Dopant-mediated oxygen vacancy tuning in ceria nanoparticles
Abstract
Ceria nanoparticles with 20 and 40 at.% RE (RE Y, Sm, Gd, and Yb) dopants were synthesized through a microemulsion method. Independently of the dopant nature and concentration, nearly monodispersed nanoparticles of size 3–5 nm were observed in high resolution transmission electron microscopic analysis. The ceria lattice either expands or contracts depending on the dopant cation ionic radii, as indicated by x-ray diffraction studies.
X-ray photoelectron and Raman spectroscopic studies were used to quantify the cerium oxidation state and oxygen vacancy concentration. The results show the tunability of the oxygen vacancy and Ce3+ concentrations based on the dopant properties. First principles simulations using the free energy density functional theory method support the observed experimental trends. The reported results establish a relationship between the oxygen vacancies and oxidation states in doped ceria required for tailoring properties in catalytic and biomedical applications.
1. Introduction
Ceria (CeO2), a rare-earth oxide, has unique properties such as high mechanical strength, oxygen ion conductivity and oxygen storage capacity, especially in the form of nanoparticles (NPs). As a result, ceria NPs have been utilized as oxygen gas sensors, catalysts, the electrolyte in solid oxide fuel cells, and high temperature oxidation resistant coatings, and in chemical mechanical polishing and biomedical applications [1–4]. Ceria has been synthesized in various morphologies such as cubes and rods by various techniques, including hydrothermal, precipitation, spray pyrolysis, electrochemical and microemulsion, in order to tune the size, structural, and surface
properties [3, 5, 6].
Many of the applications of ceria NPs arise from the low redox potential between the Ce3+ and Ce4+ states and the high mobility of oxygen vacancies in the nanosize regime [7]. On reducing the particle size of ceria, the lattice constant and Ce3+ concentration increases [8]. To maintain the charge neutrality, oxygen vacancies are created, which increase oxygen diffusion and thereby increase the ease with which the material can absorb and release oxygen. The concentration of Ce3+, and hence of oxygen vacancies, is important for tuning the required catalytic activity for both biomedical and energy applications. Apart from reducing the nanoparticle size, the addition of trivalent elements can also enhance the formation of oxygen vacancies by replacing two Ce4+ ions with two trivalent cations in the ceria lattice.
Although doped ceria systems have been investigated, the influence of doping in ceria nanoparticles from the dilute to heavy ranges (higher than 20–25 at.%) with smaller nanoparticle sizes has not been studied [9–11]. In the present work, a simple room temperature process was used for doping trivalent rare-earth cations of Yb, Y, Gd, and Sm differing in ionic radii into the matrix of ceria to study the associated lattice, surface, and defect modifications. Trivalent cations of Yb, Y, Gd, and Sm have ionic radii of 0.0985, 0.1019, 0.1053 and 0.1079 nm, respectively, which are smaller than that of Ce3+ (0.1143 nm) but higher than that of Ce4+ (0.097 nm) [12].
2. Experimental details
2.1. Synthesis
Ceria NPs were synthesized by a microemulsion method. Sodium bis(2-ethylhexyl) sulfosuccinate (AOT) and toluene were used as the surfactant and the organic phase. All the chemicals were purchased from Sigma Aldrich Inc., and used in the as-received condition. An aqueous solution of cerium nitrate mixed with the corresponding nitrate salt of the dopant was added to AOT in toluene to form the microemulsion. To the microemulsion solution 30% hydrogen peroxide solution was added and stirred for 4 h. The resultant yellow solution was allowed to stand until complete separation into organic and water phases occurred. To the organic phase, 1 N ammonium hydroxide solution was added to precipitate ceria NPs. The extracted particles were repeatedly washed with acetone and water and centrifuged to remove any surfactant and inorganic impurities present. The doped ceria samples were labeled as X REDC, where X is the dopant concentration (20 or 40 at.%) and RE represents Y, Sm, Gd, or Yb. A similar procedure was used to prepare undoped nanoceria (NC).
