- Email: [email protected]
Functional composites based on refractories produced by pressure slip casting
G Model
ARTICLE IN PRESS
JECS-10516; No. of Pages 9
Journal of the European Ceramic Society xxx (2016) xxx–xxx
Contents lists available at www.sciencedirect.com
Journal of the European Ceramic Society journal homepage: www.elsevier.com/locate/jeurceramsoc
Functional composites based on refractories produced by pressure slip casting Stefan Schafföner ∗ , Lisa Freitag, Jana Hubálková, Christos G. Aneziris Institute of Ceramic, Glass and Construction Materials, TU Bergakademie Freiberg, Agricolastrasse 17, 09599 Freiberg, Germany
a r t i c l e
i n f o
Article history: Received 19 December 2015 Accepted 3 February 2016 Available online xxx Keywords: Refractories Composites Pressure slip casting X-ray computed tomography Spinels
a b s t r a c t The present study investigates the sequential pressure filtration of two filter cake layers with a maximum grain size of 3 mm and 1 mm, respectively, using a water soluble additive system consisting of xanthan and guar gum. The pressure slip casting of the resulting composites was optimized with a factorial experimental design. The best combination was a lower filtration pressure for the first filter cake layer and a higher filtration pressure for the second one. Furthermore, a shorter filtration time of the first filter cake layer together with the application of a filter aid of a xanthan/guar gum solution resulted in an easy demolding of crack-free graded composites with an excellent bonding. The results of the full factorial experimental design were verified by X-ray computed tomography. The demonstrated pressure slip casting of graded refractories offers the possibility to produce functional composites with tailored compositions, microstructure and thus functionality. © 2016 Elsevier Ltd. All rights reserved.
1. Introduction Pressure slip casting is state of the art for the production of silicate ceramics with complex shapes such as whiteware and sanitary ware [1–6]. Compared with conventional slip casting, the process time for pressure slip casting is much shorter. The durability of the used polymer molds is also much longer than plaster of Paris molds and they do not need any drying between the casting cycles. Furthermore, pressure slip casting can be automated to a large extent. Thus, pressure slip casting offers a significantly higher efficiency compared to conventional slip casting [1,3–5,7,8,]. Besides, technical ceramics produced by pressure slip casting also exhibit an improved microstructure with less agglomerates and superior mechanical properties [2–4,9,10]. Conventional and pressure slip casting was also repeatedly investigated to produce fine graded composites [5,11–19,20]. By changing the slip composition discrete or continuous gradients of the filter cakes can be achieved. The investigated compositions ranged from oxide ceramics including alumina–zirconia mixtures [5,11–13] over non-oxide ceramics [14–16] to ceramicmetal composites [17,18,20]. Besides changing the ceramic phase composition along the gradient in the filter cakes it is also possible to tailor other characteristics such as the porosity using layered pressure slip casting [21].
In recent years pressure slip casting was also studied as a new forming technique for coarse grained oxide ceramics such as refractories [6,22]. By combining coarse and fine grain fractions the thermal shock, corrosion, and erosion resistance of refractories is optimized. Furthermore, the sintering shrinkage is also reduced to a minimum, which allows the production of large sized components. Klippel et al. [22] focused especially on the rheology of suitable slips and demonstrated the casting of coarse grained ceramics using a pressure filtration cell. In a further study Schafföner and Aneziris [6] reported the production of large plate shaped oxide ceramics with only minimal gradients caused by sedimentation. The addition of larger particle sizes had the effect of a reduced Young’s modulus, which is an often observed feature of refractories. However, only the production of coarse grained oxide ceramics from a single slip composition was reported and the production of coarse grained composites by pressure slip casting remains unstudied. The purpose of this study is to describe and examine the pressure slip casting of functional composites based on refractories. The approach in this study is to use sequential pressure slip casting of magnesium aluminate spinel slips. Thereby the overall aim was to produce refractories with a high quality bonding of filter cakes with different particle size distributions.
2. Experimental ∗ Corresponding author. E-mail address: [email protected] (S. Schafföner).
In the present study graded composites made of filter cakes of the same material but different particle size distributions were
http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008 0955-2219/© 2016 Elsevier Ltd. All rights reserved.
