Introduction

The development of high-energy-density storage devices is extremely urgent for the sake of advanced microelectronics and communications1,2,3. Of great importance is to search for appropriate dielectric materials with high permittivity4,5. Inorganic ceramic materials with high dielectric constant have received extensive attention, due to their application in multilayer ceramic capacitors (MLCC)6 and inorganic-organic hybrid flexible composite films7. The dielectric properties of inorganic ceramic materials can be improved via chemical modification8,9 and/or grain-size engineering10. Among them, rare-earth element(s) doped BaTiO3 (RE: BaTiO3) has been considered as one of the most suitable materials for ferroelectric capacitors, because of its colossal dielectric constant (CDC)11. The incorporation of trivalent RE ions (such as La3+ and Nd3+) can effectively enhance the room temperature permittivity of BaTiO312,13, thereby improving the performance of relevant energy storage devices. In addition, the dielectric properties of BaTiO3 can be tuned by varying the grain size; that is, with the decreasing grain size, the permittivity initially increases and then decreases after reaching a maximum at a critical grain size10.

The permittivity of RE: BaTiO3 is intimately related to the structural distortion and chemical defects surrounding the dopants. Theoretical calculations in terms of tolerance factor have indicated that the occupation of the exotic ions depends on their radius14,15. Since the radius of RE ions is usually between Ba2+(1.35 Å) and Ti4+(0.68 Å), larger RE3+ ions such as La3+ (1.15 Å) and Nd3+(1.08 Å) prefer to substitute for the Ba2+ -site (A-site), and the smaller ones such as Yb3+(0.87 Å) may locate exclusively at the Ti4+ -site (B-site), while the intermediate ones such as Y3+(0.93 Å) and Er3+(0.96 Å) may occupy both the A- and B- sites16,17. Accompanying with the specific doping, chemical inhomogeneity is often introduced. For example, the partial replacement of the A-site by La3+would give rise to either the formation of Ti vacancies or the reduction of Ti18. These defects could significantly affect the dielectric properties of BaTiO3-based ceramics. As demonstrated by Guillemet-Fritsch et al.19, a room-temperature colossal permittivity (εeff ~800,000) was obtained in Ba0.95La0.05TiO3-x ceramics sintered by the spark plasma sintering (SPS) method, in which Ti3+/Ti4+ acted as polaron carriers.

Although Nd: BaTiO3 ceramics have been studied13,20,21,22, the mechanism for their improved permittivity (~300,000) remains unclear13. In addition, Nd: BaTiO3 powders in the previous reports are often prepared by solid-state reaction (SSR), and high-temperature treatment (above 800 °C) is usually required, which results in large grain size and poor sintering activity. To achieve high bulk density and improved dispersion in organic matrix, ultrafine Nd: BaTiO3 nanoparticles with pure phase and uniform particle size are highly desirable yet challenging. Although it is well known that hydrothermal synthesis can yield high-purity nanoparticles with narrow size distribution23,24, the synthesis of RE: BaTiO3 nanoparticles has been rarely reported25,26. Recently, we have synthesized monodispersed BaTiO3 nanoparticles via the sol-hydrothermal method27. In this work, xNd: BaTiO3 (x = 0 ~3 mol%) nanocrystals were prepared by a similar process and their crystalline structure and defect chemistry were investigated elaborately. Subsequently, xNd: BaTiO3 ceramics were fabricated by a conventional pressureless sintering method. It’s found that the dielectric constant was dramatically modified upon Nd doping. Especially, the colossal dielectric constant was observed in the sample with 0.5 mol% Nd at room temperature, and possible mechanisms for such effect were provided.

Results and Discussion

Crystalline structure analysis of xNd: BaTiO3 nanocrystals

The as-prepared xNd: BaTiO3 nanocrystals were characterized by various techniques. Figure 1a shows the transmission electron microscopy (TEM) image of pure BaTiO3. The size distribution of these spherical particles is displayed in Fig. 1b, indicating that the diameter of majority particles (~90%) is in the range of 60~100 nm. Zooming into an individual nanoparticle (Fig. 1c), a single crystal character is revealed, as demonstrated by the fast Fourier transform (FFT) pattern (the inset of Fig. 1c) and well-defined atomic arrangement in the HR-TEM image in Fig. 1d. The interplanar spacing of the lattice is 0.397 nm, corresponding to (100) plane in tetragonal BaTiO3.

