Sensing properties of barium titanate nanoceramics tailored by doping and microstructure control

Abstract

BaTiO3 nanopowders doped with La and co-doped with La/Mn were prepared by auto-combustion and Pechini methods, respectively. The influence of the synthesis methods, dopants and sintering temperature on the BaTiO3 structure and its potential to be used as humidity and/or H2 gas sensor were studied. The optimization of all process parameters was performed to obtain adequate microstructure for the development of good sensor properties. The difference in the grain size between the La-doped and La/Mn co-doped samples and the formation of different types of defect structures in these ceramics were found to be significant for the desired electrical and ferroelectric properties. The La-doped ceramics with a pseudo-cubic structure showed the highest potential for gas sensors. The materials obtained by the Pechini method had a tetragonal structure and showed the best response, i.e., the change in electrical resistivity by four orders of magnitude in the humid atmosphere.

Introduction

Barium titanate (BT) is one of the most investigated ferroelectric materials, but the interest for its properties and application is still increasing. Due to a high flexibility of the perovskite lattice, different types of dopants can be incorporated in the BT lattice. Even very low concentration of ions of different sizes can change barium titanate crystal structure and microstructure significantly. Therefore, doping can have an effect on modification of electrical properties of the material and thus provide a great variety of potential applications. During the years, barium titanate properties have been tailored by doping. Depending on the properties that one wants to achieve, the most important is appropriate selection of the dopant type and amounts of dopants. It was reported that some of dopants were able to change the symmetry of the BT lattice from tetragonal to pseudo-cubic [1, 2], while some of them inhibited grain growth [3, 4] or induced exaggerated [5, 6] grain growth. Based on previous research, it was noticed that the dielectric properties of BT can be significantly modified by doping with La3+, Sb3+, Nb5+, Hf4+, Zr3+, Mn3+ [7,8,9,10]. The characteristic phase transitions can be shifted to lower temperatures; broadening of the dielectric permittivity peaks and certain degree of relaxation can appear. Also, the doping can lead to the appearance of semiconductivity in the material, while reaching a critical concentration of added dopant reverts the material into insulator again. A critical concentration is when A or B site ions cannot be further substituted by dopant ions and dopants are being located in the interstitial positions, at the grain boundaries, etc., inducing the changes in the electrical properties of such materials [11]. Thus, the solubility of dopants in the BT lattice was found to be a very important factor.

Porous ceramics sensors have been extensively used in the industry and scientific research. Nowadays, a large variety of applications of porous materials in devices like filters, acoustic absorbers, catalytic components, selective membranes, gas and humidity sensors can be found [12]. The basic principle of humidity sensing by ceramic sensors is change in electrical conduction or capacitance due to water chemisorptions and/or capillary conduction in the pores. It is suggested that once the H2O molecules are adsorbed on the surface, they can release the electrons to the conduction band and consequently increase the electronic conductivity. The materials used as humidity sensors usually have a quite fast response time. Ceramic materials used as humidity sensors are Al2O3, TiO2, SnO2, ZnO, In2O3, Mg2Al2O4, ZnC2O4, ZnC2O4, perovskites (BaMO3, where M = Ti, Zr, Hf, Sn), etc. [12,13,14,15]. Humidity sensors have been so far used in many areas such as in industries for sophisticated integrated circuits, in automobile industry at motor assembly lines or as defoggers, in the pharmaceutical and medical field, in agriculture for control of air conditioning in green house, soil moisture, cereal storage and in domestic application such as intelligent control of laundry, etc. [13].

On the other hand, the gas sensors are mostly n-type semiconductors with operating temperature from 200 to 600 °C. The grains of the sintered ceramics are usually covered by adsorbed oxygen. It withdraws electrons from the bulk, forming O2− ions at the surface. The electrical conductivity decreases by the oxygen adsorption. The molecules of a reducing gas interact with the adsorbed oxygen by lowering potential barrier and increasing the conductivity of the sensor. These sensors can detect reducing gases in air, e.g., leak detection in gas pipelines, indication of petrol vapor in filling stations, etc. [12, 13].

Humidity and gas sensors are usually used as porous sintered ceramics. Therefore, it is very important to control the porosity and surface activity in order to obtain the material with good gas/humidity sensing properties.

Having this in mind and based on the previous research [3, 10, 16], La incorporated in the BT lattice, which can influence the formation of n-type semiconductor and form very porous microstructure, was investigated as a potential candidate for gas and vapor sensor. In this study, two sets of samples were prepared, barium titanate doped with lanthanum and co-doped with lanthanum and manganese. The ceramics were prepared using two kinds of soft chemical synthesis methods (sol–gel Pechini and auto-combustion). The optimization of processing parameters was conducted in order to obtain porous ceramics suitable for application as humidity and hydrogen sensors.

