Flash Sintering Samaria-Doped Ceria–Carbon Nanotube Composites

Abstract

Composite ceramic green pellets were prepared by attrition milling a mixture of (CeO₂)_0.8(Sm₂O₃)_0.2 (samaria-doped ceria, SDC) ceramic powder and carbon nanotubes (CNTs), followed by uniaxial and isostatic pressing. The pellets were sintered inside a dilatometer by applying AC electric fields at 850 °C and limiting the electric current to 1 A, achieving 20.2% final shrinkage. The SDC samples reached 13.3% shrinkage under the same conditions. Higher average grain sizes were measured in specimens flash sintered with CNTs. Impedance spectroscopy analyses show that the specimens flash sintered with addition of CNTs have higher electrical conductivity. Higher delivered Joule heating at the interfaces due to the presence of the electronic conductors (CNTs) are proposed as the main reason for that improvement of the electrical behavior.

1. Introduction

Electric field-assisted pressureless sintering, usually named flash sintering when the phenomenon occurs in few seconds, has been successfully applied to several inorganic compounds since it was first reported in 2011 [1]; for example in ZrO₂: 3 mol% Y₂O₃ [1,2,3,4,5,6], ZrO₂: 8 mol% Y₂O₃ [7,8,9,10,11], ZrO₂: 20 mol% Sc₂O₃ [12], CeO₂: 20 mol% Sm₂O₃ and CeO₂: 20 mol% Gd₂O₃ [13,14,15,16,17,18], BaCe_0.8X _0.2 O_3-δ (X=Y, Sm, Gd) [19,20], BaTiO₃ [21,22,23], BiFeO₃ [24], CaCu₃Ti₄O₁₂ [25], ZnO [26,27,28], SnO₂ [29,30], Al₂O₃ [31,32], SrTiO₃ [33,34], ThO₂ [35], UO₂ [36,37], TiO₂ [38], SiC [39], Y₂O₃ [40]. The flash sintering method consists of applying a DC or AC electric field to a ceramic green compact either under heating (dynamic flash sintering) or at a temperature usually below the conventional sintering temperature (isothermal flash sintering) [41]. In both cases, the main features of the sintered compounds are fast sintering times at temperatures well below those usually required, and inhibition of grain growth [41,42,43]. Even though Joule heating has been recognized as the primary consequence resulting from the application of an electric field (see Figure 1), the discussion of the mechanism(s) is still under debate [41,42,43,44,45]. Observation of electroluminescence, phase transition, chemical diffusion, and intergranular melting during flash events led to proposals of other mechanisms besides Joule heating, including point defect creation [46,47,48,49,50,51,52,53,54,55]. Further experimental work is suggested to understand the stages of the flash sintering and the mechanisms in each stage, particularly those related to the specimens (impurity content, average particle size, distribution of particle size), to electrical parameters (field amplitude, current density, field frequency), and to experimental configuration (furnace temperature, electrode, pressure) [42].

In this paper the main concern is to explore the Joule heating action at the inter-particle region of a ceramic green pellet. Samaria-doped ceria (SDC), a promising material for use as solid electrolyte in solid oxide fuel cells (SOFCs) operating at intermediate temperatures (< 800–1000 °C required for SOFCs with yttria-stabilized zirconia solid electrolytes), was chosen as the matrix. Due to the fact that the electrical conductivity of carbon nanotubes (CNTs) is several orders of magnitude higher than that of SDC (850 S.cm⁻¹ [54] versus 0.01 S.cm⁻¹ at 800 °C [55]), they were chosen as SDC inter-particle fillers for providing an improvement of the electric current percolation path, enhancing the Joule heating and consequently the shrinkage of the SDC pellets. Since the first report of graphitic sheets, the precursor of modern carbon nanotubes [56], a significant number of applications have been found for CNTs [57]. Multi-walled carbon nanotubes were reported recently to enhance the densification of alumina through localized Joule heating using spark plasma sintering at 1500 °C under 40 MPa within 10 min [58].