Figure 1
figure 1

(a) TEM image of pure BaTiO3 nanoparticles; (b) Particle size distribution of pure BaTiO3 nanoparticles; (c) TEM image of an individual particle and its fast Fourier transform pattern in the inset; (d) HR-TEM image of the selected area marked by the yellow square in (c); (e) XRD patterns of xNd: BaTiO3 nanoparticles with indication of nominal Nd concentration; (f) Raman spectra of xNd: BaTiO3 samples with the inset showing the Nd-doping dependence of the relative intensity of peaks at 307 and 518 cm−1.

The tetragonal structure of nanoparticles has been further confirmed by X-ray diffraction (XRD) patterns. As shown in Fig. 1e, all XRD peaks obtained from pure BaTiO3 nanoparticles are well-matched with the tetragonal structure corresponding to Joint Committee on Powder Diffraction Standards (JCPDS) Files No. 05-0626. These peaks remain in the xNd: BaTiO3 nanocrystals (see Fig. 1e). Inductively coupled plasma optical emission spectrometry (ICP-OES) was further carried out to quantify the chemical composition of samples. As listed in Table 1, the actual Nd concentration is found to be 0, 0.49, 0.95 and 1.43 mol% for nominal xNd: BaTiO3 samples with x = 0, 0.5, 1.0 and 1.5 mol%, respectively. The result indicates that the Nd ions have almost completely incorporated into the BaTiO3 structure.

Table 1 Nominal versus ICP-OES determined composition of xNd: BaTiO3 nanocrystals.

However, additional peaks (as indicated by diamond symbol) appear in the sample with x = 2 mol%, which can be assigned to Nd(OH)3 impurity phase (JCPDS No. 06-0601). Also, the energy dispersive X-ray spectroscopy (EDX) analysis results in Supplementary Fig. S2 confirm that rod-like impurity can be attributed to Nd(OH)3.

To confirm that the BaTiO3 nanocrystals form only tetragonal structure at room temperature, we further measured Raman vibrational spectroscopy, as it is sensitive to the structure symmetry1,28. As can be seen from Fig. 1f, all Raman scattering spectra consist of bands around 184, 254, 307, 518 and 713 cm−1, which are the characters of perovskite BaTiO3. While the small peak at 1061 cm−1 denotes the presence of BaCO3 and the intensity decreases with the doping concentration. It is well-known that all phonons of cubic Pmm symmetry are inactive to Raman modes, due to the isotropic distribution of electrostatic forces. The tetragonal P4mm space group exhibit eight Raman active modes described by 3[A1(TO) + A1(LO)] + 4[E(TO) + E(LO)] + B1. Among these, peaks at 184, 254, 518 cm−1 are assigned to the fundamental TO mode of A1 symmetry, which generally exist both in cubic and tetragonal BaTiO329. The presence of bands at 184 cm−1 indicates a decoupling between the A1(TO) phonons, which can be induced by internal stress or lattice defects; whereas the asymmetry in the bands at 518 cm−1 suggests the existence of coupling of the TO modes associated with the tetragonal phase. The sharp peak at 307 cm−1 corresponding to [B1, E(TO + LO)] modes is attributed to the non-centrosymmetric regions arising from the displacement of titanium atoms from TiO6 octahedra30,31, suggesting the intrinsic structural distortion in the tetragonal BaTiO3. The appearance of peak at 713 cm−1 is considered to the highest frequency longitudinal optical mode (LO). Besides, the higher relative intensity of the band versus other tetragonal bands for nanoparticles can be related to Ba vacancies in the BaTiO332. Herein, the relative intensity of the peaks at 307 and 518 cm−1 has been calculated to evaluate the tetragonality of as-synthetized xNd: BaTiO3 nanoparticles (the inset of Fig. 1f), which indicates a reduction trend of tetragonal distortions with the introduction of Nd ions.