Materials and methods

Barium titanate nanopowders were prepared by two soft chemical methods. A modified Pechini process was used to prepare nanopowders of barium titanate doped with 0.3 mol% of lanthanum and 0.05 mol% of manganese (BTLM). Barium acetate (Ba(CH3COO)2, Alfa Aesar, 99.0–102.0%), titanium isopropoxide (Ti[OCH(CH3)2]4, Alfa Aesar, 99.995%), lanthanum nitrate (La(NO3)3·6H2O, Alfa Aesar, 99.99%) and manganese oxide (MnO2, Merck, 99.8%) were used as starting materials. Solutions of titanium citrate and barium citrate were prepared using ethylene glycol (EG) and citric acid (CA) as solvents (M (metal ion):CA:EG = 1:4:16). After mixing of the obtained citrate solutions, lanthanum nitrate salt and manganese oxide were added. The transparent yellow solution was heated up to 140 °C until it changed to a dark-brown glassy resin. Decomposition of organic resin was performed at 250 °C for 1 h and 300 °C for 4 h, when a black solid mass was formed. Thermal treatment of the formed precursor was performed at 500 °C/4 h, 700 °C/4 h, 800 °C/2 h and 850 °C/2 h [3].

The second chemical method used for the preparation of barium titanate doped with 0.3 mol% of lanthanum was so-called auto-combustion method. Titanium isopropoxide, citric acid, barium nitrate and ammonia were used as the starting materials. Firstly, the solution of titanium ortho-titanate was obtained by the addition of titanium isopropoxide into the solution of the citric acid and heated at 60 °C, while further reaction with barium nitrate enabled the formation of titanium orthonitrate. The citric acid was used as a fuel agent and for converting TTIP into titanium orthonitrate. The pH solution was adjusted between 6 and 7 using an ammonia solution. The solutions were heated at 90 °C until the gel was formed and self-propagation reaction was induced by heating at 120 °C. The precursor powder was calcinated at 900 °C/2 h, with a heating rate of 5 °C/min, and the barium titanate nanopowder was obtained [17].

The nanopowders were uniaxially pressed into disks of 12 mm in diameter at a pressure of 196 MPa. The sintering was performed in air at 1300 and 1350 °C for 4 h with a heating rate of 10 °C/min. The samples prepared by Pechini method were denoted as BTLM1300 and BTLM1350, while the ceramics obtained by the auto-combustion method as BTL1300 and BTL1350 (numerals referring to the sintering temperatures).

X-ray diffraction measurements were taken in order to determine the formed crystal structure (Philips PW1710 diffractometer). Scanning electron microscopy (Tescan VEGA TS 5130MM) was used to analyze the microstructure of obtained ceramics. The average grain size was determined from SEM micrographs using linear intercept technique. The density of barium titanate ceramics was calculated geometrically. The samples were prepared for the electrical measurements by polishing and applying gold electrodes on both sides of the samples. Ferroelectric and leakage current measurements were taken in the electric field range of 1–60 kV/cm on a Precision Multiferroic Test System with High Voltage Interface (Radiant Technologies, Inc.). Impedance measurements of BTLM ceramics were taken in 350–475 °C, with a step of 25 °C and in the frequency range of 42 Hz–1 MHz using a HIOKI 3532-50 LCR HiTester. Collected data were analyzed using the commercial software package Z view. Hydrogen sensitivity of the obtained ceramics was analyzed using a custom-built setup, schematically presented in Fig. 7, and the dc resistivity was measured at 450 °C using a multimeter. Humidity sensing was analyzed in a climate chamber where the relative humidity (RH%) was changed from 40 to 90% and the temperature at which the measurements were taken was 40 °C. The dc resistivity was measured using a megaohmmeter. Sensing properties were investigated by two-point resistivity measurements due to high resistivity of samples.

Results and discussion

Structural analysis

The XRD results of both BTLM- and BTL-sintered ceramics are presented in Fig. 1. The formation of the barium titanate tetragonal crystal structure was identified by the appearance of its characteristic diffraction peaks (according to JCPDS files no. 05-0626). La3+ substituted Ba2+ in the barium titanate lattice, and doping commonly induced formation of defects such as vacancies VTi and VO as well as VBa at lower concentrations [10]. The doping with manganese was rather complex due to tendency of Mn to change its valence. In the sintered samples, Mn ions exist mostly in + 3 state, but small quantities of Mn2+-2V O and Mn2+-2V ′′O defect complexes are also possible [18]. However, the existence of manganese in such a small quantity was not possible to detect using standard XRD equipment. It can be seen from Fig. 1 that characteristic splitting of diffraction peaks is more evident for the BTLM ceramics, indicating stronger tetragonality of the structure.

Figure 1
figure1

X-ray diffractograms of BTL1300, BTL1350, BTLM1300 and BTLM1350 ceramics

The microstructures of BTLM1300, BTLM1350, BTL1300 and BTL1350 ceramics are presented in Fig. 2. According to the previous research, lanthanum as a donor dopant was found to be suitable for the preparation of porous ceramics [10]. Not only does it inhibit the grain growth but it also makes barium titanate structure less dense enabling its use for sensing devices. As it was expected, the SEM micrographs showed the formation of a highly porous microstructure by both chemical methods used for the preparation of the powders. In the case of the Pechini method, which was used for the preparation of the BTLM powders, polygonal grains were formed in co-doped ceramics and the grain size varied depending on the sintering temperature. In the BTLM1300 ceramics, the grain size was 0.4–0.8 μm, and in the BTLM1350, it was 0.6–1 μm, indicating the influence of the higher sintering temperature on the grain growth. On the other hand, the grain size of the ceramics obtained by the auto-combustion method and doped only with lanthanum was much smaller, i.e., around 200–300 nm in both ceramics. A very slight difference in grain size was observed for both BTL1300 and BTL1350 ceramics. The density of the doped ceramics was around 81–85% for the BTL ceramics and 84–88% for BTLM ceramics, depending on the sintering temperature. Apart from the sintering temperature and the presence of dopants, the influence of the synthesis method was quite significant on the microstructure. As it was shown in our previous research, lanthanum inhibited the grain growth greatly but ceramic prepared by the Pechini method and doped only with La possessed grains around 1 μm in comparison with grains of ceramics obtained by the auto-combustion method which were found to be much smaller [10].

