1 Introduction

High dielectric constant materials are useful for miniaturization of electronic devices. In this situation, calcium copper titanate (CaCu3Ti4O12) has attracted a lot of interest in the past few years. This material exhibits colossal dielectric constant of εr ~10,000 with the temperature range 100–600 K [1–3]. Temperature independent colossal dielectric constant of CaCu3Ti4O12 (CCTO) has drawn the attention of many researchers worldwide. But high dielectric loss (tanδ), low breakdown voltage and high leakage current have certainly become a barrier to its industrial applications [4]. Among the previous works, the internal barrier layer capacitor (IBLC) model, which considers CCTO as consisting of semi-conducting grains with insulating grain boundaries, has been widely accepted to describe the origin of the unusually high dielectric constant of CCTO [5–7]. This electrostatic barrier obstacles the current flow through the bulk semi-conducting grains in presence of an applied field resulting huge dielectric constant value.

It is known that the radius, valancy and coordination number of an element are important parameters to determine the site it occupies in the parent compound. Up to now, there are several reports on the giant dielectric properties of CCTO [8, 9] and related oxides of the type ACu3Ti4O12 where A = Sr [10], Cd [11, 12], La2/3 [13, 14]. Some doping ions can reduce significantly dielectric loss of CCTO ceramics. La3+ substitution into CCTO can reduce dielectric loss to be less than 0.03 [15]. Both Nb and Fe doping lowered the dielectric loss [16]. The dielectric loss at low frequencies (f <104 Hz) in CaCu3Ti4O12 could be reduced by adding CaTiO3 and ZrO2 [17, 18]. However, the reduction of dielectric loss is usually accompanied by a decrease in dielectric constant value. Dielectric loss of CCTO is too high due to conducting crystalline grains/sub grains. The conductance of the barrier leads to the leakage loss. It has been revealed that addition of CaTiO3 increases the resistance of the barrier layers [19, 20].

CCTO was generally prepared by the traditional solid-state method [1], which suffers from the disadvantages of tedious work, relatively long reaction times, high calcined temperatures and inhomogeneity [21]. In recent years, there were a lot of reports on the sol–gel processing to synthesize CCTO [22–24]. Compared with traditional solid-state method, the sol–gel method affords intimate and homogeneous mixing of the metal ions at the molecular scale, thus reducing the diffusion path length required. Shorter diffusion lengths lead to shorter reaction times and lower temperatures [23]. In addition, many different processing conditions can be selected in the sol–gel technique, such as reagent, solvent, temperature, etc. In this work, an attempt has been made to synthesize a few compositions in the Ca(1−3x/2)Nd x Cu3Ti4O12 (CNCTO) by sol–gel method and study its microstructure and dielectric properties.

2 Experimental

All of employed chemicals were analytical reagent grade and supplied by ShangHai LingFeng Chemical Reagent Co. Ltd. Ca(1−3x/2)Nd x Cu3Ti4O12 (x = 0, 0.1, 0.2 and 0.3, designated as CCTO, CNCTO1, CNCTO2 and CNCTO3, respectively) powders were prepared by sol–gel method. The weighted Ca(NO3)2·4H2O, Cu(NO3)2·3H2O and Nd(NO3)3·6H2O were dissolved into absolute ethyl alcohol and deionize water(the volume ratio of absolute ethyl alcohol to water = 3:1). Meanwhile, stoichiometric Ti(OC4H9)4 liquids were dissolved into CH3CH2OH to form 0.5 M solution. Acetic acid was added to adjust the acidity of solution (pH 3) in order to form the gels slowly. Both above solutions were mixed with vigorous stirring. Amount of citric acid was added into the mixture and the blue gel was gradually formed. Subsequently, the gel was dried at 80 °C to obtain xerogels. Finally, the xerogels was calcined at 800 °C for 2 h to yield Ca(1−3x/2)Nd x Cu3Ti4O12 powders.

The as-prepared powders were thoroughly pulverized in an agate mortar and pressed into discs of 10 mm in diameter at around 50 MPa and sintered at 1,000 °C for 6 h. After polishing, silver paste was used to coat both sides of sintered discs, and then fired at 600 °C for 10 min to form electrodes in order to measure dielectric properties.

