Hydrothermal stability of CeO₂–Al₂O₃ composites, a structural study and oxygen storage capacities

Abstract

A series of CeO₂–Al₂O₃ composites with different Ce content were prepared. The uniformity of the CeO₂ dispersion was confirmed using H₂-TPR and HR-TEM. Two aging treatments were conducted, and the CeO₂–Al₂O₃ composites show superior hydrothermal stability. The sintering of CeO₂ and Al₂O₃ are independent of each other based on XRD and HR-TEM results. On the other hand, the dynamic oxygen storage capacity (DOSC) is mostly activated after 20 h of aging at 750 °C, and deactivated after 10 h of aging at 1050 °C. Combining the results of structural and DOSC studies, the interaction between CeO₂ and Al₂O₃ can be divided into two parts, (1) a chemical interaction which negatively impacts the DOSC, and (2) a spatial limitation which benefits the sample stability. The former interaction is eliminated after hydrothermal aging at 750 °C, while the later one exists even after hydrothermal aging at 1050 °C.

---

1. Introduction

The control of exhaust emissions has attracted more and more public attention, bringing a great number of studies including controlling strategies and catalyst design.¹ Catalysts used in emission control always suffer significant deactivation as a result of high temperature and redox oscillations,² which makes it important to study deactivation processes. However, previous studies have mostly focused on the evolution of the active component,^3,4 while consideration of the support oxide is lacking.⁵ In this work, a CeO₂–Al₂O₃ composite was prepared, and the intrinsic properties and aging mechanism were studied, as a complement to common studies about catalyst aging. TWC, the commonly used oxygen storage material, mainly contains ceria, zirconia and alumina, and is known as the third-generation ceria–zirconia developed by Toyota.⁶ The sample system is simplified to only ceria and alumina in our study, because ceria presents an obvious oxygen storage capacity (although lower than Ce_1−xZr_xO₂), which is good enough to understand the interaction between the OSC material and alumina, as well as its evolution during hydrothermal aging. This makes catalyst design easier and clearer.

The combination of ceria and alumina has previously been studied, which gave a catalyst with better low temperature activity and good stability, compared with the two pure oxides.^7–11 The published synthesis methods can be divided by the mixing strategy, (1) atomic mixing, mixing the Ce and Al source as a salt solution (sol method or co-precipitation),⁷ (2) molecular, mixing crystalline CeO₂ with Al₂O₃ sol,⁸ and (3) impregnation, impregnating the Ce salt on Al₂O₃ powder.^9–11 Atomic mixing is effective but difficult to control, because of the difference in the dynamics of nucleation. On the contrary, impregnation is easily controlled but less effective compared with the others. Therefore, a molecular mixing strategy was chosen, in order to achieve a controllable synthesis and good dispersion of CeO₂. The properties of the sample and their evolution during hydrothermal aging were studied.

2. Experimental methods

2.1 Sample preparation

CeO₂ precipitate and an Al₂O₃ sol were prepared separately. CeO₂ was prepared using a precipitation method. 0.1 M Ce(NO₃)₃ solution was dropped into ammonia aqua (analytical grade, KEWEI), when air was bubbled into the reactor to oxidize Ce³⁺. The NH₃·H₂O used in the reaction was in 100% excess compared to the reaction stoichiometry. After the reaction, the precipitate changed to light yellow through further gas bubbling, and a decay process was conducted with stirring at 90 °C for 1 day. After that, CeO₂ precipitate was filtered out, washed and re-dispersed into distilled water. Al₂O₃ sol was prepared by precipitation of 0.1 M Al(NO₃)₃ and 3 M NH₃·H₂O. Ammonia was added into the Al(NO₃)₃ solution in a dropwise manner. At the end of the reaction, the pH of the sol was tuned to 7. This sol was decayed at 60 °C for 1 day and at 90 °C for another day. Then the ceria precipitate and Al₂O₃ sol were mixed together, and PEG-4000 (poly ethylene glycol) was added in a mole-ratio of 0.03 based on the amount of oxide (CeO₂ + Al₂O₃). After one hour of ultrasonic treatment, the mixture was transferred into a spraying drying device, where the drying process (particle preparation) was finished in several seconds. The obtained sample was calcined at 300 °C for 2 h and at 600 °C for 3 h. Samples with a Ce mole ratio of 50, 25, 15 and 10% were prepared and named CA1, CA2, CA3 and CA4, respectively. Pure ceria and alumina were also prepared using the same method as references, and named C and A, respectively.