Figure 2
figure2

Micrographs of barium titanate specimens sintered at 1300 and 1350 °C for 4 h

Ferroelectric properties and leakage current

The PE hysteresis loops were measured at room temperature for all barium titanate compounds and presented in (Fig. 3). The presented loops were obtained by applying the electric field in the range of 1–60 kV/cm, depending on the type of ceramics and its breakdown field. The ferroelectric behavior of all ceramics was determined through the formation of typical hysteresis loops. Compared with the literature data, the ceramics showed a lower value of Pr in comparison with pure BT obtained from the polymeric precursors method [19]. From the presented diagrams, it can be observed that the co-doped ceramics possessed higher Pr values in comparison with other two ceramic samples doped only with lanthanum. The coercive field of these ceramics was also higher. These results correlated well with the XRD results, whereas the ceramic samples prepared by the Pechini method showed higher tetragonality in comparison with the other two samples. The BTL ceramics possess pseudo-cubic structure and therefore weaker ferroelectric properties. The explanations for this behavior can be also found in defect chemistry. Two types of compensation mechanisms are possible for added concentration of lanthanum [10]. First, lanthanum donor doping could occur via electronic compensation mechanism where free electrons are formed:

Figure 3
figure3

Polarization versus electrical field hysteresis loops for all ceramics at E = 25 kV/cm, inset at breakdown field

$$ {\text{Ba}}_{\text{Ba}}^{x} \to {\text{ La}}_{\text{Ba}}^{ \cdot } + e^{{\prime }} $$
(1)

and second, the ionic, titanium vacancy compensation mechanism is also expected for the concentrations of La up to 0.3 mol% and it is presented by:

$$ 4{\text{Ba}}_{\text{Ba}}^{x} + {\text{ Ti}}_{\text{Ti}}^{x} \to \, 4{\text{La}}_{\text{Ba}}^{ \cdot } + \, V_{\text{Ti}}^{{{\prime \prime \prime \prime }}} $$
(2)

The presence of La on Ba site deteriorates the tetragonal structure, and possible generation of Ti vacancies (VTi) destroys Ti–O–Ti linkages. In the BTL ceramics, compensation mechanism which can lead to a partial reduction of Ti4+ to Ti3+ and may cause a transition from an insulating behavior to a semiconducting behavior is plausible and presented as:

$$ {\text{Ba}}_{\text{Ba}}^{x} + {\text{Ti}}_{\text{Ti}}^{x} \to {\text{La}}_{\text{Ba}}^{ \cdot } + {\text{ Ti}}_{\text{Ti}}^{{\prime }} $$
(3)

On the other hand, manganese commonly substitutes Ti4+ due to small ionic radius [20]. Some authors proposed the model where the movement of manganese ions was always carried on the Ti–Ti chains in BT structure presented by the following equations [11]:

$$ {\text{Ti}}_{\text{Ti}}^{x} + e^{{\prime }} \leftrightarrow {\text{ Ti}}_{\text{Ti}}^{{\prime }} $$
(4)
$$ {\text{Mn}}_{\text{Ti}}^{{{\prime \prime }}} \leftrightarrow {\text{Mn}}_{\text{Ti}}^{{\prime }} + e^{{\prime }} $$
(5)

The presence of Mn on Ti site leads to disrupting of Ti–O–Ti linkages responsible for ferroelectricity. However, the oxygen loss during sintering in air could produce oxygen vacancies as well as free electrons according to Eq. 6.

$$ {\text{O}}_{\text{O}}^{x} \to 1/2{\text{O}}_{2} + \, V_{\text{O}}^{ \cdot \cdot } + \, 2e^{{\prime }} $$
(6)

In general, Ti4+ is chemically less stable and it easily changes into Ti3+ consequently forming the oxygen vacancies. Accumulation of the oxygen vacancies and electronic charge carriers at domain boundaries complicates polarization reorientation under an electric field and causes the effect of so-called domain pinning. Besides, there is a possible formation of defects such as V ··O -Mn Ti and V ··O -Mn ′′Ti as described by the following equation:

$$ 2{\text{Mn}}_{\text{Ti}}^{{{\prime \prime }}} + \, 2V_{\text{O}}^{ \cdot \cdot } + \, 1/2{\text{O}}_{2} \leftrightarrow 2{\text{Mn}}_{\text{Ti}}^{{\prime }} + V_{\text{O}}^{ \cdot \cdot } $$
(7)

where the oxygen vacancies could be trapped, causing the opposite effect. According to this, it was proposed that when the concentration of defects decreases (especially Vo), in the BTLM ceramics by the formation of above-mentioned defect complexes with Mn, the remnant polarization increases.