The thermal decomposition characteristic of xerogel was investigated by TG/DSC (TG 209 F3, Germany) in air atmosphere with heating rate of 10 °C/min from 30 to 800 °C. The phase composition of the powders and the sintered ceramics were analyzed by X-ray powder diffraction (D/max 2,500 PC, Japan) using a Cu Kα radiation (λ = 0.15406 nm) form 20 to 80°. The microstructure of powders and ceramics was observed using SEM (JSM-6360LA, Japan) and energy dispersive spectroscopy (JSM-6360LA, Japan). The relative density of ceramics was measured by the Archimedes method. Dielectric properties of sintered ceramics were measured as a function of temperature using a LCR meter (AT821, China).

3 Results and discussion

TG/DSC was carried out on xerogel to study the thermal decomposition process. Figure 1 shows the representative TG/DSC plots of the precursor of Ca(1−3x/2)Nd x Cu3Ti4O12 with x = 0.2. It is observed that there are three stages of weight loss corresponding to an endothermic peak and two exothermic peaks in the temperature range up to 700 °C. The DSC curve appears a small endothermic peak at 192 °C, accompanied by about 26.6 % weight loss in TG curve, which resulted from the removal of absorbed water and solvent. At the elevating temperature, two exothermic peaks appear at 247 and 295 °C in DSC curve, accompanied by 16.9 and 7.7 % weight loss in TG curve, respectively, which may be due to the main decomposition of nitrates and organic groups. There is almost no weight loss in TG curve above 400 °C, which indicates non-crystalline oxides slowly reacted and transferred to polycrystallines.

Fig. 1
figure 1

TG/DSC plot of Ca(1−3x/2)Nd x Cu3Ti4O12 for (x = 0.2)

Full size image

To help understand the evolution of the powders, XRD patterns of Ca(1−3x/2)Nd x Cu3Ti4O12 for (x = 0.2) xerogels with different heat-treatment temperatures are shown in Fig. 2. It can be seen in the figure that xerogels are amorphous without heat-treatment. The powders calcined at 300 °C shows the peaks corresponding to CuO. The diffraction peaks corresponding CCTO crystalline gradually appear and small amount of CaTiO3 and CuO are also present in this stage when the calcination temperature increases to 600 °C. With the increase in the calcination temperature from 600 to 800 °C, the peaks of CCTO increase in intensity and the peaks corresponding to CuO and CaTiO3 decrease and vanish completely when the calcinations temperature is 800 °C. This indicates that CaTiO3 react with CuO to form CCTO powders above 600 °C. And we observed a small amount of rutile TiO2 phase formation at 800 °C. No secondary phases containing Nd were detected because XRD is not sensitive to concentrations under 0.5 wt% [25]. The Nd3+ was most likely substituted in Ca lattice sites. These two occurrences are attributed to the lack of Nd found in the CNCTO2 XRD spectra.

Fig. 2
figure 2

XRD patterns of CNCTO2 xerogels with different heat-treatment temperatures

Full size image

Figure 3 shows SEM photographs of pure and Nd doped CCTO powders at various levels after calcining at 800 °C for 2 h. SEM observation indicates that all samples exhibit a good dispersity with uniform particle size. It decreases to 120 nm for x = 0.3, from 250 nm for x = 0. It can be assumed that Nd has effect on CCTO particle size. Meanwhile, the particle size decreases with increase in dopant concentration, also reported by other authors for sol–gel methods [14].

Fig. 3
figure 3

SEM photographs of a CCTO, b CNCTO1, c CNCTO2, d CNCTO3 powders

Full size image

X-ray diffraction patterns of Ca(1−3x/2)Nd x Cu3Ti4O12 (where x = 0, 0.1, 0.2 and 0.3) ceramics are shown in Fig. 4. Comparing these patterns with those of the standard powder XRD pattern of CCTO (JCPDS Card No.75–2188), the diffractogram shows that the sintered ceramics are polycrystalline and no secondary phase is formed for doping by Nd up to x = 0.3, leading to conclude that Nd3+ is replacing Ca2+ in its lattice position in the doped ceramics, without distorting the crystal structure. The lattice parameter was determined using least square refinement method show an increasing trend with the enhancing amount of Nd3+ substitution. It increases to 0.73907 nm for x = 0.3, from 0.7320 nm for x = 0. This lattice parameter enhancement with increasing amount of Nd3+ doping is expected since the ionic size neodymium (Nd3+) is bigger than that of calcium (Ca2+). This is called ionic size effect. Researchers have proved the existence of Ti3+ along with Ti4+ from the photoemission spectroscopy of CCTO [26]. Therefore this reduction of lattice parameter may also be due to increasing amount of Ti3+, formed by reduction of Ti4+ to Ti3+ by Nd3+ due to its oxidizing property.

