Atomically dispersed Cu on Ce_1−xRE_xO_2−δ nanocubes (RE = La and Pr) for water gas shift: influence of OSC on catalysis

Abstract

To elucidate the effect of CeO₂ shape and doping on activity, Cu_0.02Ce_0.98O_2−δ and Cu_0.02Ce_0.86RE_0.12O_2−δ (RE = La and Pr) were synthesized by a molecular precursor approach. The materials showed distinct activities depending on the shape and composition of CeO₂, which was well correlated with their different oxygen storage capacities.

---

Owing to its intriguing properties, including high oxygen storage capacity (OSC) and oxygen mobility, ceria (CeO₂) has been extensively studied as a metal-oxide support for numerous catalysis reactions.^1–4 In CeO₂-supported catalysts, the OSC of CeO₂ plays a crucial role in catalyzing industrially important reactions, particularly those associated with oxygen ion (O²⁻) transport, such as water gas shift (WGS) and partial oxidation (POX) reactions.^1–4 Significant efforts have been made to improve the OSC to facilitate a number of catalytic reactions. One strategy to enhance the OSC of CeO₂ is through morphology control. CeO₂ nanocubes with (100) facets have increased oxygen vacancies compared with CeO₂ irregular nanoparticles (NPs) with (111) facets.⁵ Alternatively, the incorporation of rare earth elements as dopants into the CeO₂ lattice can increase the OSC.^1–4 The doping of CeO₂ with rare earth metal elements such as Pr, La, and Tb increases the OSC, preventing significant carbon coking and ultimately enhancing the catalyst durability in the steam methane reforming reaction.⁶ In addition, atomically dispersed metals on oxide supports, Au/M-doped CeO₂ (M = La), were recently reported,^7,8 demonstrating the importance of active site distribution for the WGS reaction and providing an insight into the design of next generation catalysts with highly dispersed active sites.

Herein, we report atomically distributed Cu catalysts supported on La or Pr-doped CeO₂ nanocubes for the WGS reaction. These catalysts were prepared by a molecular precursor method using a Cu-containing organometallic compound. This new synthetic approach for CuCeREO_2−δ (RE = La or Pr) in conjunction with shape control and doping strategies provided the uniformly dispersed active sites with increased OSC needed to achieve high activity for WGS.

To improve the OSC of CeO₂, we employed the following synthetic approaches: (i) morphology control and (ii) doping with rare earth elements. Method (i) yielded CeO₂ irregular nanoparticles (NPs) (1) and CeO₂ nanocubes (2), whereas method (ii) gave Ce_0.88La_0.12O_2−δ (3) and Ce_0.88Pr_0.12O_2−δ (4) nanocubes (vide infra). To analyze the shape of the prepared CeO₂ supports, morphological studies of 1 and 2 were conducted by HR-TEM. 1, prepared by the co-precipitation method, showed irregular shapes with d-spacing of 0.315 nm (Fig. 1a), indicating the formation of CeO₂ (111) planes at the surface.⁹ In contrast, 2, obtained by the hydrothermal method, exhibited well-defined cubic structures with uniform sizes of 20–40 nm. The HR-TEM image of 2 shows apparent (100) lattice fringes with an interplanar spacing of 0.272 nm (Fig. 1b), suggesting that CeO₂ (100) fringes are dominant at the surface of 2.⁹ Structure determination for 3 and 4 by TEM indicated that doping of La or Pr into 2 did not affect the shape (Fig. 1c and d).

Fig. 1 HR-TEM images: (a) CeO₂ irregular NPs (1), (b) CeO₂ nanocubes (2), (c) Ce_0.88La_0.12O_2−δ nanocubes (3), and (d) Ce_0.88Pr_0.12O_2−δ nanocubes (4).