Defect chemistry of xNd: BaTiO3 nanocrystals

It is well-known that the properties of BaTiO3 are intimately related to the oxidation state of constituents. As-prepared xNd: BaTiO3 nanoparticles were characterized by X-ray photoelectron spectroscopy (XPS) to determine the binding state and chemical environment of elements. Figure 2a shows photoemission spectra of Ba 3d, Ti 2p, and O 1 s in xNd: BaTiO3 nanoparticles. Peak deconvolution has been obtained by fitting the curves through the Gauss-Lorentz function. Peaks at 778 eV and 793.3 eV, respectively corresponding to the Ba 3d5/2 and Ba 3d3/2, are assigned to the perovskite structure of BaTiO3. It is worth pointing out that the higher shoulders at 779.4 eV and 794.6 eV are usually associated with Ba vacancy point defects33. Note that the peak intensity at both 779.4 eV and 794.6 eV increases with the increase in Nd concentration, suggesting that Ba vacancy would boost with the introduction of Nd ions. In order to reach charge balance, the substitution of Nd3+ for Ba2+ would either convert Ti4+ to Ti3+ or generate Ti vacancies 16,34,35. As reported in the literature36,37, the existence of Ti3+ ions would lead to the broad of the Ti 2p peak and a lower binding energy shoulder belonging to 2p3/2 peak at ~456 eV. While such a feature has not been detected in our samples (see Fig. 2a), thus excluding the existence of Ti3+. Besides, the energy shift of Ti ions will extend over more than 4 eV in case of different coordination numbers38. Therefore, it can be concluded that the TiO6 octahedron is preserved. Figure 2a also presents the XPS spectra of O 1s valence state, which shows one main peak corresponding to oxygen in BaTiO3 (528.8 eV) and a broad peak caused by chemisorbed species and/or oxygen vacancies (531.9 eV)39. Previously, Lewis Wasson et al.40 stated that the residual BaCO3 could hardly be detected by XPS, as the carbonate took the form of discrete particles rather than the continuous surface layer. Combining our previous FT-IR results27, it can be thus speculated that there exist chemisorbed OH ions.

Figure 2
figure 2

(a) XPS spectra for Ba 3d, Ti 2p and O 1s of xNd: BaTiO3 samples (x = 0, 1.0 and 3.0%): the black solid lines are the experimental data and the grey lines are the simulated curves; (b) Room-temperature EPR spectra of xNd: BaTiO3 (x = 0, 1%); (c) UV-vis spectra in the absorbance mode for indicated xNd: BaTiO3 samples; (d) Photoluminescence spectra of xNd: BaTiO3 powders, where circles (grey) are the experimental data, the solid lines represent simulated spectra, and dashed lines are simulated individual emission peaks.

Electron paramagnetic resonance (EPR) is a powerful technique to detect the presence of Ba and/or Ti vacancies41. Shown in Fig. 2b are the room-temperature EPR spectra of the pure BaTiO3 and 1%Nd: BaTiO3 nanopacrystals. Note that paramagnetic centers locate at g ~1.976 for both samples, implying the existence of Ba vacancies42. According to the previous report43, the formation of Ti vacancies would present a paramagnetic center at g = 2.004~2.005. Nevertheless, this signature of Ti vacancies has not been observed in our samples. From the XPS and EPR results, it can be concluded that the charge compensation mechanism primarily involves the formation of Ba vacancies induced by Nd substitution into A-site.

In general, any imperfection (vacancies, lattice defects, impurities/doping, and local bond distortion44) can yield density of states within the band gap of an insulator. Figure 2c shows the UV-vis absorption spectra of xNd: BaTiO3 nanoparticles, with well-defined peaks and exponential tails. The Nd content dependent band gap (Eg) is plotted as the inset, from which we can see the band gap (Eg) initially decreases, then increases after x > 1% and eventually keeps constant at x > 1.5 mol%. This trend goes along with the Nd concentration dependent structural distortion observed in Raman spectra, where the structure of xNd: BaTiO3 turns toward pseudo-cubic structure with the incorporation of Nd (see the inset of Fig. 1f). Probably, it is the structural transition that results in the constant band gap for x > 1.5 mol%.