Fig. 4
figure 4

a X-ray diffraction pattern of Ca(1−3x/2)Nd x Cu3Ti4O12 ceramics, b Variation of lattice parameters with the extent of Nd doping

Full size image

Figure 5 presents the variation of relative density of Ca(1−3x/2)Nd x Cu3Ti4O12 ceramics as a function of Nd concentration. It can be seen that the relative densities increase as Nd concentration increase. CNCTO3 ceramics shows the highest relative density of 98.1 %, which may be due to less pores and uniform particle size of powders. It is difficult to prepare fully dense ceramics due to a lot of factors which can affect on producing fine-grained materials with high density, such as powder preparation process, powder agglomeration and sintering conditions [27]. In addition, many researchers reported that dopants can also have effect on the densification and microstructure of perovskite [28].

Fig. 5
figure 5

Relative density of pure and Nd doped CCTO ceramics sintered at 1,000 °C for 6 h

Full size image

Figure 6a, b, c presents SEM images of fracture surface for pure and Nd doped CCTO ceramics. It is clearly seen that microstructure of CNCTO ceramics changes significantly with increase in concentration of Nd. The average grain size becomes small with increase in Nd doping. Pure CCTO shows the average grain size about 2.0 μm, whereas for subsequent doping concentrations, the average grain size decreases to 0.8 μm for CNCTO3 ceramics. This result indicates that Nd inhibits grain growth of CCTO ceramics.

Fig. 6
figure 6

SEM images of Ca(1−3x/2)Nd x Cu3Ti4O12 ceramics: a x = 0, b x = 0.1, c x = 0.3, d gives EDS analysis of CNCTO (x = 0.3) for different regions

Full size image

Energy spectrum of elemental analysis performed on different regions of CNCTO3 gives insight into the proportion of different elements in grain and grain boundary. It can be seen from Fig. 6d that Nd appears in the grain and grain boundary of CNCTO3 as well as Ca, Cu, Ti, and O (the peaks of Pt were originated for conducting of specimen). Major content of Nd was found in grain (10.99 %) than in grain boundary region (9.57 %). Lesser Nd content in the grain boundary may be the major cause for the refinement in grain size [29].

Figure 7a illustrates temperature dependence of the dielectric properties for pure and Nd doped CCTO ceramics at the frequency of 10 kHz. The dielectric constant (εr) and dielectric loss (tanδ) increased continuously by increasing temperature, although the rate of these enhancements was not the same. It can be seen that the thermal stability of CNCTO2 and CNCTO3 is considerably higher than that of CCTO and CNCTO1. It is obvious that the dielectric constant and dielectric loss was found to drastic decreases with increase in Nd dopant concentration. CCTO exhibits the highest dielectric constant (2.9 × 104) and dielectric loss (0.24) at 20 °C, while CNCTO3 shows the lowest dielectric constant (1.12 × 104) and dielectric loss (0.12) at 20 °C. The results indicate that Nd doping can decrease dielectric constant as well as dielectric loss. According to the literature [30], the dielectric constant of CaCu3Ti4O12 is closely related to its grain size, and larger grains result in higher dielectric constants for CaCu3Ti4O12. Our results lead to the same conclusion, whereby with larger grain size, a larger dielectric constant was observed for pure CCTO. It can be easily understood with the internal barrier layer capacitor (IBLC) model, which consider materials consist of conducting grains and insulating grain boundaries. As a greater amount of crystal defects are able to exist in larger grains than that in smaller ones. Thus, there must be insulating planar defects in these grains. Dopants frequently concentrate at lattice defects. The Nd may be concentrating on the important planar defects in CCTO. This could increase the electrical resistivity of the internal barrier and decrease the dielectric loss.

Fig. 7
figure 7

Temperature dependence of dielectric properties for Ca(1−3x/2)Nd x Cu3Ti4O12 ceramics at the frequency of 10 kHz

Full size image

4 Conclusion

Sol–gel method was employed to synthesize pure and Nd-doped CaCu3Ti4O12 ceramics. XRD patterns shows that CNCTO ceramics (sintered at 1,000 °C for 6 h) were single phase with no Cu-rich phase. The lattice parameters show an increasing trend with the enhancing amount of Nd3+ substitution. SEM graphs show that the particle size decreases with increase of Nd concentration. Dielectric constant and dielectric loss is found to decrease with the increase of Nd incorporation. CNCTO3 ceramics exhibits the lowest dielectric constant of 1.12 × 104 as well as the lowest dielectric loss value of 0.12 at 20 °C (10 k Hz). Nd doping has obvious effects on grain growth and dielectric properties of CCTO ceramics.