Flash sintering of zircon: rapid consolidation of an ultrahigh bandgap ceramic

ABSTRACT Zircon (ZrSiO4) is a refractory structural ceramic material difficult to consolidate because of its thermal dissociation into ZrO2 and SiO2. Addition of sintering aids can improve its densification, but with detrimental effects on high temperature mechanical properties and corrosion resistance. In this work, zircon was consolidated by employing the Flash Sintering (FS) technique at a furnace temperature of 1250°C under an electrical field of 1000 V cm−1. The decomposition of zircon was significantly reduced by lowering sintering time and current density. Unlike conventional sintering methods, FS approach allowed to track the degree of dissociation by measuring the electrical resistivity, providing a promising route for the consolidation of such materials. Although the obtained zircon ceramics are characterized by lower density and hardness/toughness than those sintered by alternative advanced techniques (like SPS of HEBM activated powders), the consolidation can be carried out at remarkably reduced furnace temperature.


Introduction
Zircon (ZrSiO 4 ) is a natural occurring mineral with low thermal expansion coefficient (2.5 ppm °C −1 ), good thermal stability (dissociation temperature of 1673 ± 10°C), and high hardness [1,2], which makes it attractive for high temperature applications in the refractory industry. Other peculiar properties of zircon are its high corrosion resistance to molten slag and glass, as well as its capability to immobilize nuclear wastes, acting as a host for uranium and thorium within its crystal lattice [1,[3][4][5][6].
Zircon has a high thermodynamic and chemical stability; its natural minerals contain traces of uranium and thorium. Such combination of properties enables its use in geochronology for dating rocks back to 4.5 billion years. It has also a relatively high dielectric constant (ε r = 9.1) and a large energy gap (6.5-7 eV), enabling its application as a high-κ gate dielectric material in the metal oxide-semiconductor technology [7][8][9]. In spite of its unique properties, ZrSiO 4 is, in general, difficult to sinter due to its high refractoriness and tendency to decompose at elevated temperatures. For conventional sintering, processing temperatures are as high as 1600°C with holding times of at least 2 h [10,11]. If higher temperatures (>1700°C) are used, zircon gradually decomposes into amorphous silica and zirconia (ZrO 2 ) [1].
Several approaches have been developed to overcome these limitations, such as mechanochemical activation of the powders [11] and the use of advanced sintering techniques such as spark plasma sintering [12,13], hot-pressing [14,15], and microwave-assisted sintering [16]. On the other hand, dense zircon ceramics can be fabricated by incorporating sintering aids [14,17,18], although the addition of such oxides can cause a partial reduction of its properties.
An innovative way to process dense zircon ceramics, never attempted before, could be Flash Sintering (FS) [19]. FS is [20] a current-assisted sintering technology characterized by rapid densification (matter of few seconds-minutes) and by the simultaneous observation of the flash event, including a thermal runaway of internally generated Joule heating, electrical resistivity abrupt drop, and strong bright light emission. Namely, an onset field/temperature combination causes a power surge (the "flash event"), resulting in nearly instantaneous densification of the material [21].
This innovative sintering technique presents several advantages when compared to conventional sintering, such as the reduction of time and costs due to the lower temperatures and shorter sintering times [20,21]. Reduced sintering time and lower furnace temperature have other associated beneficial effects reflected in the fine grained/nanometric microstructure [22], and the consolidation of out-of-equilibrium phases [20,23]. The absence of constrained sintering upon FS makes this technique suitable for the preparation of multilayered and composite ceramics [24].
It is important to remind that the internal heat generation by the Joule effect is one of the main features of this technique [25,26]. Though the initial works on FS were carried out on ionic conductors [19,27], during the last years, materials with far different electrical properties have been sintered with FS technique. Electronic conductors require high currents to produce enough heating to trigger the flash event, and this is typically achieved using flash spark plasma sintering approach without the need for preheating [28][29][30][31].
Being the conduction across the sintering material a key parameter to sustain Joule heating, high bandgap materials like zircon are hard to flash sinter because they require the application of very high fields in combination with furnace heating. For instance, high-purity alumina (99.999% pure) cannot be flashed because the internal heat generation by the Joule effect is not high enough to trigger the flash event [32,33].
Zircon, because of its very wide bandgap (and the absence of significant ionic conduction) falls into the latter category of materials, as it presents low conductivity values (4.17x10 −4 S cm −1 ) even at temperatures as high as 1478°C [34].
One of the drawbacks of FS is the development of hot-spots during the flash transition due to the electric power spike [35]. These hot-spots can lead to nonhomogenous microstructures [36]. In order to reduce hot-spots formation, some authors developed flash processes where the current limit is raised stepwise during the process [37][38][39].
The aim of this work is to provide insights into the following questions: (i) can zircon be flash sintered despite its low electrical conductivity? (ii) Can we attain a dense microstructure using a current-ramp FS approach? (iii) Which are the main microstructural features of flash sintered zircon?