Materials 1–4 likely have different OSCs depending on the morphology or dopant. The relative quantities for oxygen vacancies of 1–4 were examined using Raman spectroscopy (Fig. 2); the sharp peaks with maxima centered at ca. 465 cm⁻¹ were attributed to the F_2g mode, characteristic of the CeO₂ fluorite crystal structure.¹ Notably, the F_2g peaks for 3 and 4 broadened slightly and shifted toward lower frequencies, indicating the formation of solid solutions upon La or Pr incorporation.¹ In addition to the F_2g mode, 4 had a dominant peak centered at ca. 570 cm⁻¹ owing to oxygen vacancies generated by doping with a trivalent cation; replacement of Ce⁴⁺ in the CeO₂ lattice with Pr³⁺ produces an oxygen vacancy to maintain charge neutrality.³ Likewise, the Raman spectrum of 3 had two additional modes at 540 and 600 cm⁻¹ associated with the local vibrations of different oxygen vacancy (V_O) complexes.¹ Particularly, the vibrational mode at 600 cm⁻¹ originates from the presence of Ce³⁺–V_O complexes in the CeO₂ lattice, referred to as the intrinsic vacancy mode, whereas that of ∼540 cm⁻¹ comes from extrinsic vacancy mode related to La doping.¹⁰ In the Pr-doped samples the extrinsic vacancy mode appears at ∼570 cm⁻¹. A higher I₅₇₅/I₄₆₅ ratio indicates a higher quantity of oxygen vacancies.¹¹ These results suggest that among these materials, the ability for the formation of oxygen vacancies for 4 is superior.

Fig. 2 Raman spectra: (a) CeO₂ irregular NPs (1), (b) CeO₂ nanocubes (2), (c) Ce_0.88La_0.12O_2−δ nanocubes (3), and (d) Ce_0.88Pr_0.12O_2−δ nanocubes (4). Inset presents Raman spectra of 1, 2 and 3 with the range of 500–650 cm⁻¹.

We also employed Ce 3d X-ray photoelectron spectroscopy (XPS) to determine the relative capability for oxygen vacancy formation of 1–4 by assessing their Ce³⁺/[Ce³⁺ + Ce⁴⁺] ratios. The XPS spectra had seven peaks corresponding to Ce 3d_3/2 (897–915 eV) and Ce 3d_5/2 (875–897 eV)¹² (u and v, respectively, Fig. S1†). The relative Ce³⁺ 3d_5/2 (v′)/[Ce³⁺ 3d_5/2 (v′) + Ce⁴⁺ 3d_5/2 (v, v′′, and v′′′)] ratios for 1–4, determined from the integrated areas of the fitted data (Fig. S1 and Table S1†), provide useful information about the oxygen vacancies. 1 and 2, which have different shapes, had Ce³⁺ fractions of 15% and 27%, respectively, indicating that 2 has increased oxygen ion mobility compared with 1. In contrast, the incorporation of La or Pr into 2 to obtain 3 or 4 did not alter the Ce³⁺/[Ce³⁺ + Ce⁴⁺] ratio, consistent with a previous result.¹³

To gain further information about OSC of the materials, we employed thermogravimetric analyses (Fig. S2†), according to a previously reported method.⁵ Since WGS reactions were conducted at <400 °C (vide infra), we determined the quantities of oxygen vacancies using the heating temperature of 400 °C. The catalysts were initially heated from room temperature to 400 °C with air flow. In this process, oxygen vacancies presented in a catalyst were filled by O₂. At 400 °C, N₂ gas was then supplied for 10 min to abstract the filled oxygen from the catalyst, which resulted in weight loss. Repeated the processes gave the quantities of oxygen vacancies in the catalyst (Table 1). Consistent with the Raman results, the measured OSCs were found to increase in the order of 4 > 3 > 2 > 1.