In this case, Pr is far lower in comparison with pure BT, indicating a strong domain wall pinning effect. On the other hand, more possible reason for poor ferroelectric properties is high porosity and small grain size of the obtained barium titanate-based ceramics. Many authors have reported the influence of density and the grain size on the dielectric, ferroelectric and piezoelectric properties of the ceramics [21, 22]. It is claimed that Ec decreases with the increase in grain size. In the present study, Ec of porous ceramics (500 nm) is similar to one obtained for pure BT (1 μm), where the grain size was much higher (presented in the previous study [19]), and this leads to the conclusion that both BTL and BTLM ceramics are easy to pole. However, the influence of the porosity and grain size on the Pr was found to be quite different. Regarding the synthesis methods that were used, in the ceramics obtained by the auto-combustion method smaller grains were obtained in comparison with BTLM ceramics prepared by the Pechini method, and according to this, lower Pr values were obtained. It was suggested that polarization reversal of ferroelectric domains is easier inside the larger grains than in the smaller ones. Buatip et al. [23] have given the explanation on how density and grain size can strongly influence the ferroelectric properties of BCZT ceramics. The theory of DE hysteresis loop based on the Kolmogorov–Avrami model (K–A model) was applied in the case of barium titanate-based ceramics. This model is expressed by the following equation: f = f0[1 − exp(− Ga·d3/kT)], where f is the ferroelectrics properties reflecting polarization or domain switchability, f0 is the initial polarization or domain of ferroelectric materials, Ga is a constant that represents grain anisotropy energy density, d is a grain size, k is a rate constant and T is temperature. From the equation, "f" is proportional to the grain size. Therefore, this indicates that the increase in grain size enhances the domain switchability and improves the ferroelectric properties [23].

From another point of view, the grain boundaries as low permittivity and poor ferroelectricity regions affect the reduction in the polarization. In the porous ceramics with small grain size, the number of grain boundaries increases, and therefore, the Pr decreases additionally. However, space charges in the grain boundary exclude polarization charge on the grain surface, and depletion layer on the grain surface can be formed. This results in polarization discontinuity on grain surface to form depolarization field, and polarization decreases accordingly [19].

The influence of sintering temperature was noticed in both types of ceramics. Pr and Ec values for the BTLM sample sintered at lower temperature were much higher in comparison with those of the ceramics sintered at higher temperature. The opposite trend was observed for the BTL samples with not so expressed difference in their Ec values. Compared with pure BT samples [19], these hysteresis loops showed a lower degree of squareness and were all unsaturated. However, the BTL samples exhibited much higher breakdown field, regardless of the sintering temperature. It means that even though these ceramics showed not so good ferroelectric properties in comparison with the BTLM samples, they exhibited better electrical strength (inset of Fig. 3). It also implied that the BTLM samples exerted more leaky ferroelectric behavior due to a higher amount of charge carriers emerged from the point defects described by Eqs. 36.

To confirm this assumption and to study the conductivity mechanism in the material, the leakage current density (j) was measured as a function of static electric field (E). jE characteristics are presented on the semi-log and log–log plots in Fig. 4a. Indeed, the BTL samples showed higher resistivity than the BTLM samples. The conduction mechanism in these samples was analyzed by plotting logj versus logE (Fig. 4b), in which nearly straight lines with two regions of different slopes were observed. These curves can be well fitted with power law: j ~ Em, where ‘m’ is the slope of the linear parts of the curve in the log–log plots which determines the nature of conduction [24]. The values m ≠ 1 represent deviation from the Ohm’s law. In all two field regions, ‘m’ is less than 1, suggesting the grain boundary limited conduction. In the polycrystalline dielectric material, the resistivity of the grain boundaries may be much higher than that of the grains. Thus, the conduction current could be limited by the electrical properties of the grain boundaries [25]. It can be seen from Fig. 4 that the leakage current is the highest for the BTLM1350 ceramics and the lowest for the BTL1350 in the whole electric field region. It is important to stress that these samples are porous and the effective distance between the electrodes can vary, which can complicate the interpretation of the experimental data obtained by both electric and ferroelectric characterization.

Figure 4
figure4

Leakage measurements presented in the form of a logjE and b logj–logE plots

Impedance analysis

The impedance analysis (IS) is a useful analysis method to study the electrical homogeneity of ceramic materials. It is known that many ceramic materials are electrically heterogeneous with semiconducting grain cores separated by insulating grain boundaries, especially when the doping process is included. The impedance analysis was used to evaluate and separate the contributions of the grain and grain boundary resistivities from the total resistivity of the material [26]. The IS analysis was performed in the air in the temperature range of 350–475 °C with a step of 25 °C. The results obtained from IS are presented in terms of impedance (Z) formalism. Variation of the imaginary part of impedance (Z′′) as a function of the real part (Z) gives the complex impedance spectrum (Nyquist plots). The shapes of these plots are temperature-dependent, and Nyquist plots for all ceramics are presented in Fig. 5. For the BTL1300 and BTL1350 samples, only one depressed semicircular arc can be detected in the whole temperature region, while a possible overlapping of two arcs that correspond to grain and grain boundary contributions can be observed in the case of the ceramics co-doped with Mn. The Z view software and equivalent circuit consisting of two parallel R-CPE elements connected in series were used to evaluate the grain (Rg) and grain boundary (Rgb) resistivity contributions for all ceramics. The extracted resistivity data have shown that Rgb values were higher than Rg. The resistivity of both types of ceramic samples sintered at higher temperature showed slightly lower values, possibly due to a lower content of grain boundaries per volume. The samples sintered at lower temperature (1300 °C) possessed the higher ratio of grain boundaries per volume and were more porous, both of which led to higher resistivity of these samples.

