Size reduction-induced properties modifications of antiferromagnetic dielectric nanocrystalline Ba 2 NiMO 6 (M 5 W, Te) double perovskites

The main objective of this work was to synthesize phase pure double perovskites Ba 2 NiTeO 6 (BNTO) and Ba 2 NiWO 6 (BNWO) in nanocrystalline form and to reveal the impact of nanocrystallinity on their magnetic and dielectric properties. The studied double perovskites were synthesized in nanocrystalline form by employing a citrate sol-gel route. A detailed investigation of their structure and properties using X-ray powder diffraction, scanning electron microscopy, Raman spectroscopy technique, energy-dispersive X-ray spectroscopy, SQUID magnetometry and electrical measurements is carefully described. Rietveld reﬁnement of X-ray powder diffraction patterns revealed phase purity of both compounds: BNTO is trigonal ( R -3 m ) while BNWO is cubic ( Fm -3 m ). Raman spectroscopy studies reveal optical phonons that correspond to vibrations of Te 6 þ /W 6 þ O 6 octahedra, while scanning electron microscopy images show irregular plate-like nanocrystals. Magnetic property measurements speak in favor of antiferromagnetic order but, in both compounds, size reduction affected their properties. BNTO has Ne´el temperature (T N ) of 10.3 K which is higher than previously reported for its bulk form. Magnetic ground state of BNWO can be explained as canted antiferromagnetism with T N ¼ 48.2 K. Room temperature measurements of dielectric constants at various frequencies suggest that these materials are high- j dielectrics with low dielectric loss. The Nyquist plot reveals depressed a semicircle arc typical for non-Debye type of relaxation phenomena for BNWO ceramic, whereas for BNTO ceramic an almost straight line of Z 00 versus Z’ has been observed, indicating its high insulating behavior. To conclude, size-dependent properties of studied double perovskites are discussed, introducing a possibility for implementation in electronic devices. at 0.1 The observed reduction of dielectric constants along with the frequency is associated with the hopping mechanism. Impedance spectroscopy results in the form of Nyquist plot revealed depressed semicircle arc suggested the non-Debye type of relaxation phenomena for BNWO ceramic, whereas for BNTO ceramic nearly straight line of Z 00 versus Z ’ has been observed indicating its high-insulating behavior.


INTRODUCTION
Applied nanomaterials have become the focus of research due to the plethora of good properties yielding to their potential application. So far, metal and semiconducting nanoparticles have shown size-and shape-dependent electronic, magnetic and optical properties [1][2][3]. This behavior has marked nanomaterials to be superior than their bulk forms in terms of producing smaller devices with enhanced functionality. So, instead of altering chemical composition, it has become more important to produce nanomaterials with different architectures [3]. Therefore, many efforts have been put into development of rational size-and shape-controlling synthesis procedures, namely solution chemistry routes [4].
Green chemistry approaches have limited the usage of chemicals to be nontoxic and environmentally friendly. Largescale industrial production demands for them also to be inexpensive, hence in choosing the right synthesis route, one should take into account all these things if possible. Simplicity of the synthesis route is also important. The citrate method was developed by Turkevich et al. [5], and nowadays, citric acid has been used with or without different additives in sol-gel synthesis of perovskite nanomaterials [6][7][8][9][10][11][12][13]. The greatest advantages of aqueous citrate sol-gel route are nontoxicity, ecofriendliness, low cost and simplicity.
Double perovskites have been widely investigated due to the possibility of magnetoelectric coupling within the singlephased material which is suitable for implementation in electronic devices and large variety of other applications, such as optoelectronics and photocatalysis [6,[14][15][16]. Ba 2 NiWO 6 (BNWO) was first prepared by Fresia et al. [17]. Its crystal structure has later been studied by Brixner [18], Nomura [19], Cox [20], Filip'ev [21] and more recently by Sahnoun [22] and Alsabah [23,24] along with its magnetic properties. Cox et al. [20] reported it to be antiferromagnetic with no further details, while Alsbah [24] reported it to be a semiconductor with indirect optical band gap value of 3.32 eV. Ba 2 NiTeO 6 (BNTO) was first prepared in 1972 by Kö hl [25] where its crystal structure was studied and later its magnetic properties were studied by Asai [26,27] and Djerdj et al. [8]. Both of them reported it to have antiferromagnetic ground state with Né el temperature, T N ¼ 8.6 K. Djerdj et al. [8] were the first group to synthesize this material in nanoscale and also to report dielectric constant value at room temperature for this compound, which was 15 when frequency of 1 MHz (1000 kHz) was applied. Dielectric loss was not reported, although it is always important to produce materials that possess high value of dielectric constant, but low dielectric loss (tan d) [6].
Although these materials are already known in their bulk forms, they have not been thoroughly investigated so far. Since these double perovskites both possess Ni 2þ as magnetically active cation on the octahedral site, one could expect similar magnetic properties in both compounds. The other octahedral site is occupied by magnetically inactive cations, either d 0 W 6þ or d 10 Te 6þ cation, which could be responsible for inducing high dielectric constant within a material to produce high-j dielectric [8,10].
On decreasing the crystallite size, surface becomes more important than the bulk and surface defects start to have an impact on already-known properties [6]. In the case of antiferromagnetic materials, effects such as spin canting might take place [28,29]. Various authors have already reported that the decrease in crystallite sizes follows the increase in T N [30][31][32][33][34][35][36]. According to Parveen et al. [37], the decrease in crystallite size results in the increase of dielectric constant at isothermal conditions. These facts speak in favor of great advantages of nanocrystalline materials over their bulk counterparts.
Herein, we present a simple, effective, low-cost synthesis method for the fabrication of nanocrystalline phase pure BNTO and BNWO double perovskites with detailed structural characterization and the study of size reduction effect on their magnetic and dielectric properties. These materials could have a promising application as antiferromagnetic high-j dielectrics in electronic devices.

