Interdependence between green compact property and powder
agglomeration and their relation to the sintering behaviour of
zirconia powder
Dean-Mo Liu
a,*
,1, Jiang-Tsair Lin
a, W.H. Tuan
baMaterials Research Laboratories, Industrial Technology Research Institute, Chutung, Hsinchu, Taiwan 31015 bInstitute of Materials Science and Engineering, National Taiwan University, Taipei, Taiwan
Received 10 March 1997; accepted 2 September 1997
Abstract
Interdependence of green density and corresponding powder agglomeration and their in¯uence on the sintering behaviour of commercial ®ne zirconia powders under a constant rate of heating (non-isothermal sintering) were investigated. Agglomeration of the powder was controlled by dierent time periods of ball-milled processing and was de®ned as the size ratio of sedimentationally-measured particle size to the size of primary particles which were microscopically-determined (hereinafter termed agglomeration parameter or AP). Green compact density shows to be approximately linearly related to powder agglomeration under identical consolidation technique, which is decreased with increasing degree of agglomeration. Both the green density and powder agglomeration aect sintering behaviour over entire sintering schedule. For a given AP the shrinkage rate reduces with increasing green compact density and vice versa, which is consistent with the literatured reports. The experimental results also showed that compacts with identical starting density showed a lower shrinkage rate when the compacts contained less agglomeration (i.e. low AP) than does for high-AP compacts. However, a higher end-point density can be obtained for low-AP compacts, suggesting a better packing structure of the powders. The use of agglomeration parameter de®ned currently, which is taken as an indication of the level of powder agglomera-tion in commercial ®ne ceramic powders, is likely to provide some useful understanding in characterising the sintering behaviour and possibly potential evolution of sintered microstructure on sintering. # 1999 Elsevier Science Ltd and Techna S.r.l. All rights reserved
1. Introduction
Dense, uniform particle packing geometry in a green ceramic powder compact is of prime criterion in achieving high end-point density body with desired microstructure and this is always the goal to be achieved by the relevant researchers and manufacturers. It is well-recognised that initial homogeneity of green microstructure strongly aects the ®nal homogeneity and integrity of the sintered microstructure. Many attempts have been used to obtain and to control green compact with physically/chemically homogeneous microstructure through a number of processing techni-ques and have received satisfactory accomplishment in some of them.
In powder processing, one important consideration is to eliminate agglomerates which are formed as clusters
of primary particles bonded strongly, i.e. ``hard'' agglomerates, (e.g. due to sintering) or weakly, i.e. ``soft'' agglomerates, (e.g. short-range surface forces) with interconnective pore network, prior to shaping. The presence of agglomerates in powders frequently reduce particle packing eciency, suppress sintering activity, and ®nally deteriorate sintered properties due to the loss of microstructural homogeneity/integrity and crack-like voids formation [1±3]. Therefore, eect of agglomerate on powder processing has been the subject of considerable interest [4±9]. In the early 1980s, Rhodes [6] observed that agglomerates, having a mean size ran-ging from 1.7 to 6.3 m, retard almost entire stages of sintering and limit the potential achievement of ®ne powder towards dense and ®ne-grained microstructure. However, Rhodes's study is somewhat qualitative, par-ticularly provides little understanding on whether the agglomerates undergo morphological change under relatively high compaction pressure (276±483 MPa). Similar eort was conducted by Sacks et al. [4] Although their work provides some understanding on 0272-8842/99/$ - see front matter # 1999 Elsevier Science Ltd and Techna S.r.l. All rights reserved
PII: S0272-8842(97)00094-1 * Corresponding author.