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008
G Model
ARTICLE IN PRESS
JECS-10516; No. of Pages 9
S. Schafföner et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx
2 Table 1 Batches of the slips. Product name
Raw material
Almatis AR 78
MgAl2 O4
Martoxid MR 70
Al2 O3
Water for binder Axilat RH 50 MD Food grade guar gum Additional water
H2 O Xanthan gum Guar gum H2 O
Grain size fraction
Slip S3
Slip S1
1–3 mm 0.5–1 mm 0–0.5 mm 0–0.045 mm –
15 wt.% 10 wt.% 30 wt .% 10 wt.% 20 wt .%
0 wt.% 11.76 wt.% 35.3 wt .% 11.76 wt .% 23.53 wt .% 12 wt .% 0.024 wt .% 0.024 wt .%
Mass relative to dry mass
produced by pressure slip casting. The casting of graded composites was optimized using a full factorial experimental design. In all experiments slip compositions similar to the ones presented by Schafföner and Aneziris [6] were applied. The used raw materials were alumina rich magnesium aluminate spinel (AR 78, Almatis GmbH, Germany) and reactive alumina (Martoxid MR 70, Martinswerke, Germany) for the finest particle fraction. For the experiments two different slip compositions with a maximum grain size of 1 mm (S1 ) and 3 mm (S3 ), respectively, were prepared. The batches are summarized in Table 1. In order to prevent a sedimentation or segregation of the coarse grain fraction, dissolved xanthan (Axilat RH 50 MD, C.H. Erbslöh GmbH & Co. KG, Germany) and food grade guar gum were used as stabilizers because these two hydrocolloids show a strong positive interaction regarding their viscosity [23]. As a rule the added amount was 0.024 wt .% for both stabilizers regarding the raw material dry mass. The xanthan/guar gum solution did not only act as a stabilizer to prevent a possible sedimentation, but also acted as a binder because the viscosity of this hydrocolloid solution depends strongly on the water content. During the filtration of the slips, the water content decreases which hence increases the viscosity of the binder/water solution. Thereby a sufficient green strength of the filter cakes can be attained. Prior to mixing the slip, the binder solution was prepared using 12 wt .% water regarding the dry mass by carefully adding the binders separately to half of the water using a high shearing laboratory mixer (RZR 2102 control, Heidolph Instruments GmbH & Co. KG, Germany). Afterwards, the two separate binder solutions were mixed together. In all experiments 3 kg of dry mass were used for each slip. For the slip preparation the dry mass was first mixed for 4 min using a standard concrete mixer (ToniMIX, Toni Technik Baustoffprüfsysteme GmbH, Germany). The slips were in general thixotropic, therefore the mixer was kept at the lowest power level to avoid a strong stirring and hence decrease in viscosity of the slips. After the dry mixing, the binder solution was added in three steps to ensure a homogeneous mixture. After each adding the slip was stirred for 3 min followed by a careful scrapping of the mixer walls to remove any lumps. Finally an amount of 3 wt .% and 4 wt .% of distilled water was added to the S1 and S3 slips, respectively. The slips were then stirred for 3 min for a last time. Shortly after preparing the slips, the water content was controlled using a thermobalance (MA 30, Sartorius AG, Germany) in order to ensure a comparable slip composition in the experiments. For all the filtration experiments a laboratory pressure filtration cell was applied. The pressure filtration cell operated similar to a pressure casting machine using pressurized air as the pressure medium. It allows the testing of a wide range of factors before performing upscaling experiments. The filter medium of the pressure filtration cell was poly(methyl methacrylate) (PMMA), which was identical to the usual filter medium material used in commercial pressure slip casting machines. The filter medium had a filtration