Table 1 Calculated oxygen vacancies for 1–4 based on thermogravimetric analysis at 400 °C

Samples	Weight loss (%)	μmol of O2^a per gcat
a The quantities of oxygen vacancy = g of decreased weight/molecular weight of O2.^5
CeO2 irregular NPs (1)	0.034	10
CeO2 nanocubes (2)	0.052	16
Ce0.88La0.12O2−δ nanocubes (3)	0.072	22
Ce0.88Pr0.12O2−δ nanocubes (4)	0.074	23

---

The enhanced OSC of the modified CeO₂ materials could provide catalysts with improved activity. We incorporated Cu into the crystal structure of CeO₂ to generate catalytically active sites for the WGS. Precipitation, wet impregnation, or co-precipitation deposition methods have widely been employed to introduce dopants into metal oxide lattices, but metal aggregates are often formed. Recently, atomically dispersed active sites supported on metal oxides were shown to be highly active for CO oxidation and WGS;^7,14 the WGS reaction was accelerated by uniformly dispersed, nonmetallic Au or Pt species strongly associated with CeO₂ that generated by removing excess Au or Pt in nanostructured Au- or Pt-CeO_x materials.⁷

We employed a different synthetic strategy using an organometallic precursor, Cu(OCH₃)₂, to dope Cu atoms into CeO₂ with increased distribution (Fig. 3). The surface hydroxyl (–OH) groups functioned as nucleophiles to react with Cu(OCH₃)₂ to yield Cu-anchored CeO₂ and CH₃OH as a byproduct (Fig. 3, reactions A and B). A reaction of 2 (0.20 g) with Cu(OCH₃)₂ (2.9 mg, 0.023 mmol) at 70 °C for 24 h clearly indicated the formation of CH₃OH, as evidenced by ¹H NMR spectroscopy (Fig. S3†). Note that Cu agglomeration was likely minimized by the steric hindrance of the organometallic precursor. The well-dispersed Cu atoms were then incorporated into the CeO₂ lattice by calcination at 400 °C (Fig. 3, reaction C). In fact, similar surface grafting methods for CeO₂ with organosilanes were employed to improve chemical mechanical polishing.^15,16 The as-synthesized catalysts are denoted as Cu_0.02Ce_0.98O_2−δ irregular NPs (Cu-1), Cu_0.02Ce_0.98O_2−δ nanocubes (Cu-2), Cu_0.02Ce_0.86La_0.12O_2−δ nanocubes (Cu-3), and Cu_0.02Ce_0.86Pr_0.12O_2−δ nanocubes (Cu-4). For comparison, CuCeO₂ was prepared by a conventional process using Cu(NO₃)₂, followed by NaBH₄ reduction: Cu_0.02Ce_0.98O_2−δ irregular NPs (Cu-5) and Cu_0.02Ce_0.98O_2−δ nanocubes (Cu-6).

Fig. 3 A schematic diagram for the generation of uniformly distributed Cu active sites based on a molecular precursor approach (yellow: Ce atom, red: O atom, and blue: Cu atom).

The CuCeO₂-based catalysts were characterized using BET and SEM-EDS (Table 2). SEM-EDS indicated that 0.58–0.74 wt% of Cu was deposited on the catalysts. STEM-EDS mapping indicated the presence of well-dispersed Cu in Cu-2 (as an example, Fig. S4†). X-ray diffraction analyses of Cu-1–Cu-4 showed only the characteristic peaks of the CeO₂ fluorite cubic structure (Fig. S5†), indicating that the Cu atoms were well dispersed and formed a Cu-doped CeO₂ solid solution. The low valence state of Cu²⁺, with its smaller ionic radius, was reported to allow facile substitution of Ce⁴⁺ in the CeO₂ lattice with simultaneous oxygen vacancy generation to produce a Cu_xCe_1−xO_2−δ solid solution.⁴

Table 2 Characterization of the prepared CuCeO₂ catalysts

Catalyst	Cu precursor	Method	Metal loading^b (wt%)	BET surface area (m^2 g^−1)
a Cu reduced using NaBH4.b Measured by SEM-EDS.
Cu-1	Cu(OCH3)2	Molecular precursor approach	0.70	56
Cu-2	Cu(OCH3)2		0.58	28
Cu-3	Cu(OCH3)2		0.62	30
Cu-4	Cu(OCH3)2		0.61	29
Cu-5	Cu(NO3)2·3H2O^a	Conventional co-reduction	0.70	56
Cu-6	Cu(NO3)2·3H2O^a		0.74	27

