Effects of Secondary Phases on the High-Performance Colossal

are driving the increasing investigations on the colossal permittivity (CP) materials. In ... Keywords: TiO2 ceramics; secondary phases; colossal perm...
3 downloads 0 Views 3MB Size
Research Article Cite This: ACS Appl. Mater. Interfaces 2018, 10, 3680−3688

www.acsami.org

Effects of Secondary Phases on the High-Performance Colossal Permittivity in Titanium Dioxide Ceramics Chunlin Zhao and Jiagang Wu*

Downloaded via UNIV OF LOUISIANA AT LAFAYETTE on September 30, 2018 at 08:56:37 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

Department of Materials Science, Sichuan University, Chengdu 610064, China ABSTRACT: The intensive demands of microelectronics and energy-storage applications are driving the increasing investigations on the colossal permittivity (CP) materials. In this study, we designed a new system of Dy and Nb co-doped TiO2 ceramics [(Dy0.5Nb0.5)xTi1−xO2] with the formation of secondary phases, and then the enhancement of overall dielectric properties (εr ∼ 5.0−6.5 × 104 and tan δ < 8%) was realized in the broad composition range of 0.5 ≤ x ≤ 5%. More importantly, effects of secondary phases on microstructure, dielectric properties, and stability were explored from the views of defect-dipoles and internal barrier layer capacitance (IBLC) effect. According to the defect-dipoles theory, the CP should mainly originate from Nb5+, and the Dy3+ largely contributes to the decreased dielectric loss. Both CP and low dielectric loss were obtained for co-doping with Dy3+ and Nb5+. Besides, the Dy enrichment induced the formation of secondary phases, which were regarded as the low loss unit dispersed into the ceramic matrix, and largely facilitate the decreased dielectric loss. In particular, the analysis of temperature-dependent complex impedance spectra indicated that a stronger IBLC effect caused by the increased grain boundary resistance can also contribute to the optimized CP and low dielectric loss under appropriate contents of secondary phases. KEYWORDS: TiO2 ceramics, secondary phases, colossal permittivity, low dielectric loss, microscopic mechanism

1. INTRODUCTION

Recently, CP TiO2 materials have been considerably investigated because of their improved dielectric properties.13 For example, by (A3+, B5+) co-doping acceptors (A3+, e.g., A = In, Bi, Al, Sm, Ga) and donors (B5+, e.g., B = Nb, Ta, Sb) into rutile TiO2, both a giant dielectric constant (∼104−105) and an improved temperature/frequency stability can be achieved.14−20 The formation of “electron-pinned defect-dipoles” and/or internal barrier layer capacitance (IBLC) effect is often adopted to interpret the physical origin for CP in this material. The former mechanism was thought to be that the defect-dipole clusters (trivalent acceptor ions, pentavalent donor ions together with oxygen vacancy and reductive Ti ions) can induce the formation of the electrons localized structure, finally resulting in large permittivity and low loss.13,15−29 Previously, the IBLC effect was often employed to illuminate the origin of high permittivity in CCTO, which can be attributed to an internal barrier layer capacitor mechanism with conductive grains and insulating grain boundaries.14,17,29−34 According to recent reports, two mechanisms should be assigned to the physical origin of CP TiO2 materials.29,35 In addition, secondary phases can be introduced into the matrix phase of rutile TiO2 ceramics due to the nonisovalent substitution or distinct ion radius.36−40 The pentavalent element ions radius (e.g., Nb5+ ∼ 0.70 Å, Ta5+ ∼ 0.69 Å) is similar to that of Ti4+ (0.745 Å) in TiO2 rutile, which is much smaller or larger than those of most

The development of microelectronic devices and high energy density storage has prompted wide investigations in search for new colossal permittivity (CP) materials.1−3 Recently, several kinds of new CP materials have been fully developed, including CaCu3Ti4O12 (CCTO),2−5 NiO,6 perovskite oxides with BaTiO3,7,8 SrTiO3,9 (BaxSr1−x)TiO3 (BST),10,11 (Pb, La)TiO3,12 etc. As shown in Figure 1, high dielectric constant (εr = ∼104) can be always achieved in these materials; nevertheless, the conflicts among permittivity, large loss, and temperature/ frequency stability still obstruct practical applications of most of these materials.

