar and Eiger milling.
It was found that steam-jet milling efficiently destroyed large, high-order aggregates with the formation of sub-micron-sized powders. However, the maximum milling depth achieved by SJM milling (~0.5 microns) was somewhat lower than in the case of jar or Eiger milling. SEM and TEM images of SJM CZ20 are shown in
Figure 7. SEM and TEM images of CZ40 before and after milling are shown in
Figure 8a,b, respectively.
The data on the surface areas and pore volumes of CZ20 and CZ40 before and after steam jet milling are summarized in
Table 2, and the pore size distribution curves for the CZ40 samples are provided in
Figure 9. As seen, steam jet milling led to a relatively minor shift in the CZ pore diameter from ~30 to ~25 nm, but resulted in the disappearance of a significant fraction of large pores (>40–50 nm). The large pores were derived from higher-order aggregates, which were removed by milling. The data in
Table 2 shows a very negligent effect of milling on the surface area of fresh CZ20 and CZ40 independent of milling depth, but a substantial loss of surface area and especially of pore volume was observed after milling. Those CZ materials aged at 1000 and 1100 °C had lower S
a and V
pore values than the starting material, and their deterioration rates strongly depended on the milling depth. Milling to a d
50 of ~0.5 microns resulted in more than a 50% loss of pore volume and up to a 30–40% loss of surface area.
According to the XRD data (
Figure 10), steam jet milling does not change the CZ crystal phase. The steam jet-milled CZ materials retained their well-defined tetragonal structures after aging at 1100 °C.
The TPR-H
2 profiles for CZ20 and CZ40 unmilled and steam jet-milled to different depths are presented in
Figure 11 and
Figure 12, respectively. The oxygen storage capacity of the unmilled CZ20 was 0.70 mMol H
2/g, and that of CZ40 was 1.05 mMol H
2/g, with the CeO
2 content being higher in CZ40. As seen, the T
max position of the CZ20 reference sample was 485 °C, which is typical for zirconium-rich CZ. The T
max of the steam jet-milled samples was significantly shifted to lower temperatures, and this shift depended on the milling depth: T
max = 220 °C for SJM-CZ with d
50 = 1.1 microns and T
max = 155 °C for CZ with d
50 = 0.5 microns. Such a significant T
max shift for 250–330 °C has only been reported previously in the case of CZ doping with precious metals (PGM) [
1]. The OSC of CZ20 milled to 1.1 and 0.5 microns was 0.68 and 0.70 mMol H
2/g, respectively. A very similar situation was observed in the case of steam jet milling of CZ40. The T
max of the unmilled CZ40 was 495 °C. The material steam jet-milled to 0.8 microns had a bimodal TPR-H
2 profile with a high temperature peak at 490 °C and a low temperature peak at 350 °C. The third CZ sample, after longer milling (d
50 = 0.5 microns), had a practically single modal TPR-H
2 profile with a T
max position of 200 °C. The OSC of the CZ40 milled to 0.8 and 0.5 microns was 1.03 and 0.99 mMol H
2/g, respectively. These data show that steam jet milling results in a significant shift of the T
max in milled materials but does not change the oxygen storage capacity.
The most plausible explanation for this phenomenon is that steam jet mechanochemical milling in a quasi-hydrothermal environment (with potential local temperature spikes over 500 °C) induces distinctive stress in the CZ lattice that facilitates oxygen mobility. This was confirmed by the HRTEM data of the unmilled and steam jet-milled CZ40. The HRTEM of unmilled CZ showed an almost perfect arrangement of atoms in all directions (
Figure 13). A similar HRTEM image of the CZ crystallite after steam jet milling showed multiple lattice distortions spread non- homogeneously through the crystal (
Figure 14). The d-values measured at multiple (hkl) locations for CZ40 and SJM-CZ40 are provided in
Table 3. The d-values for the same (hkl) location for the SJM-CZ40 material varied from approximately 2 to 5% from the d-value measured for a reference cerium-zirconium material at the same (hkl) location. Such lattice distortions created by mechanical stress influence the metal–oxygen bonds, making them weaker and the oxygen more mobile in some cases.
