Obtaining highly dense YSZ nanoceramics by pressureless,unassisted sintering

Ramp and hold sintering

The most common sintering schedule in a conventional furnace is simply heating up the sample to the desired peak temperature at a constant rate (ramp) and maintaining it at this temperature for a specific period of time (hold). Typical heating rates range from 2 to 10°C min^− 1, and peak temperatures from 1000 to 1400°C with holding times from 30 min to 80 h. Several authors have reported on conventional pressureless RHS of YSZ nanopowders. Nevertheless, the most of them achieved nanoceramics with densities below 99% TD; when denser bodies were produced, a significant grain growth was observed, as described in the next paragraphs.

Theunissen et al. ⁸⁷ obtained 97% TD nanoceramics from 3YSZ powders prepared by gel precipitation with a mean grain size of 60 nm after sintering at 1050°C for 7 h. Boutz et al. ⁸⁸ produced 96% TD YSZ nanoceramics from hydrothermally synthetised powders with a grain size of ∼95 nm after sintering at 1050°C for 5 h. Sagel-Ransijn et al. ⁸⁹ synthesised YSZ by coprecipitation and obtained specimens with 96% TD and an average grain size between 120 and 130 nm after sintering at 1070°C for 10 h. In another work from the same group, ⁹⁰ YSZ sintered bodies with 98% TD and a grain size of 180 nm were produced from hydrothermal nanopowders. Li et al. ⁹¹ sintered 3YSZ with 98% TD and an average grain size of 90 nm from nanoparticles obtained by heating of alcohol-aqueous salt solutions. Liu ⁷⁸ sintered 3YSZ compacts at 1500°C with nearly full density and grain size of 140 nm in average, after a proper colloidal processing control of a commercial powder. Boulch et al. ⁹² and Djurado et al. ⁹³ prepared spray-pyrolysed nanometric 2·5YSZ powders (∼6 nm) that lead to 70 nm primary crystallites after sintering at 1500°C for 2 h with densities around 96% TD, though. Similar results were obtained by the same group ⁹⁴ for zirconia doped with other rare-earth elements. Tadokoro and Muccillo ⁹⁵ prepared 3YSZ by coprecipitation and obtained specimens with 99·5% TD and an average grain size of 130 nm after sintering at 1200°C for 5 h. Gaudon et al. ⁵² produced 8YSZ bodies with grain sizes in the range of 50–200 nm up to 97% TD after sintering at 1400°C during 5 h from nanopowders obtained by spray pyrolysis (5–6 nm for powders processed at 600°C and 15–20 nm at 1100°C). Benavente et al. ⁹⁶ obtained dense nanostructured 3YSZ compacts by slip casting of binary mixtures of commercial powders. Green densities reached 57% TD but final densities were not reported. Homogeneous microstructure was obtained after sintering at temperatures of 1300°C with the average grain size of 160 nm.

A few works have succeeded to achieve highly dense, nanocrystalline YSZ bodies through RHS. Table 4 summarises processing and sintering parameters for YSZ nanoceramics with minimum final densities of 99% TD. A sintering elapsed time is included, which was calculated from the heating ramp (time necessary for the sample to heat up from room temperature to the peak temperature) and hold period (dwell time at peak temperature). Moreover, a grain growth factor was obtained from the ratio of the grain size of the sintered body and the crystallite size of the starting powder.

Maca et al. ⁵⁵ prepared zirconia (1·5 and 3YSZ) with particle size  < 10 nm by a sol-gel synthesis. Isostatically pressed green bodies presented pore radii  < 5 nm. After pressureless sintering at 1100°C, the bodies reached density of ∼99% TD and grain size  < 80 nm. The sintering activation energy of zirconia nanoceramics was 237 kJ mol^− 1, less than the half the value measured for submicrometric zirconia (550 kJ mol^− 1). Later, Trunec and Maca ³⁷ compared three commercially available 3YSZ nanopowders to a self-synthesised sol-gel powder. Ceramic green bodies were prepared by CIP at pressures of 300–1000 MPa. Only homogeneous green bodies with pores  < 10 nm could be sintered into bodies (>99% TD) to keep the grain sizes  < 100 nm.

Mayo and co-workers ^60,61 produced nanocrystalline 3YSZ from dry pressed coprecipitated powders (starting size 13 nm). The pore sizes of the green compacts were ∼5 nm and no interagglomerate pores were detected. Samples were sintered at 1050°C for 5 h with starting green density of 58% TD to achieve 99·9% TD with an average grain size of 85 nm.