Received: December 3, 2017 Accepted: January 12, 2018 Published: January 12, 2018

Figure 1. Dielectric constant and dielectric loss of several kinds of CP materials. © 2018 American Chemical Society

3680

DOI: 10.1021/acsami.7b18356 ACS Appl. Mater. Interfaces 2018, 10, 3680−3688

Research Article

ACS Applied Materials & Interfaces trivalent element (e.g., In3+ ∼ 0.94 Å, Bi3+ ∼ 1.17 Å, Sm3+ ∼ 0.958 Å, and Ga3+ ∼ 0.62 Å).14,37,41 Therefore, the obvious difference in ionic radius of trivalent acceptors should be responsible for the secondary phases in the TiO2 ceramics.36−40 In addition, the excellent dielectric performances as well as temperature/frequency insensitivity can be observed in the CP TiO2 systems with impure phases.36,38−40 Also, appropriate amount of secondary phases can positively contribute to the enhanced dielectric properties,38,42 similar to the CCTO ceramics.43,44 However, there are few systematic studies on the concrete effects of secondary phases and corresponding microscopic mechanism on the dielectric properties of CP TiO2 ceramics. In this work, the (Dy0.5Nb0.5)xTi1−xO2 ceramics were prepared by the solid-state method, and a new trivalent element of Dy was doped to TiO2 as the acceptor, facilitating the formation of secondary phases. However, the pentavalent Nb was considered as the substitution donor. We studied the composition dependence of dielectric properties (e.g., permittivity, dielectric loss, temperature, and frequency stability) of (Dy0.5Nb0.5)xTi1−xO2 ceramics, and the effects of secondary phases on dielectric performance and stability were investigated. The relationships between dielectric properties and secondary phases were studied in detail. According to the analyzed results, both CP and low dielectric loss were obtained by co-doping with Dy3+ and Nb5+. Also, the secondary phases caused by the enriched Dy can be regarded as the low loss unit and simultaneously facilitate the IBLC effect. The multieffects result in colossal permittivity and decreased dielectric loss for the ceramics.

2. EXPERIMENTAL PROCEDURE

Figure 2. XRD patterns of (a) (Dy0.5Nb0.5)xTi1−xO2 ceramics and (b) effect of Dy contents on XRD patterns of (Dy0.5Nb0.5)xTi1−xO2 ceramics with x = 5%.

Rutile TiO2 (99%), Dy2O3 (99.9%), and Nb2O5 (99.5%) were used as the raw materials in this work. The samples were prepared by the conventional solid-state method and calcined under 1100 °C for 4 h and then sintered at 1470 °C for 3 h in air. The detailed experimental procedure and the related measurements can be found in our previous works.15,28

To figure out microstructures and the formation of impurity phases, their field-emission scanning electron microscopy (FESEM) images are recorded (Figure 3a−f). All of the samples of Dy- and/or Nb-doped TiO2 feature a relatively dense microstructure. For TiO2 ceramics, single-doping with Dy or Nb possesses larger grain sizes than co-doping, and doping only with Dy attains the largest grain size. Meanwhile, the grain sizes of (Dy, Nb) co-doped TiO2 decrease with increasing x. In addition, single-doping with Dy can result in the formation of secondary phases (Figure 3b), which do not appear for the addition of only Nb (Figure 3a). Also, one can know from Figure 3b,c that the doped Nb can facilitate Dy element merging into the ceramics because the amount of secondary phases at x = 0.5% is apparently less than the ones with doping only 0.5% Dy. Besides, the amount of secondary phases increase with increase in the co-doped contents, as shown in Figure 3c−f. Also, the gradual decrease in grain sizes with rising x should originate from the increased content of secondary phases, which restrict the growth of grain. For the sake of analyzing the elements’ distribution and secondary phases, element mapping is conducted (Figure 4a− f). The Ti, O, and Nb are almost homogeneously distributed. For Dy single-doped ceramics (Figure 4b), the Dy presents apparently the enrichment phenomenon. However, the enrichment becomes less after co-doping Nb (Figure 4c). In addition, the secondary phases exist in the whole ceramics matrix (Figure

