Journal of Solid State Chemistry 206 (2013) 217–225
Contents lists available at ScienceDirect
Journal of Solid State Chemistry journal homepage: www.elsevier.com/locate/jssc
Cu-doping effect on dielectric properties of organic gel synthesized Ba4YMn3 xCuxO11.5 7 δ Tristan Barbier a,n, Cécile Autret-Lambert a, Pascal Andreazza b, Antoine Ruyter a, Christophe Honstettre a, Sébastien Lambert c, François Gervais a, Marc Lethiecq a a
Université François Rabelais de Tours, CNRS, CEA, ENIVL, GREMAN UMR 7347, 37200 Tours, France Centre de Recherche sur la matière divisée (CRMD), Université d′Orléans, CNRS, FRE3520, 1B rue de la Férollerie, 45071 Orléans, France c CEA—DAM, Le Ripault, 37260 Monts, France b
art ic l e i nf o
a b s t r a c t
Article history: Received 22 May 2013 Received in revised form 30 July 2013 Accepted 1 August 2013 Available online 7 September 2013
Copper doped-Ba4YMn3 xCuxO11.5 7 δ samples were synthesized by an organic gel assisted citrate process. X-ray diffraction of compositions with x ¼0.002, 0.005, 0.01, 0.02 and 0.04 does not reveal any change of hexagonal perovskite structure on doping. The effects of Cu-doping on the microstructure and dielectric properties were investigated. Cu doping modifies the electrical properties at the level of the impedance characteristics of both grain and grain boundary and to understand these different behaviours, we have carried out high-resolution transmission electron microscopy analysis. Among the Ba4YMn3 xCuxO11.5 7 δ specimens studied, the composition x ¼ 0.002 shows a permittivity (ε′r) higher than the undoped compound and a lower loss tangent (tanδ) over several orders of magnitude of frequency. & 2013 Elsevier Inc. All rights reserved.
Keywords: Hexagonal perovskite Colossal permittivity X-ray diffraction Transmission electron microscopy Stacking faults
1. Introduction Materials with high dielectric constant are used in technological applications such as multilayer capacitors. High dielectric constant allows the miniaturization of capacitive components, thus offering the opportunity to decrease the size of microelectronic devices. Materials based on ferroelectric BaTiO3 are commonly used but they raise the problem in the high temperature range due to their large variation in permittivity near their Curie temperature. Materials having general formula ACu3Ti4O12 (A¼Ca, Ba, Sr) [1] seem to be good candidates to replace BaTiO3 due to their colossal dielectric constant, up to 105 [1], which is almost constant over a wide range of temperatures (100 K to 600 K) and frequencies (102 Hz to 106 Hz) [2,3]. It has been suggested that the high permittivity of CaCu3Ti4O12 (CCTO) was extrinsic. According to impedance spectroscopy measurements of CCTO ceramics, Sinclair et al. [4] suggested that these ceramics contain semiconducting grains and insulating grain boundaries. They explained that the high dielectric constant is due to the internal barrier layer capacitance (IBLC) effect from grain boundaries. Furthermore, Fang et al. [5] published a detailed investigation on the role of boundary barrier layers in CCTO ceramics. Their results are consistent with the IBLC model suggesting that both grains and grain boundaries
n
Corresponding author. Tel.: þ 33 619634223. E-mail address:
[email protected] (T. Barbier).
0022-4596/$ - see front matter & 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jssc.2013.08.004
are responsible for dielectric response of CCTO. Many studies were carried out on CCTO in order to optimize their properties, especially with the addition of different dopants such as nickel, copper or iron [6–9]. To improve this theory and confirm the first results obtained by Kuang et al. [10] who describe that the high permittivity of Ba4 YMn3O11.5 7 δ is due to the IBLC effect, thanks to impedance spectroscopy measurements, Ba4YMn3 xCuxO11.5 7 δ compounds with x¼0, 0.002, 0.005, 0.01, 0.02 and 0.04 have been synthetized using the organic gel-assisted citrate process and investigated. For undoped and Cu-doped Ba4YMn3O11.5 samples, we provide a detailed structural analysis using X-rays diffraction, inductively coupled plasma atomic emission spectrometry (ICP-AES), and X-ray photoelectron spectroscopy (XPS). The dielectric properties of all samples were carefully determined, and the magnetic properties of all samples were also investigated. More specifically, a search for correlations between structural (XRD and HTREM imaging), magnetic and dielectric properties has been undertaken. 2. Experimental 2.1. Synthesis Undoped and doped Ba4YMn3 xCuxO11.5 7 δ powders were prepared by an organic gel assisted citrate process [11]. This synthetic
218
T. Barbier et al. / Journal of Solid State Chemistry 206 (2013) 217–225
route allows single-phase compounds to be obtained with large grain sizes and therefore very high dielectric properties [12]. The starting materials (Ba(NO3)2, Y(NO3)3, Mn(NO3)2, Cu(NO3)2) taken in stoichiometric proportions and ammonium citrate, acrylamide and NN′-methylenediacrylamide were used to obtain the gel. This gel was then calcined in air at 500 1C for 20 h in a muffle furnace. The resulting powders were grounded and pressed into cylindrical pellets (diameter ¼13 mm; thickness¼1–2 mm) by using successively uniaxial and isostatic hydraulic pressing (3 t). Then, pellets were sintered at 1400 1C for 30 h in air. The densities of the obtained samples were measured by the Archimede′s method. Different thermal treatments were investigated to enhance the density and evaluated using dilatometry measurements. The optimal density was only 83% of the theoretical one.