Durán et al. ⁶² prepared a 3YSZ powder by coprecipitation with a crystallite size of ∼9 nm, which was cold isostatically pressed (43% TD) into compacts with a pore of ∼6 nm and narrow pore size distribution. Compacts were sintered at 1150°C for 4 h or at 1200°C for 20 min hold times to obtain fully dense ceramics with a grain size  < 100 nm. From a constant-rate heating sintering, at least three densification steps could be established. A first step, comprising a particle rearrangement, was detected up to ∼800°C with activation energy of 130 ± 40 kJ mol^− 1. A second step, in which rapid densification occurs ( ≥ 95% TD) with slow grain growth, took place between 800 and 1180°C with an activation energy of ∼300 kJ mol^− 1, which was well correlated to grain boundary diffusion. Finally, a third densification step was considered to exist above 1180°C in which almost fully dense Y-TZP ceramics were achieved, in which grain growth is enhanced. Later, the same group obtained 99·9% TD with a grain size of 72 nm after 2 h holding time at 1070°C. ⁶³ When a temperature limit of 1000°C was used, the same samples took 80–100 h to reach full density, and the grain sizes remained below 100 nm (Fig. 6). ⁶⁴ Finally, when nanocrystalline 3YSZ powders synthesised by seeding-assisted chemical coprecipitation were used, nearly full dense sintered bodies were obtained at 1050°C for 5 h with a grain size  < 100 nm. ⁷³

OPEN IN VIEWER

6

(a) 99·9% dense 3 mol% YSZ (3YSZ) with grain size of ∼80 nm sintered at 1070°C for 2 h (SEM). ⁶⁴ (b) 99·9% dense 3YSZ with grain size of ∼100 nm sintered at 1200°C for 30 min (TEM) ⁵⁹

Vasylkiv and Sakka ⁵⁸ produced nanosized 3YSZ powder (∼10 nm) by hydrothermal precipitation, which was slip cast and pressed by CIP. Sintering at 1150°C for 60 h resulted in a nanograined (∼95 nm) ∼99% dense ceramic body. Later Vasylkiv et al. ⁹⁷ extended the investigation to 1·5–3YSZ and Y-TZP nanocomposites with 1–5 wt-% alumina also produced by colloidal processing. Bulk 3YSZ with an average grain size of 112 nm was shown to reach a hardness of 12·2 GPa and a fracture toughness of 9·3 MPa m^1/2. An alumina/zirconia nanocomposite with an average grain size of 94 nm was obtained with increased hardness of 16·2 GPa. Nanograined YSZ ceramics with a reduced yttria-stabiliser content were shown to reach fracture toughness of ∼14 MPa m^1/2 (2YSZ) and ∼12·5 MPa m^1/2 (1·5YSZ).

Zych and Haberko ⁵⁹ prepared green compacts by filter pressing of a nanopowder (∼8 nm) and sintered 30 min at 1200°C resulting in 99·9% TD dense, 3YSZ with mean grain size of ∼100 nm. Pradhan and Kapur ⁸² also attained nanograined fully dense 3YSZ from commercial nanopowders by colloidal processing. Aggregates were removed by centrifugation. Organics were eliminated by heating in steps from 100°C to 400°C with 30 min soaking at each temperature. Samples were sintered at 5°C min^− 1 up to 1200°C for 30–240 min resulting in full dense nanograined YSZ ceramics ( < 100 nm). Sintered ceramics exhibited Vickers hardness of 12·8 GPa as compared to 10·8 GPa for those produced from aggregated powders.

Finally, it is worth mentioning that a transparent, fully dense ZrO₂ ceramic with a grain size of 60 nm was obtained by Srdic et al. ⁵⁴ by sintering under vacuum at temperature as low as 950°C from nanocrystalline powder (∼5 nm) synthesised by the chemical vapour method. Nevertheless, this achievement was not included in Table 4, since it is a pure, undoped zirconia and the sintering was not performed at room pressure.

In summary, slow heating schedules in RHS have the desirable characteristics of simple setup, complex shapes handling, and low cost. However, densification rate is slow and a long time is needed to achieve highly dense bodies, which subsequently leads to grain growth. The most successful approaches, where densities above 99% TD were achieved and grain sizes are below 100 nm, rely on effective powder processing to eliminate agglomeration and allow homogenous green bodies. Low temperatures of calcination to avoid strong powder aggregation associated with milling with organic solvents to break agglomerates and an optimal compaction to induce small and homogenously dispersed porosities in the green body are therefore key elements in RHS to maintain grain sizes in the nanosized range. Other approaches to improve the performance of RHS for nanoparticles include adding sintering aids that form liquid phase at lower temperatures, or modifying the sintering schedule (for instance to FF or TSS cycles). The first option is out of the scope of this work. The second approach will be further discussed.

