- Email: [email protected]
The effect of powders homogenisation conditions on the synthesis of yttrium aluminium garnet (YAG) by a solid-state reaction
Author’s Accepted Manuscript The effect of powders homogenisation conditions on the synthesis of yttrium aluminium garnet (YAG) by a solid-state reaction Łukasz Zych, Radosław Lach www.elsevier.com/locate/ceri
PII: DOI: Reference:
S0272-8842(16)32144-7 http://dx.doi.org/10.1016/j.ceramint.2016.11.142 CERI14230
To appear in: Ceramics International Received date: 14 September 2016 Revised date: 12 November 2016 Accepted date: 21 November 2016 Cite this article as: Łukasz Zych and Radosław Lach, The effect of powders homogenisation conditions on the synthesis of yttrium aluminium garnet (YAG) by a solid-state reaction, Ceramics International, http://dx.doi.org/10.1016/j.ceramint.2016.11.142 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
The effect of powders homogenisation conditions on the synthesis of yttrium aluminium garnet (YAG) by a solid-state reaction Łukasz Zych*, Radosław Lach AGH-University of Science and Technology, Faculty of Materials Science and Ceramics, Mickiewicza 30, 30-059 Krakow *
Corresponding author: AGH-University of Science and Technology, Faculty of Materials Science and Ceramics, Mickiewicza 30, 30-059 Krakow, Poland. Tel.: +48 12 6172480; fax: +48 12 6334630. [email protected] Abstract The work presents results of research on the influence of the state of homogenisation of reagent powders i.e. aluminium oxide and yttrium oxide on the synthesis of yttrium aluminium garnet. Fine powders of commercial aluminium oxide and yttrium oxide were mixed in the same manner in a laboratory ball mill but under four different conditions. Three batches of powders were prepared in water of various pH values, selected on the basis of the zeta potential measurements, and one batch was prepared in propanol. They were consolidated by filter pressing and then sintered in air at a temperature from 900 C to 1700 C for 1 hour, with heating rate of 5 C/min. Samples were subjected to dilatometric studies, phase composition determination, measurement of apparent density and pore size distribution, as well as a microscopic examination. Mixing environment affected, to a given extent, almost all the features of green, and partially sintered materials as well as course changes of linear dimensions. As a result, the relative densities of samples sintered at a maximum temperature ranged from 58.31 to 98.25 %. The highest density was achieved for the material originating from the suspension having a pH of 9.5, in which heterocoagulation occurred. Keywords: mixing (A); powders: solid-state reaction (A); suspensions (A); yttrium aluminium garnet
1.
Introduction
1
Yttrium aluminium garnet (Y3Al5O12, YAG) is one of the most popular transparent ceramic material which main application is the host of solid state lasers. For this purpose, YAG can be doped with rare-earth elements such as: Nd, Er, Yb or Tm [1-3]. The main types of YAG synthesis are: the solid state reaction between Y2O3 (yttria) and Al2O3 (alumina) powders [4], and the direct synthesis of powders using methods such as co-precipitation [5, 6], citrate process [7, 8] and a sol-gel method [9, 10]. Translucent YAG ceramics was first obtained by the solid state reaction by De With and Van Dijk [4]. The production of highquality polycrystalline YAG and Nd-doped YAG ceramics by reaction in the solid state between powders of yttria and alumina was first described by Ikesue et al. in 1995 [1, 11]. Since that time, attempts have been made to better understand the reaction and the factors controlling its course. One of the most important factors is the characteristics of the initial powders [12-14], and this parameter is related to the homogeneity of powder mixture, which in turn governs the course of YAG synthesis. The homogeneity of the powders mixture can be improved when the mixing is carried out under heterocoagulation (or heteroflocculation) conditions, which means that particles of dissimilar electric charge are attracted to each other [15]. A measure of the electric charge of the powder surface is zeta potential, which the sign and the value is dependent on the pH of the suspension and the presence of some adsorbing species in the suspension. Suitable conditions for heterocoagulation to take effect occur within a specific pH range, and in the processing of ceramics this can be used the manufacturing of composites [16, 17] as well as for the preparation of powders mixtures for solid state reactions [18, 19]. On the other hand, Konsztowicz et al. showed that in zirconia alumina composites heteroflocculation of the constituent powders detrimentally affects homogeneity of their microstructure which in turn influences their mechanical properties [20]. Problems in the wider application of this effect are created by: availability of powders with the adequate zeta potential characteristics, too 2
narrow pH range etc. They may be partially bypassed by modifying the powders with additions of organic molecules adsorbing on the powder surface [20]. The paper presents the results of studies on the influence of heterocoagulation effect on properties of YAG synthetized by a solid-state reaction. To the authors’ best knowledge, such investigations have not been conducted so far. In order to observe the effects of mixing conditions only, no additives were used to enhance the sintering of YAG, such as SiO2 or TEOS [4, 11], and the samples were sintered in the air.
2.
Materials and methods Tests were performed using a commercial powders with a purity of 99.99 %: yttrium
oxide (32-36 nm APS, #5650YS, Nanostructured & Amorphous Materials Inc.) and aluminium oxide (Taimicron TM-DAR, Taimei Chemicals Co. Ltd.). Characterisation of the powders included measurement of specific surface area by the BET method (Nova 1200e, Quantachrome Inc.), observation of the powders by a transmission electron microscope (JEM 1011, Jeol.) and the determination of particle size distribution by a laser diffraction method (Mastersizer 2000, Hydro S, Malvern Inc.). The quantitative analysis of particle size distributions of the powders was performed in the diluted suspensions prepared by high-energy ultrasonication with a small addition of a dispersant (Dispex N40). Thermal analysis was used to determine mass changes of the powders occurring during heating. Measurements were performed using DTA/TG method and STA 449 F3 apparatus (Netzsch) with a heating rate of 10 C/min and the maximum temperature of 1200 C. The relationship of the zeta potential on the pH was determined by a Zetasizer NanoZS equipped with a MPT-2 autotitrator unit (Malvern Inc.). In order to fix the ionic strength of the suspensions, measurements were performed in 0.01 M NaCl solution. The pH was 3
controlled by addition of 0.5 M HCl or 0.5 M NaOH solution (POCh, Poland), respectively. The results of the measurements of zeta potential enabled to select three different homogenisation conditions of the powders in aqueous suspensions. Additionally, in order to completely eliminate water from the system, one batch of powders was mixed in iso-propanol (POCh, Poland). The powders were mixed in a ratio of 42.94 wt% of Al2O3 and 57.06 wt% of Y2O3 which leads to the synthesis of yttrium-aluminium garnet. Mixing of the powders was performed overnight in a laboratory ball-mill using a high-quality alumina balls of 5 mm diameter. The particle size distribution of the powders mixtures was determined by laser diffraction method (Mastersizer 2000, Malvern Inc.). Before the measurement, the suspensions were diluted and dispersed using a high-energy ultrasounds during 3 minutes. The suspensions were consolidated using the technique of pressure filtration and samples of 30 mm diameter and 5 mm thickness were produced [21]. A given powder suspension was pressed against the ceramic porous support covered with a nylon membrane with a pore diameter of 0.2 μm (Whatman). Filtering pressure was 10 MPa and it was kept constant until no leakage of supernatant. The green samples were dried for a few days at room temperature in a dessicator over silica gel, and then overnight at 110 C in oven. The sintering of samples was studied using a DIL 402 C dilatometer (Netzsch) with a heating rates of 5 C/min and 10 C/min and a maximum temperature of 1550 C. The pore size distribution of materials was determined by means of mercury porosimetry using Poremaster 60 (Quantachrome Inc.). In order to exclude the effect of the presence of water in the samples on their pore size distribution the measurements were carried out after heating the samples at 900 C for 30 min. The microstructure of the samples was observed using a scanning electron microscope Nova Nano-SEM 200 (FEI Co). In order to investigate the changes in phase
4
composition during the heat treatment, the selected material was heated at a temperature ranged from 900 to 1700 C for 1 hour, with a heating rate of 5 C/min. The final sintering of samples was conducted in air at 1700 C with heating rate of 5 C/min and the dwelling time of 3 hours. The apparent density of the samples was measured by the Archimedes method, and the microstructure of the sintered samples was observed using SEM method. The phase composition was determined by X-ray diffraction method (XRD, Empyrean, PANanalytical) and Rietveld refining method.
3.