Experimental procedures
Commercial zircon powder (Kreutzonit Super Extra Weiß, Mahlwerke Helmut Kreutz GmbH, Germany) was used in this study. As informed by the manufacturer, it has an average particle size of 0.8 µm and its chemical composition is (wt %): ZrO 2 64-65.5, SiO 2 33-34, Fe 2 O 3 ≤ 0.1, Al 2 O 3 ≤ 0.1, TiO 2 ≤ 0.15 [40]. The powder was dispersed in acetone and 1% of polypropylene carbonate binder (QPAC 40, Empower materials, USA) was added. After 24 h drying at room temperature, the resulting powder was grounded by mortar and pestle and then uniaxially pressed at 100 MPa in a heated (40°C) dog-bone shaped die (gage length = 20 mm; cross-section = 3 x 3 mm 2 ). Then, the samples were presintered at 1200°C in an electric furnace in static air atmosphere. Lower temperatures of 800°C and 1000°C were tried for pre-sintering the dog-bone samples, but the obtained specimen crumbled upon handling to insert the FS electrodes. The relative green density of the samples was 55%, while the green density of the presintered specimen was 57%. For this study, as usual, when using this type of die, only the middle section (i.e. the constant crosssection region of the dog bone) was analyzed.
Flash sintering of the pre-sintered samples was carried out within a tubular furnace (Nabertherm P330) kept at a constant temperature of 1250°C in air atmosphere. Two 90Pt-10Rh wires were forced within the holes on the opposite sides of the dog-bone specimen. Pt paste (C60903P5 Gwent) was spread on the metal/ ceramic interface to improve the electric contact. The samples were introduced within the pre-heated furnace and connected to a DC power source (Glassman EW series 5 kV-120 mA). The electric data were recorded at 1 Hz using a Keithley 2100 digital multimeter. Two minutes after the samples had been introduced in the furnace, the flash experiment started. The voltage limit of the power source was set at 1000 V cm −1 . The current limit was increased with a step function. Current needed to be gradually increased, otherwise sparks appeared making the flash sintering impossible. Samples were subjected to three different sintering conditions and they were labeled accordingly: (i) low current density values of 4 mA mm −2 (L-ZS) with no holding time, (ii) same conditions with 2 min holding time at 4 mA mm −2 (LH-ZS) and (iii) high current density of 11 mA mm −2 (H-ZS). The sample names are described in Table 1. Materials produced using a single step of current produced arching and are not discussed in this paper.
An estimate of the temperature of the samples upon FS was carried out according to Raj [41] by assuming black body radiation and no temperature gradients across the sample. By equating the electrical power dissipation to the radiative heat loss, the sample temperature (T s ) is obtained as: Table 1. Current density applied on flash sintered samples. See also Figure 1.

