Abstract
A combined electron spin resonance (ESR) and atomic force microscopy (AFM) study on Ce1−xGdxO2−x/2 samples is here presented, aimed at investigating the evolution of the ESR spectral shape as a function of in a wide composition range. At low , the spectrum is composed of features at ; 2.8; 6. With increasing , this pattern merges into a single broad ESR curve, which assumes a Dysonian-shaped profile at , whereas, at these values, AFM measurements show an increasing surface roughness. It is suggested that the last could cause the formation of surface polaritons at the origin of the particular ESR spectral profile observed at these high Gd doping levels.
1. Introduction
Gadolinium-doped ceria Ce1−xGdxO2−x/2 are considered one of the most promising electrolytes for many potential practical applications such as solid oxide fuel cells (SOFCs) [1]. In particular, Ce1−xGdxO2−x/2 showed ionic conductivity higher than conventional Yttria-Stabilized-Zirconia-based electrolytes and were able to operate at lower (500–700°C) temperatures [1, 2]. However, the deep understanding of the phenomena which separately contribute to ionic and to electronic conductivity mechanisms is far from trivial because of an intricate interplay between the structural effect induced by doping and the transport properties of ceria.
Neutron and synchrotron radiation powder diffraction data, analyzed both in the reciprocal and real space, have supplied important information on the local, mesoscopic, and average structure, as revealed by some of the present authors for Ce1−xGdxO2−x/2 [3, 4] and for other rare earth (RE) doped ceria samples [5–9]. It was shown that nanoscopic compositional fluctuations exist in doped ceria, but Ce and dopant ions tend to maintain the structure of the parent CeO2 and rare earth sesquioxides (RE2O3). Ce and RE rich zones can extend over several nanometres, depending on samples composition and RE dopant [6, 9].
Due to the paramagnetic nature of Gd3+ ions, also the electron spin resonance (ESR) is considered a suitable technique to study the nature of these systems. In particular, the ESR line-shape of Gd-containing samples has supplied details of electronic structure of Gd based magnetoresistive perovskites [10].
In Ce1−xGdxO2−x/2 compounds, the interplay between crystal structure and magnetic properties and sample performances has been already examined by us elsewhere [11, 12].
In the present paper we would like to focus the attention on some problems encountered during these investigations and concerning the ESR line-shape and its dependence on the gadolinium doping level (Figure 1). Indeed, the ESR line changes markedly with . In particular, a single ESR line appears at , broadening with further increasing. Two different phenomena could account for the last observation, namely, (a) a shortening of the spin-spin relaxation time due to increasing spin-spin random interactions and (b) the increasing of the internal magnetic field intensity due to formation of ferromagnetic domains. Phenomenon (a) would act on both the lobes of the first-derivative ESR pattern taking away one from the other at the same extent. Phenomenon (b) would act on its left lobe only, moving it towards lower magnetic field values. Furthermore, Gd3+ substitution for Ce4+ means the introduction both of a negative charge and of a magnetic ion into the lattice. Therefore, both charge and magnetic effects could be at the origin of the ESR line changes with . We can conclude that the correct interpretation of the observed ESR line broadening is quite intriguing, because many different phenomena can contribute to it.
The situation simplifies at higher values when a single mechanism seems to prevail to form the spectral profile. Indeed, we will show that at the ESR pattern assumes the shape of a single Dysonian line. This is not new in the literature and it was found also by us in different systems, like LaMnO3+δ, [13] and GdBaCo2O5+δ [10, 14]. In those cases, that spectral profile was attributed to the electron mobility. However, it has been also reported [11] that electrons in Ce1−xGdxO2−x/2 have scarce or null mobility at . Therefore, combining ESR and atomic force microscopy (AFM) measurements on selected samples, a different explanation is here suggested for the Dysonian profile revealed by the ESR spectra at .
2. Materials and Methods
2.1. Sample Preparation and Characterisation
Microcrystalline Ce1−xGdxO2−x/2 samples with Gd concentrations , 0.125, 0.313, 0.500, and 0.875 were prepared with the Pechini sol-gel method [15]. For each gram of desired product, cerium nitrate, Ce(NO3)3·6H2O (Aldrich 99%), and gadolinium nitrate Gd(NO3)3·6H2O (Aldrich 99.9%) were dissolved into about 150 mL of water in the wanted Gd/Ce stoichiometric ratio. 24 grams of citric acid (Aldrich 99%) and 6 mL of ethylene glycol (Aldrich ≥ 99%) were added as polymerization agents. Then the solution was heated and stirred to remove water thus forming a gel. The gel was cast into a ceramic crucible and burned in an ashing furnace (Nabertherm), heated with a heating ramp of about 3 K/min up to 773 K, and then held at the same temperature for 3 hours. The powder was then pressed into pellets and fired at 1173 K for 72 hours in air.