Two hydrothermal aging processes were conducted, at 750 °C for 20 h and at 1050 °C for 10 h (both in 10% H₂O/air) to simulate the aging in diesel and gasoline engine vehicles, respectively. The samples were named f (fresh sample), -750 h and -1050 h, respectively.

2.2 Characterizations

The mole ratios of CeO₂ in the CeO₂–Al₂O₃ composites were determined using X-ray fluorescence (XRF). Surface area measurements were performed on an F-sorb 3400 chemisorption apparatus using the BET method. Powder X-ray diffraction (XRD) was conducted on a Bruker D8-Focus diffractometer operated at 40 kV and 40 mA with nickel filtered Cu Kα radiation (λ = 0.15418 nm). The scanning was conducted between 20 and 90°, with a step speed of 5° min⁻¹. The diffraction peak at 2θ = 28.6°, according to the CeO₂ (111) face, was used to calculate the average crystal size using the Scherrer equation. SEM (scanning electronic microscopy) images were collected on an S-4800 scanning electron microscope working at 15 kV. Gold deposition was conducted to improve the conductivity of the samples. (HR) TEM images were collected on a Tecnai G2 F20 field emission transmission electron microscope (TEM) working at 200 kV. 0.05 g of the sample was dispersed into 50 ml alcohol for 1 h, and the dispersion was dropped on an ultra-thin carbon film and dried before the test. H₂-TPR was conducted on a PX200 gas adsorber equipped with a TCD detector. 50 mg of the sample was packed in a U-type quartz tube reactor. The sample was pre-oxidized with 20% O₂/N₂ at 500 °C for 30 min and cooled down to room temperature. Then it was heated up to 900 °C at a rate of 10 °C min⁻¹, with 5% H₂/N₂. The total flow rate of the reactant gas was 30 ml min⁻¹.

Dynamic oxygen storage capacity was tested on a self-designed apparatus. 25 mg of the sample blended with 40 mg of quartz sand was packed in a tubular reactor. Pulses of CO (4% CO/1% Ar/He, 400 ml min⁻¹) and O₂ (2% O₂/1% Ar/He, 400 ml min⁻¹) were purged into the reactor alternately at a frequency of 0.1 and 0.05 Hz. The gas outlet was analyzed using a Hiden 2.0 mass spectrometer.

3. Results and discussion

3.1 CeO₂ dispersion in the Al₂O₃ framework

The molar ratio of CeO₂/(CeO₂ + Al₂O₃) is listed in Table 1. For all of the samples, the detected ceria content is slightly larger than the designed value, but is within the experimental error. The microstructures of the CA particles were observed using SEM, and most of the particles are spheres with sizes of 1–10 microns. Particle size distributions are displayed in the insets of Fig. 1. Most of the particles have sizes of 1–4 μm. No obvious differences are found among the different samples, indicating that the physical structure is similar for each sample.

Table 1 BET surface area and CeO₂ crystal size before and after aging

Sample	Ce ratio (mol%)	SBET (m^2 g^−1)			D-CeO2 (nm)
		F	750 h	1050 h	F	750 h	1050 h
C	100	140	32	3	5.5	18.1	>100
CA1	52	288	157	49	4.5	8.2	33.3
CA2	26	389	224	75	4.6	8.5	29.9
CA3	16	469	224	92	5.1	8.6	29.1
CA4	12	444	225	98	4.7	8.6	24.8
A	0	469	240	42	—	—	—

---

Fig. 1 Structures of the samples under SEM (top, CA1-f; bottom, CA4-f).

Since the density of CeO₂ is larger than Al₂O₃, the CeO₂–Al₂O₃ composites in the present work can be considered as crystalline CeO₂ dispersed in a Al₂O₃ framework. Therefore, the dispersion is of great importance to the sample properties, which were investigated using H₂-TPR. As displayed in Fig. 2, there are three peaks centered at 450, 550 and 720 °C for the CA samples. The reduction temperature is independent of the CeO₂ content, while the peak intensity is roughly proportional to the cerium content (mole ratio). Meanwhile, pure CeO₂ shows reduction peaks at 390, 507 and 810 °C, for surface, subsurface and bulk oxygen, respectively.¹² The deviation of the CA composite from pure CeO₂ indicates that the chemical environment of CeO₂ in the CA composite is different from that of pure CeO₂. However, the reduction temperature of each peak is independent of the cerium content (only the intensity is changed). Therefore, the chemical environment of CeO₂ is independent of the CeO₂ ratio, which infers a uniform dispersion in all of the samples.