To further characterize structural imperfection, we measured the photoluminescence (PL) emission spectroscopy at room temperature excited by laser with the wavelength of 355 nm. In general, the PL emission occurs when there is polarization within the structure and some localized states in the band gap, e.g. free exciton levels, self-trapped excitons and defects or impurity levels45. Figure 2d displays the emission spectra of as-prepared xNd: BaTiO3 samples, where two peaks are observed at 446 nm and 554 nm, respectively. The former is considered to originate from a direct band excitation, intimately related to the distortion of TiO646, as shown in blue dashed curves (blue wavelength region). The other peak is much broader (dashed green curves), which belongs to the yellow wavelength region according to Gaussian fit. This can be attributed to defects state within the band gap of the material47. The oxygen vacancies are considered as highly localized sensitive centers to trap electron from valance band and then the interaction of electron trapped with holes form self-trapped excitations. The radioactive recombination of the self-trapped excitations thus contributes to the emission of the yellow region. Compared to the low wavelength emission, this emission intensity increased gradually, indicating the oxygen vacancies defects increase with the incorporation of Nd ions.

Based on the above analysis, it can be concluded that when Nd is introduced into BaTiO3, Ba vacancies as well as oxygen vacancies would simultaneously form to compensate the charge balance, which can be described as Equation (1):

Previously, such an occupation mechanism has been taken into consideration in the study of RE-doped BaTiO3 ceramics. However, the existence of the Ba-vacancy induced by donor-doping mechanism(s) has never been demonstrated experimentally due to the mixtures of Ba1-yLayTi1-y/4O3 and other Ti-rich phase(s) such as Ba6Ti17O4048. In our work, the formation of Ba-vacancy can be facilitated benefiting from the hydrothermal condition, where barium deficiencies are easily generated due to the introduction of protons49.

Dielectric properties of xNd: BaTiO3 ceramics

It is well-known that the partial substitution of Ba2+ by RE3+ would result in the decrease of Curie temperature TC (corresponding to a cubic-to-tetragonal structure transition). In our work, the tetragonal structure observed at room temperature (Fig. 1d–f) indicates that the Curie temperatures TC of xNd: BaTiO3 nanocrystals still remain above room temperature, i.e. TC > 300 K. In order to measure dielectric properties of our samples, xNd: BaTiO3 ceramics were fabricated from as-synthesized xNd: BaTiO3 nanocrystals by conventional pressureless sintering method at 1300 °C in air for 2 h.

As demonstrated in Fig. 3a, XRD patterns of xNd: BaTiO3 ceramics corresponds to the tetragonal structure with lattice parameters a = b = 0.3994 nm and c = 0.4038 nm (JCPDS: No.05-0626). The peak splitting of (002)/(200) observed at 2θ ~46° for the samples with x ≤ 1 mol% suggests their tetragonal structures. At higher Nd contents (x > 1 mol%), the peak splitting at around 2θ ~46° cannot be recognized apparently. The results reveal that the introduction of Nd ions gives rise to the decrease in tetragonality, and induces the transition to pseudo-cubic structure. Cross-sectional FE-SEM images shown in Fig. 3b indicate that ceramic grain size decreases with the Nd content increasing. It’s observed that xNd: BaTiO3 ceramics with x ≤ 1 mol% display compact microstructure, and their grain size is in the range of 1~5 μm. While the grain size of ceramic samples with x >1% is comparable to its counterpart nanocrystals, and loosen microstructures are observed in these ceramics.