---

We determined the catalytic activities of Cu-1–Cu-6 for the WGS reactions. First, we verified the influence of CeO₂ facets on the WGS (Fig. 4). Cu-2 is expected to show higher activity than Cu-1 since the CeO₂ nanocubes have a higher OSC than the CeO₂ irregular NPs owing to the high surface energy of the (100) facet, which originates from the instability of the top layer oxygen atoms that bridge two cerium atoms.¹⁷ Recently, the OSC of CeO₂ was determined using thermogravimetric analysis and, although nanocubes have smaller surface areas than irregular NPs, the OSC of CeO₂ nanocubes was nearly 2.6 times higher than that of CeO₂ irregular NPs at 400 °C.¹⁸ Similarly, the CO conversion using Cu-2 (42%) was greater than that of Cu-1 (30%) at 300 °C (Fig. 4a). Next, we examined the influence of doping by comparing Cu-2, Cu-3, and Cu-4. Compared with Cu-2, Cu-3 and Cu-4 had increased activities, presumably owing to the enhanced OSC caused by doping CeO₂ with rare earth elements;^1–4 the CO conversion of Cu-4 (74%) was 1.8 times higher than that of Cu-2 at 300 °C. Given the influence of both facet control and doping, Cu-4 has significantly increased activity (2.5 times higher than that of Cu-1 at 300 °C); this result correlates well with the relative OSCs of 1–4. In contrast, Cu-5, with dominant Cu (111) facets, had a higher catalytic activity than the Cu-6 nanocubes, with CO conversions of 32% and 16% at 300 °C, respectively (Fig. 4b). The CO conversion appears to be linear but it is expected to have a sigmoidal shape in an expanded temperature range. An Arrhenius plot obtained using the temperature dependent kinetic data for Cu-4 gave an activation energy of 22 kJ mol⁻¹, which is considerably lower than that of Cu-1 (Fig. S6 and Table S2†). We further conducted DFT calculations to determine the influence of the dopants on oxygen vacancy formation at the materials. Our calculations strongly support that the Pr-doped CeO₂ likely has the highest oxygen vacancy mobility (Fig. S7 and S8†).

CO + Cu-Ce-(O)-RE → CO₂ + Cu-Ce-( )-RE (1)

H₂O + Cu-Ce-( )-RE → H₂ + Cu-Ce-(O)-RE (2)

Overall: CO + H₂O → CO₂ + H₂ (3)

Fig. 4 CO conversions during the WGS over the catalysts: (a) Cu-1 (, pink), Cu-2 (, black), Cu-3 (, red), and Cu-4 (, blue); (b) Cu-5 (, purple) and Cu-6 (, green).

Based on the experimental results, a plausible mechanism involves oxygen transfer from CuCeREO₂. First, CO reacts with oxygen atoms in the CeO₂ surface lattice to generate an oxygen vacancy (eqn (1)). The resulting lattice abstracts an oxygen atom from H₂O to produce hydrogen (eqn (2)). The process is highly dependent on the OSC of the catalyst.

In summary, CuCeREO_2−δ (RE = La or Pr) catalysts were prepared using the molecular precursor approach to generate Cu active sites well dispersed on CeO₂-based supports. These catalysts were superior to those synthesized by the conventional co-reduction method. The catalytic activities for the WGS were enhanced owing to the increased OSC of CeO₂; the catalyst with exposed (100) facets had better activity than that with exposed (111) facets. In addition, the doped materials with CeO₂ (100) facets had even higher catalytic performance than the undoped CuCeO₂ material. Our synthetic approach using an organometallic precursor provides insights into the design of new types of catalysts applicable to numerous chemical transformations.