Materials
All chemicals were commercially available and used as purchased. Citric acid monohydrate pro analysis (T.T.T., Croatia) was used to prepare starting reaction solution. Precursor solution of metal cations was prepared by dissolving ammonium tungsten oxide hydrate or ammonium tellurate 99.5% (Alfa Aesar, Germany), Barium(II) nitrate ! 99% and Nickel(II) nitrate hexahydrate ! 98.5% (Sigma Aldrich, Germany). Concentrated ammonia solution pro analysis (Gram-Mol, Croatia) was used for pH value adjustments.

Synthesis
Synthesis procedure used in this research was previously reported by our group [9,10]. Stoichiometric amounts of Ba 2þ , Ni 2þ and Te 6þ /W 6þ salts were dissolved in a 10% solution of citric acid (10 g of citric acid in 100 ml of MiliQ water) and pH value was adjusted to 5 in order to deprotonate citric acid. Asprepared solution was evaporated and constantly stirred at 95 C on a magnetic hotplate stirrer until black mixture was formed which was later dried for 24 h at 120 C in a drying oven. Grinded mixture was further calcined in two steps: 8 h at 600 C and 12 h at 950 C for BNTO and 1000 C for BNWO. The heating rate was kept at 2 C/min for both calcination steps.
All characterization techniques including measurements of magnetic properties were conducted on as-prepared materials in powder form. However, for measurements of dielectric properties, it was necessary to convert prepared powders into pellets because powder could not be placed between electrodes. The sintering of the green pellets of dimension (thickness: 1.5 mm and diameter: 10 mm) for dielectric measurements was conducted at 1250 C using tube furnace. The heating rate was 5 C/ min and soaking time 4 h. Pellets were then rubbed with sandpaper to obtain a smooth parallel side and later a silver paste was painted on both the sides which act as an electrode layer during the electrical measurements. Archimedes method was used to measure the density of prepared pellets. It was confirmed that densities are above 91% theoretical density if sintering temperatures are !1250 C.

Characterization techniques
Crystal structure, microstructure, chemical composition and morphology determination. X-ray powder diffraction (XRPD) analysis was performed on a Panalytical X'Pert PRO diffractometer (h-2h geometry) with monochromatized CuK a radiation source (40 kV, 40 mA) at 292 (2) K. Data were collected with the step size of 0.02 in a 2h range of 10-90 . Computer software FULLPROF [38] was used for the Rietveld refinement of collected XRPD data. In order to perform size-microstrain analysis, the XRD profile has been modeled by employing modified Thompson-Cox-Hastings pseudo-Voigt function. The main premise was that the line broadening of the deconvoluted profile originated from the individual contributions of the line broadening caused by lattice microstrain and small crystallite sizes. Hence, half-width parameter X (V ¼ W ¼ 0) was fixed at the values determined by using LaB 6 as a crystalline standard. The quality of the Rietveld refinement was conveyed with discrepancy factor (R wp ) and the goodnessof-fit (v 2 ). Computer software VESTA 3 [39] was used to visualize the crystal structure.
Unpolarized Raman spectroscopy was used to complement the powder X-ray diffraction results. It was recorded on powders placed on microscope slides at room temperature on Renishaw inVia Raman microscope system with a backscattering geometry, HeNe laser (633 nm, 2 mW) in the range of 50-1100 cm À1 .
Scanning electron microscope (M/S FEI Nova NanoSEM 450) was used for examination of morphology of powdered samples. Energy-dispersive X-ray spectroscopy detector coupled with scanning electron microscope was used for calculation of chemical composition.