Figure 5
figure5

The complex impedance plane plots of all investigated BT ceramics

The total resistivity of BTL ceramics was a bit higher in comparison with the manganese co-doped ceramics. Usually, in the BT ceramics only doped with manganese, Mn can exist in three valence states (Mn2+, Mn3+ and Mn4+) and the electrons can bond to Mn ions and form ions with a neutral charge Mn xTi and Mn ions on the Ti sites with single and double negative charge Mn Ti , Mn ′′Ti . The generated electrons are preferentially trapped by the Mn ions. In the present study, the amount of manganese ions is very low and the motion of electrons from one Mn site to another can be effectively suppressed also by the formation of stable donor–acceptor defect complexes between lanthanum and manganese, 2[La ·Ba ]–[Mn ′′Ti ] as well as with defect complexes formed with the oxygen vacancies [27, 28]. Accordingly, Mn obviously influenced the resistivity of the ceramics by the formation of defects, and induced the decrease in the resistivity of the BTLM ceramics.

Data obtained for Rg and Rgb were used to evaluate the grain (σg = 1/Rg) and grain boundary (σgb = 1/Rgb) conductivities. The activation energies for various conduction processes were estimated from the slopes of the ln σ versus 1000/T plots presented in Fig. 6 (from the Arrhenius equation σ = σ0 exp (−Ea/kBT)) [17]. Literature data have shown that the band gap for intrinsic electronic conduction in the pure barium titanate is 3.0 eV. However, when both activation energies (for the grain and grain boundary conductions) are much smaller than a half of this value, σg and σgb are attributed to an extrinsic conduction mechanism [17, 29, 30]. The activation energy for the conduction process in the ceramics sintered at 1300 °C is lower in comparison with Ea for ceramics thermally treated at higher temperature, indicating the dominance of the intrinsic conduction mechanism, which is temperature-dependent. Contrariwise, the Ea (grain) that characterizes the BTLM ceramic samples possessed much higher values, pointing on the dominant effect of the impurities in the material.

Figure 6
figure6

Arrhenius plots of σg and σgb for all investigated ceramics

Hydrogen and humidity sensitivity measurement

Gas sensors are usually n-type semiconductors, and doping of barium titanate with lanthanum and manganese makes this material semiconducting [10]. In order to determine the suitability of the porous barium titanate ceramics for hydrogen sensing, the measurements of a change in the electrical resistivity during the hydrogen introduction in the system were taken. Since the used samples were made as ceramic compacts, it was quite important that their free surface was large enough to enable better contact of the material with the hydrogen gas. (Pellets diameter was around 11 mm.) For this measurement, a special custom-built setup was used and depicted in scheme presented in Fig. 7. The working temperature of these sensors is usually in the range from 200 to 600 °C, and their use is very significant in devices for detecting leakage of reducing gases [32].

Figure 7
figure7

Schematics of the experimental setup for hydrogen sensing properties

Here, the operation temperature was chosen to be 450 °C, considering the quick response time and reactivity of the oxygen adsorbed on the surface with hydrogen at high temperatures. Due to safety reasons, the mixture of nitrogen and hydrogen was used during the measurements. The working principle of hydrogen sensors is based on the modification of electrical conductivity or capacitance of material, resulted from the interactions between O2, O and O2− species and gas molecules [31]. According to the reaction: H2 + O2− → H2O + e, the potential barrier is lowered and conductance of the sensor increases. Figure 8 shows the instant change in the electrical resistivity in all ceramics after the introduction of hydrogen in the system. It can be noticed that the resistivity value rapidly decreased, which clearly indicated the sensitivity of these materials to the hydrogen. According to the literature data, much better response and recovery times could be obtained when the measurements were taken in the air atmosphere [33], which was not the case in this study due to safety reasons. The recovery time after removing the hydrogen was found to be much lower for the ceramics obtained by the auto-combustion method and doped only with lanthanum. By comparing the hydrogen sensitivity results for the investigated samples, it was found that the BTL1350 ceramics possessed the best sensing properties. In addition, measurement of the change in resistivity as a result of the presence of hydrogen gas at room temperature for BTLM1350 was also taken and is presented in Fig. 9. As it can be noticed the resistivity of the BTLM1350 ceramics at room temperature before the introduction of hydrogen was 1.6 MΩ and it started to increase with addition of hydrogen. There is also a difference between the trends of resistivity change at room and elevated temperatures. Upon the exposure to reducing gas such as hydrogen at high temperatures, the caught electrons are released by the reactions between the reducing gas and the negatively charged oxygen adsorbates, leading to the decrease in the resistance. However, completely opposite behavior was observed for gas sensing at room temperature where the resistance of the material increased upon exposure to hydrogen gas as presented in Fig. 9. As proposed by other authors [31], the desorption rate of absorbed gas greatly depended on the bound interaction energy, and the recovery time became longer in the case of having the stronger interaction/longer recovery. This conclusion can be valid for this study, since the shortest response time was noticed for the BTLM1300, in which the resistivity changed from 3500 Ω down to 350 Ω within only 1 min. It can be also prescribed to the reduction of Mn3+ into a Mn2+ and formation of electrons, which caused the rapid decrease in the resistivity value. However, the recovery process was found to be much longer in comparison with the other samples. Yuasa et al. [33] also studied the influence of synthesis methods on the electrical properties of La-doped ceramics and hydrogen detection. They concluded that chemical oxalic acid method had the advantage over the solid-state method, since much better control of the composition, morphology and properties was achieved by the wet chemical synthesis. Therefore, it is quite important to find out which synthesis method and preparation conditions will give desired properties of the sensor material. Besides, in our experiments, Au was used as an electrode material and the response time was found to be longer in comparison with the material on which Pt electrodes were applied. It is well known that Au has lower catalytic activity for combustion of hydrogen in comparison with Pt and due to this Pt electrodes will be used in the future study.