3. RESULTS AND DISCUSSION Figure 2a shows the X-ray diffraction (XRD) patterns of Dy and Nb co-doped TiO2 ceramics with 2θ = 20−60°. A main phase of rutile TiO2 (PDF#21-1276) is cited to identify all of the ceramics. All of the samples consist of the impure rutile phases, except for the ones only with low Nb (0.5%) content, and the secondary phases mainly appear in the vicinity of 30, 35, and 50°. In addition, the peaks for more secondary phases appear with an increase in doping contents. The standard diffraction peaks citing from Y-TiO2 (PDF#85-0001) is used to indicate the impurity phases caused by doping Dy3+ because of the similar radius between Y3+ (0.90 Å) and Dy3+ (0.912 Å).40 As shown in Figure 2a, the ceramics only with 0.5% Nb present a pure rutile phase; however, some impure phases appear at 30, 35, and 50° for 0.5% Dy, resulting from a larger ion radius of Dy3+ than that of Ti4+ (0.745 Å). This phenomenon implies that the secondary phases of (Dy, Nb) co-doped TiO2 ceramics mainly are due to the doped Dy. The similar phenomenon is also shown in Figure 2b, and the effects of Dy contents on the XRD patterns of (Dy, Nb) co-doped ceramics with x = 5% can be observed. With increasing Dy contents from less than 20% to more than 30%, the XRD patterns show stronger peaks of impurity phases and increased number of peaks for impurity phases. These impure phases will be further illuminated later. 3681

DOI: 10.1021/acsami.7b18356 ACS Appl. Mater. Interfaces 2018, 10, 3680−3688

Research Article

ACS Applied Materials & Interfaces

Figure 3. FE-SEM images of (Dy0.5Nb0.5)xTi1−xO2 ceramics with (a) 0.5% only Nb, (b) 0.5% only Dy, (c) x = 0.5%, (d) x = 5%, (e) x = 10%, and (f) x = 20%.

Figure 4. Surface element mapping of (Dy0.5Nb0.5)xTi1−xO2 ceramics with (a) 0.5% only Nb, (b) 0.5% only Dy, (c) x = 0.5%, (d) x = 5%, and (f) x = 20%. (e) Cross-sectional element mapping of the ceramics at x = 5%.

Dy, the inadequate amount of Ti and Nb, and a certain amount of O. Furthermore, the excessive Dy mainly occupies the Ti sites. That is, the secondary phases may be assigned to the composite of Dy-only doped TiO2 matrix and a few (Dy, Nb) co-doped TiO2. On the contrary, due to the Dy enrichment, the main phase of (Dy, Nb) co-doped TiO2 ceramics should be regarded as a sample consisting of co-doped TiO2 and a certain amount of Nb-only doped TiO2.

4d,e), and are basically distributed homogeneously into the ceramics. To analyze the components of secondary phases, the local sectional and line-scanning element mappings of the ceramics with x = 5% are carried out, as shown in Figure 5a,b. As previously discussed, the secondary phases largely result from Dy enrichment. Compared with the main rutile phase, one can see from Figure 5 that the secondary phases consist of excessive 3682

DOI: 10.1021/acsami.7b18356 ACS Appl. Mater. Interfaces 2018, 10, 3680−3688

Research Article

ACS Applied Materials & Interfaces

Figure 5. Element mapping of the ceramics (x = 5%) with (a) local section and (b) line scanning.

Figure 6. (a) εr and tan δ of all of the ceramics as a function of x, measured at 1 kHz and room temperature. The valence state of 5% (Dy, Nb) TiO2 ceramics with (b) Dy 4d, (c) Nb 3d, (d) O 1s, and (e) Ti 2p. The fitting results (solid lines) are given by CasaXPS software. (f) The images of the pellets.

TiO2 has a low dielectric constant (∼103).45 However, the addition of (Dy, Nb) can greatly enhance εr of the ceramics (Figure 6a), and their εr dramatically increases to (5.0−6.5) ×

Figure 6 shows the composition dependence of dielectric properties (εr and tan δ) of Dy- and/or Nb-doped TiO2 ceramics, measured at 1 kHz and room temperature. Pure 3683

DOI: 10.1021/acsami.7b18356 ACS Appl. Mater. Interfaces 2018, 10, 3680−3688

Research Article

ACS Applied Materials & Interfaces

Figure 7. (a−e) Dielectric properties and effects of Dy contents on SEM images of (Dy, Nb) ceramics with x = 5%. (f) Microschematic diagram of high dielectric properties induced by the composite of low loss unit and CP unit.