3 wt .%
4 wt .%
area of 33.7 cm2 and was supported by a porous metal plate with a relative open area of 40.2%. In all cases the slips were first filled into the pressure filtration cell, then sealed before the pressure was abruptly increased using a lever valve. The pressure was held constant during the whole filtration time. In preliminary experiments the filtration properties of both slips were first separately investigated. For that purpose the slips were filled with a height of 20 mm into the pressure filtration cell and were then filtrated with a pressure of 5 bar and 7 bar, respectively, until the dewatering of the filter cake started. The dewatering starts at the point when the cake filtration of the slip ceases, i.e. the filter cake is completely built up and no slip remains on the filter cake surface. With the beginning of the dewatering the remaining water of the filter cake is then displaced by air [22,24,25]. The start of the dewatering was clearly perceivable by the time that a fizzling sound became evident, which was caused by the breakthrough of pressurized air. The start of the filter cake dewatering was defined as the termination criterion for the filtration and the necessary time to reach the dewatering was thus recorded for later experiments at each pressure. The following experiments regarding the graded filter cake composites were mainly conducted using a 24 full factorial experimental design because factorial designs have several advantages. They allow the simultaneous study of several factors, i.e. independent variables, and also of so called interactions, i.e. the influence of one factor on the other, with only a minimum of experiments [26]. The overall aim was to achieve a composite of high quality which can be easily demolded at the same time. The factorial experimental design is summarized in Table 2 in conjunction with Table 3. All of the following statistical analyses were conducted with the software package R [27]. The general procedure for the casting of graded filter cakes was always to filtrate the slip S1 (dmax = 1 mm) on top of a filter cake of the S3 slip (dmax = 3 mm). In the following the chosen factors of the factorial experimental design will be described. The first factor (A) was the use (+1) or nonuse (−1) of a so called binder filter aid. This binder filter aid had the aim to improve the demolding of the filter cake and to prevent the penetration and thus possible clogging of the filter medium
Table 2 Factors of the 24 full factorial experimental design. Factor
Identifier
Lower level (−1)
Higher level (+1)
Binder filter Filtration pressure of first (coarse) slip S3 Filtration pressure of second (finer) slip S1 Filtration time relative to beginning of filter cake dewatering
A B
Without 5 bar
With 7 bar
C
5 bar
7 bar
D
75%
100%
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008
G Model
ARTICLE IN PRESS
JECS-10516; No. of Pages 9
S. Schafföner et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx
3
Table 3 24 Full factorial experimental design for the pressure slip casting of graded filter cakes including results for the investigated responses, the filter medium and filter cake resistances, and the apparent porosity. Run
Order
Factors A
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
14 16 12 10 2 8 4 6 13 15 11 9 1 7 3 5
−1 +1 −1 +1 −1 +1 −1 +1 −1 +1 −1 +1 −1 +1 −1 +1
˛av,3
Rm,3
˛av,1
Rm,1
Y1 (1–12)
Y2 (1–12)
Y3 (1–9)
Apparent porosity
(m kg−1 )
(m−1 )
(m kg−1 )
(m−1 )
(%)
10 12 5 8.5 4.5 12 7 7 6.5 10 5 8 6 12 5 8
11 11 6 11 10 11 10 9 11 9 10 11 10 10 9 8
7 8 7 5 9 8 8 9 4 4 4 3 6 6 6 6