Fig. 2 H₂-TPR of the CA composite, using C as a reference.

On the other hand, the peak shift of the surface/subsurface oxygen indicates that the crystalline surface becomes more difficult to be reduced, as a result of a CeO₂–Al₂O₃ interaction. But the peak for the bulk oxygen shifts in the reverse direction. According to the literature, Eleonora et al.¹² studied H₂-TPR on CeO₂, and found a significant decrement of the BET surface area at the beginning of the last reduction peak. So they concluded that the last peak, according to the bulk-like oxygen, is related with a re-construction of crystalline CeO₂ during the experiment. Therefore, the evolution of the last peak (weakened and shifted to lower temperature) can be explained by the reconstruction of crystalline CeO₂ being limited by the presence of Al₂O₃.

The microstructure of the CA composite was detected using HR-TEM, using CA2-f as the example (Ce mole ratio 26%). As shown in Fig. 3, the particle is a spherical shape at the TEM resolution. The non-uniformity of the particle colour indicates the existence of CeO₂-rich and Al₂O₃-rich areas. The edge of the particle in Fig. 5A was amplified step by step. At the highest resolution (Fig. 5D), the crystalline CeO₂ fringes can be identified (0.31 nm), which are attributed to the CeO₂ (111) interplanar distance. However, no crystalline fringes can be identified for Al₂O₃, as a result of the poor crystallinity of alumina. In Fig. 3D, several CeO₂ crystals are presented together, while another one is several nanometres from them. Considering that the excess surface energy makes the crystalline aggregate spontaneously, the isolated CeO₂ indicates the existence of a CeO₂–Al₂O₃ interaction which stabilizes the crystalline CeO₂ from its neighbours (Fig. 4).

Fig. 3 Sample structure under (HR) TEM (CA2-f, images (B–D) are amplifications of the cubic area in the former images).

Fig. 4 XRD patterns of the samples before and after hydrothermal aging.

3.2 Evolution of microstructure in hydrothermal aging

The evolution of the microstructure after aging was studied. As shown in Table 1, the surface areas of CeO₂ (C) and Al₂O₃ (A) are 140 and 469 m² g⁻¹, respectively, while the value for the CA composite increased as the ratio of CeO₂ decreased. After hydrothermal aging at 750 °C, the surface areas of the pure oxides decreased to 32 (CeO₂) and 240 m² g⁻¹ (Al₂O₃), and the value for the CA sample is between 157 and 225 m² g⁻¹, which increased as the CeO₂ ratio decreased. After aging at 1050 °C, the surface areas of the pure oxides decreased to 3 (CeO₂) and 42 (Al₂O₃) m² g⁻¹, but the value for CA is between 49 and 98 m² g⁻¹. This indicates a superior hydrothermal stability of the CA composite compared with the pure oxide.

XRD patterns are displayed in Fig. 4. For pure CeO₂, only the cubic fluorite phase is observed, and the diffraction peaks of the aged sample are sharper after aging, which indicates the increase of the CeO₂ crystal size. For pure Al₂O₃, the main phase of the fresh and 750 h samples is γ-Al₂O₃, which changes to the α-phase after aging at 1050 °C. For the CeO₂–Al₂O₃ composite, the main peaks of the fresh samples are the cubic fluorite phase of CeO₂, even when the Ce mole ratio is 12%. Hydrothermal aging at 750 °C does not induce an obvious difference in the diffraction patterns, and the CeO₂ crystal size calculated using the peak width (2θ = 28.6°, (111)) shows limited growth compared with the results of the fresh samples (from 5 to 8 nm). Meanwhile, γ-Al₂O₃ was only observed in C3 and C4, as a wide shoulder at 2θ = 46°. This can be attributed to the low crystallinity of alumina, which is covered by the presence of CeO₂. After aging at 1050 °C, the crystalline growth of CeO₂ becomes obvious, and a smaller Ce content induces a smaller crystal size. Meanwhile, the relative intensity for α-Al₂O₃ becomes stronger (based on the Ce (111) diffraction peak) as the Ce content decreases. According to the above results, the decrement of the surface area induced by hydrothermal aging at 750 °C (from ∼400 to ∼200 m² g⁻¹) can be attributed to either the phase transition (from amorphous to the γ-phase) of Al₂O₃, or the crystalline growth of CeO₂. On the contrary, the further decrement of the surface area in aging at 1050 °C indicates significant sintering, which is accompanied by the crystalline growth of CeO₂ and a γ–α phase transition of Al₂O₃. It is noteworthy that the crystalline growth of CeO₂ is in a reverse correlation with the formation of α-Al₂O₃, indicating that the sintering of the two oxides is independent of each other.