Fast firing

The objective of FF is to increase the densification rate in detriment to the coarsening rate by rapidly approaching the sintering temperature. Because coarsening mechanisms (e.g. surface diffusion and vapour transport) usually prevail over densification mechanisms (e.g. lattice and grain-boundary diffusion) at lower temperatures, it has been suggested that rapid heating to higher temperatures can be beneficial to achieve high density allied to fine grain size. ⁹⁸ In this case, the shorter time spent at lower temperatures serves to reduce the extent of coarsening, while densification is activated at high temperature conditions, resulting in high densities and fine microstructures. This mechanism is expected to be even more important when dealing with nanoparticles as the starting materials simply owing to the large surface areas of the powders, which tend to benefit surface diffusion and sample evaporation (this last due to the increased vapour pressure).

The conventional FF entails the quick insertion of a green body in and out of a preheated oven at the peak temperature and is arguably the simplest way to perform rapid densification. ⁹⁹ This technique has been employed since the 1970s to sinter both traditional ¹⁰⁰ and advanced ceramics ⁹⁸ and proven compatible to laboratory and industrial scale furnaces. ^101,102 Fast firing of ceramics has been successfully accomplished in continuous zone-sintering furnaces and in intermittent resistive box or tube furnaces. Recently, a commercial bottom-loaded, box furnace has been traded as ‘the world's fastest sintering furnace’, which was specially designed for sintering zirconia dental restorations in a total cycle of 10 min. ¹⁰³

In a typical laboratory box furnace schedule for FF designed to achieve nearly fully dense YSZ, the furnace is preheated at the peak temperature and the green body is rapidly moved into the hot zone. According to a numerical simulation performed for fast-fired alumina, the temperature field as a function of time for a middle cut of the compact is shown in Fig. 7. ¹⁰⁴ Considering a ceramic green bar (20 × 20 × 100 mm) placed into a furnace at 1500°C, the compact's centre takes about 120 s to reach 98% of the furnace temperature. Owing to the elevated temperature difference between the centre of the body and the furnace, the sample heating occurs under non-isothermal conditions as shown by the isotherms. The higher the temperature difference between the centre of the body and the furnace, the higher the energy transferred to the body. Thus, at the beginning of the process, with the insertion of the body at room temperature into the hot furnace, the heat rate is at its maximum. The radiative heat transfer is higher than the convective as expected due to the high temperature of the furnace and the low ?ow velocities inside the furnace. Radiation exchanges counts for more than 95% of the total heat rate transfer through the process.

OPEN IN VIEWER

7

Temperature field on the middle plane of a fast-fired ceramic body ¹⁰⁴

The thermal gradient inside the sample can therefore cause cracking problems during FF. That is, due to sintering, the outer regions of the sample will shrink faster than the inner regions, creating an internal stress that can lead to cracks and which will persist till the gradient of temperature vanishes. Note that even if cracks are not observed, large samples may present different microstructures when comparing outer and inner regions. Of course, materials with low thermal expansions and higher heat conductivity will be more suitable for FF.

The benefits of FF have encouraged applications of this technique to many advanced ceramics. Successfully fast-fired products have been reported for alumina, ^105–107 ceria, ¹⁰⁸ ferrites, ^109,110 barium titanate, ^111,112 indium tin oxide (ITO), ¹¹³ lead iron niobate (PFN), ¹¹⁴ lead magnesium niobate (PMN), ¹¹⁵ lead zirconate titanate (PZT), ¹¹⁶ yttrium aluminium garnet (YAG), ¹¹⁷ as well as composites, ^118,119 and glass ceramics. ^120,121

It is worth noting that some firing schedules referred to in the literature as ‘FF’ or ‘rapid sintering’ should not be strictly considered ‘fast’. That is, in some cases, although the heating rate is relatively high (e.g. 200°C min^− 1), it lies much below the highest rate reported for this process (∼500°C min^− 1); in other cases, although the rates are that high, the corresponding holding times at the top temperature are relatively long (e.g. 1 h) as compared to 1 min used in some cases. In addition, the cooling rate, or the time spent for the sintered products to reach room temperature, has received likewise much less attention. This longer runs disable most of the benefits of FF in terms of retaining grain growth and energy savings, making it more similar to RHS in many senses.

In the case of FF specifically on YSZ nanoceramics, only a few works have been reported in the literature. Table 5 presents a summary of the results from FF studies utilising nanosized powders and where densities of the final product are above 99% TD. The sintering elapsed times might be used as a criterion to evaluate how ‘fast’ the firing actually is.

Feng et al. ⁵⁶ used a hydrothermal processing to synthesise nanoscale and monodispersed 3YSZ powder with an equiaxed shape and an average particle size of ∼15 nm. The bulk density by FF at 1200°C for 1 min was 99·2% TD with a grain size of ∼90 nm. When fired at 1300°C for 1 min, the density was 99·7% TD with a grain size of ∼150 nm. In a second series of experiments, ⁵⁷ a 22 nm, non-agglomerated ZrO₂+2·8 mol% Y₂O₃ powder was synthesised, pressed at 100 MPa and rapidly heated at 500–1300°C min^− 1. The fast-fired samples yielded a bulk density 99·8% TD and a grain size of ∼100 nm. However, the disc shape specimens were very thin and the homogeneity of the final density was not shown.