1 Present address: Department of Materials Science & Engineering,
the eect of agglomeration on sintering kinetics, further analysis of how ``extent'' of the agglomeration on sin-tering kinetics has not been reported. A later evaluation by Dynys et al. [5] presented a better understanding on the eect of agglomerate on sintering of alumina by controlled amount of agglomerate blended with agglomerate-free powder. They indicated a signi®cant decrease in shrinkage rate by as much as a factor of 10 under the presence of agglomerate at early stage of sin-tering. Green compact density has long been recognized to aect sintering behaviour, an earlier observation for Al2O3 by Brush [10] indicated that sintering rate increases as green density is reduced for the ``normal'' sintering regime. This has also been observed by Vasilos et al. [11] in the sintering of ZrO2. Therefore, on this basis, one may conclude that sintering behaviour is essentially microstructure-dependent. This investigation is attempting to realise the interrelationship between green density and agglomerate, and how sintering behaviour interplays between these two factors on a commercial ®ne zirconia powder.
Agglomerates are commonly observed in commercial ®ne ceramic powders. The ``agglomeration'' in the starting powders may frequently be determined with diculty in a quantitative manner. (This may be one of the primarily unsolved problems that appeared in many related studies; instead, controlled agglomerates, i.e. for both content and size, were intentionally added to fur-ther characterise the resulting behaviour/property of interest.) Moreover, it becomes even more ambiguous and uncertain, for either commercial or intentionally-prepared powders when these agglomerate-contained powders were consolidated by high-pressure compac-tion (a pressure of few hundreds of MPa is frequently used to consolidate the powders), e.g. die pressing, cold-isostatically pressing, and pressure ®ltration, primarily because the agglomerates may undergo deformation and collapse into smaller fragments and primary parti-cles under high compressive forces. This causes a less reliable prediction or assessment on agglomerate-related sintering behaviour and sintered property. Unfortu-nately, high-pressure consolidation techniques have been widely-used in many agglomerate-related studies such as those publications mentioned above. This is one reason for the current study to adopt a colloidal casting without or with a slightly external force (when the cast is still wet) to consolidate ceramic powders. This tech-nique ensures to a large extent the retention of initial powder agglomeration and is believed to be practically and physically meaningful in the interpretation of experimental results. Since it is impossible to accurately determine the ``degree of agglomeration'' (particularly in terms of the quantity, size, and geometry) in com-mercial powders, one simple and probably physically meaningful method is to compare the relative size of the actual particle (or agglomerate) size of a given powder
(usually determined by sedimentation method) and pri-mary particle size (determined microscopically) as pre-viously being used by Roosen et al. [12] This relative size is de®ned as agglomeration parameter (abbreviated as AP) and will be tentatively used as an indication for the ``degree of agglomeration'' for the ceramic powder in current investigation. Accordingly, the greater value of the AP, the more extensive agglomeration of the powder is. The use of nonisothermal technique (i.e. sintering under constant rate of heating) for the study of sintering kinetics has been well-documented and has been con-®rmed to provide suciently accurate kinetic informa-tion in comparison to the conveninforma-tionally time-consuming isothermal method [13,14]. In this technique, the speci-men is ®red under a constant heating rate to the preset temperature. The shrinkage or shrinkage rate can therefore be accurately determined using a computer-equipped dilalometrical instrument as a function of tem-perature. One important consideration in kinetics study with nonisothermal technique is the in¯uence of surface diusion particularly at initial-stage sintering [13]. How-ever, since surface diusion is usually expected to occur at temperature at or below initial shrinkage tempera-ture, it is thus possible to avoid such interference by carefully selecting and analyzing the obtained sintering data as had been treated previously [13,15].
Brie¯y, the basic assumption for nonisothermal tech-nique is identical to that for isothermal method at any given shrinkage and corresponding temperature, iso-thermal sintering theory is then applicable to the former technique. Thus for nonisothermal sintering, the densi-®cation behaviour can be generally expressed under a constant heating rate by,
d L=Lo=dt K= L=Lon 1
where K Koexp ÿQ=RT and L=Lo stands for spe-cimen shrinkage at any temperature. Young and Cutter [13] estimated the sintering activation energy of oxides with success through an Arrhenious-type plot, i.e. Td L=Lo=dT vs 1/T, used a derivation by Johnson [16] for intermediate-stage sintering. In principle, the plot of Td L=Lo=dT vs 1/T yields a straight line with a slope of ÿnQ=R, where n is the characteristic of sin-tering mechanism, e.g. n 1=2 for volume diusion and n 1=3 for grain-boundary diusion, and Q is the acti-vation energy for corresponding diusion mechanism. The term nQ is representative of the eective activation energy of sintering.