2.2. Characterization
Powder x-ray diffraction (XRD) was performed with monochromatized Cu Kα radiation (λ 1.5418 A˚ ). The size and morphology of the powder were measured at 300 keV with a Philips (FEI Tecnai F30 TEM) high resolution transmission electron microscope (HRTEM). X-ray photoelectron spectroscopy (XPS) studies were carried out using a Perkin-Elmer 5400 ESCA system to calculate the concentration of oxidation states. To quantify the oxygen vacancy concentration, Raman spectra were measured with a Horiba Jobin Yvon LabRam IR micro-Raman system at an excitation of 633 nm. Density functional theory was used to calculate the lattice parameter variation with dopant nature using the CPMD program package [14].
3. Results and discussion
3.1. Size and structural characteristics of the nanoparticles
The bright field HRTEM images (figure 1) of NC, 20YbDC, and 20SmDC reveal polyhedral shaped 3–5 nm particles with a mean size of 3.7 nm. The mean size of the nanoparticle was found to be independent of dopant concentration or nature. Lattice fringes can be seen clearly, with an interplanar spacing of 0.312 nm, corresponding to the (111) reflection plane of crystalline ceria. The reflection planes from the selected area electron diffraction (SAED) patterns (inset in figure 1) correspond to the fluorite structure of ceria, indicating that the structure and size of the NPs remain the same upon doping.
The XRD patterns of the particles, shown in figure 2(a), exhibit broad peaks that can be indexed to the single-phase fluorite structure of Ce1—x REx O2—y type solid solution for the entire doping range. The change in lattice constant with dopant concentration for the particles measured by fitting the XRD peaks using PeakFit (Version 4.0) software is shown in figure 2(b) [23]. From the Scherrer formula, the mean size was found to be 3.5 nm [13]. The lattice parameter of the NC was calculated to be 0.5436 nm, but it varied from
0.542 nm (for 40YbDC) to 0.5449 nm (for 40SmDC). The difference in the lattice parameter shift can be related directly to the ionic radii of the dopants. As the ionic radii of Gd3+ (0.1053 nm) and Sm3+ (0.1079 nm) are larger than that of Y3+ (0.1019 nm), the lattice constant is larger for all Sm and Gd doped samples compared to the corresponding Y doped samples. It also indicates that doping smaller trivalent ion such as Yb3+ (0.0985 nm) results in ceria lattice contraction, and the magnitude of the lattice parameter changes with trivalent dopant concentration.
We also used density functional theory as implemented in the CPMD program package to study the dopant dependent lattice parameter modification in the matrix of ceria NPs. The local spin density approximation, free energy density functional method [15], and Goedecker separable norm- conserving pseudopotentials were used [16]. The valence electron wavefunctions were expanded in a plane wave basis set with an energy cutoff of 50 Hartrees, and a 2 2 2 Monkhorst–Pack mesh was used for integration in reciprocal space. A similar approach employing free energy density functional theory was recently applied to modeling of cerium– zirconium mixed oxide nanocrystals [17]. This work is the first ab initio prediction of the dopant effect on the lattice parameter in the fluorite type structure of ceria, although a similar study with the classical interatomic potentials was published earlier [18]. In our calculations, we used the supercell that was derived from the face centered cubic ceria structure by relaxing the symmetry constraints. Two out of four metal ions in the unit cell were replaced by dopant (Y, Yb or Sm). In order to keep the correct stoichiometry (2CeO2 X2O3, X Y, Yb or Sm), one oxygen atom was also removed. This supercell corresponds to 50% level of doping, which is close to the 40% studied in our experiment. We assumed complete substitution by dopant atoms, resulting in a variation of the lattice parameter. Simulated ceria NPs had a lattice parameter of 0.542 nm,which is in agreement with experimental values (figure 2(b)). For 50 at.% Yb, Y, and Sm doped ceria, the calculated lattice parameter was found to be 0.5352, 0.5474, and 0.551 nm, respectively. This is in qualitative agreement with experimental trend (lattice expansion for Y and Sm, lattice contraction for Yb). However, the absolute shifts in the lattice parameter were overestimated compared to the experimental results. We trace this overestimation to the smaller bulk moduli, predicted with the pseudopotential method. When explicit core electrons are replaced with pseudopotentials, the interatomic repulsion is substantially weaker than in reality and the lattice parameter is more sensitive to expansion or contraction under pressure or admixture of the dopant. Supercells with much larger sizes would be required for modeling the experimentally investigated doping levels (20 and 40%), which would require the investigation of equilibrium configuration for both the dopant atom and oxygen vacancy arrangements in the supercell. The large number of variations in combination with the increased number of atoms in the system made this route practically unfeasible. The supercell used in the present study does not suffer from these drawbacks and represents the 50% doped ceria unambiguously. The study of larger supercells with more flexible concentrations of Ce3+ and oxygen vacancies is possible when lower plane wave cutoffs and ultrasoft pseudopotentials are used. We also attempted simulations of Gd doped ceria, but self consistent field convergence techniques, available in the CPMD package, were not successful at achieving convergence. Gd3+ presents the most difficult case of all lanthanides, located at the border between the localized f electrons in the later lanthanides series and delocalized f electrons in early lanthanide series.