Kim and Kim ⁸³ performed rapid sintering experiments at varying heating rates comparing a tube furnace to a microwave to elucidate the factors that hinder the densification of commercial zirconia powders. Three commercial YSZ nanopowders were used: tetragonal (3 mol% Y₂O₃), and cubic (6·6 and 8 mol% Y₂O₃). Compacts were first calcined at 600°C for 4 h and then sintered at 1430°C for 1 h. For those conditions, rapid heating resulted in the decrease of sintered densities in all three powders. The limited densification was attributed to entrapped gases caused by residual chlorine in the zirconia powders made by the chloride process. ¹²² In a second experimental run, the compacts were previously calcined at 1100°C for 1 h, and then sintered at 1430°C for 1 h. In this case, ∼99% TD were obtained for all three fast-fired YSZ powders. Kim and Kim ¹²³ also investigated the pore shrinkage behaviour during FF of 3YSZ powders. At temperatures above 1200°C, it was proposed that all the pores became unstable due to the grain growth and then started to shrink and disappear. No data on grain growth was presented, though. The more pronounced effect of rapid heating on the pore size distributions is that no pore growth was observed. Pores of smaller sizes rapidly shrunk and disappeared, and pores of larger sizes also started to shrink during FF.

Moreover, Qiu et al. ¹²⁴ synthesised 3YSZ nanopowder by coprecipitation and produced dense fine-grained ceramics either by RHS (heating rate of 3°C min^− 1 to 1250°C, and hold for 30 min) or by FF (500°C min^− 1 for both heating and cooling rates with 2 min hold). The samples presented 99·5% TD and ∼200 nm grain size, or 97·5% TD and ∼120 nm grain size, respectively for the slow and fast sintering schedules, suggesting that grain growth could not be separated from the densification at the final stage of sintering.

Interestingly, Chen and Mayo ¹²⁵ developed a similar work on nanocrystalline 3YSZ, which although presented an opposite trend. In this case, two starting nanopowders were used: one synthesised by coprecipitation and a commercial product. The powders were uniaxially (5 MPa) and isostatically pressed (250–300 MPa) to compacts with 47–50% TD and then pressureless sintered in air. The heating rates ranged from 2 to 200°C min^− 1. In the case of the highest rate, the samples were preheated in a box furnace and then pushed manually into a tube furnace to the peak temperature area with a dwell time of 4 min. The results showed that the heating rates had virtually no effect on the densification or grain growth of 3YSZ powder compacts. Nevertheless, the heating rates were considered only up to 200°C min^− 1 and the pretreatment carried out (500°C for 4 h) might have not been enough to vaporise impurities entrapped in the compact.

Chen and Mayo ¹²⁵ discussed the problem concerning the gradient of densification induced by the high heating rates. Figure 8a shows SEM micrographs of a 3YSZ nanocrystalline sample partially sintered by heating at 20°C min^− 1 to 1300°C and held for 2 h. A differential densification can be seen where the surface of the sample is nearly dense, while there is increasing porosity when approaching the sample centre. While the authors suggest that the dense shell could constrain the interior, decreasing densification, other works have shown that this is not the case for very fast heating rates. For instance, Fig. 8b shows a TEM micrograph of a 2·8YSZ nanocrystalline sample fully sintered by FF at 500°C min^− 1 to 1300°C and held for 1 min with a grain size of about 100 nm. This suggests that particle packing and agglomeration control are key elements to avoid the observed constrained sintering. Nevertheless, FF is certainly more suitable for processing of small or thin walled articles. When this method is applied to large parts, a thermal gradient is more likely expected to develop between the surface and the core of the sample, subsequently retarding the densification of the interior of the body. ¹²

OPEN IN VIEWER

8

(a) Sintering front in a partially sintered 3 mol%YSZ (3YSZ) (SEM). ¹²⁵ (b) Full dense 2·8YSZ with grain sizes of ∼100 nm (TEM) ⁵⁶

Though FF is still considered a technologically powerful tool for YSZ densification, being used in orthodontic prosthesis in-office manufacturing, data collected in Table 5 suggests a significant grain growth is still observed when high densities are of interest. This is because grain growth is always present and with high motilities due to the high temperatures of processing. To go around this, FF for very short times has been used as a starting point to another process to achieve dense nanograined samples, as described in the next Section.