2. Experimental procedures
Ceramic suspensions containing 75 wt% zirconia powder (Daiichi-HSY3.0, 3 mole% yttria-stabilized powder) in distilled water, 1 wt% (based on the
pow-ders) plastisizer and 0.5 wt% dispersing agent. The sus-pension was ball-milled in a polyethylene jar with a ®xed ratio of zirconia milled ball weight to powder weight. In order to obtain the powders of dierent levels of agglomeration, the suspension was milled over a time period from 0 to 48 h. Owing to the intimate inter-dependence between agglomeration and packing density of the slip-cast compacts, i.e. a higher-agglomerated powder usually results in lower-density compacts, a pressure (up to 50 MPa) was used to further consolidate the casts when they were still wet to form powder com-pacts with higher densities with the same level of agglomeration. A follow-up analysis of particle size in these pressure-cast compacts ensures the retention of starting agglomeration under such pressure range. Green density as well as the pore size distribution of the compacts were determined using mercury porosimetry (Autopore, II-9220). Three to four specimens were used to determine the average green density of the compacts, which has an average error of measurement less than 0.45%. A small portion of the suspension was used to measure the resulting particle size distribution (Horiba, LA-910) and the rest of which was cast to form powder compacts. Specimens with 555 mm3 were cut from the cast and subjected to a dilatometer (Netzsch, Model 415) for shrinkage measurement at a constant heating rate of 300C hÿ1 up to 1500C. The primary particle size (dprimary) was determined using SEM (Cambridge Instruments, Model 360) by averaging over 500 particles from a dilute, dispersed suspension. Powder morphol-ogy and green microstructure of the powder compacts were examined microscopically.
3. Results and discussion
3.1. Characterization of powder and powder compacts Fig. 1 shows the morphology of the starting powders. The powders show somewhat irregular geometry and
some clusters of particles can clearly be seen, indicating the presence of agglomeration. The particle size dis-tribution of the powders after varying periods of ball milling is representatively shown in Fig. 2, which is characteristic of uni-model distribution. A shift in the distribution towards a smaller-sized region is evident as ball-milled time is increased; however, the slow reduc-tion in the mean particle size (d50) of the powders with milling time (Table 1) suggests a strong interparticle bonding in the agglomerates, i.e. ``hard'' agglomerates. Table 1 listed the mean (d50) particle/agglomerate size for varying time periods of milling and corresponding agglomeration parameter in terms of the measured dprimary. The greater value of the parameter indicates a greater level of agglomeration.
The in¯uence of agglomerates on powder packing eciency is signi®cant and has been an important subject of considerable interest [1,9,17]. The packing eciency in a powder compact can be explicitly and directly demonstrated in terms of pore size distribution of the green compacts. Fig. 3 showed the selected pore distributions
Table 1
The measured mean particle/agglomerate size (d50) and corresponding
agglomeration parameter (AP) of the zirconia powder after dierent time periods of ball milling in this investigation
Milling time (h) d50(mm) AP (=d50/dprimarya)
As-received 1.208 5.49 0.5 1.091 4.96 2 1.071 4.87 4 1.028 4.67 8 0.995 4.52 10 0.992 4.51 16 0.933 4.24 24 0.823 3.72 48 0.681 3.1 a d primary=0.22 mm.
Fig. 2. Particle (or agglomerate) size distribution of the powders after varying time periods of ball milling.
of the powder compacts after varying milling times. The pore distributions are uni-model and show somewhat sharper/narrower characteristic, particularly pro-nounced when the milling time is increased. This clearly indicates an improved particle packing eciency can be reached by extensive milling. The green densities of the compacts increased from 41.6 to 54.2% of theoretical density with increased milling time from 0.5 to 48 h. An extensive agglomeration (e.g. for short milling-time powder) causes green microstructure inhomogeneity by developing large interagglomerate pores (Fig. 4).