1.34 × 1013 8.82 × 1012 2.22 × 1013 1.35 × 1013 1.41 × 1013 7.81 × 1012 1.01 × 1013 9.48 × 1012 1.34 × 1013 9.81 × 1012 1.87 × 1013 1.01 × 1013 1.24 × 1013 1.19 × 1013 1.82 × 1013 7.78 × 1012
2.16 × 1014 2.29 × 1014 5.32 × 1013 1.74 × 1014 1.73 × 1014 1.53 × 1014 2.16 × 1014 1.54 × 1014 1.99 × 1014 3.16 × 1014 2.18 × 1014 2.25 × 1014 2.50 × 1014 1.24 × 1014 1.65 × 1014 1.99 × 1014
2.09 × 1013 1.11 × 1013 2.34 × 1013 1.56 × 1013 1.74 × 1013 1.11 × 1013 1.67 × 1013 1.23 × 1013 1.79 × 1013 1.40 × 1013 1.67 × 1013 1.20 × 1013 1.82 × 1013 8.20 × 1012 1.82 × 1013 6.32 × 1012
4.40 × 1014 6.40 × 1014 6.93 × 1014 5.87 × 1014 5.34 × 1014 4.35 × 1014 3.95 × 1014 4.93 × 1014 7.28 × 1014 6.45 × 1014 8.70 × 1014 8.70 × 1014 7.50 × 1014 5.82 × 1014 5.82 × 1014 7.39 × 1014
16.9 17.4 16.0 16.3 15.4 17.2 15.6 16.0 17.4 17.9 16.6 17.2 15.8 17.8 16.3 16.5
Responses B −1 −1 +1 +1 −1 −1 +1 +1 −1 −1 +1 +1 −1 −1 +1 +1
C −1 −1 −1 −1 +1 +1 +1 +1 −1 −1 −1 −1 +1 +1 +1 +1
D −1 −1 −1 −1 −1 −1 −1 −1 +1 +1 +1 +1 +1 +1 +1 +1
by the fine particle fraction of the slips. For this purpose a small amount of the xanthan/guar gum binder solution (10 ml) was first filtrated with 1 bar before filtrating the S3 slip. Thus, a thin skin of the filtrated binder solution was formed on the filter medium. The second investigated factor (B) was the filtration pressure of the first slip, i.e. the coarse grained slip S3 , because during filtration the filter cakes might exhibit some degree of compressibility. In case of compressible filter cakes a higher filtration pressure does not necessarily significantly increase the filtration speed because highly compacted filter cakes increase the filtration resistance [28]. Therefore, the filtration pressure was varied between 5 bar (−1) and 7 bar (+1). The third investigated factor (C) was accordingly the filtration pressure of the second layer, i.e. the S1 slip. The filtration pressure was again varied between 5 bar (−1) and 7 bar (+1). The fourth and last factor D was the filtration time of the first slip layer S3 . At the lower level (−1) the filtration time was only 75% of the time needed until the filter cake dewatering started, therefore a slip layer was still present on the top of the slip. At the higher level (100%, +1) the slip was filtrated until the filter cake dewatering stage was reached. The necessary filtration time as a function of the applied pressure and the usage of the binder filter aid was determined in the preliminary experiments described above. In order to determine the filtration properties of the slips, the filtrate mass was always registered with a scale below the pressure filtration cell and recorded with a computer in an interval of 1 s. To reduce the time offset until the first filtrate was registered, the filter medium was always immersed into distilled water before mounting. Thus, the filtrate always displaced other water in the filter medium. The data were later used to determine the filtration properties of the filter cakes using the conventional cake filtration theory (see also Eq. (1)) [28]. After filtration the filter cakes were carefully demolded using a circular die which fitted exactly in the tubular pressure filtration cell. Not once this demolding caused any cracking. The filter cakes were first evaluated regarding the bottom surface quality and the effort to remove the filter cakes from the filter medium (Y1 ). Furthermore, the filter cakes were examined on the overall quality of the filter cake, henceforth called response Y2 . As a third response (Y3 ) the joining of the composite, i.e. the connection of the top (FC1 ) and bottom layer (FC3 ) of the filter cake, was analyzed. In all cases an ordinal scale was applied to evaluated each response. The ordinal scale ranged from 1 (worst) to 12 (best) for Y1 and Y2 and 1 (worst) to 9 (best) for Y3 .