Two-step sintering

Two-step sintering (TSS) is a thermal schedule that has been proposed to promote high densification of ceramics with hindered grain growth. The key features in this method are: exposing the sample to a temperature (T ₁) enough to activate densification mechanisms and allow sample to achieve >75% TD; and lower the temperature of about 100–150°C (T ₂) to induce densification with minimal grain growth. ²⁷ The limited grain growth during the second step in TSS while densification continues has been attributed to a grain boundary network pinned by triple-point junctions that have higher activation energy for migration than the grain boundaries. ⁸¹ Thus the feasibility of densification without grain growth was thought to rely on the suppression of grain-boundary migration while keeping grain-boundary diffusion active. Fig. 9 presents a typical TSS curve compared to a conventional RHS cycle or one-step sintering.

OPEN IN VIEWER

9

Typical furnace schedule for two-step sintering (TSS) of near fully dense yttria-stabilized zirconia (YSZ) compared to conventional ramp and hold sintering (RHS)

Wang et al. ²⁸ suggested a ‘kinetic window’ separating grain-boundary diffusion and grain-boundary migration. At temperatures above the kinetic window, grain growth occurs for which the driving force diminishes as the grain size increases. At temperatures below the kinetic window, sintering is exhausted before full density is achieved. Thus, TSS can be used to reach TD >98% by exploiting this kinetic window. Once the critical density at T ₁ is reached, the lower temperature hold at T ₂ will place samples within the ‘kinetic window.’ When conditions for TSS fall below this point, values of TD >96% cannot be achieved even if a starting density of 70% is obtained at T ₁, as grain growth may still be suppressed but densification will be exhausted. Above the ‘kinetic window,’ grain growth is likely to occur. ²⁷

Two-step sintering has been efficiently used to fabricate various dense advanced ceramics such as Y₂O₃, ^27,28 SiC, ¹²⁶ Al₂O₃, ¹²⁷ BaTiO₃, ^29,128 ZnO, ^129,130 lead titanates, ^131,132 and alkaline niobates. ^133,134 A typical TSS schedule is usually performed in conventional resistive furnaces, but some two-step experiments in alternative techniques such as microwave sintering have been reported.

It has been reported that the grain growth can be retarded and full density can be achieved by using TSS schedules for nanocrystalline ceramics as well. According to the original TSS schedule proposed by Chen and Wang, ²⁷ fully dense Mg-doped Y₂O₃ was obtained by TSS at 1000°C for 20 h, after reaching 76% TD in an initial sintering at 1080°C. In this case, starting powders of 10 nm were used and a final dense ceramic grain size of 60 nm was reached.

Numerous authors have reported on TSS of zirconia nanopowders. In most cases, however, neither the densities obtained were >99% TD nor the nanocrystalline nature of the green body was preserved. ¹³⁵ Lee ⁸⁵ used 3YSZ commercial powder, which was injection-moulded and sintered at 1500°C (T ₁), held for 5 min, and then rapidly cooled down and maintained at 1300°C (T ₂) for 10 h. The final density of the two-step sample was 99% TD, but the average grain sizes reached 590 nm. Yu et al. ⁸⁶ applied TSS (T ₁ = 1100–1500°C for 6 min; T ₂ = 900–1000°C for 1 h) for injection-moulded 3YSZ parts. The sintered specimens presented a grain size of ∼500 nm and density from 94–99% TD. Chen et al. ¹³⁶ sintered a nanosized 3YSZ powder at 1200°C for 1 min and then at 1050°C for 35 h, and reached a relative density of ∼98% TD and a grain size of 100 nm. Xiong et al. ¹³⁷ applied TSS in the consolidation of 3YSZ commercial nanopowder. Relative densities of 97% TD and grain sizes of ∼170 nm were obtained by RHS (1250°C for 1 h) and TSS (T ₁ = 1275°C for 1 min, T ₂ = 1150°C for 30 h). Mazaheri et al. ³⁹ obtained nearly fully dense 3YSZ compacts (∼99% TD) with an average grain size of 110 nm after sintering at 1300°C (T ₁) and 1150°C (T ₂) for 30 h. Suárez et al. ¹³⁸ sintered slip-cast 3YSZ bodies with 99·9% TD and an average grain size of 125 nm after an optimised TSS schedule (T ₁ = 1300°C for 1 min, T ₂ = 1200°C for 15 h).

Mæland et al. ⁶⁵ sintered 4YSZ nanoparticles using a modified TSS proposed by Han. ¹³⁹ The optimised schedule (T ₁ = 1400°C; T ₂ = 1300°C for 20 h) corresponded to 96% TD and a grain size of 200 nm. Lena et al. ⁶⁶ submitted a 4·5YSZ powder synthesised by the Pechini method to a TSS schedule with heating rate of 8°C min^− 1 to 1550°C (T ₁), cooling rate of 50°C min^− 1 to T ₂ : 1300°C (2 h), 1300°C (10 h) or 1400°C (2 h). The highest density (97% TD) was observed for samples treated at 1400°C (T ₂) for 2 h, which showed a submicrometric grain size.