Figure 8
figure8

Responses transients to hydrogen for all ceramics at 450 °C

Figure 9
figure9

Responses transients to hydrogen for the BTLM1350 at 25 °C

The humidity sensitivity of all ceramic samples was analyzed in the climate chamber where the RH% was changed from 40 to 90% and temperature at which the measurements were taken was 40 °C. The basic working principle of the humidity sensors in porous perovskite ceramics is the change in electrical conduction or capacitance due to water chemisorptions and/or capillary conduction in the pores [14]. The water molecules that are adsorbed on the surface can release electrons to the conduction band, resulting in the increase in the electrical conductivity. The change of electrical resistivity with relative humidity was analyzed and is shown in Fig. 10. The decrease in resistivity was noticed with RH% increase, indicating the sensitivity of the material to the humid atmosphere. It can be seen that the fastest response by four orders of magnitude was found for the both BTL ceramics and the slowest response for the ceramics obtained by the Pechini method. However, a quicker response was noticed for the BTL1350 ceramics, whereas the RH% change from 40 to 50% induced the change in the resistivity for almost two orders of magnitude. In comparison with the results obtained for pure barium titanate by other authors [14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34], these results are very promising. Basically, when RH is low, water molecules chemisorb on the grain surfaces due to formation of chemical bond between the surface oxygen with the first water layer. This bond is not affected by the further change in humidity. Further layers of water molecules are physically adsorbed, and they can dissociate under the high electric field effect, according to the reaction: H2O → H3O+ + OH. The charge transport occurs when the hydroxyl groups as a surface charge carriers release a proton to a neighboring water molecule, which upon receiving it releases another proton and so on [12,13,14]. Besides the schematic illustration of the above-described humidity sensing mechanism, the defects reactions (presented using Kröger–Vink notations), dominant in each type of material and responsible for humidity sensing, are shown in Fig. 11a. Obviously, the concentration of the surface oxygen plays important role in the humidity sensing, and higher concentration of oxygen enables quicker response of the sensor. As it can be seen in the BTL ceramics, the electronic compensation mechanism is dominant but in the BTLM, along with this mechanism, there is also ionic compensation mechanism in which manganese changes titanium ion with formation of oxygen vacancies. Consequently, the BTLM samples possess lower concentration of the surface oxygen, which is crucial for the formation of initial chemisorbed water layer. Therefore, the BTLM ceramics showed longer response time and not so good sensing properties in comparison with the BTL ceramics.

Figure 10
figure10

Electrical resistivity–RH% relationships for all types of materials measured at 40 °C

Figure 11
figure11

a Schematic illustration of the humidity sensing mechanism, b influence of porosity on conductivity and vapor/gas sensitivity properties

It has to be point out that the presence of porosity as well as pore size distribution is also among the determinative factors for humidity sensors and obviously a selected chemical method can significantly influence the properties of barium titanate ceramics and enable the microstructure tailoring. Figure 11b illustrates how different types of formed structures (porosity and density of ceramics) can influence the current flow and access of vapor and gas into the structure. As presented in SEM images, the BTL ceramics possess open porosity and narrow grain necks providing easier access of vapor/gas into the material. The current has to flow via connected areas, and these samples appear to be highly resistive. In the BTLM ceramics, the vapor/gas access is quite limited due to lower porosity and formation of broader necks, which enables an unaffected current flow, shorter path of the charge carriers through the sample and consequently lower resistivity. Furthermore, the existence of open porosity in the BTL samples makes the process of desorption much quicker in comparison with BTLM ceramics.

Humidity sensing properties of the BTL1300 sample were successfully tested on a laboratory made device as presented in Fig. 12.

Figure 12
figure12

Laboratory testing device for humidity sensing

Conclusions

The barium titanate nanopowders doped with lanthanum and co-doped with lanthanum and manganese were successfully prepared by the auto-combustion and by the Pechini method, respectively. Sintering of the powder compacts was performed at 1300 and 1350 °C for 4 h. SEM analysis showed that ceramics obtained by the auto-combustion method and doped only with La possessed smaller grains in both sintering regimes.