104 when the compositions change from 0.5 to 5% and then drops slowly as x further increases. Meanwhile, a low dielectric loss (5 × 104) and low tan δ ( 4 × 104 and tan δ < 15%). To identify the chemical valence and illustrate the origin of colossal permittivity and low loss, the X-ray photoelectron spectroscopy (XPS) of 5% (Dy, Nb) co-doped TiO2 ceramics are recorded (Figure 6b−e). Two peaks corresponding to Dy 4d3/2 and 4d5/2, respectively, were observed in Figure 6b at the binding energies of 159.0 and 153.1 eV, and the energy of spin−orbit splitting is 5.9 eV, confirming the existence of Dy3+. Meanwhile, one can find from Figure 6c that the peaks consist of the binding energies of 209.5 and 206.8 eV, and the spin− orbit splitting of 2.7 eV should be consistent with Nb5+.13,15 Figure 6d shows the O 1s XPS data and three fitting peaks in this spectrum. The energy peak at 529.6 eV should correspond to the bulk Ti−O bond, and the lowest energy peak at 530.8 eV is assigned to the presence of oxygen vacancies and surface hydroxyl, whereas the peak at 531.7 eV is consistent with an adsorbed surface H2O.13,15 For Ti, the 2p1/2 and 2p3/2 binding energies at 464.0 and 458.3 eV, respectively, correspond to that of pure rutile TiO2 (Figure 6e). Besides, noticeable Ti3+ signal is also detected, and there are two small peaks at 457.6 and 460.1 eV, corresponding to the three valence states of Ti.13,15 In general case, the addition of Dy3+ needs a positive charge to keep the charge balance, and hence a negative charge will be

generated for compensation by simultaneously doping with Nb5+. It is well known that Nb5+ cannot generate oxygen vacancies,13,46 and the Dy3+ can induce oxygen vacancies for charge compensation (eq 1). In addition, an extra electron can be induced when the Nb5+ occupies the position of Ti4+, and this electron is captured by Ti4+ to form Ti3+ (eqs 2 and 3). As shown in the following equation 2TiO2

Dy2O3 ⎯⎯⎯⎯⎯→ 2Dy ′Ti + V •• O + 3OO 4TiO2

Nb2O5 + 2TiO2 ⎯⎯⎯⎯⎯→ 2Nb•Ti + 2Ti′Ti + 8OO +

(1)

1 OO↑ 2 (2)

4+

Ti

+ e → Ti

3+

(3)

To further demonstrate the role of Dy3+ and Nb5+, the images of Dy and/or Nb TiO2 ceramics are recorded (Figure 6f). The TiO2 ceramics with 0.5% Dy are dark yellow, which is similar to the pure TiO2 ceramics. However, the 0.5% Nb-only and (Dy, Nb) TiO2 ceramics are black, which should result from the formation of same color center (electrons captured by oxygen vacancies or the combination of oxygen vacancies and Ti3+ ions). This result is well consistent with their dielectric behaviors (see the inset in Figure 6a), i.e., donor (e.g., Nb5+) doping is responsible for the CP behavior and acceptor (e.g., Dy3+) doping mainly contributes to the low tan δ. For studying the effects of secondary phases on the dielectric properties of (Dy, Nb) co-doped ceramics, we carry out the dielectric properties and their surface morphologies of x = 5% ceramics with different Dy contents, as shown in Figure 7. With rising Dy, the number of secondary phases gradually increases, and it begins to generate long cylindrical secondary phases besides granular ones when the Dy content exceeds 10%. In addition, the variation in Dy contents slightly affects the permittivity of the ceramics, whereas dielectric loss can be apparently influenced. As shown in Figure 7a, few secondary phases are found due to 20% Dy deletion, but the ceramics present a high tan δ. With the increase in the quantity of secondary phases, the tan δ apparently decreases (see the table in Figure 7). Hence, the secondary phases with Dy enrichment 3684

DOI: 10.1021/acsami.7b18356 ACS Appl. Mater. Interfaces 2018, 10, 3680−3688

Research Article

ACS Applied Materials & Interfaces

Figure 8. Temperature-dependent εr and tan δ of (a, b) all (Dy, Nb) TiO2 and (c, d) changed Dy contents on the x = 5% ceramics in the range of −150−200 °C at 1 kHz. Frequency-dependent εr and tan δ of (e, f) all of the ceramics and (g, h) changed Dy concentration on the x = 5% ceramics in the range of 102−106 Hz.