Results and discussion
3.1. Powders characterisation The primary particle size (crystallites) of powders was determined from their specific surface area (SSA), which was 36.6 and 12.9 m2/g for Y2O3 and Al2O3, respectively. The equivalent diameters of the spherical particles, calculated on this basis (dBET), were 33 and 117 nm, respectively. Crystallite sizes, determined on the basis of X-ray diffraction lines broadening (dXRD), were 35 nm and 107 nm for Y2O3 and Al2O3, respectively. Good agreement between values of dBET and dXRD of both powders indicated that there were no broad phase contacts formed between the primary particles (crystallites). The yttria powder tended to agglomerate (Fig. 1a) and the modal diameter of the agglomerates, determined from the particle size distribution was 3 μm (Fig. 2). In contrast, the alumina powder was practically non-agglomerated, and consisted of single crystallites of submicrometric size (Fig. 1b). In this case, the modal particle diameter i.e. 129 nm (Fig. 2), corresponded well to the diameter of the primary particles, calculated on the basis of the specific surface area of the powder, i.e. 117 nm.
5
It means that the actual particle size of yttria (agglomerates) involved in the synthesis of YAG was at least 100 times greater than its crystallite size. In general, the nanoparticles exhibit a strong tendency to form agglomerates during the synthesis and handling, and a significant degree of agglomeration observed in yttria powder was probably caused by a smaller crystallite size (32 nm) compared with alumina powder (107 nm). The modal diameter of the agglomerates of yttria was approximately 30 times greater than the diameter of the alumina particles, which in turn can affect the efficiency of mixing of powders and finally the course of the reaction of synthesis. Another issue, related to nanopowders, is the amount of bounded water, which is particularly important in the case of powders used as reagents for the solid state synthesis [22]. It was found, that the weight loss of yttria powder during heating up to 1200 °C, as determined from the DTA / TG measurement, was 9.67 %, and the removal of water proceeded in a two-step process (Fig. 3). Most of the water (ca. 6 wt%), which was rather loosely adsorbed on the surface of the powder particles was removed up to 300 C. Another, smaller mass loss (c.a. 2 wt%) occurred between 800 C and 1200 C, which may be related to the removal of some strongly bound water which origin was not clear. In the case of alumina powder, probably due to its lower specific surface area, the total amount of water (mass loss) was only 0.60 % (not shown here). Preliminary experiments showed that in order to synthesise a pure phase of YAG, it was crucial to use an amount of reagent powders of the total weights corrected for mass loss. As the mixing of reagent powders was carried out in aqueous environment of different pH values, it influenced also the measured values of zeta potential (Fig. 4).
6
The zeta potential vs. pH relationships for the powders were different, with the isoelectric points (iep) at pH 8 and pH 10.9 for Al2O3 and Y2O3, respectively. In most cases, isoelectric point of Y2O3 at a pH of about 8 has been reported, however, the iep at a pH of 11 has been also found [23, 24]. This implies, that the pH window, in which the zeta potential of the alumina and yttria powders had opposite signs and wherein heterocoagulation could occur, was in the range between pH 8 and 10.9. The maximum difference between the values of the zeta potential of powders occurred at pH 9.5. The observed small decrease in the value of the zeta potential at pH 6 accompanied with a much higher variability of results probably resulted from the partial dissolution of the yttria powder. The yttria powder had a nanometric size and thus was more reactive and readily soluble under acidic conditions. Finally, the powders were mixed under four different conditions: three aqueous solutions of different pH values and iso-propanol which was used to completely exclude water from the system (in this latter case, the zeta potential was not measured, Table 1.).
Table 1. Mixing conditions of the reactive powders Batch number
mixing conditions
zeta potential of Y2O3 particles, mV
zeta potential of Al2O3 particles, mV
1
water, pH 7
+ 51
+ 45
2
water, pH 9.5
+ 48
- 45
3
water, pH 12
- 30
- 50
4
iso-propanol
n/a
n/a
All batches of powders were prepared under the same conditions, i.e. by mixing overnight in a laboratory ball mill with high purity alumina balls of 5 mm diameter. The volume fraction of
7
solid phase in the suspensions was ca. 12 %, and the weight ratio of powders to the balls was 1:10. The initial pH of the suspension of the reagent powders was about 9, and it was adjusted to the desired level by addition of 1 M HCl solution or 10 % TMAH solution (tetramethylammonium hydroxide, Acros Organics). Both, HCl and TMAH should evaporate without a trace from the material during heating. The pH of suspensions was checked several times at the beginning but also at the end of the mixing process. Generally, the pH of the suspensions was stable during the mixing process, with the exception of the suspension having a pH of 7. In this case, it was difficult to stabilise the pH value, as it quickly returned to a value of 9, indicating that small additions of HCl resulted in some changes to the reagent powders. Most probably, the HCl additions led to a slight dissolution of the yttria powder, because the instantaneous values were much lower than pH 7, and the powder consisted of nanometric, and thus reactive particles. The volume fractions of the powders used in the YAG synthesis were similar and equalled to 48.7 % and 51.3 % for Al2O3 and Y2O3, respectively. Consequently, the particle size distribution of the mixture should be similar to the superposition of the particle size distributions of individual powders (Fig. 2). However, the particle size distributions of the powder mixtures showed distinct differences (Fig. 5). The main populations of particles present in the powder mixtures corresponded to those present in the starting powders (Fig. 2) but their proportions varied between the mixtures (Fig. 5). Both populations of powders were clearly visible in the mixtures prepared at pH 7 and 12 and in iso-propanol. In the case of mixtures prepared at pH 12 and in iso-propanol, the particle size distributions were almost the same and reassembled the superposition of the distributions of starting powders. Such a distribution indicated that no interaction between the particles of yttria and alumina occurred. A certain level of inter-particle interaction became visible in the case of powders mixture 8
prepared at pH 7, as here the proportions between the populations have changed, and the population of larger particles started to be dominant. The reason of such behaviour was not clear but it probably was related to the observed partial dissolution of the yttria powder. It is also probable, that the effect of a partial agglomeration of the alumina particles appeared. A possible explanation for this is that, due to the increasing ionic strength of the solution caused by the presence of yttrium ions, the absolute value of the zeta potential of the powders decreased, thus resulting in the agglomeration of alumina powder. In contrast to other powders mixtures, the particle size distribution of the mixture prepared at a pH of 9.5 was monomodal, with the modal size corresponding to the size of yttria powder particles (Fig. 2). It may indicate, that in such pH conditions, heterocoagulation effect occurred (Fig. 4) and the smaller particles of alumina were attached to larger particles of yttria (Fig. 2). In summary, mixtures of powders had different particle size distributions resulting from the conditions of their mixing and possible interactions between the particles. 3.2. Materials characterisation The microstructure of filter pressed samples prior to the reaction of yttria and alumina was observed by means of the pore size distribution and SEM observations. Preliminary tests showed that the samples heat-treated at 900 C for 30 min were suitable for such characterisation. The pore size distributions of all samples were monomodal with both modal pore diameter (Fig. 6b) and total pore volume (Fig. 6a) depending on the type of the powders mixture. The modal pore diameters were much smaller than the size of the agglomerates of yttria powder (~ 3 μm) (Figs 1a, 2), which means that the measured pores existed inside the agglomerates and the inter-agglomerate pores are not present. This in turn indicates that the compressive forces acting during the pressure filtration are sufficient to deform/compress 9
agglomerates of yttria powder. The smallest modal pore size (62 nm) and the lowest total pore volume (~ 0.35 cc/g) were present in a sample obtained from the suspension of pH 7. Total pore volumes of other samples were similar and ranged from 0.49 to 0.55 cc/g for samples derived from the propanol suspension, and the suspension of pH 12, respectively (Fig. 6b). Real density of the materials heat treated at 900 C was 4.59 g/cm3, which was calculated based on the volume fractions of the powders and their real densities. Relative densities of the materials corresponded to their total pore volumes and they were as follows; 37.9 % for the samples derived from suspension of pH 7, 33.0 % for the suspension of pH 9.5, 32.0 % for the propanol suspension and 30.5 % for suspension of pH 12. The modal pore diameters of these samples were similar and close to 100 nm (Fig. 6b), which corresponded well to the appropriate particle size distribution of suspensions (Fig. 5). Samples originating from suspension of pH 9.5 had the highest modal pore diameter of ca. 200 nm and its total pore volume was similar to the sample derived from a suspension in propanol. The pore size distributions of samples, to some extent corresponded to the size distributions of particle in the subsequent suspensions. The suspensions of similar particle size distributions i.e. in propanol and at pH 12 led to obtaining the samples with similar pore size distributions. The smaller modal pore diameter than in the case of sample prepared from a suspension of pH 9.5 can be a result of the presence of a higher number of small particles (Fig. 5). The reason why the microstructure of the sample originating from the suspension of pH 7 is characterised by the smallest modal pore diameter and total pore volume is unclear. One possible explanation is that the mechanical strength of the partially dissolved yttria agglomerates was lower than the strength of the agglomerates in the other suspensions, and they were much extensively deformed/compressed during the filter pressing.