The structure of our samples has been investigated by means of synchrotron radiation powder diffraction analysis. The experimental patterns and the structural analysis are reported elsewhere [4]. All the samples were single phase with CeO2 type structure (space group Fm-3m) for and and with Gd2O3 type structure (space group Ia-3) for , 0.500, and 0.875.
2.2. ESR Measurements
ESR measurements on the , 0.125, 0.313, 0.500, and 0.875 samples were performed by a Bruker ELEXSYS spectrometer equipped with an ER4102ST standard rectangular cavity at band (9.4 GHz) frequency at room temperature. Powdered sample was placed into quartz tube and the derivative of power absorbed was recorded as a function of the static magnetic field .
2.3. AFM Imaging
Samples with , 0.313, 0.5, and 0.875 were dispersed in isopropanol and sonicated for about 4 h. For AFM imaging, a 5 μL drop was pipetted on a freshly cleaved mica disk and after ~30 s the sample was blown dry with nitrogen gas. AFM imaging was performed using a Nanoscope Multimode IIId system (Bruker, Santa Barbara, CA, USA) operating in air in tapping mode. AFM images were collected using the RMS amplitude of the cantilever as the feedback signal for the vertical sample position. The RMS free amplitude of the cantilever was approximately 15 nm and the relative set-point above 95% of the free amplitude. Rectangular silicon probes with nominal spring constant around 2.5 N/m (NSG01, NT-MDT, Russia) and cantilever length of 120 μm were used. The cantilever resonance frequency was about 130 kHz. The mica support (Ted Pella, CA, USA) was glued to a metal disk that was magnetically fixed to the AFM sample holder. Images were recorded at ~1 Hz line rate and a resolution of 512 × 512 pixels per image was chosen. AFM images were subject to a line-by-line subtraction of linear background to eliminate sample tilt from the images and correct for stepwise changes between individual scan lines.
3. Experimental Results
3.1. The ESR Spectra
All the ESR spectra are reported in Figure 1. At the lowest () Gd3+ concentration the spectrum is composed of features at , 2.8, and 6 in agreement with what is reported in the literature [16] with diluted solid solutions of this kind. This pattern merges into a single () broad resonance line at , with peak-to-peak linewidth () increasing with gadolinium concentration (see inset of Figure 1). This trend is steeper in the CeO2-like solid solution zone than in the Gd2O3-like one. Attempts to fit the spectral profile by Lorentzian, Gaussian, and Dysonian functions were carried on for all the experimental spectra at . The best fit was always obtained by a Dysonian function according to is the linewidth, is dispersion-to-absorption ratio, and is the resonance field. The profile fitting performed by using (1) is reported in Figure 2, for samples with . However, only at , a single analytical curve of this kind was able to describe the experimental data with good accuracy.
(a)
(b)
(c)
(d)
3.2. AFM Imaging
AFM topography images of samples with and 0.875, absorbed on a freshly cleaved mica disk and visualized in air operating in tapping mode, have been reported in Figure 3, as an example. The two surfaces appear quite different from each other. Indeed, the sample has a more squared and planar appearance whereas the surface of the sample with the higher is rougher with rounded features.
(a)
(b)
In addition, AFM imaging allowed quantitatively the characterization of the surface morphology of all the samples. In particular, the sample surface roughness has been calculated as root-mean-square (rms) data using the following equation: where is the average value within the examined area, is the local value, and indicates the number of points within the area.
The rms roughness experimental results, measured on 30 × 30 nm2 areas and obtained analysing about 20 AFM topography images taken from several different samples, have been reported in Table 1.
It is worth noting that the surface of all the samples appears to be quite flat with corrugations of 1-2 nm. Interestingly, the rms surface roughness values are increasing with . In fact, while the sample has a rms surface roughness, on average, of 1.11 nm, the sample with the highest value shows a surface corrugation, on average, of 2.22 nm. A different trend can be observed analysing the distribution of the rms surface roughness values as reported in Figure 4. In particular, the histograms show a narrow and peaked distribution for the samples with low , whereas a broader distribution is exhibited by the samples with the highest Gd content.