It has been well-recognised that agglomerates fre-quently reduce green compact density of a given powder by inhibiting particle packing. How such inhibition is related to the degree of powder agglomeration in a given powder would be critically important in powder processing control. An attempt to relate the powder agglomeration with respect to the green compact den-sity is made and the resulting relation is given in Fig. 5.
The compacts used to construct Fig. 5 were made under identical green-shape processing, i.e. casting without pressing, those with pressuring were excluded. The green density exhibits a roughly linear relation against the agglomeration parameter over the range of approximately 3±5.5. This is consistent with that observed by compaction of agglomerate-containing alumina powder reported by Dynys et al. [5]. Although the AP range is narrow (it represents essentially a status of current limitation on powder processing and mostly ascribed to the given nature of the powder used), as a ®rst approximation, the green compact density follows a linear dependence on the as-de®ned agglomeration parameter of the powders. In other words, the powder in the suspension is likely to deposit in a linear fashion (at least within the AP range of current study) related to the starting level of agglomeration in the powders. This ®nding implies that a potential correlation can be deduced between the de®ned AP value and particle packing eciency for the commercial powders and similar concept may be applicable to other ceramic powders. By extrapolating to AP=1, i.e. ®nely-divided particles, the obtained density is about 71.3%, which is relatively closed to the theoretical value (74%) for close-packing spheres of identical diameter obtained by McGeary [18]. A small dierence between these values may result from the fact that the powders currently-used are not exactly spherical in shape and identical in size. Applying pressure causes an increase in green density of the compacts by a density increment of 4.5±8% in comparison to those without pressing (will be seen in a later discussion). However, such an improvement in packing density is observable only for high-agglomer-ated compacts, i.e. for AP4.51, as illustrhigh-agglomer-ated in terms of the variation in pore/pore volume distribution in Fig. 6 for AP=4.96 compacts, and the density has little or no improvement for low-agglomerated compacts
Fig. 5. Green compact density as a linear function of powder agglomeration.
Fig. 3. Pore size distribution of the powder compacts consolidated with the powders under varying ball-milled times.
(higher starting density). As illustrated in Fig. 6, a higher compaction pressure tends to eliminate larger voids which were formed due to agglomerates, and shifts the pore distribution toward a small-sized region, which substantiates a further removal of larger voids. Similar observation was also reported by Zheng et al. [19] on the dry compaction of alumina powder. The pore size at maximum frequency is reduced by about 12% from 121.8 to 107 nm when 50 MPa pressure was applied. A calculation based on this shift indicated a reduction of approximately 9 vol% of voids greater than the pore at maximum frequency is eliminated. The fractional change in pore volume for those pore size smaller than 0.155 r (where r represents the radius of primary particle and the value of 0.155 r indicates the largest radius of a void developed by closed packing of three identical spheres) which is assumed to be con-structed by the agglomerated particles (i.e. intraagglo-merate pores) is relatively unchanged. This observation further supports the retention of original agglomerate structure as previously discussed. The plastically ``com-pressible'' characteristic for high-AP compacts indicates that the powders consolidate in a ``loose'' random packing structure within which the voids constructed by either interagglomerate or agglomerate-particle contact can be further removed by deformation or fracturing the ``weak'' structure (not agglomerates themselves as con®rmed by almost identical particle distribution and invariant intraagglomerated pore volume before and after pressing) or alternatively forcing the agglomerates/ particles to slide over one another to a certain extent upon pressing. On the contrary, the seemingly plasti-cally ``incompressible'' low-AP compacts suggest a ``strong'' interagglomerate or agglomerate-particle con-tact structure; in other words, the agglomerates and
particles were lock somewhat ®rmly (because the pres-sures used to further consolidate the cast are not rela-tively high and a higher pressure may again cause the cast to become ``compressible'') in position during initial ®ltration.