10
SEM microphotographs were taken on fracture surfaces of the samples calcined at 900 C (Fig. 7). The microstructures of all the samples were homogenous and consisted of particles with a diameter of 150 – 200 nm, and there was no trace of large yttria powder agglomerates. Destruction of the agglomerates was previously confirmed by the pore size distributions (Fig. 6), and consequently it was impossible to distinguish between the alumina and yttria particles. It can be concluded, based on the microstructure of the annealed samples, that the yttria powder exhibited two levels of agglomeration, and its crystallites formed agglomerates of about 150 – 200 nm, which were then grouped into larger structures of a few microns in diameter (Figs 1a, 2). The microstructures of the samples obtained from the propanol suspension (Fig. 7d) and the suspension of pH 12 (Fig. 7c) were almost identical. The sample prepared from a suspension having a pH of 9.5 consisted of slightly smaller particles, but the microstructure was not significantly different from the above two mentioned samples (Fig. 7b), in contrast to the microstructure of a sample prepared from a suspension of pH 7, which consisted of more densely packed particles (Fig. 7a). 3.3. Dilatometry studies Dimensional changes occurring during heating in the investigated system could be caused by: removal of water, phase transitions and sintering. Synthesis of YAG phase (Y3Al5O12) during solid state reactions between Al2O3 and Y2O3 is preceded by a synthesis of phases with a lower Al to Y ratio i.e.; YAM (Y4Al2O9) and then YAP (YAlO3) [11, 14]. The phases differ in crystallographic structure and true density and consequently in a specific volume (Table 2), which means that certain phase transformations might result in dimensional changes observed in the dilatometric curves.
11
Table 2. Basic characteristics of phases present in the Y2O3 – Al2O3 system Crystallographic True density Specific volume 3 structure g/cm cm3/g Y4Al2O9, YAM monoclinic 4.56 0.219 YAlO3, YAP orthorhombic 5.33 0.188 Y3Al5O12, YAG cubic 4.51 0.222 Y2O3 + Al2O3 cubic/hexagonal 4.59* 0.218* *calculated on the basis of volume fractions, and true density of the powders Phase
In order to monitor the phase changes occurring in the system, samples made of suspension having pH of 9.5 were chosen. The samples were heated at a temperature from 900 to 1700 C with 100 C interval for 1 hour and with a heating rate of 5 C/min. Their phase composition was determined by the XRD method with the Rietveld refining (Fig. 8). The reaction of yttria and alumina started with the formation of a YAM phase at temperature higher than 900 C (Fig. 8) and thus no dimensional changes attributed to the phase transition below this temperature were present. The changes taking place in the materials during the thermal treatment were studied by dilatometric measurements (Fig. 9). The figure presents the dilatometric curves of the mixtures of reactive powders as well as the alumina and yttria powders. Courses of the dilatometric curves for all powder mixtures were similar, and could be divided into four regions. The temperature of start and end, as well as the rate and range of changes of linear dimensions were depended on the type of material. The shrinkage of samples observed in the first region in the range between room temperature and ca. 650 C could be attributed to the removal of water, mainly present in the yttria powder (Fig. 3). The shrinkage of samples prepared from the aqueous suspensions ranged from 5 to 6 %, while for the propanol-derived sample it was only 2 %. It was probably caused by the fact that the part of water was extracted from the yttria powder during mixing.
12
Next shrinkage region started at ca. 650 C and finished at 1050 – 1170 C, depending on the material. In this region, the shrinkage of samples stopped, so their densities did not change. As it was stated in the paragraph about the pore size distributions, relative densities of all samples heated at 900 C were similar and ranged from 30.5 to 37.9 %. Generally, the densities of materials were rather low, because they contained yttria powder composed of highly porous agglomerates which were not compressed/deformed during the filter pressing. The third period of shrinkage started at c.a. 1050 C and lasted to the temperature of YAG phase synthesis i.e. 1300 – 1320 C. Most of the phase transformations occurred in this temperature region (Fig. 8). However, the observed shrinkage was the result of a sintering process occurring between various phases present in the system. Estimated specific volume of the mixture of reagent powders and specific volume of the YAM phase were almost identical i.e.; 0.219 and 0.218 cm3/g, respectively (Table 2). For this reason, no changes related to the formation of YAM appeared in the dilatometric curves (Fig. 9). The appearance of the YAP phase over 1000 C is associated with a distinct change in the specific volume from 0.219 to 0.188 cm3/g (Table 2), but the related shrinkage overlapped with the sintering shrinkage. Although the sintering of yttria was also possible, as it was the nanometric powder and its crystallites started to sinter as low as at 900 C (Fig. 9), but it was consumed by the solid-state reaction below 1100 C (Fig. 8). The alumina was present in the system up to 1300 C, but over 1100 C the YAP become dominant phase, which sintering was revealed by slower shrinkage rate observed in the dilatometric curves approximately above this temperature. The final phase transformation i.e. from YAP to YAG took place about 1300 -1320 C and was associated with the change of a specific volume from 0.188 to 0.222 cm3/g (Table 2). This effect could be observed as the small expansion in dilatometric curves for all materials (Fig. 9). This effect was the smallest in case of the sample prepared from a suspension of pH 13
12, and it was shifted towards higher temperature. The phase composition of the material heat treated at 1300 C contained 16 % of YAP and 84 % of YAG (Table 3). It means, that in this case the solid-state reaction proceeded more slowly than in the material prepared from a suspension of pH 9.5. An onset of the fourth, last stage of shrinkage could be placed above the YAP – YAG transformation effect i.e. at temperatures higher than 1320 °C. This stage was related with transformation of residual YAP to YAG (Table 3) and onset of sintering of YAG phase. The YAG sintering rate, which can be estimated by a slope of the dilatometric curves in the fourth shrinkage stage, depended on the material’s type. Due to increasing sintering rate the materials can be ordered as follows; the propanol suspension, suspension of pH 7, suspension of pH 12 and suspension of pH 9.5 (Fig. 9). Because the only difference between the samples was conditions of mixing of the reactive powders, the sintering rate can be attributed to a different level of their homogenisation. This in turn affects kinetics of the solid-state reactions occurring in the system, as well as its geometry reflected in the pore size distribution.
3.4. Sintering studies In order to investigate the intermediate stage of sintering, the samples heat treated at 1500 C for 1 hour were included, and their pore size distribution and apparent density were determined. Pore size distributions in the materials sintered at 1500 C were characterised by similar values of the modal pore diameters in the range from ca. 700 to 800 nm (Fig. 10b). Apart from a sample prepared from a suspension having a pH of 9.5, other materials showed 14
the same sequence of modal pore diameters as the samples heat treated at 900 C (Fig. 6). The smallest pores were observed in the sample prepared from a suspension of pH 7 (691 nm) and the highest in the sample prepared from the propanol suspension (818 nm); the sample made of suspension at pH 12 showed the intermediate value (779 nm). In comparison with other materials, the relative modal pore size in the sample prepared from suspension of pH 9.5 and sintered at 1500 C decreased (Figs. 6b, 10b). Relative total pore volume of the sample prepared from suspension of pH 9.5 was changing in a similar manner. Eventually, the sample obtained at pH 9.5 and sintered at 1500 C had the lowest total pore volume (0.2 cc/g) and the corresponding highest value of the relative density of 51.0 % (Fig. 10a). Such a change in pore size distribution as compared to the other materials was a result of faster sintering of the material, which could be observed as the highest shrinkage rate in the fourth region in the dilatometric curve (Fig. 9). This is probably due to complete transformation to the YAG phase at a lower temperature sintering, which was not slowed down by the transformation of the residual YAP into YAG. The relative density of other materials sintered at 1500 C was between 42.6 and 49.1 % (Table 3). The samples were finally sintered at 1700 C for 1 hour with a heating rate of 5 C/min. The relative densities of samples sintered at a maximum temperature varied significantly depending the type of material (Table 3). The highest density i.e. 98.2 % was achieved for the material originating from suspension of pH 9.5, while the lowest i.e. 58.3 % was observed for the sample derived from the propanol suspension. All materials sintered at this temperature were composed in 100% of YAG phase (Table 3).