2.5. Electron spin resonance Electron spin resonance (ESR) measurements have been performed with a BRUKER EMX 6/1 spectrometer operating in the Xband (E9.5 GHz). This spectrometer can be used with a small amount of polycrystalline samples (E10 mg) in a wide range of temperatures (5–300 K) thanks to its helium flux cryostat. Three parameters were deduced from the ESR signals: the Lande factor (geff ¼hυ/mB Hres), the line width (ΔHpp), and the double integration normalized (D.I.N.). ΔHpp is estimated from the peak-to-peak distance between the maximum of the ESR absorption and the minimum of the derivative function. Note that the D.I.N. is proportional to the magnetic susceptibility χESR. 2.6. Magnetic measurements
2.2. X-ray diffraction Powder diffraction data for Rietveld refinements were collected using a PanalyticalX′Pert PRO diffractometer (Bragg–Brentano mode) with a X′Celerator detectorusingthe CuKα radiation (λ E 1.540 Å). X-ray diffraction (XRD) data were collected over a 2θ range of 10–1501 with a step size of 0.0131, and a step time 12 s. Experimental profiles were modelled using a pseudo-Voigt profile shape function. With respect to the crystallographic structure, the lattice parameters, atomic positions, isothermal temperature factors (Biso), and site occupancies were refined using the FullProf software [13].
2.3. Chemical analysis Chemical compositions of the different samples were checked by inductively coupled plasma atomic emission spectrometry (ICP-AES) measurements. X-ray photoelectron spectroscopy (XPS) surfaces of Ba4YMn3 xCuxO11.5 7 δ samples were taken with a Thermo VG ESCALAB 250 spectrometer with a monochromatic source of Al radiation (Kα¼1486.6 eV) at room temperature coupled with a hemispherical energy analyser between 10 and 1200 eV. This spectrometer worked under a base pressure of 5.10 10 Torr in the multi-technical analysis chamber. The binding energy (BE) was calibrated with respect to the C (1s) value of a contaminated carbon, on the surface of the pellets, as 285.0 eV. The spectrum was obtained for each sample just after they have been polished and degased. A thermo advantage data system was used for data acquisition and processing. Determinations of core-level peak positions and spectral intensities (peak areas) were done after smoothing and subtracting a background signal (Shirley′s model [14]).
2.4. Electron microscopy The microstructure was analysed by scanning electron microscopy (SEM) using a Hitachi 4160-F microscope at 200 kV. SEM images also show that ceramics are not optimally sintered with the presence of some residual open-porosities which can explain the low density value. More, the distribution of grain sizes exhibits two peaks centered at 1 and 3 mm, respectively. Concerning the morphology, we have noticed that some grains are plate like with a hexagonal surface which is characteristic of this hexagonal perovskite structure. Transmission electron microscopy (TEM) investigations were performed at an accelerating voltage of 200 kV on a JEOL 2100F instrument equipped with a double tilt sample holder (7201). HRETM multislice simulations were performed using JEMS software.
AC magnetic susceptibility measurements (ACMS) were also performed in order to analyse magnetic properties using a physical property measurement system (PPMS—Quantum design). AC magnetic susceptibility was measured in the temperature range from 5 K to 300 K for an external field B¼ 1 T in order to determine the Neel temperature. Two kinds of measurements have been performed: a zero field cooling (ZFC) and a field cooling (FC) to verify the absence of spin glass. No differences have been observed confirming the absence of magnetic cluster which could have been at the origin of a spin glass behaviour. 2.7. Impedance spectroscopy Finally, the dielectric constant, loss tangents and conductivity were measured using the Agilent 4294A and its holder for pellets (Agilent 16451B). These pellets were polished and then Au electrodes were sputtered (E350 nm) onto their faces before all electrical measurements. Impedance measurements were performed at room temperature in a frequency range of 40 Hz to 100 MHz with an AC amplitude of 500 mV.