As the primary concern of sintering, a complete elimination of pores within a compact is one of impor-tant requirements in modern technical ceramics. A recent investigation by Zheng et al. [20] indicated that a ratio of pore size to particle size as one relatively important parameter in determining whether the pores can be completely removed after sintering, leaving a full-dense (also for small-grained) ceramic body. The smaller the ratio, the easier the compact's pore can be eliminated. Therefore, the pore size/particle size ratio is essentially crucial to the green compact (here we do not intend to determine what the critical ratio is in obtaining full-dense ZrO2body). Attempt is thus made to see how the ratio can be related to green density for compacts with a variety of ``agglomeration''. Fig. 7 showed the resulting trend. The pore size used here is the average pore dia-meter calculated by intrusion volume-pore diadia-meter data from mercury porosimetry. In general, compacts with higher green density exhibits lower pore size/parti-cle size ratio, suggesting the pores are more likely to be eliminated than those in lower-density compacts. With-out pressing (circle symbols), compacts with higher AP value have lower density and higher pore size/particle size ratio. A further compaction improves packing den-sity and reduces the ratio even when the AP value is high, e.g. for AP=5.13 compacts. Under pressing, lar-ger voids removed and the agglomerates appeared to have no signi®cant eect on resulting packing structure. This limited inhibition of agglomerates on particle packing under pressing might be due to somewhat small-sized agglomerates in the powders, generally dis-tributed over the range of 0.5±3 m.
Fig. 7. Green density of the compacts as a function of pore size/par-ticle size ratio.
Fig. 6. The variation of pore size distribution of the compacts with AP=4.96 after pressing at 50 MPa, where the larger voids were removed.
3.2. Shrinkage behaviour
Fig. 8 showed the shrinkage rate of the powder com-pacts of varying milling times (expressed in terms of green densities g) with respect to the relative sintered density. Compacts with a lower starting density exhi-bit a higher shrinkage rate and this is consistent with those reported in the literature [10,21]. As commonly observed in sintering experiments, Lange [2] recently suggested that the maximum shrinkage rate corre-sponds to an in¯ection in relative density (max) to be a transition from densi®cation kinetics to coarsening (i.e. grain-growth) kinetics. Below max, the sintering is dominated by densi®cation kinetics and above which coarsening kinetics dominates over further sintering schedule. This suggestion implies that the sintering pro-®le enables to re¯ect the potential microstructure evo-lution as well as the corresponding starting packing structure within a given powder compacts. The higher the sintered density achieved at corresponding max-imum shrinkage rate appears to ensure the potential evolution of desired sintered microstructure, i.e. ®ne-grained, dense ceramic body. This further indicates the potential bene®t for ceramics with desired micro-structure for the powder compacts with lower AP and higher g. Although the shrinkage rate is higher for low-g compacts, a lower end-point density usually results under current sintering schedule. This observation on which the low-gcompacts densi®es faster than high-g compacts is in accordance with an earlier observation by Bruch [10] in the sintering of Al2O3for the as-de®ned ``normal'' sintering regime and also consistent with that experimentally observed by Vasilos et al. [11] in ZrO2. However, a close examination of the compacts used to characterise Fig. 8, one can ®nd that two green compact parameters are essentially involved, i.e. green density and agglomeration, which may cause dierent eects on
the sintering. Therefore, a separate understanding of each parameter on sintering behaviour, particularly for a commercial ®ne ceramic powder as model material, is essentially important as can be seen in a number of publications.
The in¯uence of green density on sintering behaviour can be easily obtained by consolidating the given powder suspension (having a ®xed value of agglomeration para-meter) with varying pressures. Figs. 9(a) and (b) showed the density±shrinkage rate curves for the powder com-pacts of dierent green densities at a constant agglom-eration parameter (AP)=4.96 and 4.51, respectively. Clearly, compacts with lower starting density exhibit higher shrinkage rate at a given sintered density below approximately the region at which maximum shrinkage rate occurred. Furthermore, the sintered density at which maximum shrinkage rate (max) occurred is increased with increasing starting density and with decreasing agglomeration (Fig. 10); however, no simple analytical correlation is likely to be deducible under investigation. If according to the suggestion of Lange
Fig. 9. Shrinkage rate-sintered density curves for the compacts of dif-ferent green densities, each has a constant AP value of (a) 4.96 and (b) 4.51.