15
Table 3. Relative density and phase composition of sintered materials pH 7
pH 9.5
pH 12
propanol
Relative density 1500 C
49.1 %
51.0 %
43.5 %
42.6 %
1700 C
73.9 %
98.2 %
83.0 %
58.3 %
YAG 90.7 %, AP 6.0 %, corundum 3.3 %
Phase composition 1300 C
YAG 91.5 %, YAP 5.3 %, corundum 3.2 %
YAG 100 %
YAG 80.3 %, YAP 13.7 %, corundum 6.0 %
1500 C
YAG 97.4 %, YAP 2.6 %
YAG 100 %
YAG 99.1 %, YAP 0.9 %
YAG 100 %
1700 C
YAG 100 %
YAG 100 %
YAG 100 %
YAG 100 %
The microstructures of samples with the highest and lowest density, sintered at 1700 C are shown in Fig. 11. The microstructure of the densified sample originating from the suspension of pH 9.5, revealed the grains having diameters of about 15 μm (Fig. 11a). Residual pores were identified to be located at the triple junctions of grain boundaries, which means that further densification of the material may be possible by the post-HIP treatment. The microstructure of the sample prepared from the propanol suspension (Fig. 11b) was very porous and composed of grains of ca. 1 μm size which means that they only slightly grown comparing to the microstructure of similar material at 900 C (Fig. 7d). Such a microstructure indicated that in this case mostly its coarsening took place. Microstructures of the other materials, which were not shown here, indicated on their partial sintering (Table 3). The probable explanation of the observed differences in the final densification of the materials (Table 3, Fig. 11) was a different homogenisation level of the reactive powders. 16
This affected geometry of the system i.e. density and pore size distribution (Table 3, Fig. 10) and kinetics of the phase transformations resulting in different sintering rates (Fig. 9). As a result, the sample prepared from propanol suspension which had the lowest density at 1500 °C (Table 3) and the lowest sintering rate (Fig. 9) was only partially densified (Table 3, Fig. 11). On contrary, the highest density was achieved by the sample derived from suspension of pH 9.5 (Table 3, Fig. 11) which had the highest intermediate density (Table 3) and showed the highest sintering rate (Fig. 9). In case of this material good homogeneity of the reactive powders was a result of the heterocoagulation effect (Fig. 4). Other two materials followed this scheme and reached intermediate densities (Table 3, Fig. 9). However, at this moment the reason of the lowest homogenisation of the reactive powders in propanol suspension is not clear and this issue requires further studies.
4.
Conclusions It was shown that the mixing conditions affected the properties of both green and
partially sintered materials as well as the course of the changes of linear dimension during the heating. As a result, the relative densities of the samples sintered at 1700 C for 1 h in air ranged from 58.31 to 98.25 %. The highest density was achieved for the material originating from suspension of pH 9.5, as a result of heterocoagulation effect occurring at this pH. This in turn led to a better homogenisation of the reagent powders, and consequently to the complete transformation to YAG phase occurring at a lower temperature (1300 C), without interfering with the sintering process. It may be generally concluded, that mixing conditions of the reactive powders have a great impact on the solid-state reaction and sintering of YAG.
17
Acknowledgements The work was financially supported by the National Science Centre (Poland) under grant number DEC-2011/03/B/ST8/06286.
References [1] A. Ikesue, T. Kinoshita, K. Kamata, K. Yoshida, Fabrication and optical properties of high-performance polycrystalline Nd:YAG ceramics for solid-state lasers. J. Am. Ceram. Soc. 78 (1995) 1033–1040. [2] L. Moreira, L. Ponce, E. de Posada, T. Flores, Y. Peñaloza, O. Vázquez, Y. Pérez, Er:YAG polycrystalline ceramics: The effects of the particle size distribution on the structural and optical properties, Ceram. Int. 41 (2015) 11786-11792. [3] A. Sidorowicz, A. Wajler, H. Weglarz, A. Olszyna, Precipitation of Tm2O3 nanopowders for application in reactive sintering of Tm:YAG, Ceram. Int. 40 (2014) 10269-10274. [4] G. de With, H.J.A. van Dijk, Translucent Y3Al5O12 ceramics, Mater. Res. Bull. 19 (1984) 1669-1674. [5] J-G. Li, T. Ikegami, J-H. Lee, T. Mori, Y. Yajima, Coprecipitation synthesis and sintering of yttrium aluminum garnet (YAG) powders: The effect of precipitant. J. Eur. Ceram. Soc. 20 (2000) 2395–2405. [6] H. Wang, L. Gao, K. Niihara, Synthesis of nanoscale yttrium aluminum garnet powder by the co-precipitation method. Mater. Sci. Eng. A288 (2000) 1–4. [7] B-J. Chung, J-Y. Park, S-M. Sim, Synthesis of yttrium aluminum garnet powder by a citrate gel method, J. Ceram. Proc. Res., 4 (2003) 145–150. [8] M. Zarzecka-Napierała, M.M. Bućko, J. Brzezińska Miecznik, K. Haberko, YAG powder synthesis and characteristics, J. Eur. Ceram. Soc. 27 (2007) 593– 597. [9] R. Manalert, M.N. Rahaman, Sol-gel processing and sintering of yttrium aluminum garnet (YAG) powders, J. Mater. Sci. 31 (1996) 3453 - 3458 [10] G. Gowda, Synthesis of yttrium aluminates by the sol-gel process, J. Mater. Sci. Lett. 5 (1986) 1029– 1032. [11] A. Ikesue, I. Furusato, K. Kamata, Fabrication of polycrystalline, transparent YAG ceramics by a solid-state reaction method, J. Am. Ceram. Soc. 78 (1995) 225–228. [12] B.L. Liu, J. Li, K. Ivanov, W.B. Liu, J. Liu, T.F. Xie et al, Solid-state reactive sintering of Nd:YAG transparent ceramics: the effect of Y2O3 powders pretreatment. Opt. Mater. 36 (2014) 1591–1597. [13] L. Esposito, A. Piancastelli, Role of powder properties and shaping techniques on the formation of pore-free YAG materials, J. Eur. Ceram. Soc. 29 (2009) 317-322. [14] E.R. Kupp, S. Kochawattana, S.-H. Lee, S. Misture, G.L. Messing, Particle size effects on yttrium aluminum garnet (YAG) phase formation by solid-state reaction. J. Mater. Res. 29 (2014) 2303-2311. [15] B.V. Derjaguin, A theory of the heterocoagulation, interaction and adhesion of dissimilar particles in solutions of electrolytes, Prog. Surf. Sci. 43 (1993) 60-73. 18
[16] P.E. Debély, E.A. Barringer, H.K. Bowen, Preparation and sintering behavior of finegrained A12O3-SiO2 composite. J. Am. Ceram. Soc. 68 (1985) C76 - C78. [17] X. Zhang, Y. Zhang, H. Chen, Effect of pH on rheology of aqueous Al2O3/SiC colloidal system. J. Adv. Ceram. 3 (2014) 125-131. [18] Y.I. Cho, H. Kamiya, H. Takano, M. Horio and H. Suzuki, Analysis of heterocoagulation structure of colloidal mullite precursor powder - experimental approach and computer simulation, in K.E. Gonsalves, M-I. Baraton, J.X. Chen, J.A. Akkara (Eds.) Proceedings of 1997 MRS Fall Meeting, 501 (1997) 255 – 261. [19] U. Aschauer, O. Burgos-Montes, R. Moreno, P. Bowen, Hamaker 2: A toolkit for the calculation of particle interactions and suspension stability and its application to mullite synthesis by colloidal methods, J. Disper. Sci. Technol. 32 (2011) 470 — 479. [20] K.J. Konsztowicz, R. Langlois, Effects of heteroflocculation of powders on mechanical properties of zirconia alumina composites, J. Mater. Sci. 31 (1996) 1633-1641. [21] Ł. Zych, R. Lach, A. Wajler, The influence of the agglomeration state of nanometric MgAl2O4 powders on their consolidation and sintering, Ceram. Int. 40 (2014) 9783-9790. [22] S.N. Bagayev, V.V. Osipov, V.I. Solomonov, V.A. Shitov, Fabrication of Nd3+:YAG laser ceramics with various approaches, Opt. Mater. 34 (2012) 1482-1487. [23] M. Kosmulski, The pH-dependent surface charging and the points of zero charge, J. Colloid. Interf. Sci. 253 (2002) 77–87. [24] M.K.M. Hruschka, W. Si, S. Tosatti, T.J. Graule, L.J. Gauckler, Processing of β-silicon nitride from water based α-silicon nitride, alumina and yttria powder suspensions, J. Am. Ceram. Soc. 82 (1999) 2039–2043.