Figure 3: Properties of the Nd: BaTiO3 ceramics sintered at 1300 °C for 2 h.
figure 3

(a) XRD pattern of the xNd: BaTiO3 samples; (b) SEM images of the xNd: BaTiO3 ceramics; (c) Temperature dependence of the permittivity and dielectric loss for the xNd: BaTiO3 ceramic samples measured at 1 kHz; (d) Permittivity versus temperature curves of 0.5%Nd: BaTiO3 ceramics at different frequencies.

Figure 3c shows the temperature dependent relative dielectric constant εeff and tangent loss tanδ. Two features are remarkable: (1) the pure BaTiO3 ceramics show a similar temperature dependence as conventional bulk ceramics with TC ~125 °C, while the magnitude of εeff is about twofold higher; (2) the Nd introduction nonmonotonically enhances εeff and the colossal εeff ~7.6×105 is observed in 0.5%Nd: BaTiO3 ceramics with tanδ ~0.8 at 12.2 °C.

In BT and rare-earth doped BT system, several explications have been proposed to explain the colossal dielectric constant, including internal barrier layer capacitance effect50, hopping polarization19, and electrode effect51. However, these ceramics either sintered in the non-oxidation atmosphere or sintered at a fast sintering rate with ultrafine grain size, which could result in the inner grain conductivity or the reduction of the Ti4+. Moreover, giant permittivity values were reported in hexagonal barium titanate (h-BaTiO3) single crystals52, and the high permittivity values ~100,000 of the oxygen deficient materials were explained by the presence of interfacial boundaries consisting of crystal defects. In addition, it is reported that in NaNbO3-doped BaTiO3 system53, colossal permittivity can be attributed to the high-energy electric state of Ba+ (or Ba2+ −e) by Ba2+ obtaining an electron, which can create electron hopping conduction and increase conductivity of the ceramics. These charged defects like Ba+ or Ba-vacancy are regarded to be responsible for the colossal permittivity53,54. However, neither the reduction of Ti4+ ions nor Ba2+ was observed in our work. In our viewpoint, the enhanced permittivity of xNd: BaTiO3 ceramics (x ≤ 1 mol%) should be on the one hand attributed to the compact microstructure, resulting from the highly-active xNd: BaTiO3 nanoparticles. As we see in Fig. 3c, the permittivity of pure BaTiO3 ceramics is about twofold higher than that of conventional bulk ceramics. In combination with defect chemistry discussed above, the enhancement of the permittivity can be on the other hand ascribed to the space-charge polarization55, where the defects (i.e. Ba vacancies and O vacancies) induced by the dopants can act as additional dipoles. Since such a polarization cannot follow the alternating field at high frequency, the dielectric constant shows apparent decrease with the frequency arising. That is why the permittivity of the 0.5%Nd: BaTiO3 sample decreases dramatically with increasing frequency, where the maximum dielectric constant of ~160,000 and ~30,000 is observed at 10 kHz and higher frequencies (≥100 kHz), respectively (Fig. 3d).

In all, previous reports19,56 about the colossal permittivity in RE-doped BaTiO3 ceramics were almost achieved with the assistant of special atmosphere and/or sintering techniques. This work demonstrate that conventional pressureless sintering condition in air is adequate to obtain xNd: BaTiO3 ceramics with colossal permittivity, as long as ceramic powders is of high sintering activity, which is a huge technique advantage in potential industrial production.

Conclusion

A series of xNd: BaTiO3 nanoparticles (x = 0 ~3.0 mol%) were successfully synthesized by a modified sol-hydrothermal method. It’s demonstrated that all samples exhibit tetragonal phase and their tetragonality gradually decreases with the increasing Nd content. Shoulder peaks at high binding energy reveal the possible coexistence of Ba and O vacancies, while the probable emergence of reductive Ti3+ is excluded. Furthermore, the existence of Ba and O vacancies is confirmed by EPR and PL analysis, respectively. It’s thereby proposed that the introduction of Nd ions into BaTiO3 nanocrystals induce the simultaneously formation of Ba and O vacancies due to the valence equilibrium. Controlled by the RE content, dielectric constant of ceramic samples initially increases and then falls. Among them, 0.5%Nd: BaTiO3 ceramics sintered from the nanopowders possess a room-temperature colossal permittivity (~760,000). The present work serves as a precedent and fundamental understanding of the crystal structure associated chemical defects of the RE-doped BaTiO3 nanopowders, and deliberate efforts are on-going to better understand the underlying mechanism for the room-temperature colossal permittivity.