3. Results and discussion 3.1. Crystallography Fig. 1 shows a typical XRD pattern of sintered Ba4YMn2.99Cu0.01O11.5 7 δ ceramics (the nominal compositions of this compounds, obtained by the ICP-AES measurements is Ba3.98(7)YMn2.99 (8)Cu0.00(9)O11.48(0), the other nominal compositions were written in Table 1). All the analysis showed that compounds are single phase. For Ba4YMn3O11.5 7 δ sample (x ¼0), lattice parameters are in good agreement with those previously published by Kuang et al. [10] although they have used the solid route. Experimental data were fitted to a hexagonal structure with a R3m space group (no. 166). The difference between the observed and the calculated X-ray diffraction patterns shows that the structure is suitably formed. Indeed, the cubic perovskite structure can be described as a stacking of close packed of AO3 layers in a cubic sequence (ABC…). On another hand, the “hexagonal” perovskite structure shows a structure characterized by a (ABA…) stacking sequence of the BaO3 layers. Between these two extreme descriptions, numerous intermediate structures, called hexagonal polytypes, may exist with different AO3 stacking sequences of a mixed hexagonal (h)—cubic (c) structure. To evaluate the stability of the “cubic” perovskite ABO3, Goldschmidt proposed a tolerance factor t ¼ (rA þrO)/(√2(rB þrO)) where rA and rB are the ionic radii of the cation located on A and B sites, respectively, and rO is the ionic radii of O2 [15]. Goldschmidt suggested that the perovskite (ABO3) family could be stable in a range of tolerance factor from
T. Barbier et al. / Journal of Solid State Chemistry 206 (2013) 217–225
219
0.8 to 1.1. Yu et al. [16] showed that this factor can also be used in the case of hexagonal polytypes. The theoretical compound Ba4YMn3O11.5 7 δ possesses a (cchh)3 stacking sequence of AO3 layers, and it can be supposed that Y and Mn are both on the B-site (octahedral) in the perovskite model. So, the tolerance factor can be rewritten as follows: t ¼(rA þrO)/(√2(1/4rY þ (3 x)/ 4rMn þx/4rCu)þrO). From this equation, tolerance factors calculated for all samples are larger than 1 and remain almost constant tE 1.0523 for 0r xr 0.04. As a result the phase remains
hexagonal for all compositions. The lattice parameters and the associated volume of these compounds were deduced from the refinement of experimental data. The a parameter remains almost constant with the Cu substitution whereas the c parameter decreases. The unit cell volume decreases with increasing Cu content up to x¼ 0.04 (Fig. 2). This figure shows two different behaviours: x o0.01 and x 40.01. These two behaviours differ by the rate of decrease in volume as a function of doping level and show that this variation does not follow the Vegard′s law.
Fig. 1. Rietveld refinement of XRD data for Ba4YMn2.99Cu0.01O11.5 hexagonal perovskite.
Fig. 2. Variation of the lattice volume with Cu content (left). Variation with the grain resistivity with Cu content (right).
Table 1 Refined structural parameters for Ba4YMn3 xCuxO11.5 7 δ. Samples
x¼ 0
x ¼0.002
x¼ 0.005
x¼ 0.01
x ¼0.02
x ¼0.04
a (Å) c (Å) Volume (ų) Ba1 (2/3, 1/3, z) (Å) Biso (Å2) Ba2 (0,0 z) (Å) Biso (Å2) Y (0,0,0) Biso (Å2) Mn1/Cu1 (1/3, 2/3,z) (Å) Biso (Å2) Mn2/Cu2 (1/3, 2/3,1/6) Å Biso (Å2) O1 x (Å) y (Å) z (Å) O2 x (Å) y (Å) z (Å) RBragg χ² Ba1–O1(n3) (Å) Ba1–O2(n6) (Å) Ba1–O2(n3) (Å) Ba2–O1(n6) (Å) Ba2–O1(n3) (Å) Ba2–O2(n3) (Å) Y–O2(n6) (Å) Mn1–O1(n3) (Å) Mn1–O2(n3) (Å) Mn2–O1(n6) (Å) Mn1–Mn2 (Å) Mn1–O1–Mn2 (1) T Grain size Curie Constante Curie–Weiss temperature θcw/(K) Compositions measured by ICP-AES
5.7910(8) 28,662(0) 832.077(8) 0.0471(2) 1.6(2) 0.1296(2) 1.5(2) 0.03(1) 0.0770(4) 1.4(5)
5.7908(2) 28.654(6) 831.991(0) 0.0477(2) 1.2(1) 0.1292(1) 0.9(1) 0.73(3) 0.078(3) 0.2(1)
5.7895(2) 28.654(1) 831.770(4) 0.0481(1) 1.1(1) 0.1291(1) 0.4(1) 0.63(8) 0.773(4) 1.2(1)
5.7907(4) 28.649(7) 831.758(5) 0.0481(2) 0.2(1) 0.1293(2) 1.0(2) 0.9(2) 0.0778(4) 0.9(3)
5.7892(9) 28.644(8) 831.632(9) 0.0498(2) 2.2(3) 0.1295(2) 0.2(2) 1.0(4) 0.0732(4) 1.2(5)
5.7904(4) 28.617(9) 831.272(6) 0.0474(3) 0.8(2) 0.1294(2) 0.1(2) 0.9(2) 0.0778(5) 0.4(2)