Fig. 1. TEM image of; a) Y2O3, b) Al2O3 powder Fig. 2. Particle size distribution of the initial powders determined by the laser diffraction method Fig.3. DTA/TG analysis of the yttria powder Fig. 4. Zeta potential of the alumina and the yttria powders vs. pH Fig. 5. Particle size distribution of the powder mixtures
19
Fig. 6. Pore size distributions of samples heat-treated at 900oC for 1 hour: a) cumulative curve, b) derivative curve Fig. 7. SEM microphotographs of samples annealed at 900oC for 1 h. The samples were derived from; a) suspension of pH 7, b) suspension of pH 9.5, c) suspension of pH 12, d) propanol suspension Fig. 8. Phase composition of samples prepared from suspension with pH 9.5 heat-treated at different temperatures Fig. 9. Dilatometric curves of samples heated with 5oC/min Fig. 10. Pore size distributions of samples heat-treated at 1500oC for 1 hour: a) cumulative curve, b) derivative curve Fig. 11. Microstructure of samples sintered at 1700oC derived from; a) suspension of pH 9.5, b) propanol suspension
20
21
22
23
24
25
26
27
28
29
30
31
32
PII: DOI: Reference:
S0272-8842(16)32144-7 http://dx.doi.org/10.1016/j.ceramint.2016.11.142 CERI14230
To appear in: Ceramics International Received date: 14 September 2016 Revised date: 12 November 2016 Accepted date: 21 November 2016 Cite this article as: Łukasz Zych and Radosław Lach, The effect of powders homogenisation conditions on the synthesis of yttrium aluminium garnet (YAG) by a solid-state reaction, Ceramics International, http://dx.doi.org/10.1016/j.ceramint.2016.11.142 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
The effect of powders homogenisation conditions on the synthesis of yttrium aluminium garnet (YAG) by a solid-state reaction Łukasz Zych*, Radosław Lach AGH-University of Science and Technology, Faculty of Materials Science and Ceramics, Mickiewicza 30, 30-059 Krakow *
Corresponding author: AGH-University of Science and Technology, Faculty of Materials Science and Ceramics, Mickiewicza 30, 30-059 Krakow, Poland. Tel.: +48 12 6172480; fax: +48 12 6334630. [email protected] Abstract The work presents results of research on the influence of the state of homogenisation of reagent powders i.e. aluminium oxide and yttrium oxide on the synthesis of yttrium aluminium garnet. Fine powders of commercial aluminium oxide and yttrium oxide were mixed in the same manner in a laboratory ball mill but under four different conditions. Three batches of powders were prepared in water of various pH values, selected on the basis of the zeta potential measurements, and one batch was prepared in propanol. They were consolidated by filter pressing and then sintered in air at a temperature from 900 C to 1700 C for 1 hour, with heating rate of 5 C/min. Samples were subjected to dilatometric studies, phase composition determination, measurement of apparent density and pore size distribution, as well as a microscopic examination. Mixing environment affected, to a given extent, almost all the features of green, and partially sintered materials as well as course changes of linear dimensions. As a result, the relative densities of samples sintered at a maximum temperature ranged from 58.31 to 98.25 %. The highest density was achieved for the material originating from the suspension having a pH of 9.5, in which heterocoagulation occurred. Keywords: mixing (A); powders: solid-state reaction (A); suspensions (A); yttrium aluminium garnet
1.
Introduction
1
Yttrium aluminium garnet (Y3Al5O12, YAG) is one of the most popular transparent ceramic material which main application is the host of solid state lasers. For this purpose, YAG can be doped with rare-earth elements such as: Nd, Er, Yb or Tm [1-3]. The main types of YAG synthesis are: the solid state reaction between Y2O3 (yttria) and Al2O3 (alumina) powders [4], and the direct synthesis of powders using methods such as co-precipitation [5, 6], citrate process [7, 8] and a sol-gel method [9, 10]. Translucent YAG ceramics was first obtained by the solid state reaction by De With and Van Dijk [4]. The production of highquality polycrystalline YAG and Nd-doped YAG ceramics by reaction in the solid state between powders of yttria and alumina was first described by Ikesue et al. in 1995 [1, 11]. Since that time, attempts have been made to better understand the reaction and the factors controlling its course. One of the most important factors is the characteristics of the initial powders [12-14], and this parameter is related to the homogeneity of powder mixture, which in turn governs the course of YAG synthesis. The homogeneity of the powders mixture can be improved when the mixing is carried out under heterocoagulation (or heteroflocculation) conditions, which means that particles of dissimilar electric charge are attracted to each other [15]. A measure of the electric charge of the powder surface is zeta potential, which the sign and the value is dependent on the pH of the suspension and the presence of some adsorbing species in the suspension. Suitable conditions for heterocoagulation to take effect occur within a specific pH range, and in the processing of ceramics this can be used the manufacturing of composites [16, 17] as well as for the preparation of powders mixtures for solid state reactions [18, 19]. On the other hand, Konsztowicz et al. showed that in zirconia alumina composites heteroflocculation of the constituent powders detrimentally affects homogeneity of their microstructure which in turn influences their mechanical properties [20]. Problems in the wider application of this effect are created by: availability of powders with the adequate zeta potential characteristics, too 2
narrow pH range etc. They may be partially bypassed by modifying the powders with additions of organic molecules adsorbing on the powder surface [20]. The paper presents the results of studies on the influence of heterocoagulation effect on properties of YAG synthetized by a solid-state reaction. To the authors’ best knowledge, such investigations have not been conducted so far. In order to observe the effects of mixing conditions only, no additives were used to enhance the sintering of YAG, such as SiO2 or TEOS [4, 11], and the samples were sintered in the air.
2.
Materials and methods Tests were performed using a commercial powders with a purity of 99.99 %: yttrium
oxide (32-36 nm APS, #5650YS, Nanostructured & Amorphous Materials Inc.) and aluminium oxide (Taimicron TM-DAR, Taimei Chemicals Co. Ltd.). Characterisation of the powders included measurement of specific surface area by the BET method (Nova 1200e, Quantachrome Inc.), observation of the powders by a transmission electron microscope (JEM 1011, Jeol.) and the determination of particle size distribution by a laser diffraction method (Mastersizer 2000, Hydro S, Malvern Inc.). The quantitative analysis of particle size distributions of the powders was performed in the diluted suspensions prepared by high-energy ultrasonication with a small addition of a dispersant (Dispex N40). Thermal analysis was used to determine mass changes of the powders occurring during heating. Measurements were performed using DTA/TG method and STA 449 F3 apparatus (Netzsch) with a heating rate of 10 C/min and the maximum temperature of 1200 C. The relationship of the zeta potential on the pH was determined by a Zetasizer NanoZS equipped with a MPT-2 autotitrator unit (Malvern Inc.). In order to fix the ionic strength of the suspensions, measurements were performed in 0.01 M NaCl solution. The pH was 3
controlled by addition of 0.5 M HCl or 0.5 M NaOH solution (POCh, Poland), respectively. The results of the measurements of zeta potential enabled to select three different homogenisation conditions of the powders in aqueous suspensions. Additionally, in order to completely eliminate water from the system, one batch of powders was mixed in iso-propanol (POCh, Poland). The powders were mixed in a ratio of 42.94 wt% of Al2O3 and 57.06 wt% of Y2O3 which leads to the synthesis of yttrium-aluminium garnet. Mixing of the powders was performed overnight in a laboratory ball-mill using a high-quality alumina balls of 5 mm diameter. The particle size distribution of the powders mixtures was determined by laser diffraction method (Mastersizer 2000, Malvern Inc.). Before the measurement, the suspensions were diluted and dispersed using a high-energy ultrasounds during 3 minutes. The suspensions were consolidated using the technique of pressure filtration and samples of 30 mm diameter and 5 mm thickness were produced [21]. A given powder suspension was pressed against the ceramic porous support covered with a nylon membrane with a pore diameter of 0.2 μm (Whatman). Filtering pressure was 10 MPa and it was kept constant until no leakage of supernatant. The green samples were dried for a few days at room temperature in a dessicator over silica gel, and then overnight at 110 C in oven. The sintering of samples was studied using a DIL 402 C dilatometer (Netzsch) with a heating rates of 5 C/min and 10 C/min and a maximum temperature of 1550 C. The pore size distribution of materials was determined by means of mercury porosimetry using Poremaster 60 (Quantachrome Inc.). In order to exclude the effect of the presence of water in the samples on their pore size distribution the measurements were carried out after heating the samples at 900 C for 30 min. The microstructure of the samples was observed using a scanning electron microscope Nova Nano-SEM 200 (FEI Co). In order to investigate the changes in phase
4
composition during the heat treatment, the selected material was heated at a temperature ranged from 900 to 1700 C for 1 hour, with a heating rate of 5 C/min. The final sintering of samples was conducted in air at 1700 C with heating rate of 5 C/min and the dwelling time of 3 hours. The apparent density of the samples was measured by the Archimedes method, and the microstructure of the sintered samples was observed using SEM method. The phase composition was determined by X-ray diffraction method (XRD, Empyrean, PANanalytical) and Rietveld refining method.
3.
Results and discussion
3.1. Powders characterisation The primary particle size (crystallites) of powders was determined from their specific surface area (SSA), which was 36.6 and 12.9 m2/g for Y2O3 and Al2O3, respectively. The equivalent diameters of the spherical particles, calculated on this basis (dBET), were 33 and 117 nm, respectively. Crystallite sizes, determined on the basis of X-ray diffraction lines broadening (dXRD), were 35 nm and 107 nm for Y2O3 and Al2O3, respectively. Good agreement between values of dBET and dXRD of both powders indicated that there were no broad phase contacts formed between the primary particles (crystallites). The yttria powder tended to agglomerate (Fig. 1a) and the modal diameter of the agglomerates, determined from the particle size distribution was 3 μm (Fig. 2). In contrast, the alumina powder was practically non-agglomerated, and consisted of single crystallites of submicrometric size (Fig. 1b). In this case, the modal particle diameter i.e. 129 nm (Fig. 2), corresponded well to the diameter of the primary particles, calculated on the basis of the specific surface area of the powder, i.e. 117 nm.