Magnetic properties measurements.
Temperature-dependent susceptibility and isothermal magnetization have been investigated with a Quantum Design MPMS-XL-5 magnetometer on a powder sample. All presented data have been corrected for a temperature-independent sample holder contribution and a diamagnetic contribution estimated from Pascal's constants [40].

Dielectric measurements.
The electrical properties like dielectric and impedance were done using a computer-operated LCR meter (Hikoi IM3570) in the frequency sweep of 0.1 kHz to 1 MHz and at room temperature.

XRD analysis
Room temperature powder X-ray diffraction patterns confirm a perovskite phase formation for all synthesized compounds ( Fig. 1a and b), and their phase purity since all observed Bragg reflections were matched with the calculated reflections based on the assumed structural models. Rietveld refinement analysis reveals that the BNTO crystallizes in centrosymmetric rhombo-hedra1 space group R-3m with a ¼ 5.7976(2), c ¼ 28.604(1) Å (c ¼ 120 ) for the unit cell in the trigonal setting, while BNWO crystallizes in cubic centrosymmetric space group Fm-3m with lattice parameter a ¼ 8.0679(2) Å . The outcome of the Rietveld refinement is displayed in Fig. 1 whilst the results of the refinement are summarized in Table 1 and Supplementary Tables S1  and S2 in Supplementary data. Selected interatomic distances for both compounds are summarized in Table 2. Trigonal BNTO consists of face-sharing trimers, Ni 2 TeO 12 , which are alternatively interconnected via TeO 6 octahedra forming a typical 12layer perovskite motif as displayed in Fig. 2 (left). Cubic BNWO is rock-salt ordered consisting of regular BaO 12 cuboctahedra and corner-sharing regular NiO 6 and WO 6 octahedra. Due to the similar size, ions located on the B sites of double perovskite structure were allowed to occupy the antisites each other [41]. For BNTO compound, occupancies of Te and Ni were refined by imposing linear constraints of the site occupancies: refers to the site occupancy of B ion at the corresponding site while Te Ni and Ni Te denote antisite ions at the 6c and 3a/3b, respectively. The outcome of such constraint refinement revealed g(Te Ni ) ¼ g(Ni Te )¼ 0.1191% or 11.92(5)% Ni/Te ions occupy antisites each other. For BNWO compound there is only one site for W ions 4a and Ni ions 4b, and the constraint refinements yielded 5.4(2)% W Ni (Ni W ) antisite disorder. The refinement of oxygen site occupancy factors did not reveal any deviation from full occupancy within the standard deviation.
The line-broadening analysis performed within the Rietveld refinement reveals that BNWO shows higher crystallinity (average crystallite size equals to 88 nm) compared to BNTO (58 nm). In both cases, the Rietveld refinement has shown that the level of microstrain is negligible, i.e. parameter U was fixed to its instrumental value. The average crystallite size for BNTO is a bit larger compared to the value reported by Djerdj et al. [8], where a similar synthesis approach was used. Rietveld refinement yields satisfactory values of fitting quality parameter, R wp implying the right phase was chosen with appropriate lattice parameter values. Goodness of fitting (GoF) parameters are low (1.4 and 1.5) indicating a good refinement.