Figure 9. Complex impedance plots of (Dy, Nb) co-doped TiO2 (x = 5%) ceramics with different Dy contents: (a−c) −20% Dy, (e−g) x = 5%, and (i−k) +20% Dy. The solid lines are the fitting curves according to the equivalent electric circuit model. (b), (f), and (j) are the expanded plots. (c), (g), and (k) are the expanded views of the high-frequency data close to the zero. (d), (h), and (l) are their corresponding temperature-dependent direct current conductivity of grain boundary. The pentastar symbols are the experimental results, and the solid lines are the best fitting results according to Arrhenius law.

should be largely responsible for the greatly reduced tan δ. Even though there are lots of secondary phases (Figure 7d,e), the ceramics also present a relatively low tan δ. As discussed above,

the Nb5+ substitutions can afford extra electrons. These electrons, localized by the nearby defect clusters (consisting of Dy3+, Ti3+, oxygen vacancy, etc.), can induce the CP, and 3685

DOI: 10.1021/acsami.7b18356 ACS Appl. Mater. Interfaces 2018, 10, 3680−3688

Research Article

ACS Applied Materials & Interfaces Dy3+ is responsible for the low tan δ due to the formation of localized structure.13,15−29 For the Dy-only doped TiO2 ceramics or the secondary phases, there should be excessive localized structure of defect clusters to trap electrons and hence weakened the freely hop loss. For Nb-only doped TiO2 ceramics or the main phase of (Dy, Nb) co-doped TiO2, a mass of electrons was generated, inducing the sharply increased permittivity. However, there are not enough localized structure to imprison these redundant electrons, and therefore a relatively high tan δ would be induced. Consequently, the secondary phases can be regarded as a low loss unit, and the matrix grain of co-doped TiO2 acts as a CP unit. The excellent dielectric properties with high permittivity and low loss in (Dy, Nb) co-doped TiO2 ceramics should be ascribed to the outcome of microscopic composite effect between the two units (Figure 7f). To reveal the temperature/frequency stability of dielectric properties in (Dy, Nb) co-doped TiO2 ceramics, their permittivity and dielectric loss as functions of temperature and frequency are measured (Figure 8). It can be observed from Figure 8a,e that the dielectric constant has similar values for −150−200 °C and 102−106 Hz, indicating an excellent temperature/frequency stability of permittivity, especially for 0.5 ≤ x ≤ 5%. In particular, the tan δ also shows a relatively good temperature/frequency stability (Figure 8b,f). For investigating the influence of secondary phases on temperature/frequency stability, the corresponding measurement is carried out (Figure 8c,d,g,h). Apparently, enhanced temperature/frequency stability is achieved with increasing amount of secondary phases, especially for 10% Dy excessive co-doped ceramics, indicating the optimized effects of secondary phases on dielectric stability. As previous reports, the IBLC effect influences the CP behavior of (A3+, B5+) co-doped TiO2 ceramics. The local interfacial polarization at the interface between conductive grain and insulating grain boundary is dominant to the permittivity enhancement.14,17,29−34 Hence, the temperaturedependent complex impedance for x = 5% (Dy, Nb) TiO2 ceramics with different Dy contents are measured to study the IBLC effect with different contents of secondary phases (Figure 9). The corresponding resistance is obtained by fitting the impedance spectra to an equivalent circuit consisting of two parallel resistance (R) and capacitance (C) elements connected in series, of which RgCg stands for the contribution from grains, RgbCgb is the contribution from grain boundaries (Figure 10a).14,17 In the temperature range of −20−200 °C, the (Dy, Nb) TiO2 ceramics almost exhibit a single semicircular arc. First, it can be seen from the expanded plots of high-frequency data (Figure 9c,g,k) that the semicircular arc shows a nonzero

but tiny intercept on the origin of real-axis Z′ (high-frequency end). This intercept is the grain resistance (Rg)17 with a small value (