0.8(4)
1.4(3)
0.62(7)
0.62(7)
1.1(5)
0.9(5)
0.48376 0.51625 0.12414 0.48748 0.51252 0.28987 5.35 3.52 2.869(7) 2.896(3) 3.023(5) 2.901(3) 2.928(4) 3.055(5) 2.185(3) 2.028(1) 1.827(1) 1.932(2) 2.577(1) 88.9(6) 1.05235 1.5(3) mm 5.31
0.48156 0.51845 0.12461 0.4886 0.5114 0.28883 7.10 1.23 2.882(4) 2.896(1) 3.069(4) 2.908(2) 2.940(3) 3.015(3) 2.198(2) 1.987(6) 1.839(5) 1.918(2) 2.525(9) 88.6(4) 1.05235 1.8(1) mm 3.81
0.48156 0.51845 0.12461 0.48860 0.51140 0.28883 7.06 1.53 2.882(2) 2.901(1) 3.103(2) 2.916(8) 2.928(2) 3.046(1) 2.275(1) 1.884(1) 1.747(1) 1.888(1) 2.544(9) 88.1(7) 1.05235 1.7(2) mm 3.79
0.46562 0.53439 0.13024 0.51411 0.48589 0.28968 9.80 1.15 3.040(2) 2.901(3) 3.175(3) 2.926(1) 2.968(2) 2.897(3) 2.174(3) 2.068(3) 1.965(2) 1.855(1) 2.546(1) 89.1(2) 1.05234 1.4(1) mm 3.30
0.48541 0.51460 0.08355 0.49466 0.50534 0.25705 9.03 1.20 2.878(3) 2.905(2) 3.130(4) 2.897(2) 2.812(3) 2.972(3) 2.177(1) 2.130(5) 1.949(4) 2.011(2) 2.537(9) 88.9(1) 1.05234 1.5(2) mm 3.10
0.48141 0.51860 0.13297 0.50231 0.49769 0.28952 8.77 1.18 2.897(2) 2.895(1) 3.147(2) 2.904(1) 2.857(1) 2.955(2) 2.152(2) 2.083(2) 1.877(2) 1.898(1) 2.543(9) 89.5(7) 1.05233 1.9(4) mm 2.29
254.46
234.11
231.76
228.76
210.23
170.07
Ba3.98(9)Y Mn2.99 (2)O11.49(9)
Ba3.99(7)YMn2.99(7) Cu0.00(2)O11.49(9)
Ba3.98(5)YMn2.99(4) Cu0.00(5)O11.48(8)
Ba3.98(7)YMn2.99(8) Cu0.00(9)O11.48(0)
Ba3.98(8)YMn2.97(7) Cu0.02(3)O11.47(2)
Ba3.98(2)YMn2.95(5) Cu0.04(3)O11.45(9)
220
T. Barbier et al. / Journal of Solid State Chemistry 206 (2013) 217–225
3.2. XPS In order to understand this evolution, XPS measurements were performed. XPS spectra of all compounds were measured in a wide range of energies. Only Ba, Y, Mn and O elements are observed on the surface of the polished samples. Cu element is not visible because, due to the low x-values, it is presumably dispersed in the bulk. More important, no additional peaks related to contaminations were detected. Fig. 3 displays only the Mn2p spectrum for different samples. The broad emission line is fitted for Mn2p3/2 with the largest peak location fixed at about 642 eV, and a distinct smaller intensity peak at 654 eV for Mn2p1/2. The binding energy difference between 2p3/2 and 2p1/2 is about 12 eV. The Mn2p lines show a complex character, due to overlapping of several lines. Indeed the Mn2p3/2 spectrum of each single oxidation state consists
of five multiplet emission lines separated from each other by 1 eV. For the undoped compound, binding energies are between 641.7 and 646.7 eV. The fit of these lines suggests that the Mn cations valence is þ 4. As the Cu ratio increases, the line centered around 643 eV broadens, indicating that Mn cations exhibit a mixed valence of þ3 and þ4 in accordance with the results of Gupta et al. (see Table 2) [17,18]. Then, an estimation of the Mn3 þ /Mn4 þ ratio has been extracted from the experimental data. It can be observed on this figure that the Mn4 þ percentage decreases when the Cu content increases. This has been explained by Li et al. [19]. They demonstrated that Cu2 þ ions reduce into Cu þ by heating. Then on the (Mn, Cu) sites, the new equilibrium becomes:
3Cu2 þ -3Cu þ þ 3=2VO€
Fig. 3. Mn2p1 and Mn2p3 spectrum for doped samples from XPS measurements at RT.
T. Barbier et al. / Journal of Solid State Chemistry 206 (2013) 217–225
221
Table 2 Spectral fitting parameters for Mn2p, for Ba4YMn3 xCuxO11.5 7 δ.
Mn2p3 A
Mn2p3 B
Mn4 þ
Mn2p3 C
Mn2p3 D
Mn2p3 E
Mn2p3 A
Mn2p3 B
Mn3 þ
Mn2p3 C
Mn2p3 D
Mn2p3 E 4þ
% Mn
Binding energy (eV) Intensity (%) FWHM (%) Binding energy (eV) Intensity (%) FWHM (%) Binding energy (eV) Intensity (%) FWHM (%) Binding energy (eV) Intensity (%) FWHM (%) Binding energy (eV) Intensity (%) FWHM (%) Binding energy (eV) Intensity (%) FWHM (%) Binding energy (eV) Intensity (%) FWHM (%) Binding energy (eV) Intensity (%) FWHM (%) Binding energy (eV) Intensity (%) FWHM (%) Binding energy (eV) Intensity (%) FWHM (%)
x ¼0
x ¼0.002
x ¼0.01
x¼ 0.02
x ¼0.04
Gupta and Sen
641.75 100.0 100.0 642.75 71.0 98.9 643.65 31.0 99.5 644.65 21.0 98.4 646.75 9.0 100.0 – – – – – – – – – – – – – – – 100%