5
It means that the actual particle size of yttria (agglomerates) involved in the synthesis of YAG was at least 100 times greater than its crystallite size. In general, the nanoparticles exhibit a strong tendency to form agglomerates during the synthesis and handling, and a significant degree of agglomeration observed in yttria powder was probably caused by a smaller crystallite size (32 nm) compared with alumina powder (107 nm). The modal diameter of the agglomerates of yttria was approximately 30 times greater than the diameter of the alumina particles, which in turn can affect the efficiency of mixing of powders and finally the course of the reaction of synthesis. Another issue, related to nanopowders, is the amount of bounded water, which is particularly important in the case of powders used as reagents for the solid state synthesis [22]. It was found, that the weight loss of yttria powder during heating up to 1200 °C, as determined from the DTA / TG measurement, was 9.67 %, and the removal of water proceeded in a two-step process (Fig. 3). Most of the water (ca. 6 wt%), which was rather loosely adsorbed on the surface of the powder particles was removed up to 300 C. Another, smaller mass loss (c.a. 2 wt%) occurred between 800 C and 1200 C, which may be related to the removal of some strongly bound water which origin was not clear. In the case of alumina powder, probably due to its lower specific surface area, the total amount of water (mass loss) was only 0.60 % (not shown here). Preliminary experiments showed that in order to synthesise a pure phase of YAG, it was crucial to use an amount of reagent powders of the total weights corrected for mass loss. As the mixing of reagent powders was carried out in aqueous environment of different pH values, it influenced also the measured values of zeta potential (Fig. 4).
6
The zeta potential vs. pH relationships for the powders were different, with the isoelectric points (iep) at pH 8 and pH 10.9 for Al2O3 and Y2O3, respectively. In most cases, isoelectric point of Y2O3 at a pH of about 8 has been reported, however, the iep at a pH of 11 has been also found [23, 24]. This implies, that the pH window, in which the zeta potential of the alumina and yttria powders had opposite signs and wherein heterocoagulation could occur, was in the range between pH 8 and 10.9. The maximum difference between the values of the zeta potential of powders occurred at pH 9.5. The observed small decrease in the value of the zeta potential at pH 6 accompanied with a much higher variability of results probably resulted from the partial dissolution of the yttria powder. The yttria powder had a nanometric size and thus was more reactive and readily soluble under acidic conditions. Finally, the powders were mixed under four different conditions: three aqueous solutions of different pH values and iso-propanol which was used to completely exclude water from the system (in this latter case, the zeta potential was not measured, Table 1.).
Table 1. Mixing conditions of the reactive powders Batch number
mixing conditions
zeta potential of Y2O3 particles, mV
zeta potential of Al2O3 particles, mV
1
water, pH 7
+ 51
+ 45
2
water, pH 9.5
+ 48
- 45
3
water, pH 12
- 30
- 50
4
iso-propanol
n/a
n/a
All batches of powders were prepared under the same conditions, i.e. by mixing overnight in a laboratory ball mill with high purity alumina balls of 5 mm diameter. The volume fraction of
7
solid phase in the suspensions was ca. 12 %, and the weight ratio of powders to the balls was 1:10. The initial pH of the suspension of the reagent powders was about 9, and it was adjusted to the desired level by addition of 1 M HCl solution or 10 % TMAH solution (tetramethylammonium hydroxide, Acros Organics). Both, HCl and TMAH should evaporate without a trace from the material during heating. The pH of suspensions was checked several times at the beginning but also at the end of the mixing process. Generally, the pH of the suspensions was stable during the mixing process, with the exception of the suspension having a pH of 7. In this case, it was difficult to stabilise the pH value, as it quickly returned to a value of 9, indicating that small additions of HCl resulted in some changes to the reagent powders. Most probably, the HCl additions led to a slight dissolution of the yttria powder, because the instantaneous values were much lower than pH 7, and the powder consisted of nanometric, and thus reactive particles. The volume fractions of the powders used in the YAG synthesis were similar and equalled to 48.7 % and 51.3 % for Al2O3 and Y2O3, respectively. Consequently, the particle size distribution of the mixture should be similar to the superposition of the particle size distributions of individual powders (Fig. 2). However, the particle size distributions of the powder mixtures showed distinct differences (Fig. 5). The main populations of particles present in the powder mixtures corresponded to those present in the starting powders (Fig. 2) but their proportions varied between the mixtures (Fig. 5). Both populations of powders were clearly visible in the mixtures prepared at pH 7 and 12 and in iso-propanol. In the case of mixtures prepared at pH 12 and in iso-propanol, the particle size distributions were almost the same and reassembled the superposition of the distributions of starting powders. Such a distribution indicated that no interaction between the particles of yttria and alumina occurred. A certain level of inter-particle interaction became visible in the case of powders mixture 8
prepared at pH 7, as here the proportions between the populations have changed, and the population of larger particles started to be dominant. The reason of such behaviour was not clear but it probably was related to the observed partial dissolution of the yttria powder. It is also probable, that the effect of a partial agglomeration of the alumina particles appeared. A possible explanation for this is that, due to the increasing ionic strength of the solution caused by the presence of yttrium ions, the absolute value of the zeta potential of the powders decreased, thus resulting in the agglomeration of alumina powder. In contrast to other powders mixtures, the particle size distribution of the mixture prepared at a pH of 9.5 was monomodal, with the modal size corresponding to the size of yttria powder particles (Fig. 2). It may indicate, that in such pH conditions, heterocoagulation effect occurred (Fig. 4) and the smaller particles of alumina were attached to larger particles of yttria (Fig. 2). In summary, mixtures of powders had different particle size distributions resulting from the conditions of their mixing and possible interactions between the particles. 3.2. Materials characterisation The microstructure of filter pressed samples prior to the reaction of yttria and alumina was observed by means of the pore size distribution and SEM observations. Preliminary tests showed that the samples heat-treated at 900 C for 30 min were suitable for such characterisation. The pore size distributions of all samples were monomodal with both modal pore diameter (Fig. 6b) and total pore volume (Fig. 6a) depending on the type of the powders mixture. The modal pore diameters were much smaller than the size of the agglomerates of yttria powder (~ 3 μm) (Figs 1a, 2), which means that the measured pores existed inside the agglomerates and the inter-agglomerate pores are not present. This in turn indicates that the compressive forces acting during the pressure filtration are sufficient to deform/compress 9
agglomerates of yttria powder. The smallest modal pore size (62 nm) and the lowest total pore volume (~ 0.35 cc/g) were present in a sample obtained from the suspension of pH 7. Total pore volumes of other samples were similar and ranged from 0.49 to 0.55 cc/g for samples derived from the propanol suspension, and the suspension of pH 12, respectively (Fig. 6b). Real density of the materials heat treated at 900 C was 4.59 g/cm3, which was calculated based on the volume fractions of the powders and their real densities. Relative densities of the materials corresponded to their total pore volumes and they were as follows; 37.9 % for the samples derived from suspension of pH 7, 33.0 % for the suspension of pH 9.5, 32.0 % for the propanol suspension and 30.5 % for suspension of pH 12. The modal pore diameters of these samples were similar and close to 100 nm (Fig. 6b), which corresponded well to the appropriate particle size distribution of suspensions (Fig. 5). Samples originating from suspension of pH 9.5 had the highest modal pore diameter of ca. 200 nm and its total pore volume was similar to the sample derived from a suspension in propanol. The pore size distributions of samples, to some extent corresponded to the size distributions of particle in the subsequent suspensions. The suspensions of similar particle size distributions i.e. in propanol and at pH 12 led to obtaining the samples with similar pore size distributions. The smaller modal pore diameter than in the case of sample prepared from a suspension of pH 9.5 can be a result of the presence of a higher number of small particles (Fig. 5). The reason why the microstructure of the sample originating from the suspension of pH 7 is characterised by the smallest modal pore diameter and total pore volume is unclear. One possible explanation is that the mechanical strength of the partially dissolved yttria agglomerates was lower than the strength of the agglomerates in the other suspensions, and they were much extensively deformed/compressed during the filter pressing.