Fig. 8. Shrinkage rate-sintered density curves for compacts with dif-ferent ball-milled times (diering in both green density and degree of powder agglomeration).
[2], the present observation indicates that a dominant densi®cation kinetics should be operative for high-g and low-AP compacts up to sintered density of 74% theoretical (in fact, a higher max is achievable for higher-g and lower-AP compacts as shown latter in Fig. 11). As generally realised that a desired sintered microstructure would usually be expected.
The in¯uence of agglomeration on the sintering behaviour is shown in Figs. 11(a) and (b) for the com-pacts with g 45 and 54%, respectively. Compacts with lower AP have generally a lower shrinkage rate, but have a higher value of max. The highest max cur-rently attainable is about 88% for the compacts with g 54% and AP=3.1. This observation clearly indi-cates the inhibition of densi®cation due to the presence of agglomerates. An investigation by measuring the sintered density (via Archemede's principle) at various sintering temperatures for compacts (g 45%) of var-ious AP values, as illustrated in Fig. 12, further justi®ed that high-AP compacts are generally more dicult to densify than do for low-AP compacts. Similar result was also observed for g 54% compacts. A dierential densi®cation between agglomerate and matrix particle causing a further re-opening of crack-like pores at agglomerate±particle interfaces may be explainable as one of the prime factors for such poor densi®cation [3]. Such agglomerate-induced inhibition not only reduces the sinterability of the powder compacts but also pro-motes the onset of the transition of sintering mechanism from densi®cation kinetics to coarsening kinetics at lower sintered density. As generally realised that under such circumstance, the voids within the sintered compacts may frequently be retained and larger grains developed, which would accordingly largely reduce a desired per-formance of the ®nal products.
Plotting the max data with respect to the corre-sponding pore size/particle size ratio, Fig. 13 showed
Fig. 10. The sintered density at maximum shrinkage rate, i.e. max, changes with green density and AP of a given powder compact.
Fig. 11. Shrinkage rate-sintered density curves for the compacts of dierent AP values, each has a constant relative green density of (a) 45% and (b) 54%.
Fig. 12. Sintered density as a function of sintering temperature for compacts of identical green density but with varying degrees of agglomeration.
the resulting relation with each data point designated with a parenthesis of (AP/starting density). Generally, the smaller the ratio in a given compacts, the larger the max can be obtained. The high-gand low-AP ensure a small-pore, uniform green powder compact (i.e. narrow pore distribution) and which is able to reach a highest max at Tmax, suggesting a greater potential of develop-ing dense, pore-free sintered ceramic body. However, it also revealed that the pore size/particle size ratio under a higher AP, the compacts with even a higher gexhibit a somewhat larger value of the ratio, suggesting a potential diculty in completely eliminating the com-pact's voids [19]. The higher pore size/particle size ratio for the higher-AP compacts designated in Fig. 12, hav-ing lower sintered densities, seems to provide further supports.
Therefore, based on a combined knowledge from these literatured reports and current observations, one can conclude that improved starting density may not be of prime consideration ensuring desired sintered micro-structure/property, instead, a complete (or relatively extensive) elimination of the agglomerates associated with a reasonably ecient particle packing structure (reducing pore size/particle size ratio) would ascertain a satisfactory green property and consequent sintered property. However, due to some diculty in obtaining compacts of identical starting density over a wide vari-ety of AP, the ranges for both green density and AP under investigation are restricted. Although current investigations provide only a limiting data basis, the in¯uence of green density and agglomeration can be comparatively dierentiated.