2. Materials and Methods

CeO₂: 20 mol% Sm₂O₃ (20SDC) agglomerated ceramic powders with >100 m²·g⁻¹, 5–10 nm primary crystallite size (Fuel Cell Materials, USA) were used as the matrix [59]. Multiwalled carbon nanotubes (CNT) were synthesized by catalytic chemical vapor deposition at 730 °C using ethylene as carbon source and a Fe–Co/Al₂O₃ catalyst [60]. The CNTs have diameters ranging from 10 to 30 nm and carbon purity of 93%. They were used without purification. The compositions were prepared with 10 vol% CNT added to 20SDC by attrition milling at 5000 rpm in a custom-made Teflon^®-lined metallic jar, adding 1 mm diameter zirconia-yttria (ZrO₂: 5 mol% Y₂O₃, 6.0 g/cm³, hardness HV1250 [61]) sintered ceramic spheres (Tosoh, Japan) and isopropyl alcohol; this procedure was repeated 4 times for 15 min with intermediate resting to avoid excessive heating and powder agglomeration. The mixed powders were uniaxially pressed into a cylindrical shape (ϕ 5 × 3 mm) at 10 MPa and isostatically (National Forge Co, USA) at 200 MPa. The geometrical density of all samples was 35–40% TD (theoretical density).

Thermal analysis of the compositions was carried out in a Simultaneous Thermal Analysis 409E Netzsch instrument (Karlsruhe, Germany) in the room temperature–1200 °C range with a 10 °C·min⁻¹ heating rate under flowing air at 5 L·min⁻¹. The gas evolving from the thermogravimetric sample chamber was analyzed in a mass spectrometer (Thermostar GSD 320T1, 1–100 amu, Pfeiffer, Germany).

X-ray diffraction analyses were performed at room temperature in a Bruker-AXS D8 Advance diffractometer with Bragg–Brentano configuration with Cu k_α radiation (λ = 1.54049 Å) in the 20–80 2θ range, 0.02° step size, 10 s step time to ascertain there was no structural phase change during sintering.

Electric field-assisted sintering was performed in samples positioned inside a dual vertical dilatometer (Unitherm model 1161, Anter, USA) especially adapted for application of electric field in the samples [8,11]. Briefly, platinum grids were inserted in both sides of cylindrical green pellets and connected by means of alumina insulated platinum wires to a custom-made power supply (AC power supply, 0–64 V, 0.1–1.1 kHz, 0.5–5 A) for the application of an electric voltage when the sample reached a pre-set temperature at a 10 °C·min⁻¹ heating rate. The shrinkage/expansion temperature profile data of the sample were collected at the dilatometer; the voltage and current data were collected with two Fluke 8050A multimeters interfaced to the output of the power supply. Apparent density of the sintered samples was evaluated by the Archimedes method.

Impedance spectroscopy data were carried out with a Hewlett Packard 4192A impedance analyzer in the 5–13 MHz frequency range with 200 mV AC input signal in samples spring-loaded between platinum disks inside a sample chamber made of Inconel 600, alumina, and platinum terminal leads. Platinum paste (ESL, USA) was deposited on the parallel surfaces of the ceramic green pellets and cured at 600 °C.15 min⁻¹ prior to measurement. For the collection of the impedance data, a model 360 Hewlett Packard Controller was used; the collection, analysis and deconvolution of the [−Z″(f) × Z′(f)] impedance diagrams were performed with special software [62]; Z′ and Z″ are the real and the imaginary components of the impedance and f is the frequency of the input signal.

Polished (down to 1 µm diamond paste) and thermally etched (1300 °C/15 min) surfaces of flash sintered specimens were observed in a scanning electron microscope (JEOL JSM 6701F).