HR-TEM images after hydrothermal aging at 1050 °C were collected for CA2-1050 h (as shown in Fig. 5). At the TEM resolution (Fig. 5A), the edge of the particle becomes more distinct, which indicates significant enlargement of the crystals. A particle fragment was found (obtained by ultrasonic treatment), which allows the investigation of the microstructure of the CA composite. As shown in Fig. 5B, the crystals presented in the image have two different shapes, polygon and spherical crystals. Phase attribution was carried out by measuring the interplanar distances of the crystals. In Fig. 5C and D, the interplanar distance is about 0.31 nm for the polygon crystals, which is attributed to the CeO₂ (111) face. In Fig. 5E, the interplanar distances of the spherical crystal are 0.21 and 0.25 nm, which can be attributed to the α-Al₂O₃ (113) and (104) faces, respectively. The crystal size of CeO₂ (polygon) agrees well with the results of XRD. However, the existence of α-Al₂O₃ in sizes of 5–10 nm indicates that the γ–α phase transition occurs without sintering. In order to confirm the effect of CeO₂ to Al₂O₃ sintering, the CA4-1050 h particles were also investigated using HR-TEM (Fig. 5F). Compared to CA2-1050 h, the well defined crystals in sizes of about 500 nm can be attributed to the sintered α-Al₂O₃, and the result agrees well with XRD, that the Al₂O₃ sintering is more serious as the Ce content decreases. Therefore, the sintering of CeO₂ and Al₂O₃ is independent of each other, and the presence of CeO₂ does not stop the γ–α phase transition, but hinders the sintering of α-Al₂O₃.

Fig. 5 (HR) TEM images of the CA composite after aging at 1050 °C. ((A), a full particle of CA2; (B–E), a fragment of CA2, (C–E) are amplifications of the cubic area in (B), and (F) a particle of CA4-GH).

Fig. 6 Temperature dependence of DOSC (normalized by CeO₂ content).

3.3 Dynamic oxygen storage capacity (DOSC)

The DOSC calculated with a 0.1 Hz frequency cycle was studied. For the fresh sample, the DOSC becomes smaller in the order of CA1 > CA2 > CA3 > CA4, which is attributed to the CeO₂ content (50–10% in CA1–CA4, while Al₂O₃ is irreducible). The hydrothermal aging does not change this order, but reduces the difference among the samples. Considering that the CeO₂ content will not be changed by aging, the efficiency of CeO₂ clearly depends on the aging treatment, and may be influenced by the CeO₂ content. So, the results were normalized by CeO₂ weight, as shown in Fig. 6. For the fresh samples, the specific DOSC (based on CeO₂ weight) decreases as the CeO₂ content decreases (including the pure CeO₂), indicating that the CeO₂ surface becomes less active when the CeO₂ content decreases. After hydrothermal aging at 750 °C, the DOSC–temperature curve becomes independent of the Ce content, indicating that the crystalline CeO₂ becomes similar for all of the CA composites. Meanwhile, the specific DOSC of the CA composite is much higher than that of pure CeO₂, which is attributed to the superiority of the hydrothermal stability (see the XRD/BET results). After hydrothermal aging at 1050 °C, the larger CeO₂ content induces lower specific DOSC, indicating that the sample with a lower CeO₂ content suffers a smaller deactivation (compared to the results for 750 h and 1050 h). The results indicate that hydrothermal aging at 750 °C improves the specific DOSC at low temperature (<500 °C), and makes it less sensitive to temperature. Also, hydrothermal aging at 1050 °C decreases the specific DOSC, while the decrement becomes limited as the CeO₂ content decreases.