642.43 100.0 100.0 643.43 77.5 100.0 644.33 82.5 100.0 645.33 17.5 100.0 647.33 45.0 100.0 641.23 100.0 100.0 641.93 101.0 100.0 642.73 117.5 100.0 643.73 91.76 100.0 645.13 43.5 100.0 85%
642.15 100.0 100.0 643.15 66.2 100.0 644.05 32.3 100.0 645.05 13.8 100.0 647.11 10.8 100.0 640.95 100.0 100.0 641.65 100.0 100.0 642.45 0.74 100.0 643.45 23.0 100.0 644.85 6.0 128.0 84%
642.42 100.0 100.0 643.42 72.3 100.0 644.32 33.3 100.0 645.32 16.6 100.0 647.38 27.7 100.0 641.22 100.0 100.0 641.92 91.0 100.0 642.72 82.0 100.0 643.72 46.0 100.0 645.12 15.0 100.0 83%
642.55 100.0 100.0 643.55 65.0 100.0 644.45 30.0 100.0 645.45 20.0 100.0 647.51 15.0 100.0 641.10 100.0 100.0 641.79 100.0 100.0 642.59 69.0 100.0 643.59 36.0 100.0 644.99 11.0 100.0 80%
641.9 100.0 100.0 642.9 66.7 100.0 643.8 33.3 100.0 644.8 13.5 100.0 646.8 23.3 100.0 640.7 100.0 100.0 641.4 100.0 100.0 642.3 135.0 100.0 643.1 70.0 100.0 644.9 30.0 100.0 62%
The charge compensation can be achieved by a partial occupation of Mn4 þ ions on the Cu site, according to the following equation: 3Cu2 þ -2Cu þ þ Mn4 þ
  r Mn4 þ ðVIÞ ¼ 0:53 e andr Mn4 þ ðIVÞ ¼ 0:39 e, and in agreement with the previous results obtained by Créon et al. in Ba4InMn3O11,5 [20]. 3.3. ESR
And during the cooling, a redox reaction leads to the Cu oxidation inducing setting of free electrons which reduce Mn4 þ to Mn3 þ on the Mn site: Cu þ -Cu2 þ þ 1e Mn4 þ þ 1e -Mn3 þ when Cu substitutes for Mn, the charge compensation could be in accordance with the following relationship: Cu þ þ Mn4 þ -Cu2 þ þ Mn3 þ Then, a decrease in the oxygen stoichiometry could exist due to the partial Cu substitution in manganese sublattice: Mn4 þ Cu2 þ þ VÖ. The formation of vacancies in anion sublattice can also be the result of this reaction: OOX21/2 O2 þVÖ þ2e . Nevertheless, the Cu2 þ insertion or the Mn3 þ presence cannot explain the decrease of cell volume because ionic radii of Mn3 þ Â and Cu2 þ are 0.58 Å and 0.73 Å, respectively r Mn4 þ ðVIÞ ¼ 0:53 e : It could be assumed that, at low doping level (o0.01) i.e. low amount of Mn3þ , the stacking sequence could remain the same. So, the cell volume decrease could be explained by the increase of the number of oxygen vacancies. But, when the rate of copper is higher (40.01), the amount of Mn3þ and oxygen vacancies is therefore more important. We propose that the large amount of oxygen vacancies could change the Mn4þ crystallographic environment from an octahedral to a tetrahedral configuration (Fig. 2) in agreement with ionic radii
ESR spectra of Ba4YMn3 xCuxO11.5 7 δ were measured from 300 K to 4 K. The main contribution to the ESR signal is due to Mn2 þ and Mn4 þ because of their unpaired electrons and not to the Cu cations because of the low content in each sample even at low temperature. However, Mn3 þ , which is assumed to be undetectable, has been already observed in special cases for several manganites [21,22]. In Fig. 4a, we plotted a typical thermal evolution of the ESR spectra measured for a powder sample and in Fig. 4b we plotted the thermal evolution of the D.I.N. The single line is centered at g E1.98, which is a typical value for the Mn4 þ , down to 35 K. Below this temperature, the absorption line becomes broader and slightly shifts to low fields indicating that a magnetic transition occurs towards an antiferromagnetic state (AF) [23]. The double integration normalized (D.I.N.) can be calculated from the temperature dependence of the ESR intensity. Its evolution in the paramagnetic phase (PM) is directly proportional to the static spin susceptibility. Starting from 300 K, the intensity increases continuously with the temperature decreases up to a maximum. This maximum (TN) is located at 26 K for the undoped sample and increases with Cu ratio up to 34 K for x ¼0.04 (Fig. 4b). Below this maximum, when temperature decreases, the intensity diminishes. This sharp drop is associated to the onset of the antiferromagnetic–paramagnetic transition. 3.4. Magnetic measurements Inverse magnetic susceptibilities of the samples measured in an applied magnetic field of 1T from 5 K to 300 K are shown in Fig. 5.