Laberty-Robert et al. ⁶⁷ obtained 98% TD 8YSZ bodies from nanocrystalline powders by TSS (T ₁ = 1500°C; T ₂ = 1450°C for 4–20 h) with grain sizes from 1 to 5 μm, though. Han ¹³⁹ employed a modified TSS to sinter 8YSZ tapes from synthesised nanopowders (10–15 nm). The temperature was raised up to 1400°C, and then decreased to 1250–1300°C for 10–20 h. Samples with grain sizes in the range of 100–400 nm and 99% TD were obtained. Ghosh et al. ⁶⁹ obtained 8YSZ ceramics with 97% TD after a TSS schedule (T ₁ = 1125°C for 3 h; T ₂ = 1090°C for 20 h). However, the grain size ranged from 150 to 220 nm. Durá and López de la Torre ¹⁴⁰ sintered 8YSZ by TSS (T ₁ = 950–1150°C; T ₂ = 750–950°C) resulting in grain sizes ∼150 nm and densities up to 92% TD. Mazaheri et al. ^70,71 employed TSS to consolidate nanocrystalline 8YSZ. The optimised schedule (T ₁ = 1250°C; T ₂ = 1050°C for 20 h) corresponded to 97% TD and 295 nm. ⁷² Muccillo and Muccillo ¹⁴¹ obtained sintered 8YSZ specimens with >97% TD and grain size of 580 nm by TSS (T ₁ = 1330°C; T ₂ = 1230°C for 20 h).

Table 6 summarises in the processing parameters for successfully obtaining >90% TD nanocrystalline YSZ ceramics through TSS. Some studies compare different sintering techniques to TSS and their data are also included. Interestingly, only compositions up to 3 mol% yttria-doped zirconia could be successfully transformed into nanocrystalline materials. This will be further discussed in the next Section on microstructural development.

Paul et al. ⁸⁰ investigated the influence of grain size and yttria content on the hydrothermal aging behaviour of YSZ nanoceramics for biomedical applications. Slip cast green bodies were heated at 10°C min^− 1 to 1150°C (T ₁) for 6 s and then cooled down to 1050°C (T ₂) where the samples were held for up to 10 h. When compared to a dry pressed commercial 3YSZ powder which was conventionally sintered (1500°C, 2 h), the TSS samples achieved nearly full densification with very limited grain growth. The nanostructured sample (Nano-3YSZ with  < 100 nm grain size) did not exhibit degradation under hydrothermal aging, whereas submicrometric 3YSZ samples underwent severe degradation.

Binner and Vaidhyanathan ⁸¹ prepared 3YSZ green bodies by die pressing (∼50% TD) and slip casting (∼55% TD), which were sintered using both a RHS and a TSS approach. The conventionally sintered samples were heated at 7°C min^− 1 to 900–1150°C and held for up to 8 h. The TSS samples were heated at 7°C min^− 1 to 1150°C and immediately cooled down to 1000–1050°C for holding periods of up to 30 h. Both radiant and hybrid-microwave/radiant furnaces were used in either dry or wet processed samples. While densities >98% TD were achievable with all procedures, an average grain size below 100 nm could only be obtained when applying TSS, being the smallest size achieved when using the hybrid heating. That is, the bulk of the grain growth observed using this approach occurred during heating to T ₁. With radiant heating, 7°C min^− 1, the final average grain sizes were 95–100 nm; with the hybrid heating, 20°C min^− 1, grain sizes were in the range of 70–80 nm for die pressed samples and ∼65 nm for the more homogeneous slip cast samples.

Santacruz et al. ³⁵ produced 3YSZ slip cast bodies from commercial nanosuspensions. After the removal of the organic dispersants at 500°C for 2 h, the green samples were sintered in an electrical furnace by TSS. This involved heating the samples to 1150°C at 20°C min^− 1 and holding them for 1 min before the temperature was reduced to 1000°C, and then holding at this temperature for 5 or 10 h. Figure 10 shows the nanostructures of the TSS sintered samples: 99% TD ceramics were obtained with an average and uniform grain size of 80 and 90 nm after 5 and 10 h of soaking time (Fig. 10a and b , respectively).

OPEN IN VIEWER

10

FEG-SEM micrographs of three yttria-stabilised zirconia (3YSZ) nanocrystalline samples prepared by TSS (T ₁ = 1150°C; T ₂ = 1000°C at with soaking times at T ₂ of: (a) 5 h, and (b) 10 h ³⁵ )

In a further work, Binner et al. ³⁴ prepared 1·5, 2 and 3YSZ green bodies by die pressing (∼54% TD) and slip casting (∼52% TD), using TSS by hybrid (microwave/conventional) heating. The optimum sintering conditions of 6 s at 1150°C (T ₁) followed by 3 h at 10501C (T ₂) for the 3YSZ slip-cast bodies yielded densities of >99·5% TD and a final mean grain size of  < 65 nm. However, for 1·5YSZ and 2YSZ, it was found that the TSS cycle needed modification to allow for the variation in yttria content. In general, the lower the yttria level, the less demanding the sintering conditions required, which was attributed to the influence and magnitude of the associated grain-growth process.