Acknowledgements

The authors thank the Global Research Laboratory (GRL) Program through the National Research Foundation of Korea funded by the Ministry of Science, ICT and Future Planning of Republic of Korea. The part of the research was also supported by the Fundamental Technology Development Programs for the Future through the Korea Institute of Science and Technology as well as by the National Research Foundation of Korea Grant funded by the Korean Government (MSIP) (University-Institute cooperation program). The computational resources were supported by the supercomputing center of Korea Institute of Science and Technology Information (KISTI) (KSC-2015-C2-002).

References

1. J. R. McBride, K. C. Hass, B. D. Poindexter and W. H. Weber, J. Appl. Phys., 1994, 76, 2435–2441.
2. D. Harshini, Y. Kim, S. Nam, T.-H. Lim, S.-A. Hong and C. Yoon, Catal. Lett., 2013, 143, 49–57.
3. M. Guo, J. Lu, Y. Wu, Y. Wang and M. Luo, Langmuir, 2011, 27, 3872–3877.
4. D. Zhang, Y. Qian, L. Shi, H. Mai, R. Gao, J. Zhang, W. Yu and W. Cao, Catal. Commun., 2012, 26, 164–168.
5. J. Zhang, H. Kumagai, K. Yamamura, S. Ohara, S. Takami, A. Morikawa, H. Shinjoh, K. Kaneko, T. Adschiri and A. Suda, Nano Lett., 2011, 11, 361–364.
6. D. Harshini, D. H. Lee, J. Jeong, Y. Kim, S. W. Nam, H. C. Ham, J. H. Han, T.-H. Lim and C. W. Yoon, Appl. Catal., B, 2014, 148–149, 415–423.
7. Q. Fu, H. Saltsburg and M. Flytzani-Stephanopoulos, Science, 2003, 301, 935–938.
8. M. Yang, L. F. Allard and M. Flytzani-Stephanopoulos, J. Am. Chem. Soc., 2013, 135, 3768–3771.
9. S. Bernal, G. Blanco, J. J. Calvino, C. López-Cartes, J. A. Pérez-Omil, J. M. Gatica, O. Stephan and C. Colliex, Catal. Lett., 2001, 76, 131–137.
10. Z. D. Dohčević-Mitrović, M. Grujić-Brojčin, M. Šćepanović, Z. V. Popović, S. Bošković, B. Matović, M. Zinkevich and F. Aldinger, J. Phys.: Condens. Matter, 2006, 18, S2061.
11. M.-F. Luo, Z.-L. Yan, L.-Y. Jin and M. He, J. Phys. Chem. B, 2006, 110, 13068–13071.
12. M. A. Henderson, C. L. Perkins, M. H. Engelhard, S. Thevuthasan and C. H. F. Peden, Surf. Sci., 2003, 526, 1–18.
13. N. Paunovic, Z. Dohcevic-Mitrovic, R. Scurtu, S. Askrabic, M. Prekajski, B. Matovic and Z. V. Popovic, Nanoscale, 2012, 4, 5469–5476.
14. B. Qiao, A. Wang, X. Yang, L. F. Allard, Z. Jiang, Y. Cui, J. Liu, J. Li and T. Zhang, Nat. Chem., 2011, 3, 634–641.
15. Z. Zhang, L. Yu, W. Liu and Z. Song, Appl. Surf. Sci., 2010, 256, 3856–3861.
16. K. Moller and T. Bein, Chem. Mater., 1998, 10, 2950–2963.
17. H. Nörenberg and J. H. Harding, Surf. Sci., 2001, 477, 17–24.
18. T. Adschiri, Y. Hakuta, K. Sue and K. Arai, J. Nanopart. Res., 2001, 3, 227–235.

---

Footnote

† Electronic supplementary information (ESI) available: Materials and detailed preparation procedures, catalyst characterization, catalytic reaction, analysis of products, and DFT calculations. See DOI: 10.1039/c5ra20557c

---