222
T. Barbier et al. / Journal of Solid State Chemistry 206 (2013) 217–225
existing in the material. θCW values obtained are very large, negative, and vary with x from 234 K to 170 K for x ¼0.002 and x ¼0.04, respectively. This result suggests strong AF interactions which should occur between local spins of the Mn cations [24]. This observation is consistent with the fact that the strong AF coupling between spins is mediated by the 901 Mn–O–Mn bonding, which are formed by the face sharing of MnO6 octahedra (Different bond angles obtained from the structural analysis are written in Table 1). For the undoped sample, the effective magnetic moment calculated from the Curie constant (3.87μB) is close to that expected for Mn4 þ (3.9μB) in the lonely spin approximation (S ¼3/2), consistent with the charge balance in the material, and the XPS measurements. Then, the AF spin arrangement leads to a total spin of 3/2 for each trimer. It must be noticed that the effective moment strongly decreases when x increases to reach a value close to 2.4μB for x¼ 0.04. The substitution ratio does not explain this decrease if it is assumed that magnetic moments of Mn3 þ (S ¼2) and Cu2 þ (S ¼1/2) are equal to 4.89μB and 1.73μB, respectively, and that Cu þ is diamagnetic. However, the introduction of these cations inside trimers could weaken magnetic interactions. Unlike the ESR, magnetization measurements do not allow to obtain a clear AF transition because this is not a local approach of magnetic interactions. Indeed, the Mn trimers inside the structure are separated by some YO6 octahedra connecting by corner sharing. Because Y3 þ ions are diamagnetic, the presence of such octahedra breaks the long range AF interactions resulting in the absence of a magnetic signature. More, it can be noticed that room temperature magnetization and Mn4 þ percentage versus Cu are almost superimposed (see inset of Fig. 5) confirming the Mn substitution by Cu in trimers. 3.5. Impedance spectroscopy
Fig. 4. (a) Typical thermal evolution of the ESR spectra measured for a powder sample; (b) thermal evolution of the double integration normalized of Ba4YMn3 xCuxO11.5 7 δ samples.
The complex impedance spectroscopy is a powerful tool to separate the grain (ρg, Cg), the grain boundary (ρgb, Cgb), and the interface contributions (ρint, Cint) [25,26]. Figs. 6–8 show the impedance spectroscopy for Ba4YMn3 xCuxO11.5 7 δ samples at room-temperature. In each figure, dots are experimental data and solid lines correspond to fits with the Cole–Cole model using the equivalent circuit shown in insert of Fig. 6 [27,28]. Instead of a “simple” capacity, we use this model with a CPE (constant phase
Fig. 5. Magnetic susceptibility versus temperature for undoped and doped B4YMn3O11.5 7 δ. Inset inverse susceptibility for these samples over an extended T range.
The χ 1(T) follows in the paramagnetic domain the Curie–Weiss law: 1/χ ¼(T θCW)/C, where C is the PM Curie constant, and θCW is the Curie–Weiss temperature corresponding to the intersection of the fit with the abscissa axis. The paramagnetic Curie–Weiss temperature is closely related to the type of magnetic interactions
Fig. 6. Complex impedance plots for Ba4YMn3 xCuxO11.5 7 δ samples at room temperature. The scatters are experimental data and solid lines are fits with the Cole–Cole model.
T. Barbier et al. / Journal of Solid State Chemistry 206 (2013) 217–225
223
Table 3 Values of RgCgαg and RgbCgbαgb obtained by fit for the undoped and doped compounds.
Fig. 7. Frequency dependence of permittivity of Ba4YMn3 xCuxO11.5 7 δ samples at room temperature. Solid lines correspond to fitswith the Cole–Cole model.
Compounds
ρg (kΩ)
Cg (nF)
αg
ρgb (kΩ)
Cgb (nF)
αgb
x ¼0 x ¼0.002 x ¼0.005 x ¼0.01 x ¼0.02 x ¼0.04
0.62 0.15 0.95 800.00 1605.00 1400.00
0.03 0.06 0.01 0.06 0.03 0.04
0.94 0.94 0.98 0.91 0.92 0.89
10.40 2.45 13.00 3000.00 1910.00 950.00
30.00 30.00 50.00 7.00 2.60 2.80
0.78 0.83 0.73 0.60 0.74 0.78
semiconducting grains and insulator grain boundaries strongly suggests that the IBLC model can be applied for weakly doped samples and, therefore, explains the colossal permittivity of these ceramics as shown in Fig. 7. Consequently, their permittivity at 100 kHz drops from 5000 down to the more current value of 50 depending on Cu doping. Due to its very low grain resistivity, Ba3.99(7)YMn2.99(7)Cu0.00(2)O11.49(9) shows the largest permittivity between 100 kHz and 10 MHz. This frequency range is of great interest for microelectronic devices. Note that the IBLC model does no longer give rise to colossal permittivity for samples with x equal or above 0.01 for which the grain becomes too resistive. Fig. 8 shows the frequency dependence of the global conductivity which is consistent with the ρg and ρgb values extracted from the Nyquist diagrams and the approach using the IBLC model. On the other hand, other dopings that have been investigated (Fe, Nb, Co, Ti) show degraded properties since ε′r is well below the values of undoped Ba4YMn3O11.5 7 δ and slightly doped samples. 3.6. High-resolution transmission electron microscopy and suggested link to grain conductivity
Fig. 8. Plots of conductivity versus frequency for Ba4YMn3 xCuxO11.5 7 δ samples.