In summary, TSS schedules may retard the grain growth of YSZ. Similarly to FF, a very short heating cycle up to the maximum sintering temperature (T ₁) is first applied. However, differently from FF, TSS needs a dwell time of at least 3 h at the lower temperature (T ₂) for complete densification, even when a microwave furnace is used instead of a conventional one.

Powder characteristics

Nanopowders are normally synthesised at low temperatures or from non-equilibrium processes, which can lead to metastable particles of different phases of the desired material. ⁶⁰ After synthesis, removal of organics and phase transformation to stable structures are typically needed. However, those processes are usually accompanied by microstructural changes, which might be detrimental to the final sintered product. For instance, decomposition or phase transformation may induce volume changes and introduce substantial amount of porosity and cause poor densification during sintering. ¹² When prolonged sintering is used to remove pores, grain coarsening may set in.

Particle size distribution is another critical element affecting microstructural evolution during densification of nanoparticles. That is, nanopowders may present wide particle size distribution with some particles even reaching submicron sizes. A few large particles do not affect the packing density substantially but may lead to abnormal grain growth during sintering. ¹⁴² On the other hand, very narrow distributions are also not desirable, as packing efficiency is decreased because of the high frictional force between nanoparticles, ¹² making more difficult to achieve high green densities needed for enhanced densification.

While nanoparticles have inherent high surface areas, theoretically enabling a high driving force for densification, the very same high reactivity may induce agglomerations and aggregations, or even partial sintering, before densification starts, creating undesirable heterogeneities and/or limiting densification. ¹⁴³ While soft agglomerates do not affect significantly densification behaviour, hard agglomerates, formed typically during powder synthesis or pre-treatments at moderate temperatures, can strongly affect sintering. In those cases, as it is sometimes impractical to break hard agglomerates even when making usage of highly energetic treatments, agglomerate sizes (and hence the interagglomerate pore size) rather than the nanoparticles size are typically expected to govern the densification of compacts. When agglomeration causes loss of contacts in a particle packing before sintering, a poorly packed region can mature into a critical defect. ¹²

Regarding specifically the fabrication of YSZ nanocrystalline ceramics, the most successful starting nanopowders employed are those synthesised by sol-gel, ⁵⁵ hydrothermal, ⁵⁹ or coprecipitation routes. ⁶³ In those cases, powders with a narrow pore size distribution and particle sizes  < 10 nm were obtained (small soft agglomerates), comparing to commercial powders which usually present crystallite sizes >20 nm and particle sizes in the range of 40–75 nm. ^38,82 When commercial powders are used, even a low grain growth factor during sintering (∼5 times) may lead to crystalline ceramics out of the nanometric range.

Packing and pore size distribution

Agglomerated powders are typically difficult to densify because the packing of large agglomerates leaves behind large and interagglomerate pores. In this way, no matter how small the crystallites inside the agglomerate are, the size of the agglomerate dictates the size of the interagglomerate pores and, ultimately, the densification behaviour of the ceramic. ¹⁴⁴

When compacted, a powder composed of agglomerates contains both small intercrystallite pores and large interagglomerate pores. The latter pores require long sintering times and/or high sintering temperatures to be eliminated, which are conditions typically associated with grain growth. On the other hand, the very same interagglomerate pores, show slow shrinkage kinetics, and therefore are usually the last pores remaining in the ceramic during final-stage sintering. As such, they act as pinning sites preventing grain boundary movement, and their spacing ultimately determines the final sintered grain size. ⁶⁰

Trunec and Maca ³⁷ showed the importance of green body microstructure for the pressureless RHS of YSZ nanoparticles. A detailed investigation of pore size evolution during sintering indicated that dense nanocrystalline zirconia ceramics could be prepared only with a ceramic green body with both small pores and a homogeneous structure. Inhomogeneous green bodies, even with small pores, prevented the densification due to local shrinkage and subsequent pore growth. Therefore, zirconia nanoparticle compacts with the required microstructure (maximum pore size 5–10 nm, homogeneous structure) were densified by pressureless sintering at 1100°C into dense nanocrystalline bodies (diameter 20 mm, thickness 4 mm) with a relative density higher than 99% and a grain size of about 85 nm.

Densification and grain growth

The success of heating schedules in controlling the microstructural evolution of ceramics during sintering comes from the fact that the different mass flow mechanisms occurring at high temperatures will show distinct activation energies. Therefore, specific mechanisms can be activated while others can be suppressed in order to achieve desirable properties (microstructures) of the final product. ¹⁴⁵ The main competitive processes are densification and grain growth, this last being of course undesirable for attaining nanograined structures.