(a)
(b)
(c)
(d)
4. Discussion and Conclusion
The ESR pattern of the sample (see Figure 1) is very similar to the “U-type” spectra observed in various disordered systems at low concentration of Gd3+ doping. In particular, Cugunov and Kliava attributed its contribution to Gd3+ ions in rhombic or lower crystal-field symmetry in glasses and the broad Gaussian-shaped contribution to clustered Gd3+ ions [17]. However, these authors were not able to explain why the former feature disappeared at higher microwave frequencies. Koopmans et al. [18] observed analogous ESR features with Gd3+ doping Ti-based perovskites and assigned them to a superposition of a large number of well-defined spectra. Furthermore, they interpreted the disappearance of the line with increasing microwave frequency, reported elsewhere also by us [19], attributing that feature to a “forbidden” transition with low probability that is reduced to zero when the Zeeman term is larger than the crystal field term. In other studies, features observed at , 2.8, and 6 were assigned to transitions within the multiplet of energy levels of Gd3+ ions [20–22]. Furthermore, the linear relationship between Gd3+ concentration and line-width of the “U” spectrum suggested that no concentration-dependent clustering effect was present, at least up to of Gd2O3 [22].
At , the merging of these features into a single broad resonance ESR curve with suggests the prevailing of a strong interaction among Gd-Gd magnetic dipoles, perhaps grouped into clusters. On the other hand, pair distribution function (PDF) measurements are in agreement with Gd clusterization since the local scale of Ce1−xGdxO2−x/2 with compounds is described by a Gd2O3 droplets embedded in globally fluoritic sample matrix [4]. Gd3+ ion clusters were also introduced by [17] and investigated by Kliava et al. [23] to explain both the magnetization and ESR measurements of Gd2O3 doped glasses.
The steeper trend of versus in the CeO2-like solid solution zone at (see Figure 1) suggests that the added Gd ions perturb the Gd-Gd interaction stronger in the embedded Gd2O3 droplets than in the Gd2O3-like phase forming [4] at .
As stated above, ESR spectra are better described by Dysonian than single Lorentzian or Gaussian functions. It is interesting to investigate why the best fits of the ESR line profile were those obtained by a Dysonian function and why the last fits better at higher values (Figure 2). In most cases, a Dysonian line-shape appears when a sample becomes an electronic conductor [24–26]. However, it is well known that Ce1−xGdxO2−x/2 solid solutions have negligible electronic conductivity, apart from pure CeO2 and slightly Gd-doped compounds when annealed in reducing conditions [2, 11, 27, 28].
Besides, the Dysonian shape has been also attributed to the formation of surface polaritons. The last are propagating coupled electromagnetic-electric dipole excitation modes, created by the diffraction of the microwave ESR field on the surface roughness and are able to mix absorption and dispersion ESR modes, leading to a Dysonian spectral profile [29]. The formation of polaritons could be further favoured by the high spin-orbit interaction characterizing the surface-doping Gd ions. Indeed, AFM data showed that the surface roughness increases and that its distribution broadens with samples characterized by higher values (Figures 3 and 4). This suggests that the Dysonian ESR profile observed with them and their surface roughness (inducing polaritons formation) are correlated to each other.
To summarize, we performed ESR and AFM measurements on Ce1−xGdxO2−x/2 as a function of . In agreement with previous reports, for low Gd concentration (), the ESR signal is composed by three resonance lines at , 2.8, and 6. For a single broad signal is observed with a linewidth which increases upon doping. This -dependent broadening is consistent with the increase of Gd-Gd interactions. Line-shape analysis reveals that the Dysonian function well describes all the spectra but the implication of electron delocalization disagrees with the compounds insulating ground state. By using AFM we observed an increase of surface roughness with increasing . This gives an alternative explanation to support the Dysonian line-shape in Ce1−xGdxO2−x/2 system through surface polariton formation.
The present study also suggests that ESR spectroscopy can reveal not only a very useful technique to characterize sample paramagnetic impurities and defects, but also a quite sensitive tool to investigate the surface properties, like roughness, which may play an important role in the surface science and related applications.
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.