3.2. Surface and defect chemistry of the nanoparticles
The surface chemistry of the doped and undoped ceria NPs was studied using x-ray photoelectron spectroscopic (PE- PHI5400) analysis. Identification and a semiquantitative estimation of dopants was carried out by using the 4d, 3d5/2, 3d and 4d lines of Gd, Sm, Y, and Yb, respectively, with reported atomic sensitivity factors [19]. The calculated dopant concentration was found to match well with the experimental results, indicating the nearly complete incorporation of dopants into ceria lattice. In order to determine the concentration of Ce3+ and Ce4+, the 3d peak positions (figure 3(a)) and FWHM values of cerium peaks were fitted using PeakFit (Version 4.0) software for the samples [8, 20]. In this, the v0, vr, u0, and ur peaks are the characteristic peaks of Ce3+, while v, vr, vrrr, u, urr, and urrr are attributed to Ce4+ ions. From the ratio of integrated peak areas of Ce3+ to that of total Ce3+ and Ce4+, the atomic fraction of Ce3+ can be calculated as shown [20]: to the increase in the lattice parameter (as well as oxygen vacancy concentration) by doping larger ions (Sm3+, Gd3+, Y3+) whereas there is a decrease in lattice parameter (as well as oxygen vacancy concentration) by doping smaller ions (Yb3+). Raman scattering is a non-destructive and rapid analysis technique for investigating the symmetry, lattice modifications, and chemical environments of materials. Raman spectra for the synthesized samples were measured with a Horiba Jobin-Yvon LabRam IR micro-Raman system with a spatial resolution of 2 μm. A He–Ne laser was used for the excitation (633 nm) with a power of 3 mW. Figure 4(a) shows the Raman spectra for the samples; the peak positions for the single Raman line were determined by fitting the data to a Lorentzian line shape using PeakFit. The inset in figure 4(b) shows the Raman shift (F2g mode) for NC (455 cm—1), 40SmDC (443 cm—1) and 40YbDC (457 cm—1). It can be seen that for dopants with concentration is expected, which holds for Sm, Gd, and Y dopants. But for Yb3+, which has smaller ionic radius, the defect and Ce3+ concentrations decrease with a corresponding decrease in lattice parameter than that of NC, indicating that Yb addition suppresses the formation of oxygen vacancies. Thus by doping with trivalent ions differing in ionic radii, the oxygen vacancy and Ce3+ concentrations can be tuned, which is important for many catalytic applications.
4. Conclusion
Nearly monodispersed Sm, Gd, Y, and Yb doped ceria NPs of size 3–5 nm were synthesized using a microemulsion method. Lattice expansion or contraction of the ceria matrix was observed depending on the ionic radius of the dopants, supported by our ab initio calculations. Due to larger ionic radii, Sm, Gd, and Y doping generates more defects and higher Ce3+ concentration in the ceria lattice compared to Yb doping. It is known that doping trivalent cations in ceria can enhance the oxygen vacancy concentration. But the present results indicate that, by the proper selection of trivalent dopants, the defect concentration can be tuned. Such doped ceria NPs with tunable redox properties can be useful for many catalytic applications,MMAE including biocatalysis.