element) to correctly fit experimental data. This model can be described by the following equation: εn ðωÞε1 ¼ ðεs ε1 Þ=ð1 þ ðωτÞα Þ where εn (ω) is the complex dielectric constant, εs and ε1 are the dielectric constants at “static” and “infinite” frequencies, respectively. ω is the angular frequency and τ is a time constant linked to relaxation phenomena. The exponent α, which takes a value between 0 and 1, allows the description of different spectral shapes due to different relaxation times (DRT). Note that when α ¼0, the Cole–Cole model reduces to the Debye model. Even though three semicircles in complex impedance plots are expected, related to (ρg, Cg), (ρgb, Cgb), and (ρint, Cint), only two are observed in most cases. The interface semicircle is not observed, perhaps because it is out of the measured frequency range. The observation of the interface semicircle would imply measurements well above 10 MHz, which are not available with the device used. The second semicircle, located at low Z′ values and related to the grain signature, is even not observed in two cases of x¼0.002 and 0.005. For these specimens, we have estimated the grain resistivity (ρg) value from the zero offset (see insert of Fig. 6). Note that the x ¼0.002 sample is the one which exhibits the lowest grain and grain boundary resistivity values. On the other hand, the grain capacitance (Cg) is calculated from fits of data measured at highest frequencies using the Cole–Cole model. The ρg values extracted from fits for x equal or larger than 0.01, are at least 1000 times larger than those of the others (Table 3). The presence of
How can one understand these abrupt changes of regimes, i.e. improvement of dielectric properties with respect to the undoped compound for very low doping, then considerable drop of both grain conductivity and effective permittivity for higher dopings? To answer this question, we performed electron diffraction and high resolution electron microscopy investigation on a selected area of Ba4YMn3 xCuxO11.5 7 δ samples (Fig. 9). All patterns showed that the structure is well rhombohedral with the expected R3m space group. However, along certain directions some structural disorder can be observed. The reciprocal zone axis [3 1 0] is the most relevant pattern and is shown in insert of Fig. 9a and b. These two images result from both compound families presenting different behaviours for: x o0.01 and x 40.01 as detailed previously. Fig. 9a (photographs on the top left and bottom left) corresponds to x ¼0.005 sample and the sample with a composition of x¼ 0.01 is presented in Fig. 9b (photographs on the top right and bottom right). Concerning Fig. 9a (on the bottom left), the pattern exhibits two superimposed sets of spots that reveal a twining structure in the crystal. Indeed, the two sets of diffraction spots can be separately indexed on the [3 1 0] and [ 3 1 0]T zone axis. This twining has been shown in the undoped compound Ba4YMn3O11.5 7 δ [29]. The twined structure results from a parallel planar fault as observed in the HRTEM image (Fig. 9a). This image can be seen in terms of stacking sequence of BaO3 layers. Then the defect can be described by the appearance of a block with four face-sharing octahedral instead of one block with three face sharing octahedral as described for the 12R polytype (Fig. 10). For the sample with x¼ 0.01, the formation of twinning domains is clearly illustrated in Fig. 9b (bottom and top right) by the systematic superimposition of the [3 1 0] and [ 3 1 0]T variants like in the previous example. However, along that direction some diffuse streaks appear which can be interpreted as the signature of “nano-twinning”. The corresponding image in Fig. 9b
224
T. Barbier et al. / Journal of Solid State Chemistry 206 (2013) 217–225
Fig. 9. HTREM images for x¼ 0.005 (a), and x ¼0.01 (b) along the zone-axis [3 1 0]. Inserts: images of the reciprocal spaces.
illustrates these nano-twinning by some disordered intergrowth along the c axis. The contrast can be interpreted as the succession of a variable number (n) of trimers. First we observe that these two different behaviors relate to the same compounds as those observed in changes in the volume as a function of doping level. So it could be assumed that for low levels of doping (i.e. xo 0.01), some oxygen vacancies can create charge carriers that increase the grains conductivities. This can then contribute to the IBLC effect and explain the giant permittivity observed for weakly-doped samples. However for doped samples with x40.01, the Mn3 þ rate and oxygen vacancies are higher. So this may change the stacking sequence of Ba4YMn3 xCuxO11.5 7 δ and create numerous defects observed in Fig. 9b. The stacking sequence can control the resistivity of grains and grain boundaries. For example Kuang et al. have synthesized a polytype of Ba4YMn3O11.5 7 δ: Ba4YMn3O10.7 [30] (Fig. 10). This structure can be described by the sequence (ch)3, the Y cations are located in the corner-sharing octahedral, whereas the Mn cations are located in the face-sharing octahedral to form Mn2O9 dimer. This stacking sequence leads to a resistivity of grains and grain boundaries of 20 MΩ cm (620 Ω cm for Ba4YMn3O11.5 7 δ) and 8.4 MΩ cm respectively (10.4 kΩ cm for Ba4YMn3O11.5 7 δ). As the permittivity of Ba4YMn3 xCuxO11.5 7 δ is generated by the IBLC mechanism, these stacking modifications could explain the
very differences between the permittivity of weakly-doped samples and the more strongly Cu doped samples.
4. Conclusion This work has been undertaken in order to find experimental evidence of the influence of substitution on structural, magnetic and dielectric properties. The substitution of Cu on Mn site in hexagonal perovskite Ba4YMn3O11.5 7 δ shows that the structure keeps a 12R type structure and occurs as a solid-solution phase. Moreover, structural and chemical analyses suggest that the samples have some oxygen vacancies. The Mn valence would be mixed with Mn3 þ and Mn4 þ cations when the Cu ratio increases. Partial replacement of Mn by Cu results in the existence of two behaviours of resistivity, and so of relative permittivity, because the IBLC model explains that the low resistivity of grains contributes to high permittivity. Indeed, the low-doped compounds exhibit a colossal permittivity of about 104 at 10 kHz whereas for the highly-doped samples (x4 0.04) the permittivity is dramatically influenced and falls down to about 50 at 10 kHz. To understand these different behaviours, we have carried out high-resolution transmission electron microscopy analysis. They show that for low levels of doping (i.e. xo 0.01), some oxygen
T. Barbier et al. / Journal of Solid State Chemistry 206 (2013) 217–225
225
Fig. 10. Three different poly-type of BaYMnO, on the left 12R, in the middle 6H and on the right 10H.