Sample
Step 1 Step 2 Step 3 Step 4 Step 5 Step 6 Step 7 Step 8 Step where T 0 is the furnace temperature, W the power dissipation in the steady stage of FS, A the surface area of the specimen (assumed to be the gauge section), ε the emissivity of the material and σ Stefan and Boltzmann constant. The emissivity value for zircon used in Equation (1)was 0.52 [42].
Manually fracture flash sintered samples were observed using a Leica S APO stereo microscope. X-ray diffraction of sintered samples (Bruker D8 Advanced equipment with Cu Kα in Ni filter at 40 kV to 30 mA, 1 s steps of 0.04° 2θ) was performed at 2θ between 15° and 80°. Apparent density and open porosity of the sintered samples were obtained by the Archimedes' method. Due to the small sample size, at least 10 measurements were taken for each sample.
Further, microstructural analysis of the sintered samples was carried out by Scanning Electronic Microscope (SEM, JEOL JMS-6000, Japan). Samples were mounted in epoxy resin, polished and coated by sputtering with a thin layer of carbon. Back-Scattered-Electrons (BSE) imaging mode was also used in order to further assess possible dissociation of zircon. Images were taken in the middle section of the dog-bone samples, and estimated values of zircon dissociation were obtained from BSE images, by correlating each phase to its corresponding intensity in the histogram of each image and quantifying zircon across the surface sample. At least five images were analyzed in this way for each sample.
Vickers hardness tests of the sintered samples were conducted with a microindenter (Buehler, USA) using a 3 kg load and 15 s dwell time. At least 10 indentations were produced on each sample.