3.3. Activation energy
As mentioned previously, the nonisothermal (con-stant-heating-rate) technique provides a simple and
reliable way to assess the sintering kinetics of powder compacts. Fig. 14 showed an Arrhenious-type plot, i.e. Td L=Lo=dT vs 1/T, for powder compacts with varying time periods of milling. Some low-temperature data scattered and the initial slopes of the lines were relatively high for both compacts (corresponding to a linear shrinkage below 0.5%). This has readily been observed in a number of related studies and is believed to be a result of several interferences by, e.g., surface diusion, thermal equilibrium, and/or instrumental limitation [13,15]. After which fractional shrinkage, the data points follow nearly a straight line up to a shrink-age of approximately 8.6% for both compacts (some up to 12%), indicating a single diusion mechanism dom-inates. A change in slope in this Arrhenious-type plot, e.g. above 8.6% shrinkage, indicates the change in mass transfer mechanism.
On checking, the straight lines in Fig. 14 for compacts of varying milling times (representing dierent values of green density and AP) showed slightly dierent slopes, indicating dierent nQ values. Such dierence might be caused by a number of factors such as measurement error and/or probably green microstructure inhomo-geneity. However, despite the possible in¯uence of above factors on sintering behaviour, the eective acti-vation energy for all these compacts is averaged to a value of 30.022.78 Kcal molÿ1 ( represents the standard deviation of the measurements), correspond-ing to a sintercorrespond-ing activation energy of 909 Kcal molÿ1 which is closely in agreement with the value obtained by Young et al. [13], 90 Kcal moleÿ1, and Jorgensen [22], 92.5 Kcal moleÿ1, in the sintering of yttria-stabilized zirconia powders under a grain-boundary diusion mechanism (n 1=3). This agreement together with reasonable scale of standard deviation implies that both green density and agglomerate may exert relatively small or negligible eect on the sintering law or
acti-Fig. 14. A Td L=Lo=dT ÿ 1=T plot for the powder compacts con-solidated by a 4-h and a 24-h milled powder.
Fig. 13. Sintered density versus pore size/particle size ratio, indicating maxincreased with decreasing ratio.
vation energy for sintering of the zirconia powder; the former factor observed presently is in agreement with a previous report by Woolfrey [23] while the latter may be considered as a new understanding.
4. Conclusion
In¯uence of green density and powder agglomeration on the sintering behaviour of zirconia powder was investigated. Both factors are interdependent and a lin-ear density±agglomeration relation is experimentally deduced over a small range of agglomeration. They have shown to aect in dierent extents the sintering behaviour over the entire sintering schedule. Low-den-sity powder compacts (usually are high-agglomerated compacts) exhibited higher initial shrinkage rate in comparison to high-density compacts (low-agglomer-ated compacts), which is consistent with the observation by other researchers. Less agglomeration in the starting powder and associated with a smaller pore size/particle size ratio after consolidation ensure a better sinterability and potential evolution of desirable sintered micro-structure of the ceramic. The variation in green density and agglomeration in the powder compacts showed a relatively small or negligible eect on activation energy or sintering law for sintering of the ceramic powders. The use of the de®ned agglomeration parameter to characterize powder agglomeration in commercial ®ne ceramic powders appeared to provide as a useful indi-cation in characterizing the sintering behaviour and possible evolution of sintered microstructure by refer-ring to the corresponding density-shrinkage rate curve upon sintering.
Acknowledgements
The authors are indebted in part to the National Sci-ence Council and in part, to the Minister of Economic Aair, Taiwan, for supporting this research work under contract no. 863KG2230.
References
[1] A. Roosen, H.K. Bowen, In¯uence of various consolidation techniques on the green microstructure and sintering behaviour of alumina powders, J. Am. Ceram. Soc. 71 (11) (1988) 970±977.
[2] F.F. Lange, Powder processing science and technology for increased reliability, J. Am. Ceram. Soc. 72 (1) (1989) 3±15. [3] F.F. Lange, B.I. Davis, I.A. Aksay, Processing-related fracture
origins: III dierential sintering of ZrO2agglomerates in Al2O3/
ZrO2composite, J. Am. Ceram. Soc. 66 (6) (1983) 407±408.