Raman spectroscopy
Raman spectroscopy has been carried out in order to investigate possible occurrences in disordered and nanocrystalline materials, such as phonon confinement effects. Also, bond length  Supplementary  Table S3. According to Kroumova et al. [42], cubic BNWO has A 1g þ E g þ 2T 2g , 4 first-order Raman active modes and trigonal BNTO has 7 A 1g þ 9 E g which is 16 first-order Raman active modes. Figure 3a shows Raman spectrum of cubic BNWO where lattice translational mode is observed at 134 cm À1 , while oxygen bending mode can be seen at 432 cm À1 . Although according to Ayala et al. [43], m 2 octahedral stretching mode is absent in most cubic perovskites, it is observed here for BNWO at 552 cm À1 . The other octahedral stretching mode, m 1 is assigned to the highest wavenumber appearing in Raman spectrum, 807 cm À1 . Asymmetric broadening of Raman shifts in both compounds occur due to phonon confinement effect in nanocrystalline compounds [44][45][46]. Raman spectrum depicted in Fig. 3b shows six modes in total for trigonal BNTO. Lattice translational (T) and librational (L) modes [47] are depicted by two peaks: at 108 and 120 cm À1 and at 220 and 250 cm À1 , respectively. Oxygen bending in TeO 6 octahedra (m 5 mode) can be seen in the range of 350-450 cm À1 . Modes m 1 (>700 cm À1 ), m 2 (470-650 cm À1 ) and m 3 (580-715 cm À1 ) are assigned to octahedral BO 6 stretching according to Silva et al. [47]. For the case of studied BNTO compound, m 1 and m 3 vibrations are detected at 745 and 683 cm À1 , respectively. The differences in the line broadening of the Raman bands are similar to Sr 3 Co 2 WO 9 [10] and Sr 3 Fe 2 WO 9 [9], where it was explained by the structural disorder which is also present in this case in the form of antisite disorder (Ni/Te and Ni/W). A similar Raman spectrum of BNTO has already been reported by Djerdj et al. [8]. A larger difference in m 1 mode position is perceived between Te 6þ -and W 6þ -based compound which amounts to $62 cm À1 . This is comparable to the previously reported observations for similar compounds [47]. Silva et al. [47] gave an explanation that the rise in wavenumber value in W-containing double perovskites compared to Te-containing correlatives is a result of the increase of the bonding energy of W-O bond in WO 6 octahedron. The reason behind this is in d-orbital occupation of Te 6þ (d 10 ) and W 6þ (d 0 ). Fully occupied d orbital avoids the formation of ptype Te-O bondings while empty d orbital allows the overlap of the t 2g orbitals [48]. Overlapping of t 2g orbitals is responsible for the increase of bonding energy of W-O bond in WO 6 octahedron [47]. Even

Morphology and chemical composition
To investigate morphology of synthesized compounds and to confirm targeted chemical composition, scanning electron microscopy (SEM) coupled with energy-dispersive X-ray spectroscopy (EDS) was performed on powder samples. EDS spectra of both BNTO and BNWO are shown in Supplementary Figs S1 and S2 and quantitative composition is given in Supplementary    Table S4. From the obtained data, empirical formulas that correspond to targeted compounds are calculated. SEM images of both compounds are shown in Fig. 4. Figure 4 shows that both compounds are indeed nanocrystalline. The shape of crystal grains for BNTO is irregular hexagonal-like, while for BNWO is rather irregular plate-like and heterogeneous in size, similar as it was reported earlier for Ba 3 Fe 2 WO 9 [9] synthesized using the same synthesis procedure. The average grain size was calculated for the powder samples by the line intercept method. The obtained values are 73 and 101 nm for BNTO and BNWO, respectively. These values are larger than the average crystallite size values determined from the line broadening of powder XRD patterns. This is, however, expected because grains imaged by SEM may consist of several crystallites. Also, the line intercept method is not fully precise. The average grain sizes were also determined by the line intercept method for pelletized samples and they amount to 125 and 142 nm for BNTO and BNWO, respectively. These values are larger in comparison to the corresponding values obtained from the materials in powder form, which is the result of the sintering process.