For coating the TWC substrate, mixed CZ oxides, together with other components (alumina and PGM), were used in the formation of aqueous slurry. Thus, the question arises if such a treatment can make an impact on the fast oxygen mobility. To check this, an aqueous SJM-CZ20 (d
50 = 0.5 microns) slurry was made, mixed continuously for 1 day, and then dried and calcined at 1000 °C for 6 h. The characterization data of the textural and redox properties are summarized in
Table 4, and the TPR-H
2 profiles of the steam jet-milled CZ20 before and after re-slurring are presented in
Figure 15. As seen, re-slurring did not change the specific surface area, but increased the pore volume by more than 35%. The latter could be related to nano-particle re-agglomeration during slurry drying, with the formation of higher order aggregates having large mesopores. The TPR-H
2 profiles of the two tested samples look similar: Very close T
max values (~160 °C) and identical OSCs of ~0.7 mM H
2/g equal to the theoretical values. This indicates that the dispersion of nano-SJM-CZ in water followed by drying and the formation of micron-sized aggregates did not release the lattice stresses, which are responsible for the fast oxygen mobility, caused by high-energy milling.
The tolerance of the fast oxygen mobility effect toward changes in the cerium oxidation state was studied in consecutive TPR-H
2 runs, mimicking changes in a rich-lean exhaust gas environment. Two materials were used in the experiment: Steam jet-milled nano-CZ20 with d
50 = 0.5 μm and the same material after re-slurring in water and drying. The TPR-H
2 profiles of six consecutive runs for nano-SJM-CZ20 are presented in
Figure 16, and those for SJM-CZ20 after re-slurring are in
Figure 17. In both cases, there were no changes in the available OSC. However, the consecutive TPR-H
2 runs had an impact on the TPR profiles. There was a gradual decrease in the low-temperature TPR peak at ~155–200 °C and an increase in the high-temperature peak at 400–450 °C. This shows that the fast oxygen mobility effect is not durable and the lattice distortions induced by high energy milling are metastable. Valence-state Ce
IV (0.097 nm) to Ce
III (0.114 nm) oscillations gradually released the lattice stresses.
However, these changes depend on the material. The data in
Table 5 show that for nano-SJM-CZ20 in the first TPR-H
2 run, the fraction of available OSC at a temperature below 300 °C was ~83% and that below 400 °C was ~94%, but it fell rapidly and reached only 17% and 44%, respectively, on the sixth run. A similar but less pronounced decay effect was observed for SJM-CZ20 after re-slurring. The fraction of available OSC on the sixth TPR-H
2 run at a temperature below 300 °C was 50% and below 400 °C was 75%, which is still very high. The data indicate that the re-slurring and re-agglomeration of nano-SJM-CZ stabilized lattice distortions/stresses.
Nano-CZ20 with d
50 = 50 microns was also impregnated with different amounts (from 0.01% to 0.1%) of rhodium or palladium. The data in
Table 6 show that the non-impregnated nano-CZ20 had a very low T
max of 165 °C and that the presence of PGM resulted in a further shift of the T
max to even lower temperatures. Only 100 ppm of Rh was sufficient to move the T
max to below 100 °C.
The typical TPR-H
2 profiles of SJM-CZ20, SJM-CZ20 doped with 0.1%Rh, and unmilled CZ20 doped with 0.1%Rh are presented in
Figure 18. It is known that the impregnation of CZ with PGM results in a significant shift in T
max from 450–550 to 150–250 °C. As seen, the impregnation of CZ20 with 0.1%Rh shifted the T
max to ~150–200 °C. The CZ20-0.1%Rh TPR-H
2 profile and the T
max value were very similar to that of the SJM-CZ20 nanomaterial without Rh. The impregnation of nano-CZ20 with 0.1%Rh shifted the T
max to 67 °C and changed the TPR profile to bimodal with two distinct sharp peaks.
The TPR-H
2 profiles of SJM-CZ20 impregnated with 0.1%Pd after five consecutive runs are shown in
Figure 19. There were no changes in the available OSC (~0.7 mM H
2/g), peak shape, or T
max, which indicates that the PGM impregnation had a stabilizing effect on the fast oxygen mobility.