vacancies can create charge carriers that increase the grains conductivities (in this case, the stacking sequence remains … cchhcc…). This can then contribute to the IBLC effect and explain the giant permittivity observed for weakly-doped samples. However for doped samples with x 40.01, the Mn3 þ rate and oxygen vacancies are higher. So this may change the stacking sequence of Ba4YMn3 xCuxO11.5 7 δ and create numerous defects. So the stacking sequence can control the resistivity of grains and of grain boundaries. As the permittivity of Ba4YMn3 xCuxO11.5 7 δ is generated by the IBLC mechanism, these stacking modifications could explain the large differences between the permittivity of weakly-doped samples and that of samples with higher Cu doping.
References [1] A. Deschanvres, B. Raveau, F. Tollemer, Bull. Soc. Chim. Fr. (1967) 4077–4078. [2] A.P. Ramirez, M.A. Subramanian, M. Gardel, G. Blumberg, D. Li, T. Vogt, S.M. Shapiro, Solid State Commun. 115 (2000) 217–220. [3] M.A. Subramanian, A.W. Sleight., Solid State Sci. 4 (2002) 347–351. [4] D.C. Sinclair, T.B. Adams, F.D. Morrison, A.R. West., Appl. Phys. Lett. 80 (2002) 2153–2155. [5] F. Tsang-Tse, S. Hsu-Kai, J. Am. Ceram. Soc. 87 (2004) 2072–2079. [6] S. Krohns, J. Lu, P. Lunkenheimer, V. Brizé, C. Autret-Lambert, M. Gervais, F. Gervais, F. Bouré, F. Porcher, A. Loidl, Eur. Phys. J. B 72 (2009) 173–182. [7] V. Brizé, C. Autret-Lambert, J. Wolfman, M. Gervais, P. Simon, F. Gervais, Solid State Sci. 11 (2009) 875–880.
[8] G. Chiodelli, V. Massarotti, D. Capsoni, M. Bini, C.B. Azzoni, M.C. Mozzati, Solid State Commun. 132 (2004) 241–246. [9] F. Liang, S. Mingrong, Z. Fengang, L. Zhenya, Y. Jing.Dielectric, J. Appl. Phys. 104 (2008) 064110. [10] X. Kuang, C. Bridges, M. Allix, J.B. Claridge, H. Hughes, M.J. Rosseinsky.Internal, Chem. Mater. 18 (2006) 5130–5136. [11] A. Douy, Int. J. Inorg. Mater. 3 (2001) 699–707. [12] T. Barbier, C. Autret-Lambert, C. Honstettre, F. Gervais, M. Lethiecq, Mater. Res. Bull. 47 (2012) 4427–4432. [13] J. Rodriguez-Carvajal, Physica B 192 (1993) 55–69. [14] D.A. Shirley, Phys. Rev. B: Condens. Matter 5 (1972) 4709–4714. [15] V.M. Goldschmidt, Geochemischeverteilungsgesetze der element, VII, VIII, 1927. [16] Z. Yue, F. Zhao, Z. Gui, L. Li., J. Alloys Compd. 395 (2005) 126–131. [17] R.P. Gupta, S.K. Sen, Phys. Rev. B: Condens. Matter 10 (1974) 71–79. [18] R.P. Gupta, S.K. Sen, Phys. Rev. B: Condens. Matter 12 (1975) 15–19. [19] J. Li, M.A. Subramanian, H.D. Rosenfeld, C.Y. Jones, B.H. Toby, A.W. Sleight, Chem. Mater. 16 (2004) 5223. [20] N. Créon, C. Michel, M. Hervieu, A. Maignan, B. Raveau, Solid State Sci. 5 (2003) 243–248. [21] S.B. Oseroff, M. Torikachvili, J. Singley, S. Ali, S.W. Cheong, S. Schultz, Phys. Rev. B: Condens. Matter 53 (1996) 6521. [22] A. Shengelaya, G.M. Zhao, H. Keller, K.A. Muller, Phys. Rev. Lett. 77 (1996) 5296. [23] C. Autret, M. Gervais, N. Raimboux, P. Simon, Phys. Rev. Lett. 6 (2004) 815–824. [24] J.J. Adkin, M.A. Hayward, Chem. Mater. 19 (2007) 755–762. [25] X. Hao, G. Xiangfeng, Y. Rong, X. Zhaoxian, J. Alloys Compd. 482 (2009) 14–17. [26] D.C. Sinclair, A.R. West, J. Appl. Phys. 66 (1989) 3850–3856. [27] K.S. Cole, R.H. Cole, J. Chem. Phys. 9 (1941) 341–351. [28] K.S. Cole, R.H. Cole, J. Chem. Phys. 10 (1941) 98–105. [29] T. Barbier, C. Autret-Lambert, F. Gervais, M. Lethiecq to be published. [30] X. Kuang, H. Zhu, M. Allix, C.A. Bridges, M.J. Rosseinsky, Y. Li, J. Mater. Chem. 22 (2012) 8103–8109.