Magnetic properties
As shown in Fig. 5, temperature-dependent susceptibility measurements were carried out while a magnetic field of 1000 Oe was applied. A maximum of susceptibility at T N ¼ 10.3 K is a clear evidence of a phase transition from paramagnetic to antiferromagnetic state. The measured T N is slightly higher than the reported 8.6 K [8,26]. The difference can be attributed to the nanosized material with the average crystallite size of 58 nm used in our research. It has been shown that the increase in T N value is caused by the decrease of the crystallite size [30][31][32][33][34][35][36].
A zero-field-cooled (ZFC) and field-cooled (FC) susceptibility was measured in BNTO with no observed differences between the two cooling regimes. In a high-temperature region (T > 200 K) a Curie-Weiss law v ¼ C/(T-#) was used in order to describe the measured susceptibility. A full line in v À1 versus T plot (inset in Fig. 5) shows the best correspondence between the measured data. We obtained the Curie constant C ¼ 1.44 emu K/mol and Curie-Weiss temperature # ¼ À155 K. The effective magnetic moment l ¼ ffiffiffiffiffiffiffiffi 8 C p ¼ 3:4 l B is in a good agreement with the theoretical value for Ni(II) ions with a total electronic spin S ¼ 1 and nonzero orbital contribution L [49], while the negative Curie-Weiss temperature confirms the presence of antiferromagnetic interaction between the magnetic ions.
The high-temperature susceptibility (above $150 K, Fig. 5) and inverse susceptibility of BNWO are similar to those of BNTO. Using data for T > 200 K in a Curie-Weiss fit, we obtained the Curie constant C ¼ 1.36 emu K/mol (effective magnetic moment 3:3 l B ) and Curie-Weiss temperature # ¼ À110 K for BNWO. These conclusions, obtained from the high-temperature susceptibility data, coincide with the previously published  research [20,50]. They reported type II antiferromagnetism in BNWO with a T N of 48 K using a neutron diffraction study.
Below 50 K we have found a large splitting between the ZFC and FC susceptibility (Fig. 5). However, the FC susceptibility sharply increases below the T c ¼ 45.5 K, which is an indication of ferromagnetic behavior below this critical temperature rather than antiferromagnetic as previously reported.
In order to obtain a deeper insight into the magnetic ground state of BNWO, additional temperature-dependent susceptibility measurements in different magnetic fields and isothermal magnetizations at several temperatures below and above the T c were carried out. The outcomes are displayed in Fig. 6.
The splitting between ZFC and FC susceptibility measured in 100 Oe is even larger than the one measured in 1000 Oe. From the data shown in Fig. 6a, we can conclude that with increasing the magnetic field in which the susceptibility was measured, the splitting between ZFC and FC susceptibility decreases. The splitting almost vanishes in magnetic field of 30 kOe where ZFC and FC curves practically coincide. The inset in Fig. 6a shows susceptibility around the transition temperature measured in 100 Oe (the smallest used magnetic field) and 50 kOe (the largest used magnetic field). While the abrupt increase of the susceptibility at T c ¼ 45.5 K when measured in 100 Oe clearly suggests the paramagnetic to ferromagnetic transition, the measured maximum of the susceptibility in 50 kOe at T N ¼ 48.2 K and no abrupt increase of the susceptibility immediately below this temperature confirm the previously reported antiferromagnetic ground state.
The magnetization curves shown in Fig. 6b are in agreement with already-described dual nature of the system. In a full scale, between -50 kOe and 50 kOe, the magnetization curves at all investigated temperatures between 50 K and 2 K are linear up to the maximal magnetic field. The magnetization of $0.07 m B per Ni 2þ ion measured in a maximal magnetic field of 50 kOe ion is almost constant below 50 K. This value is considerably smaller than the theoretical saturation value for Ni(II) ions (J % S ¼ 1) of gÁJÁm B ¼ 2 m B . So this 'full range' picture suggests the antiferromagnetic ground state.
However, when the magnetization is observed in a small enough magnetic field (inset in Fig. 6b), clear hysteresis loops are seen for the measurements at 40 K and below. The remanent magnetization at 2 K is 2.4Á10 À3 m B /Ni and almost linearly decreases down to zero at 45 K. We tentatively attribute this weak ferromagnetic signal to a non-completely colinear antiferromagnetic arrangement of nickel magnetic moments. Thus the magnetic ground state of BNWO nanocrystalline system may be classified as a canted antiferromagnetism.