10
SEM microphotographs were taken on fracture surfaces of the samples calcined at 900 C (Fig. 7). The microstructures of all the samples were homogenous and consisted of particles with a diameter of 150 – 200 nm, and there was no trace of large yttria powder agglomerates. Destruction of the agglomerates was previously confirmed by the pore size distributions (Fig. 6), and consequently it was impossible to distinguish between the alumina and yttria particles. It can be concluded, based on the microstructure of the annealed samples, that the yttria powder exhibited two levels of agglomeration, and its crystallites formed agglomerates of about 150 – 200 nm, which were then grouped into larger structures of a few microns in diameter (Figs 1a, 2). The microstructures of the samples obtained from the propanol suspension (Fig. 7d) and the suspension of pH 12 (Fig. 7c) were almost identical. The sample prepared from a suspension having a pH of 9.5 consisted of slightly smaller particles, but the microstructure was not significantly different from the above two mentioned samples (Fig. 7b), in contrast to the microstructure of a sample prepared from a suspension of pH 7, which consisted of more densely packed particles (Fig. 7a). 3.3. Dilatometry studies Dimensional changes occurring during heating in the investigated system could be caused by: removal of water, phase transitions and sintering. Synthesis of YAG phase (Y3Al5O12) during solid state reactions between Al2O3 and Y2O3 is preceded by a synthesis of phases with a lower Al to Y ratio i.e.; YAM (Y4Al2O9) and then YAP (YAlO3) [11, 14]. The phases differ in crystallographic structure and true density and consequently in a specific volume (Table 2), which means that certain phase transformations might result in dimensional changes observed in the dilatometric curves.
11
Table 2. Basic characteristics of phases present in the Y2O3 – Al2O3 system Crystallographic True density Specific volume 3 structure g/cm cm3/g Y4Al2O9, YAM monoclinic 4.56 0.219 YAlO3, YAP orthorhombic 5.33 0.188 Y3Al5O12, YAG cubic 4.51 0.222 Y2O3 + Al2O3 cubic/hexagonal 4.59* 0.218* *calculated on the basis of volume fractions, and true density of the powders Phase
In order to monitor the phase changes occurring in the system, samples made of suspension having pH of 9.5 were chosen. The samples were heated at a temperature from 900 to 1700 C with 100 C interval for 1 hour and with a heating rate of 5 C/min. Their phase composition was determined by the XRD method with the Rietveld refining (Fig. 8). The reaction of yttria and alumina started with the formation of a YAM phase at temperature higher than 900 C (Fig. 8) and thus no dimensional changes attributed to the phase transition below this temperature were present. The changes taking place in the materials during the thermal treatment were studied by dilatometric measurements (Fig. 9). The figure presents the dilatometric curves of the mixtures of reactive powders as well as the alumina and yttria powders. Courses of the dilatometric curves for all powder mixtures were similar, and could be divided into four regions. The temperature of start and end, as well as the rate and range of changes of linear dimensions were depended on the type of material. The shrinkage of samples observed in the first region in the range between room temperature and ca. 650 C could be attributed to the removal of water, mainly present in the yttria powder (Fig. 3). The shrinkage of samples prepared from the aqueous suspensions ranged from 5 to 6 %, while for the propanol-derived sample it was only 2 %. It was probably caused by the fact that the part of water was extracted from the yttria powder during mixing.
12
Next shrinkage region started at ca. 650 C and finished at 1050 – 1170 C, depending on the material. In this region, the shrinkage of samples stopped, so their densities did not change. As it was stated in the paragraph about the pore size distributions, relative densities of all samples heated at 900 C were similar and ranged from 30.5 to 37.9 %. Generally, the densities of materials were rather low, because they contained yttria powder composed of highly porous agglomerates which were not compressed/deformed during the filter pressing. The third period of shrinkage started at c.a. 1050 C and lasted to the temperature of YAG phase synthesis i.e. 1300 – 1320 C. Most of the phase transformations occurred in this temperature region (Fig. 8). However, the observed shrinkage was the result of a sintering process occurring between various phases present in the system. Estimated specific volume of the mixture of reagent powders and specific volume of the YAM phase were almost identical i.e.; 0.219 and 0.218 cm3/g, respectively (Table 2). For this reason, no changes related to the formation of YAM appeared in the dilatometric curves (Fig. 9). The appearance of the YAP phase over 1000 C is associated with a distinct change in the specific volume from 0.219 to 0.188 cm3/g (Table 2), but the related shrinkage overlapped with the sintering shrinkage. Although the sintering of yttria was also possible, as it was the nanometric powder and its crystallites started to sinter as low as at 900 C (Fig. 9), but it was consumed by the solid-state reaction below 1100 C (Fig. 8). The alumina was present in the system up to 1300 C, but over 1100 C the YAP become dominant phase, which sintering was revealed by slower shrinkage rate observed in the dilatometric curves approximately above this temperature. The final phase transformation i.e. from YAP to YAG took place about 1300 -1320 C and was associated with the change of a specific volume from 0.188 to 0.222 cm3/g (Table 2). This effect could be observed as the small expansion in dilatometric curves for all materials (Fig. 9). This effect was the smallest in case of the sample prepared from a suspension of pH 13
12, and it was shifted towards higher temperature. The phase composition of the material heat treated at 1300 C contained 16 % of YAP and 84 % of YAG (Table 3). It means, that in this case the solid-state reaction proceeded more slowly than in the material prepared from a suspension of pH 9.5. An onset of the fourth, last stage of shrinkage could be placed above the YAP – YAG transformation effect i.e. at temperatures higher than 1320 °C. This stage was related with transformation of residual YAP to YAG (Table 3) and onset of sintering of YAG phase. The YAG sintering rate, which can be estimated by a slope of the dilatometric curves in the fourth shrinkage stage, depended on the material’s type. Due to increasing sintering rate the materials can be ordered as follows; the propanol suspension, suspension of pH 7, suspension of pH 12 and suspension of pH 9.5 (Fig. 9). Because the only difference between the samples was conditions of mixing of the reactive powders, the sintering rate can be attributed to a different level of their homogenisation. This in turn affects kinetics of the solid-state reactions occurring in the system, as well as its geometry reflected in the pore size distribution.
3.4. Sintering studies In order to investigate the intermediate stage of sintering, the samples heat treated at 1500 C for 1 hour were included, and their pore size distribution and apparent density were determined. Pore size distributions in the materials sintered at 1500 C were characterised by similar values of the modal pore diameters in the range from ca. 700 to 800 nm (Fig. 10b). Apart from a sample prepared from a suspension having a pH of 9.5, other materials showed 14
the same sequence of modal pore diameters as the samples heat treated at 900 C (Fig. 6). The smallest pores were observed in the sample prepared from a suspension of pH 7 (691 nm) and the highest in the sample prepared from the propanol suspension (818 nm); the sample made of suspension at pH 12 showed the intermediate value (779 nm). In comparison with other materials, the relative modal pore size in the sample prepared from suspension of pH 9.5 and sintered at 1500 C decreased (Figs. 6b, 10b). Relative total pore volume of the sample prepared from suspension of pH 9.5 was changing in a similar manner. Eventually, the sample obtained at pH 9.5 and sintered at 1500 C had the lowest total pore volume (0.2 cc/g) and the corresponding highest value of the relative density of 51.0 % (Fig. 10a). Such a change in pore size distribution as compared to the other materials was a result of faster sintering of the material, which could be observed as the highest shrinkage rate in the fourth region in the dilatometric curve (Fig. 9). This is probably due to complete transformation to the YAG phase at a lower temperature sintering, which was not slowed down by the transformation of the residual YAP into YAG. The relative density of other materials sintered at 1500 C was between 42.6 and 49.1 % (Table 3). The samples were finally sintered at 1700 C for 1 hour with a heating rate of 5 C/min. The relative densities of samples sintered at a maximum temperature varied significantly depending the type of material (Table 3). The highest density i.e. 98.2 % was achieved for the material originating from suspension of pH 9.5, while the lowest i.e. 58.3 % was observed for the sample derived from the propanol suspension. All materials sintered at this temperature were composed in 100% of YAG phase (Table 3).