Results and discussion
The flash event was successfully reproduced on zircon. All samples were flash sintered with a current step regulation (CSR) approach. Electrical data ( Figure 1) revealed that each current step up was accompanied by a small voltage/field spike, each spike decreasing in height with increasing current. This observed phenomenon is attributed to Ohm's law. The increased electrical conductivity with the temperature promoted the thermal runaway effect [25].
From the electrical data of the flash sintering of the samples, different parameters were obtained, as shown in Table 2. Both L-ZS and LH-ZS samples show similar values of peak power dissipation, temperature and resistivity. Sample H-ZS shows a higher value of peak power dissipation, a slightly higher temperature and a resistivity of an order of magnitude less with respect to the other samples. Zircon can undergo partial dissociation at high temperatures, being ZrSiO 4 the major stable phase up to ~1500°C [11,13], according to the following reaction: Both purity and grain size strongly influence the onset temperature for zircon dissociation [1]. Sample H-ZS was subjected to FS for a longer period at high temperature, which presumably can account for a larger degree of dissociation with respect to L-ZS and LH-ZS. This higher degree of dissociation could also explain the drop in resistivity at the end of the process in sample H-ZS, as ZrSiO 4 electrical resistivity at high temperatures is higher than that of silica glass and ZrO 2 . The surface of the samples was observed (Figure 2  The cross-section of the anode and cathode region is also shown, revealing a blackening in the center of samples L-ZS and H-ZS. This effect could be another indication of a partial dissociation of zircon into silica and ZrO 2 phases (Equation (2)), with the electrochemical blackening of ZrO 2 -like phases being responsible for such behavior as discussed in [43] and [44].
As explained in previous works, blackening develops from the cathode and moves toward the anodic region, analogously to the well-known electrochemical blackening associated with the partial reduction of ZrO 2 [45]. This black "core" surrounded by a white surface layer is explained by the rapid re-oxidation of the surface upon cooling after FS.
The blackening phenomena is certainly of interest to understand the flash sintering mechanisms. As summarized in Table 2, the blackening can be qualitatively associated with the electrical resistivity of the sample. The lower the sample resistivity, the more significant is the blackening. LH-ZS material was heavily blackened because of the high current density (11 mA/ mm 2 ) and the final resistivity of 1.28 kΩ cm. For samples L-ZS and LH-ZS, the blackening effect was rather mild and the sample resistivity at the end of FS was in both cases 11.3 kΩ cm. To be more precise, L-ZS sample was more blackened and more prone to cracking than sample LH-ZS ( Figure 2). The latter effect can be explained by looking at the power profiles ( Figure 1) The L-ZS sample reached a powder dissipation of 200 mw/mm 3 after 70 s while 110 s was needed for the LH-ZS. The smooth LH-ZS raise in power might justify the absence of cracking and blackening. The current and power   increase effects in FS are particularly important and they have been discussed in [46] Figure 3 shows XRD plots of the sintered zircon samples. Interestingly, although not all samples blackened, there is a degree of dissociation in all of them, evident by the presence of monoclinic ZrO 2 (m-ZrO 2 ) peaks. Although the ZrO 2 formed by partial dissociation of zircon is usually tetragonal ZrO 2 (t-ZrO 2 , in agreement with its thermodynamic stability range) it can transform spontaneously to m-ZrO 2 at room temperature [1,13,47,48]. Besides ZrSiO 4 and m-ZrO 2 , no other crystalline phases were found, thus confirming that the SiO 2 derived from the dissociation went into an amorphous (glassy) phase. Samples were later exposed to thermal annealing at 800°C for 1 h and XRD measurements do not reveal any other crystalline phases, thus confirming that the blackening of the samples was attributable to ZrO 2 partial reduction [45].  There is a decrease in density with increasing current (Table 1), this being an unusual effect in flash sintering since higher current usually leads to Joule effect heating and sintering. Although low open porosity values were found, in general, apparent density is quite lower than zircon theoretical density (4.65 g cm −3 ), with H-ZS material showing the lowest value.
Possible explanations for these effects could be a combination of closed porosity and partial decomposition of zircon. As stated before (Equation (2)), the products of the reaction are amorphous silica glass (density 2.2 g cm −3 ) and either m-ZrO 2 or t-ZrO 2 (density equal to 5.83 and 6.10 g cm −3 , respectively), which give a total theoretical density of the products between 3.78 and 3.87 g cm −3 , this value being dependent on the type of ZrO 2 phase formed. The products of the reaction give a total theoretical density lower than that of pure ZrSiO 4 , which explains the lower densities.  SEM micrographs taken on samples L-ZS (Figure 4(a,  b)) and LH-ZS (c,d) show a homogeneous microstructure, low porosity and minor or no defects, while a non-homogeneous microstructure filled with cracks and defects is revealed in sample H-ZS (Figure 4(e,f)).
The dissociation in samples LH-ZS and H-ZS was comparatively higher than in sample L-ZS, as seen by the presence of higher amounts of ZrO 2 (light gray) and silica (dark gray) derived from this reaction. Also, the presence of porosity in all samples not revealed by the Archimedes' method further suggests that this closed porosity is derived from defects originated upon sintering.
From SEM images and considering the distinct color of each phase present, analysis of the histogram of the images yields an estimate of zircon's degree of decomposition in each sample, as shown in Table 4. For this calculation, the theoretical density of each phase was considered and the mass fraction was calculated accordingly. As expected, decomposition rate increases with temperature and sintering times. Table 4 shows the measured Vickers hardness for samples L-ZS and LH-ZS. High current values in sample H-ZS (associated with an overheating of the samples) produced a much weaker structure, as samples tended to crack when handled. As such, Vickers hardness was not measured for this sample. Vickers hardness is slightly lower for sample LH-ZS when compared with sample L-ZS. Gauna et al. [11] well correlated the hardness of different zircon samples with their densities. As explained before, zircon's dissociation brings about a decrease in density, which explains the obtained values when compared to those obtained on fully sintered materials. With the aim to compare the obtained results with the literature, the relative density of the sintered samples was calculated against zircon theoretical density, giving values of 83% and 85% for LH-ZS and L-ZS, respectively. The comparison is shown in Table 5.
In Table 5, a comparison is made between the zircon material sintered in the present study and those reported in the literature. Vickers hardness of flash sintered samples is comparable to values found in previous works on materials with similar density [11,52], although lower than those found in the literature for dense zircon ceramics produced by SPS or hot pressing. In spite of the obtained lower density and hardness, FS technique allows to consolidate zircon at a much lower furnace temperature. The resulting open porosity is well below 1.5%, value only obtained with

Conclusions
• Zircon samples were successfully flash sintered by a current ramp regulation approach. • Flash sintered zircon samples show a variable degree of dissociation and moderate Vickers hardness. Although the relative density is low, the open porosity is small. • Higher current and longer sintering time cause a higher dissociation of zircon into m-ZrO 2 and SiO 2 .
Although completely dense, single-phase materials were not obtained. This work reports the potential path not only for the study of flash sintering of zircon, but also to other high bandgap materials by using a stepwise current increase to avoid the formation of undesired electrical arcs.

Disclosure statement
No potential conflict of interest was reported by the authors.