Later the filter cakes were carefully dried at 40 ◦ C, 60 ◦ C, 80 ◦ C, and 110 ◦ C for 6 h at each temperature. The drying process was not optimized, however it was known that a carefully drying is necessary to prevent eventual cracking. After drying the filter cakes were fired in air for 4 h at a temperature of 1650 ◦ C with a heating and cooling rate of 3 K min−1 . After firing all filter cakes were investigated on their apparent porosity and apparent density using the Archimedes method according to DIN EN 993-1 with water as the immersion fluid. In order to evaluate the inner structure, the four filter cakes which achieved the best overall values regarding Y1 to Y3 were examined with a micro focus X-ray computed tomograph (XCT) (CT-ALPHA, ProCon X-Ray GmbH, Germany) similar to Schafföner et al. [29]. The XCT was equipped with a transmission X-ray tube (YXLON International Feinfocus GmbH, Germany) and a detector (Dexela, PerkinElmer Inc., USA) with a total area of 120 × 150 mm2 and a pixel size of 75 m. The whole samples were scanned with a voltage of 150 kV and a current of 90 A; the target power was 11.5 W. The edge length of the voxels after reconstruction was about 98 m. The reconstructed XCT data were processed and visualized with the software package VGStudio Max 2.2 (Volume Graphics GmbH, Germany). Although beam hardening artifacts occurred, no regions of interest (ROI) were extracted in order to obtain a full image of the graded filter cakes. Finally, to investigate the interface between the two coarse grained filter cake layers, cut samples were investigated with a digital microscope (VHX-2000, Keyence Corp., Japan). 3. Results and discussion The aim of the present study is to produce graded functional composites based on refractories using the pressure slip casting technique. For pressure slip casting the filtration properties can be explained by t/V vs. V plots. Fig. 1, for example, presents the t/V vs. V values of S3 slips as a function of the applied pressure. The plotted graphs express the filtration properties according to the conventional cake filtration equation at constant pressure: (˛av cV + ARm ) dt = dV A2 · p
(1)
where t is the filtration time, V the filtrate volume, A the filter area, the dynamic filtrate viscosity, ˛av the average specific resistance of the filter cake, c the effective concentration of solids in the
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008
G Model
ARTICLE IN PRESS
JECS-10516; No. of Pages 9
S. Schafföner et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx
1e+13
2.5e+08 1.0e+08
High pressure (7 bar) Cake filtration
0e+00
2e−06
4e−06
6e−06
8e−06
Fig. 1. Comparison of the filtration of the slips S3 of the runs 13 (low filtration pressure, 5 bar) and 15 (high filtration pressure, 7 bar). Both slips were filtrated without using the binder filtering aid and until the cake dewatering was reached.
slip, p the applied filtration pressure, Rm the resistance of the filter medium. Integrating Eq. (1) with the initial conditions t = V = 0 results in a linear equation: K1 t = V + K2 V 2 where ˛av c K1 = 2 A p
(2)
(3)
and Rm Ap
(4)
Eq. (2) in conjunction with the Eqs. (3) and (4) can then be used to determine ˛av and Rm of each filter cake by approximating the t/V vs. V data with a linear equation. The necessary effective solid concentration in the slip can be calculated according to the following equation: c=
l s 1 − ms
ME
A
B
D
AB
AD
BD
ABD
Factors
Filtrate volume in m3
K2 =
5e+12 −1e+13
5.0e+07
Filter cake dewatering
0e+00
Filter cake dewatering
Effects in m kg
2.0e+08 1.5e+08
Cake filtration
0.0e+00
Time per filtrate volume in s m3
Low pressure (5 bar)
ME
−5e+12
4
(5)
where l is the liquid of the dispersing medium of the slip, i.e. water, s the mass fraction of solids in the slip, and m the ratio of the mass of the wet cake to the mass of the dry cake. To determine m of the graded filter cakes, the same value was applied for both cake layers because only the wet and dry mass of the graded filter cakes was determined. Due to the large filtration resistance of the filter cakes and due to the setup of the filtration pressure cell, the time offset between the time of pressure increase to start the filtration and the registration of the first filtrate volume was often quite long. Therefore the calculations according to Eq. (2) took only the time after registrating the first filtrate into account and ignored the time offset. The steady-state cake filtration in Fig. 1 can be approximated by a line expressed in Eq. (2). Only values that fell in a 95% confidence interval around the approximated line were included to approximate the line. The coefficients of the line, i.e. slop and intercept, were thus determined by iteration. All determined values regarding the filter medium resistance Rm , the average filter cake resistance ˛av , and the concentration of the slips are listed in Table 3. In the following the results of the full factorial experimental design for these responses will be evaluated. Fig. 2 presents the effects of the full factorial experimental design for the average filter cake resistance of the filter cakes made up of the S3 slips (˛av,3 ). The effects are plotted in a so called
Fig. 2. Lenth plot of the effects regarding the average filter cake resistance made up of the S3 slips (˛av,3 ). Only effects that fall above or below the line of the margin of error (ME) are considered as possibly active. In this case, however, not a single effect was identified as active.