The comparison of hydrogen sensitivity results for all types of ceramics in this study showed that the BTLM ceramics possessed the best sensing properties. It was found that even very small addition of Mn lead to quicker response (conductivity drop of 90% in the first minute), due to reduction of Mn3+ into Mn2+ ion. However, faster and higher degree of recovery time was observed for the ceramics obtained by the auto-combustion method. When it comes to humidity properties, the BTL1350 sample, in which RH% change from 40 to 50% induced a change in the electrical resistivity by two orders of magnitude, exhibited the fastest response when material was placed in the humid atmosphere.

The analysis of all results lead to the conclusion that both dopants nature, sintering regime and microstructure obtained by the selected synthesis methods were important for the development of adequate sensing properties and application of materials. The ceramics obtained by the auto-combustion method and doped only with lanthanum showed the highest potential for gas and humidity sensing.

References

  1. 1

    Buscaglia MT, Buscaglia V, Viviani M, Nanni P, Hanuskova M (2000) Influence of foreign ions on the crystal structure of BaTiO3. J Eur Ceram Soc 20:1997–2007

    Article  Google Scholar 

  2. 2

    Moulson AJ, Herbert JM (2003) Electroceramics, 2nd edn. Wiley, London

    Google Scholar 

  3. 3

    Vijatović Petrović MM, Bobić JD, Ramoška T, Banys J, Stojanović BD (2011) Electrical properties of lanthanum doped barium titanate ceramics. Mater Charact 62:1000–1006

    Article  Google Scholar 

  4. 4

    Vijatovic Petrovic MM, Bobic JD, Ramoska T, Banys J, Stojanovic BD (2011) Antimony doping effect on barium titanate structure and electrical properties. Ceram Int 37:2669–2677

    Article  Google Scholar 

  5. 5

    Kumar P, Singh S, Spah M, Juneja JK, Prekash C, Raina KK (2010) Synthesis and dielectric properties of substituted barium titanate ceramics. J Alloy Compd 489:59–63

    Article  Google Scholar 

  6. 6

    Guo HZ, Mudryk Y, Ahmad MI, Pang XC, Zhao L, Akinc M, Pecharsky VK, Bowler N, Lin ZQ, Tan X (2012) Structure evolution and dielectric behavior of polystyrene-capped barium titanate nanoparticles. J Mater Chem 22:23944–23951

    Google Scholar 

  7. 7

    Vijatovic Petrovic MM, Bobic JD (2018) Perovskite and Aurivillius: types of ferroelectric metal oxides. In: Stojanovic BD (ed) Magnetic, ferroelectric, and multiferroic metal oxides. Elsevier Publisher, Amsterdam, pp 35–49

    Google Scholar 

  8. 8

    Brzozowski E, Castro MS (2005) Grain growth control in Nb-doped BaTiO3. J Mater Process Technol 168:464–470

    Article  Google Scholar 

  9. 9

    Zhao C, Wu B, Cheng Thong H, Wu J (2018) Improved temperature stability and high piezoelectricity in lead-free barium titanate-based ceramics. J Eur Ceram Soc. https://doi.org/10.1016/j.jeurceramsoc.2018.08.004

    Google Scholar 

  10. 10

    Vijatovic Petrovic MM, Bobic JD, Grigalaitis R, Stojanovic BD, Banys J (2013) La-doped and La/Mn-co-doped barium titanate ceramics. Acta Phys Pol A 124:155–160

    Article  Google Scholar 

  11. 11

    Chatterjee S, Stojanovic B, Maiti H (2003) Effect of additives and powder preparation techniques on PTCR properties of barium titanate. Mater Chem Phys 78:702–710

    Article  Google Scholar 

  12. 12

    Farahani H, Wagiran R, Nizar Hamidon M (2014) Humidity sensors principle, mechanism, and fabrication technologies: a comprehensive review. Sensors 14:7881–7939

    Article  Google Scholar 

  13. 13

    Chen Z, Lu C (2005) Humidity sensors: a review of materials and methods. Sensor Lett 3:274–295

    Article  Google Scholar 

  14. 14

    Viviani M, Buscaglia MT, Buscaglia V, Leoni M, Nanni P (2001) Barium perovskites as humidity sensing materials. J Eur Ceram Soc 21:1981–1984

    Article  Google Scholar 

  15. 15

    Wang J, Wang X, Wang X (2005) Study on dielectric properties of humidity sensing nanometer materials. Sensors Actuators B 108:445–449

    Article  Google Scholar 

  16. 16

    Vijatovic MM, Stojanovic BD, Bobic JD, Ramoska T, Bowen P (2010) Properties of lanthanum doped BaTiO3 produced from nanopowders. Ceram Int 36:1817–1824

    Article  Google Scholar 

  17. 17

    Dzunuzovic AS, Vijatovic Petrovic MM, Stojadinovic BS, Ilic NI, Bobic JD, Foschini CR, Zaghete MA, Stojanovic BD (2015) Multiferroic (NiZn)Fe2O4–BaTiO3 composites prepared from nanopowders by auto-combustion method. Ceram Int 41:13189–13200