3. Results and Discussion

Figure 2 shows the results of the thermogravimetric analysis of the (90 vol% CeO₂: 20 mol% Sm₂O₃–10 vol% carbon nanotube) composite in the room temperature–1200 °C range. The decrease of mass up to approximately 300 °C is ascribed to the release of physisorbed water. Mass loss is observed up to approximately 570 °C, corresponding to the removal of carbon nanotubes due to reaction with oxygen producing CO₂. The detection of CO₂ was confirmed by analysis of the output gas with the mass spectrometer. From 250 °C to 450 °C, CO₂ is released from both solid electrolyte and the solid electrolyte–CNT mixture due to CO₂ adsorbed in the samples during handling in the laboratory atmosphere. From 450 °C to 700 °C, a larger amount of CO₂ is detected due to the reaction of the CNTs with oxygen pertaining to the synthetic air used in the experiment. The difference in the temperature ranges for the CNTs annihilation is due to the different location of the detection: the thermocouple of the TG equipment is close to the sample, but the travelling distance to the probe of the mass spectrometer is away (1 m) from the TG sample chamber gas outlet.

In good agreement with the results of mass loss in the thermogravimetric analysis (Figure 2), Figure 3 shows the dependence of the DC resistance of the 20SDC–10 vol% CNT composites during heating to and cooling from 750 °C. The electrical resistance of the composite is relatively low (3.5 kohm) at 250 °C due to the electronic conduction of the CNTs, and increases as the composite loses CNTs due to reaction with oxygen of the atmosphere; when the composite reaches approximately 450 °C, the resistivity decreases due to the oxide ion conduction of the solid electrolyte (electrolytic region).

Figure 4 shows the dilatometric curves featuring the thickness shrinkage, from room temperature to 850 °C, of 20SDC + 10 vol% CNTs cylindrical pellets without and with the application of 210 V·cm⁻¹ AC (1.1 kHz) electric field, limiting the current to 1 A to avoid thermal runaway that could occur with continuous increase of Joule heating. There is a 3% shrinkage of the sample with CNTs in the 400–450 °C range due to loss of CNTs, in agreement with the thermogravimetric data. The shrinkage levels attained were 8.6% and 20.2% without and with applying the electric field. Similar experiments were conducted in 20SDC samples without CNTs. The shrinkage levels were 8.1% and 13.3%, respectively, evidence of the enhancement of the densification by introducing CNTs in the 20SDC green pellets. Difference was also observed in the voltage–current temperature profiles, shown in the same figure. In the sample without CNTs, it takes 60 s for the flash event, i.e., for the electric current to reach the programmed value (1 A). The oscillation in the electric current might be due to the partial electronic conductivity of 20SDC, induced by the increase in temperature promoted by Joule heating. In the sample with CNTs, on the other hand, the applied voltage oscillates to yield a pulsed electric current through the sample. This might be due to the intermittent reaction of the CNTs with oxygen, yielding CO₂ release.

The Archimedes densities of the flash sintered 20SDC and 20SDC + 10 vol% CNT were 96.1% and 95.0% of the theoretical density. The scanning electron microscopy images of the flat surfaces of these samples, after polishing and thermally etching, are shown in Figure 5. The average grain size of the specimen with addition of CNTs is larger than in the specimens without CNTs (similar results were reported recently on SDC with graphite additions [63]); relatively larger intergranular pores are detected in the former, probably due to the enhanced Joule heating delivered to these specimens.

Figure 6 shows the [−Z″(f) × Z′(f)] impedance spectroscopy diagrams of the two flash sintered samples: CeO₂: 20 mol% Sm₂O₃ and CeO₂: 20 mol% Sm₂O₃ + 10 vol% carbon nanotubes. The improvement of the electrical conductivity with mixing carbon nanotubes with samaria-doped ceria is evident. The electrical resistivity of the former is 16.4 kohm, much lower than that of the latter, which is 3 Mohm.

4. Conclusions

The insertion of carbon nanotubes at the interparticle region of green pellets of ceria–samaria solid electrolytes promoted an increase of the thickness shrinkage of those composites when submitted to electric field-assisted (flash) sintering. It is proposed that the carbon nanotubes improve the percolation path for the electric current pulses, derived from the application of the electric field, enabling higher values of Joule heating with a consequent higher degree of densification, relatively larger grain sizes and intergranular pores. A pronounced decrease of the grain boundary resistivity is measured in specimens flash sintered with carbon nanotube addition, suggesting that the enhancement of Joule heating by the addition of carbon nanotubes greatly improved the grain-to-grain contact, being an alternative for promoting high densification of electroceramics.