Dielectric and impedance properties
Figures 7a and b show the dielectric constant and loss factor versus frequency plot, respectively, for BNTO and Figs 7c and d for BNWO ceramics at room temperature. For both BNTO and BNWO ceramics, the dielectric constant value is higher in the low-frequency area which is considered to be <10 kHz. The high-frequency area is >10 kHz. The dielectric constant value in the low-frequency region has several polarization contributions whereas the effect of polarization fades at a higher frequency region, while in the high-frequency region its value becomes almost constant [51]. As the frequency increases, the switching polarity period is reduced as the electric field keeps on deviating very quickly and the space charge polarization also fades in this region. Therefore, the charge carriers fail to act in response to the fast-varying electric field. The reduction of the dielectric constant along with the frequency is associated with the hopping mechanism. The W 6þ has an effective positive charge in BNWO which leads to conduction of electrons leading to higher dielectric constant of 290 than BNTO which is 96 at 0.1 kHz. The frequency dependence of the loss factor follows the dielectric constant behavior for both the BNTO and BNWO samples. It is  noteworthy that loss factor for BNTO compound is significantly lower compared to BNWO: 0.21 for BNTO compared to 1.6 for BNWO at 0.1 kHz. The loss factor increases in the low-frequency region whereas in the high-frequency region it reduces. The reason behind this can be the rise in leakage current from various ionic charge balance or dislocation [52]. Also, average crystallite size and average grain size are larger in BNWO than in BNTO. This might cause the difference in values of dielectric constant aside from chemical composition. Various authors have shown that dielectric constant increases with the increase of grain size [53][54][55].
To further understand the electrical properties of synthesized compounds, the complex impedance (Z*) has been studied in order to disclose the resistive or/and capacitive contribution to conductivity of the material after applying AC electric field [56]. The impedance spectroscopy is convenient to investigate the conduction mechanism and electrical properties. Cole-Cole plots (Nyquist plots) are represented by the variation of imaginary part (Z 00 ) against the real part (Z 0 ) of the complex impedance (Z*), as shown in Fig. 8. Nyquist plot (Z' vs Z'') displayed in Fig. 8 sheds light upon the morphology, for instance, bulk effect and extrinsic effects like grain boundaries formed after the sintering, specimen, electrode interfaces, defects, etc. The relaxation process is resembled by means of a semicircle in the Nyquist plot. If Debye-type conduction occurs then the semicircle center lies on the real Z' axis while for the non-Debye-type process, the center will lie down under the real Z' axis making a depression angle. A perfect semicircular arc with a single relaxation mechanism is depicted in homogeneous materials. As a consequence, arcs can be described in relation to the parallel RC circuit which is known as Debye-type behavior. For BNTO compound depicted in Fig. 8b, one can observe nearly straight line of Z 00 versus Z'. This indicates a high insulating behavior in BNTO double perovskite. BNWO shows depressed semicircle arc (Fig. 8a) representing the non-Debye type of relaxation mechanism most probably due to inhomogeneity in grain sizes of ceramic samples. This is confirmed by SEM images of pelletized samples in insets of Fig. 4. The fitting of the experimental impedance data with the equivalent electrical circuit was performed by using demo version of ZSWIMP WIN software which is limited to only fitting features but restricts to achieve the values of the resistances and capacitances. For both materials BNTO and BNWO, the Nyquist plot shows that there is a contribution of grain and grain boundary in the materials at room temperature. The experimental impedance data are fitted with the RQC-RC equivalent circuit model depicting the contribution of grain and grain boundary.

CONCLUSIONS
In this study two nanocrystalline double perovskites, BNTO and BNWO were successfully synthesized by employing simple, low-cost and nontoxic aqueous citrate sol-gel method. According to the powder XRD measurements and Raman spectroscopy, they are single-phased. Synthesized double perovskites crystallize in centrosymmetric cubic space group Fm-3m (BNWO) and centrosymmetric rhombohedral space group R-3m (BNTO). Both materials are nanocrystalline having an average crystallite size of 58 nm for BNTO and 88 nm for BNWO with negligible microstrain. Partial disorder has been observed for both compounds in terms of Te/W Ni antisite disorder which is a bit more pronounced for BNTO [11.92(5) %]. Both compounds are antiferromagnetic with pronounced size-effect modifications. BNTO has a larger value of T N ¼ 10.3 K than previously reported in literature due to crystallite size reduction to 58 nm. For BNWO, Né el temperature has been measured in the field of 50 kOe and it amounts to 48.2 K. However, in the field of 100 Oe there is a clear ferromagnetic transition with T c ¼ 45.5 K. This behavior speaks in favor of canted antiferromagnetism. Measurements of room temperature dielectric constant versus frequency reveal that both double perovskites possess relatively large dielectric constants at low frequencies (BNWO 290, BNTO 96 at 0.1 kHz). The observed reduction of dielectric constants along with the frequency is associated with the hopping mechanism. Impedance spectroscopy results in the form of Nyquist plot revealed depressed semicircle arc suggested the non-Debye type of relaxation phenomena for BNWO ceramic, whereas for BNTO ceramic nearly straight line of Z 00 versus Z' has been observed indicating its high-insulating behavior.

SUPPLEMENTARY DATA
Supplementary data is available at Oxford Open Materials Science online.

ACKNOWLEDGMENTS
The authors are thankful to Dr Pascal Cop and Sebastian Werner from the AG Smarsly, Institute for Physical Chemistry, Justus Liebig University of Giessen for all discussions, useful advices, motivation and efforts to conduct powder X-ray measurements. J.B. and I.D. are also thankful to their master student Marina Sekuli c for assisting with materials synthesis and various measurements and last but not least to Sugato Hajra for measurement of dielectric properties.