Lenth plot [30]. Lenth plots have the advantage that the results of an unreplicated factorial experimental design can be effectively presented. Furthermore, active and inactive effects can be easily distinguished. For the analysis of the resistance of the filter medium (Rm,3 ) and of the average filter cake resistance of the composition S3 (˛av,3 ) the filtration pressure of the second slip, i.e. the factor C, was not taken into account because the filtration of the second slip took always place after the measurements of the filtration of the S3 slips. On the other hand, excluding the factor C for the analysis of Rm,3 and ˛av,3 increases the degrees of freedom for the statistical tests and thus their power. The overall mean of ˛av,3 was 1.26 × 1013 m kg−1 . Such a specific filtration resistance is considered as very high [28] and is most probably caused by the broad particle size distribution and by the considerable fine particle content of the slips. As can be seen in Fig. 2 no factor was identified as significantly active regarding ˛av,3 . Especially, since the effect of the filtration pressure (B) was determined as inactive in the full factorial design on the chosen pressure levels, the compressibility of the filter cakes of S3 composition (FC3 ) can be regarded as negligible. Filter cakes of a high compressibility, on the other hand, would have been highly influenced by the filtration pressure. A low compressibility of the filter cakes FC3 can be regarded as an advantage because filter cakes with a low compressibility have generally a more homogeneous microstructure [28]. Fig. 3 provides the Lenth plot for the filter medium resistance of S3 composition (Rm,3 ). Similar to ˛av,3 no effect was identified as significant. This result can be again attributed to the lack of compressibility of the filter cake layer FC3 . On the other hand, in case of compressible filter cakes the pressure drop over the filter cake and filter medium are influenced by the filtration pressure. Thus, compressible filter cakes often exhibit a changing filter medium resistance as a function of the filtration pressure [28]. Furthermore, the analysis also revealed that the binder filter aid (A) did not influence Rm,3 . Fig. 4 presents the Lenth plot of the average filtration resistance (˛av,1 ) of the filter cake layers of the S1 composition (FC1 ). The binder filter aid (A) and the filtration pressure of the FC1 filter cake layer (C) had a negative effect on the filtration resistance of the second filter cake layer. Thus, the binder filter aid and a higher filtration pressure of the second filter cake layers helped to reduce the average filtration resistance of the second layer of the graded filter cake.
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008
G Model
ARTICLE IN PRESS
JECS-10516; No. of Pages 9
S. Schafföner et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx
−5e+13
Effects in m−1 0e+00
5e+13
ME
ME
A
B
D
AB
AD
BD
ABD
Factors Fig. 3. Lenth plot of the effects regarding Rm,3 of the first filter cake layer FC3 . Again not a single effect was identified as active, either.
The overall mean of ˛av,1 was 1.50 × 1013 m kg−1 and thus on average 21.5% higher than the overall mean of ˛av,3 . A paired Student’s t-test also indicated a significant difference in the means of the two average filtration resistances (p < 0.01). The results of the full factorial design regarding ˛av,1 seem to contradict the above presented result regarding ˛av,3 because no effect was identified as significant for ˛av,3 . Yet, comparing the Lenth plot of the effects for both filter cake layers in Figs. 2 and 4 it can be observed that for the average filtration resistance of both filtration cake layers the effect of the binder filter aid (A) was quite pronounced. But, for ˛av,3 the so called margin of error (ME) was much larger than for ˛av,1 . According to Lenth [30] ME is calculated from the 0.975th quantile of the t distribution on d degrees of freedom (t0.975;d ) and the pseudo standard error (PSE): ME = t0.975;d · PSE
(6)
The PSE, on the other hand, is calculated from the estimates of the contrasts of interest cj : PSE = 1.5 · median |cj |
(7)
|cj |<2.5s0
whereas s0 is defined as s0 = 1.5 · median|cj |
(8)
j
5