[4] M.D. Sacks, J.A. Pask, Sintering of mullite-containing materials II. Eect of agglomeration, J. Am. Ceram. Soc. 65 (2) (1982) 70±77. [5] F.W. Dynys, J.W. Hallron, In¯uence of aggregates on sintering,
J. Am. Ceram. Soc. 667 (9) (1984) 596±601.
[6] W.H. Rhodes, Agglomerate and particle size eect on sintering yttria-stabilized zirconia, J. Am. Ceram. Soc. 64 (1) (1981) 19±22. [7] F.F. Lange, Sinterability of agglomerated powders, J. Am.
Ceram. Soc. 67 (2) (1984) 83±89.
[8] R.E. Mistler, R.L. Coble, Microstructural variation due to fabri-cation, J. Am. Ceram. Soc. 51 (4) (1968) 237.
[9] M.A.C.G. van de Graaf, K. Keizer, A.J. Burggraaf, In¯uence of agglomerate structures on ultra®ne substituted zirconia powders on compaction and sintering behaviour, Sci. Ceram. 10 (1979) 83±92. [10] C.A. Brush, Sintering kinetics for the high density alumina
pro-cess, Am. Ceram. Soc. Bull. 41 (12) (1962) 799±806.
[11] T. Vasilos, W.H. Rhodes, Fine particulates to ultra®ne-grain ceramics, in: Ultra®ne-Grain Ceramics, J.J. Burke, N.L. Reed, V. Weiss, (Eds.), Syracuse University, Syracuse, NY, 1970, pp. 137±172.
[12] A. Roosen, H. Hausner, Sitering kinetics of ZrO2powders, Adv.
Ceram. 12 (1984) 714±726.
[13] W.S. Young, I.B. Cutler, Initial sintering with constant rates of heating, J. Am. Ceram. Soc. 53 (12) (1970) 659±663.
[14] J. Wang, R. Raj, Estimate of the activation energies for boundary diusion from rate-controlled sintering of pure alu-mina, and alumina doped with zirconia or titania, J. Am. Ceram. Soc. 73 (5) (1990) 1172±1175.
[15] J.L. Woolfrey, M.J. Bannister, Nonisothernal techniques for studying initial-stage sintering, J. Am. Ceram. Soc. 55 (8) (1972) 390±394.
[16] D.L. Johnson, New method of obtaining volume, grain-bound-ary, and surface diusion coecients from sintering data, J. App. Phys. 40 (1) (1969) 192±200.
[17] I.A. Aksay, Microstructure control through colloidal consolida-tion. In Advance in Ceramics, Vol. 9, Forming of Ceramics, J.A. Mangles (Eds.), American Ceramics Society, Columbus, OH, 1984, pp. 94±104.
[18] R.K. Mcgeary, Mechanical packing of spherical particles, J. Am. Ceram. Soc. 44 (10) (1961) 513±522.
[19] J. Zheng, J.S. Reed, The dierent roles of forming and sintering on densi®cation of powder compacts, Am. Ceram. Soc. Bull. 71 (9) (1992) 1410±1416.
[20] J. Zheng, J.S. Reed, Eects of particle packing characteristics on solid-state sintering, J. Am. Ceram. Soc. 72 (5) (1989) 810±817. [21] M.A. Occhimero, J.W. Holloran, The in¯uence of green density
upon sintering, in: Materials Science Research, Vol. 16, Sintering and Heterogeneous Catalysis, G.C. Kuczynski, A.E. Miller, G.A. Sargent (Eds.), Plenum Press, 1984.
[22] P.J. Jorgesen, in: Sintering and Related Phenomena, G.C. Kuczynski, N.A. Hooton, C.F. Gibbon, (Eds.), Gordon and Breach, New York, 1967 pp. 407±418.
[23] J.L. Woolfrey, Eect of green density on the initial-stage sinter-ing kinetics of UO2, J. Am. Ceram. Soc. 55 (8) (1972) 383±389.