    Article  Google Scholar 

  18. 18

    Dwivedi RK, Parkash O, Kumar D, Srivastava KK, Singh P (2007) Effect of simultaneous substitution of La and Mn on dielectric behavior of barium titanate ceramics. J Mater Sci 42:5490–5496 10.1007/s10853-006-0985-8

    Article  Google Scholar 

  19. 19

    Vijatovic Petrovic MM, Bobic JD, Ursic H, Banys J, Stojanovic BD (2013) The electrical properties of chemically obtained barium titanate improved by attrition milling. J Sol Gel Sci Technol 67:267–272

    Article  Google Scholar 

  20. 20

    Li T, Li L, Zhao J, Gui Z (2000) Modulation effect of Mn2+ on dielectric properties of BaTiO3 -based X7R materials. Mater Lett 44:1–5

    Article  Google Scholar 

  21. 21

    Cai W, Fu C, Lin Z, Deng X (2011) Vanadium doping effects on microstructure and dielectric properties of barium titanate ceramics. Ceram Int 37:3643–3650

    Article  Google Scholar 

  22. 22

    Tan Y, Zhang J, Wu Y, Wang C, Koval V, Shi B, Ye H, McKinnon R, Viola G, Yan H (2015) Unfolding grain size effects in barium titanate ferroelectric ceramics. Sci Rep 5:9953

    Article  Google Scholar 

  23. 23

    Buatip N, Promsawat N, Pisitpipathsin N, Namsar O, Pawasri P, Ounsung K, Phabsimma K, Rattanachan ST, Janphuang P, Projprapai S (2018) Investigation on electrical properties of BCZT ferroelectric ceramics prepared at various sintering conditions. Integr Ferroelectr 187:45–52

    Article  Google Scholar 

  24. 24

    Godara S, Kumar B (2015) Effect of Ba–Nb co-doping on the structural, dielectric, magnetic and ferroelectric properties of BiFeO3 nanoparticles. Ceram Int 41:6912–6919

    Article  Google Scholar 

  25. 25

    Chien Chiu F. (2014) A review on conduction mechanisms in dielectric films, advances in materials science and engineering, Hindawi Publishing Corporation, 578168

  26. 26

    Sharma S, Shamim K, Ranjan A, Raib R, Kumari P, Sinha S (2015) Impedance and modulus spectroscopy characterization of lead free barium titanate ferroelectric ceramics. Ceram Int 41:7713–7722

    Article  Google Scholar 

  27. 27

    Moriwake H, Fisher C, Kuwabara A (2010) First-principles calculations of electronic structure and solution energies of Mn-doped BaTiO3. Jpn J Appl Phys 49:09MC01

    Article  Google Scholar 

  28. 28

    Chikada S, Kubota T, Honda A, Higai S, Motoyoshi Y, Wada N, Shiratsuyu K (2016) Interactions between Mn dopant and oxygen vacancy for insulation performance of BaTiO3. J Appl Phys 120:142122

    Article  Google Scholar 

  29. 29

    Thakur OP, Feteira A, Kundys B, Sinclair DC (2007) Influence of attrition milling on the electrical properties of undoped-BaTiO3. J Eur Ceram Soc 27:2577–2589

    Article  Google Scholar 

  30. 30

    Devi S, Jha AK (2009) Phase transition and electrical characteristics of tungsten substituted barium titanate. Phys B 404:4290–4294

    Article  Google Scholar 

  31. 31

    Kim Y, Ha SC, Kim K, Yang H, Choi SY, Kim YT, Park JT, Lee CH, Choi J, Paek J, Lee K (2005) Room-temperature semiconductor gas sensor based on nonstoichiometric tungsten oxide nanorod film. Appl Phys Lett 86:213105

    Article  Google Scholar 

  32. 32

    Farahani H, Wagiran R, Nizar Hamidon M (2014) Humidity sensors principle, mechanism, and fabrication technologies: a comprehensive review. Sensors 14:7881–7939

    Article  Google Scholar 

  33. 33

    Yuasa M, Nagano T, Tachibana N, Kida T, Shimanoe K (2013) Catalytic combustion-type hydrogen sensor using BaTiO3-based PTC thermistor. J Am Ceram Soc 96:1789–1794

    Article  Google Scholar 

  34. 34

    Caballero AC, Villegas M, Fernandez JF, Viviani M, Buscaglia MT, Leoni M (1999) Effect of humidity on the electrical response of porous BaTiO3 ceramics. J Mater Sci Lett 18:1297–1299

    Article  Google Scholar 

Download references

Acknowledgements

The authors gratefully acknowledge the Ministry of Education, Science and Technological Development of Republic of Serbia for the financial support of this work (Projects III45021, III45007). The authors are thankful to Dr Zeljko Despotovic for experimental support.

Author information

Affiliations

Authors

Corresponding author

Correspondence to M. M. Vijatović Petrović.

Ethics declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Vijatović Petrović, M.M., Radojkovic, A., Bobić, J.D. et al. Sensing properties of barium titanate nanoceramics tailored by doping and microstructure control. J Mater Sci 54, 6038–6052 (2019). https://doi.org/10.1007/s10853-018-03308-4

Download citation