Examining Eqs. (7) and (8) it becomes clear that ME becomes inflated in case of several possibly active effects, or in case of a larger variance in the response values. Thus, whether an effect is active depends on the size of the estimate of the effect, i.e. cj , and on the value of ME. This also explains the difference in Figs. 2 and 4 because for ˛av,1 ME was much smaller compared to ˛av,3 (2.76 × 1012 m kg−1 and 9.40 × 1012 m kg−1 , respectively). A possible reason for the smaller ME value for ˛av,1 could be a smaller variance in the cake buildup because the FC1 filter cake layers were filtrated on top of the first filter cake layers FC3 . What is more, all active effects regarding ˛av,1 reduced the specific filtration resistance. Especially the negative effect of the filtration pressure of the second slip (C) is somewhat unusual because in case of a compressible filter cake layer FC1 an increasing filtration pressure would be expected on a higher level of the factor C, i.e. on a higher filtration pressure. The activity of the effects A and C regarding ˛av,1 can be interpreted in view of the fact that the filtrate from S1 slips has to flow through the already consolidated filter cake layers FC3 . Thus, according to the conventional filtration theory (Eq. (1)) these results imply that a higher specific filtrate volume is registered at the higher levels of the factors A and C. Fig. 5 presents the filter medium resistance of the filter cakes FC1 (Rm,1 ). In case of Rm,1 C and D are possibly active. A higher filtration pressure of the S1 slips, i.e. a higher level of the factor C, causes a decreasing Rm,1 . A higher pressure for the filtration of the second filter cake layer, as present at the higher level of the factor C, might contribute to overcome the filter medium resistance provided by the first filter cake layer FC3 . By contrast, a higher consolidation of the filter cake layer FC3 , as present at the higher level of the factor D, acts as a higher filter medium resistance for the filtration of the second filter cake layer FC1 . The overall mean of Rm,1 (6.24 × 1014 m−1 ) was 3.79 times higher than the overall mean of Rm,3 (1.92 × 1014 m−1 ). The much higher values for Rm,1 than for Rm,3 are caused by the sequential filtration of the two slips, whereas the filter cake layer FC3 acts as the filter medium for the filtration of the slip S1 . In the following the results of the full factorial experimental design on the properties of the graded filter cakes will be evaluated. The result for each run is presented in Table 3. As can be seen from Fig. 6 the surface quality of the demolded filter cakes and the effort to remove the filter cakes from the filter medium improved with the application of the binder filter aid (see Fig. 7). By contrast, the surface quality deteriorated and the effort to
1.5e+14
B A
D C
AC AB
AD BC
CD BD
ABD ABC
BCD
ACD
ABCD
Factors
ME
Effects in m−1 −5.0e+13 5.0e+13
SME
−1.5e+14
−2e+12
ME
−6e+12
Effects in m kg
2e+12
ME
B A
D C
AC AB
AD BC
CD BD
ME
ABD BCD ABC ACD ABCD
Factors Fig. 4. Lenth plot of the effects regarding average filtration resistance (˛av,1 ) of the filter cake layer made up by the S1 slip (FC1 ). The application of the binder filter aid prior to filtrating the slip S3 (A) and the filtration pressure of the second filter cake layer (C) were identified as probably active.
Fig. 5. Lenth plot of the effects regarding the resistance of the filter medium of the second filter cake layer (Rm,1 ). An increasing pressure of the second filter cake layer (C) reduced Rm,1 , whereas a longer filtration time of FC3 (D) increased Rm,1 .
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008
G Model
ARTICLE IN PRESS
JECS-10516; No. of Pages 9
S. Schafföner et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx
3
6
3
ME
−2 B A
D C
AC AB
AD BC
CD BD
ME
ABD BCD ABC ACD ABCD
−3
−3
−2
−1
−1
Effects 0
Effects 0 1
1
2
2
ME
B
D
A
C
AC AB
AD BC
Factors Fig. 6. The surface of the filter cakes and the effort to remove the filter cakes from the filter medium improved substantially with the application of the binder filter aid (A). On the other hand, a higher filtration pressure of the first filter cake layer FC3 (B) frequently impaired the surface quality of the filter cakes.
CD BD
ME
ABD BCD ABC ACD ABCD
Factors Fig. 8. No effect was determined as active regarding the integrity of the graded filter cakes.
Fig. 7. Plan view on the bottom of the graded filter cake no. 14. The binder filter aid (factor A) together with a low filtration pressure of the first filter cake layer (factor B) resulted in an easy separation of the filter cake from the filter medium and in an excellent surface at the bottom of the filter cake.