As the evolution of the DOSC agrees well with the sintering of crystalline CeO₂, the oxygen storage capacity was calculated based on the CeO₂ surface area. The CeO₂ surface area was calculated using the average crystal size obtained in XRD (Table 1), using a spherical assumption as shown in the following equation:

where S and m are the surface area and weight of the sample, respectively; S*, V* and ρ indicate the surface area, volume and density of crystalline CeO₂, 7.1 g cm⁻³. The results for pure CeO₂ (fresh and 750 h) are displayed as references. As shown in Fig. 7, the specific DOSC decreases as the CeO₂ content decreases for the fresh samples, and increases as the sintering degree increases. Furthermore, compared with the pure CeO₂, the specific DOSC per surface area of the CA composites is lower than for pure CeO₂, which increases to a similar value after hydrothermal aging at 1050 °C. It indicates that the smaller CeO₂ content and more significant sintering (crystalline growth) benefits the specific DOSC.

Fig. 7 The specific DOSC based on the CeO₂ surface area (a) and the activation energy as a function of reduction degree (b).

In order to further understand the evolution of DOSC, the activation energy was calculated as a function of reduction degree. The fundamental assumption of this calculation is that all of the CeO₂ is effective,^13,14 and the details of the calculation are presented in the ESI.† As displayed in Fig. 7 (bottom), the correlation between E_a and reduction degree only depends on the aging treatment, which is independent of the CeO₂ content. For fresh samples, E_a increases monotonically as the reduction degree increases from ∼60 to ∼120 kJ mol⁻¹, indicating that oxygen release becomes more difficult as the sample is reduced. After hydrothermal aging at 750 °C, the E_a becomes larger, and E_a becomes independent of the reduction degree after a certain reduction.

In the oxygen storage materials, the oxygen release will become more difficult as more oxygen is released, which can be identified by an increased activation energy.¹⁴ Furthermore, it can be attributed to the more significant influence of the oxygen transition from the bulk to the surface, which has a higher activation energy.¹⁵ Therefore, a more sensitive E_a indicates that the transition of oxygen from the bulk is easier, the insensitive part indicates that the oxygen transition is very easy, and the oxygen release is the limiting step. As shown in Fig. 7b, the sensitivity of E_a to the reduction degree (the slope) becomes smaller after aging at 750 °C and 1050 °C. It indicates that the supplementation of oxygen to the sample surface becomes easier in the aged samples. On the other hand, the 1050 h samples show similar E_a–reduction degree curves to those of the 750 h samples. Therefore, the sintering does not hinder the supplementation of surface oxygen, but reduces the total amount of oxygen released as a result of the sintering of CeO₂.

3.4 CeO₂–Al₂O₃ interaction

The combination of support materials (strong interaction with the active component but low stability) and wall materials (stronger stability) at the nano-scale has been applied to improve the stability of catalysts.^16,17 However, the possible interaction between the two materials is speculated, which can be further understood by combining the results of structural and DOSC studies.

The evolution between fresh and 750 h samples indicates a strong interaction between CeO₂ and Al₂O₃. The interaction between CeO₂ and Al₂O₃ was identified through the uniform chemical environment of CeO₂ (H₂-TPR) and the adhesion of CeO₂ crystals on the Al₂O₃ framework (HR-TEM). On the other hand, hydrothermal aging at 750 °C does not induce a significant increase of the CeO₂ crystal size, but the E_a of DOSC becomes insensitive to the reduction degree after a certain reduction. Considering that 750 °C is only high enough for CeO₂ sintering (not for Al₂O₃), the evolution of the aging at 750 °C is attributed to a reconstruction of CeO₂ with the restriction of the Al₂O₃ framework. Therefore, the strong interaction between CeO₂ and Al₂O₃ has a negative effect on the DOSC of the sample, but it can be easily eliminated.

On the other hand, the results of the two different aged samples indicate that the reason for improved hydrothermal stability is attributed to the spatial limitation, while chemical interactions should not be included. Hydrothermal aging at 750 °C only allows the sintering of CeO₂, and aging at 1050 °C allows the sintering of both CeO₂ and Al₂O₃. Although the hydrothermal aging at 1050 °C induces more serious sintering, the correlation between E_a and reduction degree is similar to the results for 750 h. Therefore, the improvement of the hydrothermal stability cannot be attributed to the chemical interaction, for the interaction only exists in the fresh sample. Considering that the sintering of the two components is independent, the good hydrothermal stability is attributed to the spatial limitation.