In general, nanoparticles show low sintering activation energy. For instance, the densification activation energy itself of nanosized YSZ (237 kJ mol^− 1) is about 50% lower than that of submicron YSZ. ⁵⁵ On the other hand, the activation energy for grain growth in nanoparticles is also much lower than that in micrometric particles (e.g. ∼20 kJ mol^− 1 for ZnO nanoparticles, which is one order of magnitude lower than the activation energy of densification for this material). ¹⁴⁶ For nanosized YSZ particles, the reported energy activation for grain growth was even lower (∼5 kJ mol^− 1 for 8, 9 and 10YSZ). ¹⁴⁷

Maca et al. ¹⁴⁸ analysed the influence of sintering schedules (TSS and RHS) on the final microstructure of commercial 3YSZ and 8YSZ powders. The results showed that the efficiency of TSS was more dependent on crystal structure than on particle size and green body microstructure. For tetragonal 3YSZ, both TSS (T ₁ = 1305°C, T ₂ = 1275°C, t ₂ = 10 h) and RHS (1405°C, no dwell time) reached a density of ∼99% TD and a grain size of 160–170 nm. On the other hand, TSS (T ₁ = 1440°C, T ₂ = 1290°C, t ₂ = 15 h) was more successful in sintering cubic 8YSZ when compared to RHS (1530°C for 1 h). Nevertheless, for a density of ∼99·3% TD, relatively large grain sizes from 3·0 μm for RHS and 1·3 μm for TSS were produced for 8YSZ ceramics.

Lourenço et al. ¹³⁵ proposed a method to optimise TSS schedules of commercial nanopowders (3YSZ and 8YSZ) that combines predictions of the sintering kinetics with Taguchi fractional experimental plans. The controlled processing parameters were T ₁, T ₂, time at T ₂ (t ₂), and the cooling rate between T ₁ and T ₂. Dense 3YSZ was obtained with an average grain size of ∼100 nm under the conditions T ₁ = 1320°C, T ₂ = 1250°C, t ₂  = 12 h and 20°C min^− 1. Grain growth is inhibited to a lesser extent for 8YSZ, with the smallest average grain size of 450 nm obtained for T ₁ = 1320°C, T ₂ = 1270°C, t ₂ = 12 h and 20°C min^− 1.

In both cases, 8YSZ is notably harder to densify at reasonably low grain sizes when compared to 3YSZ. The reason may be on local phase transitions of the partially stabilised 3YSZ that can increase reactivity and pin boundary from growing, or certainly different levels of impurities in each sample (such as traces of alumina). Nevertheless, TSS schedules were applied with relative success to hinder grain growth in cubic zirconia ceramics; less success was reported for hexagonal alumina and no significant effect of refining the microstructure of sintered body was recorded in the case of tetragonal zirconia ceramics subjected to TSS. ¹⁴⁸ In that particular study, it was suggested that the grain growth factor was lower for increasing ceramic crystal symmetry, which can be linked to interfacial energies and grain growth driving forces and activation energies.

A further investigation on the effect of zirconia phase structure (monoclinic, tetragonal and cubic) on the sintering behaviour of zirconia nanopowders (∼13 nm) was performed by Trunec et al. ¹⁴⁹ Green bodies compacted by CIP were densified by pressureless RHS at a heating rate of 5°C min^− 1 up to 1150°C with no dwell at the peak temperature. In this case, for highly dense bodies (99% TD), tetragonal 1·5YSZ and cubic 8YSZ presented respective mean grain sizes of 170 and 180 nm, whereas monoclinic pure zirconia exhibited a more difficult densification and coarser microstructure (∼570 nm). The similarity between the tetragonal and cubic phases is remarkable and contrasts with TSS results. However, caution should be exercised when direct comparing those results as they start with different nanopowders.

Indeed, the distinct densification behaviour for samples with different yttrium contents can be related to the defect concentration, nature, evolution during the densification routine, and particularities of the defect interactions at the nanoscale. While these are typically difficult parameters to quantify, recently positron annihilation spectroscopy (PAS) has been proposed as a non-destructive investigation technique that can shed some light into this hypothesis. ¹⁵⁰ For instance, PAS has been used for estimating the mean grain size from conventional positron lifetime technique ¹⁵¹ and displaying trends in microstructure evolution which take place during sintering of YSZ ceramics at different temperatures. ¹⁵²

Earlier studies on the control of grain growth during sintering of YSZ have also shown that a transient stage of abnormal grain growth, which is determined by the microstructural situation after compaction, occurs in the non-isothermal temperature range, between 1100 and 1200°C. ¹⁴² In order to maintain the nanocrystalline nature of YSZ ceramics produced from nanograined powders, the stage of abnormal grain growth should be kept as short as possible. In this regard, FF schedules seem to be a promising method to reach fine-grained, dense YSZ ceramics.