15
Table 3. Relative density and phase composition of sintered materials pH 7
pH 9.5
pH 12
propanol
Relative density 1500 C
49.1 %
51.0 %
43.5 %
42.6 %
1700 C
73.9 %
98.2 %
83.0 %
58.3 %
YAG 90.7 %, AP 6.0 %, corundum 3.3 %
Phase composition 1300 C
YAG 91.5 %, YAP 5.3 %, corundum 3.2 %
YAG 100 %
YAG 80.3 %, YAP 13.7 %, corundum 6.0 %
1500 C
YAG 97.4 %, YAP 2.6 %
YAG 100 %
YAG 99.1 %, YAP 0.9 %
YAG 100 %
1700 C
YAG 100 %
YAG 100 %
YAG 100 %
YAG 100 %
The microstructures of samples with the highest and lowest density, sintered at 1700 C are shown in Fig. 11. The microstructure of the densified sample originating from the suspension of pH 9.5, revealed the grains having diameters of about 15 μm (Fig. 11a). Residual pores were identified to be located at the triple junctions of grain boundaries, which means that further densification of the material may be possible by the post-HIP treatment. The microstructure of the sample prepared from the propanol suspension (Fig. 11b) was very porous and composed of grains of ca. 1 μm size which means that they only slightly grown comparing to the microstructure of similar material at 900 C (Fig. 7d). Such a microstructure indicated that in this case mostly its coarsening took place. Microstructures of the other materials, which were not shown here, indicated on their partial sintering (Table 3). The probable explanation of the observed differences in the final densification of the materials (Table 3, Fig. 11) was a different homogenisation level of the reactive powders. 16
This affected geometry of the system i.e. density and pore size distribution (Table 3, Fig. 10) and kinetics of the phase transformations resulting in different sintering rates (Fig. 9). As a result, the sample prepared from propanol suspension which had the lowest density at 1500 °C (Table 3) and the lowest sintering rate (Fig. 9) was only partially densified (Table 3, Fig. 11). On contrary, the highest density was achieved by the sample derived from suspension of pH 9.5 (Table 3, Fig. 11) which had the highest intermediate density (Table 3) and showed the highest sintering rate (Fig. 9). In case of this material good homogeneity of the reactive powders was a result of the heterocoagulation effect (Fig. 4). Other two materials followed this scheme and reached intermediate densities (Table 3, Fig. 9). However, at this moment the reason of the lowest homogenisation of the reactive powders in propanol suspension is not clear and this issue requires further studies.
4.
Conclusions It was shown that the mixing conditions affected the properties of both green and
partially sintered materials as well as the course of the changes of linear dimension during the heating. As a result, the relative densities of the samples sintered at 1700 C for 1 h in air ranged from 58.31 to 98.25 %. The highest density was achieved for the material originating from suspension of pH 9.5, as a result of heterocoagulation effect occurring at this pH. This in turn led to a better homogenisation of the reagent powders, and consequently to the complete transformation to YAG phase occurring at a lower temperature (1300 C), without interfering with the sintering process. It may be generally concluded, that mixing conditions of the reactive powders have a great impact on the solid-state reaction and sintering of YAG.
17
Acknowledgements The work was financially supported by the National Science Centre (Poland) under grant number DEC-2011/03/B/ST8/06286.
References [1] A. Ikesue, T. Kinoshita, K. Kamata, K. Yoshida, Fabrication and optical properties of high-performance polycrystalline Nd:YAG ceramics for solid-state lasers. J. Am. Ceram. Soc. 78 (1995) 1033–1040. [2] L. Moreira, L. Ponce, E. de Posada, T. Flores, Y. Peñaloza, O. Vázquez, Y. Pérez, Er:YAG polycrystalline ceramics: The effects of the particle size distribution on the structural and optical properties, Ceram. Int. 41 (2015) 11786-11792. [3] A. Sidorowicz, A. Wajler, H. Weglarz, A. Olszyna, Precipitation of Tm2O3 nanopowders for application in reactive sintering of Tm:YAG, Ceram. Int. 40 (2014) 10269-10274. [4] G. de With, H.J.A. van Dijk, Translucent Y3Al5O12 ceramics, Mater. Res. Bull. 19 (1984) 1669-1674. [5] J-G. Li, T. Ikegami, J-H. Lee, T. Mori, Y. Yajima, Coprecipitation synthesis and sintering of yttrium aluminum garnet (YAG) powders: The effect of precipitant. J. Eur. Ceram. Soc. 20 (2000) 2395–2405. [6] H. Wang, L. Gao, K. Niihara, Synthesis of nanoscale yttrium aluminum garnet powder by the co-precipitation method. Mater. Sci. Eng. A288 (2000) 1–4. [7] B-J. Chung, J-Y. Park, S-M. Sim, Synthesis of yttrium aluminum garnet powder by a citrate gel method, J. Ceram. Proc. Res., 4 (2003) 145–150. [8] M. Zarzecka-Napierała, M.M. Bućko, J. Brzezińska Miecznik, K. Haberko, YAG powder synthesis and characteristics, J. Eur. Ceram. Soc. 27 (2007) 593– 597. [9] R. Manalert, M.N. Rahaman, Sol-gel processing and sintering of yttrium aluminum garnet (YAG) powders, J. Mater. Sci. 31 (1996) 3453 - 3458 [10] G. Gowda, Synthesis of yttrium aluminates by the sol-gel process, J. Mater. Sci. Lett. 5 (1986) 1029– 1032. [11] A. Ikesue, I. Furusato, K. Kamata, Fabrication of polycrystalline, transparent YAG ceramics by a solid-state reaction method, J. Am. Ceram. Soc. 78 (1995) 225–228. [12] B.L. Liu, J. Li, K. Ivanov, W.B. Liu, J. Liu, T.F. Xie et al, Solid-state reactive sintering of Nd:YAG transparent ceramics: the effect of Y2O3 powders pretreatment. Opt. Mater. 36 (2014) 1591–1597. [13] L. Esposito, A. Piancastelli, Role of powder properties and shaping techniques on the formation of pore-free YAG materials, J. Eur. Ceram. Soc. 29 (2009) 317-322. [14] E.R. Kupp, S. Kochawattana, S.-H. Lee, S. Misture, G.L. Messing, Particle size effects on yttrium aluminum garnet (YAG) phase formation by solid-state reaction. J. Mater. Res. 29 (2014) 2303-2311. [15] B.V. Derjaguin, A theory of the heterocoagulation, interaction and adhesion of dissimilar particles in solutions of electrolytes, Prog. Surf. Sci. 43 (1993) 60-73. 18
[16] P.E. Debély, E.A. Barringer, H.K. Bowen, Preparation and sintering behavior of finegrained A12O3-SiO2 composite. J. Am. Ceram. Soc. 68 (1985) C76 - C78. [17] X. Zhang, Y. Zhang, H. Chen, Effect of pH on rheology of aqueous Al2O3/SiC colloidal system. J. Adv. Ceram. 3 (2014) 125-131. [18] Y.I. Cho, H. Kamiya, H. Takano, M. Horio and H. Suzuki, Analysis of heterocoagulation structure of colloidal mullite precursor powder - experimental approach and computer simulation, in K.E. Gonsalves, M-I. Baraton, J.X. Chen, J.A. Akkara (Eds.) Proceedings of 1997 MRS Fall Meeting, 501 (1997) 255 – 261. [19] U. Aschauer, O. Burgos-Montes, R. Moreno, P. Bowen, Hamaker 2: A toolkit for the calculation of particle interactions and suspension stability and its application to mullite synthesis by colloidal methods, J. Disper. Sci. Technol. 32 (2011) 470 — 479. [20] K.J. Konsztowicz, R. Langlois, Effects of heteroflocculation of powders on mechanical properties of zirconia alumina composites, J. Mater. Sci. 31 (1996) 1633-1641. [21] Ł. Zych, R. Lach, A. Wajler, The influence of the agglomeration state of nanometric MgAl2O4 powders on their consolidation and sintering, Ceram. Int. 40 (2014) 9783-9790. [22] S.N. Bagayev, V.V. Osipov, V.I. Solomonov, V.A. Shitov, Fabrication of Nd3+:YAG laser ceramics with various approaches, Opt. Mater. 34 (2012) 1482-1487. [23] M. Kosmulski, The pH-dependent surface charging and the points of zero charge, J. Colloid. Interf. Sci. 253 (2002) 77–87. [24] M.K.M. Hruschka, W. Si, S. Tosatti, T.J. Graule, L.J. Gauckler, Processing of β-silicon nitride from water based α-silicon nitride, alumina and yttria powder suspensions, J. Am. Ceram. Soc. 82 (1999) 2039–2043.
Fig. 1. TEM image of; a) Y2O3, b) Al2O3 powder Fig. 2. Particle size distribution of the initial powders determined by the laser diffraction method Fig.3. DTA/TG analysis of the yttria powder Fig. 4. Zeta potential of the alumina and the yttria powders vs. pH Fig. 5. Particle size distribution of the powder mixtures
19
Fig. 6. Pore size distributions of samples heat-treated at 900oC for 1 hour: a) cumulative curve, b) derivative curve Fig. 7. SEM microphotographs of samples annealed at 900oC for 1 h. The samples were derived from; a) suspension of pH 7, b) suspension of pH 9.5, c) suspension of pH 12, d) propanol suspension Fig. 8. Phase composition of samples prepared from suspension with pH 9.5 heat-treated at different temperatures Fig. 9. Dilatometric curves of samples heated with 5oC/min Fig. 10. Pore size distributions of samples heat-treated at 1500oC for 1 hour: a) cumulative curve, b) derivative curve Fig. 11. Microstructure of samples sintered at 1700oC derived from; a) suspension of pH 9.5, b) propanol suspension
20
21
22
23
24
25
26
27
28
29
30
31
32