4. Conclusions

CeO₂–Al₂O₃ composites were prepared using a two step protocol, which included the separate preparation of a crystalline CeO₂ dispersion and a Al₂O₃ sol, and a quick drying process of the mixture using a spraying drying method. The sample has a BET surface area of about 79–90 m² g⁻¹ even after 10 h of hydrothermal aging at 1050 °C, which is good enough for the application of exhaust aftertreatment. Based on the characterization and DOSC tests, the interaction between CeO₂ and Al₂O₃ can be separated into two parts: (1) a CeO₂–Al₂O₃ interaction which decreases the specific activity of the DOSC, and (2) a spatial limitation which benefits the hydrothermal stability. The first one can be eliminated by 20 h of aging at 750 °C.

Acknowledgements

The work was financially supported by the introduction of talent and technology cooperation plan of Tianjin (14RCGFGX00849) and the National Natural Science Foundation of China (Grant No. 21576207).

References

1. C. W. Sun, H. Li and L. Q. Chen, Energy Environ. Sci., 2012, 5, 8475.
2. W. S. Epling, L. E. Campbell, A. Yezerets, N. W. Currier and J. E. Parks, Catal. Rev.: Sci. Eng., 2004, 46, 163.
3. M. Cargnello, V. V. T. Doan-Nguyen, T. R. Gordon, R. E. Diaz, E. A. Stach, R. J. Gorte, P. Fornasiero and C. B. Murray, Science, 2013, 341, 771.
4. S. K. Matam, E. V. Kondratenko, M. H. Aguirre, P. Hug, D. Rentsch, A. Winkler, A. Weidenkaff and D. Ferri, Appl. Catal., B, 2013, 129, 214.
5. C. Tyrsted, K. M. Ø. Jensen, E. D. Bøjesen, N. Lock, M. Christensen, S. J. L. Billinge and B. B. Iversen, Angew. Chem., Int. Ed., 2012, 51, 9030.
6. H. R. Chen, Z. Q. Ye, X. Z. Cui, J. L. Shi and D. S. Yan, Microporous Mesoporous Mater., 2011, 143, 368.
7. Q. Y. Wang, Z. G. Li, B. Zhao, G. F. Li and R. X. Zhou, J. Mol. Catal. A: Chem., 2011, 344, 132.
8. K. M. S. Khalil, J. Colloid Interface Sci., 2007, 307, 172.
9. T. A. Maia, J. M. Assaf and E. M. Assaf, Fuel Process. Technol., 2014, 128, 134.
10. M. H. Jiang, B. W. Wang, Y. Q. Yao, Z. H. Li, X. B. Ma, S. D. Qin and Q. Sun, Appl. Surf. Sci., 2013, 285P, 267.
11. H.-J. Sédjame, C. Fontaine, G. Lafaye and J. Barbier Jr, Appl. Catal., B, 2014, 144, 233.
12. E. Aneggi, A. Trovarelli, C. de Leitenburg, G. Dolcetti and A. Trovarelli, J. Alloys Compd., 2006, 408–412, 1096.
13. C. E. Hori, A. Brenner, K. Y. Simon Ng, K. M. Rahmoeller and D. Belton, Catal. Today, 1999, 50, 299.
14. M. Yang, M. Q. Shen, J. Wang, J. Wen, M. W. Zhao, J. Wang and W. L. Wang, J. Phys. Chem. C, 2009, 113, 12778.
15. M. Zhao, M. Shen and J. Wang, J. Catal., 2007, 248, 258.
16. M. Nakamura, H. Wakamatsu, K. Suga, T. Sekiba, Y. Hiramoto and K. Shibata, SAE Int., 2009, 2009-01-1069.
17. Y. Aoki, S. Sakagami, M. Kawai, N. Takahashi, T. Tanabe and T. Sunada, SAE Int., 2011, 2011-01-0296.

---

Footnote

† Electronic supplementary information (ESI) available: Details of the calculation of oxygen storage capacity. See DOI: 10.1039/c5ra15731e

---