Methods

Synthesis of xNd: BaTiO3 nanoparticles

Nd-doped BaTiO3 nanocrystals (xNd: BaTiO3, x = 0, 0.5, 1.0, 1.5, 2.0 and 3.0 mol%) were synthesized via a modified sol-hydrothermal method. To prepare TiO2 sol, Ti(OC4H9)4 and ethanol were fully blended with continuous stirring, and then a solution containing ethanol, HNO3 and deionized water was slowly added. The as-prepared TiO2 sol (10 ml) was then added to the solution of Ba(Ac)2 in 40 ml deionized water. Simultaneously, appropriate amount of Nd(NO3)3 solution was added to control the doping concentration. During the process, the alkalinity (2 mol/L) was regulated by a KOH solution. The resulting mixture was stirred for 20 min, and then the Teflon vessel was put into a stainless-steel autoclave. The sealed autoclave was heated to 200 °C for 16 h, and then cooled to room temperature naturally. After the synthesis, the precipitates were washed with deionized water and ethanol in sequence several times. The resulting precipitates were collected and frozen in a refrigerator, and then dried by a freeze drying devices (Alpha 1-2LD, Christ, Germany).

Preparation of xNd: BaTiO3 Ceramics

The as-prepared xNd: BaTiO3 powders were mixed with 3 wt% PVA, pulverized using a mortar and pestle, and then pressed into pellets of 15 mm diameter under a uniaxial pressure of 8 MPa. After de-binding at 650 °C for 5 h, the green pellets were sintered at 1300 °C for 2 h in air by the conventional solid sintering method with a heating rate of 100 °C/h. Finally, the sintered ceramics were polished and coated with silver electrodes for electrical measurements.

Characterization

Powder X-ray diffraction (XRD) was collected on a Bruker D8 Advance diffractometer operating at 40 kV and 40 mA using Cu Kα radiation (λ = 1.54178 Å) to determine the structure of obtained samples. Raman spectra were recorded in the wavenumber range of 100~1100 cm−1 using a Jobin Yvon T64000 (Jobin Yvon, France) excited by the laser with a wavelength of 800 nm. The elemental composition was obtained by inductively coupled plasma-optical spectroscopy (ICP-OES) on the Optima 5300DV (PE, USA). The morphology and microstructure were obtained using a Hitachi S-4800 (Hitachi, Japan) field emission scanning electron microscope (FE-SEM). High-resolution transmission electron microscopy (HR-TEM) and selected area electron diffraction (SAED) images were obtained with the use of Tecnai G2 F30 S-TWIN (FEI, USA) microscope operated at 200 kV. The valence states of elements were analyzed by X-ray photoelectron spectroscopy (XPS) using an Escalab 250Xi (ThermoFisher Scientific, USA). Electron paramagnetic resonance (EPR) measurements were performed at room temperature using a Bruker A300-10/12 spectrometer operating at 9.85 GHz. Photoluminescence (PL) spectra were recorded at room temperature by exciting the samples through a 355 nm He–Cd laser on a QM40-NIR (PTI, USA). UV-vis diffuse reflectance spectra were recorded on a UV-visible spectrophotometer TU-1901 (PGeneral Instrument Inc., China) at room temperature with BaSO4 as the reference and then converted into absorption spectra via Kubelka–Munk transformation. The dielectric properties of the samples were determined using the HP 4294 A (Hewlett-Packard, USA) impedance analyzer connected with a dc powder supply.

Additional Information

How to cite this article: Sun, Q. et al. Crystalline Structure, Defect Chemistry and Room Temperature Colossal Permittivity of Nd-doped Barium Titanate. Sci. Rep. 7, 42274; doi: 10.1038/srep42274 (2017).

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.