Effects
−1
0
1
ME
−2
ME
−3
remove the graded filter cakes increased significantly with a higher filtration pressure of the first filter cake layer FC3 . When the first filter cake layer FC3 was filtrated with a higher filtration pressure and without the binder filter aid, the filter cake layers frequently stuck on the filter medium which at times caused some damage of the graded filter cakes. By contrast, a lower filtration pressure of the first filter cake layer (lower level of factor B) together with the binder filter aid generally allowed the demolding of filter cakes with a superior surface. According to Fig. 8 no effect was active regarding the integrity of the graded filter cakes as the filter cakes were generally of high quality. The integrity was evaluated regarding the uniformity of the filter cakes, the occurrence of cracks or other flaws, and the quality of the rim. The evaluation regarding the integrity of the filter cakes did not include the bottom surface adjacent to the filter medium and the interface between the two filter cake layers. The Lenth plot in Fig. 9 contains the effects regarding the composite quality of the two filter cake layers. The bonding between the two filter cake layers was especially evaluated regarding the difference in diameter of the FC3 and FC1 filter cake layers after demolding, drying, and firing. Furthermore, a possible delamination of the two filter cake layers was evaluated after each process
B A
D C
AC AB
AD BC
CD BD
SME ABD BCD ABC ACD ABCD
Factors Fig. 9. Lenth plot of the effects regarding the composite and bonding quality of the two filter cake layers. Especially a higher filtration pressure of the second filter cake layer FC1 (C) and a shorter filtration time of the first filter cake (D) improved the bonding of the graded filter cakes.
step. As a rule the two filter cake layers had the same diameter and did not exhibit any delamination in green or fired state. However, it was often possible to clearly distinguish the filter cake layers (see Fig. 10). Yet, this clearly visible interface between the two filter cake layers became almost negligible in case of a higher filtration pressure of the second filter cake (C) combined with shorter filtration time of the first filter cake layer FC3 (D), see Fig. 11. The activity of these effects can be attributed to a higher
Fig. 10. Lateral view on the graded filter cake no. 2. The filter cake layer FC1 containing a dmax = 1 mm is on the top, while the filter cake layer FC3 with a dmax = 3 mm is at the bottom. Unfortunately, the bonding was less than perfect and the two filter cake layers can be clearly distinguished.
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008
G Model
ARTICLE IN PRESS
JECS-10516; No. of Pages 9
S. Schafföner et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx
7
Fig. 11. Lateral view on the graded filter cake no. 6. In this case the bonding between the two filter cake layers was considerably better and the transition between the filter cake layer was much less evident.
Fig. 13. Lateral XCT view (2D) on the filter cake of run 1 with the filter cake layer FC3 on the left and the filter cake layer FC1 on the right. Some beam hardening was noticed. Nevertheless, a very homogeneous distribution of the coarse grains can be observed in both filter cake layers. Both filter cakes contained only a negligible amount of large pores. The bonding between both filter cake layers was excellent and without cracks.
design, i.e. the samples of the runs 1, 2, 6, 14, were further investigated using XCT. Fig. 13 shows a lateral slice of the graded filter cake of run 1, which had the best bonding between the FC3 and FC1 filter cake layers. Meanwhile, for all other three investigated graded filter cakes the area of the bonding can be still clearly distinguished (see Fig. 14). That is, the bonding was less than ideal. Furthermore, the XCT analyses also revealed some curvature of the bonding, which was likely caused by friction at the pressure filtration cell walls (see Fig. 14). In the XCT analyses it was also observed that both filter cake layers were very dense and contained only a negligible amount of large sized pores caused by entrapped air. This is a significant
1.0
compression in case of a higher filtration pressure of the second filter cake layer FC1 . A shorter filtration time of the first filter cake layer, on the other hand, resulted in a softer top layer of the FC3 filter cake layer, which consequently improved the bonding between the two filter cake layers. Finally, Fig. 12 presents the Lenth plot regarding the apparent porosity. The overall mean of the apparent porosity was 16.66% and thus significantly lower than reported previously for large plates produced by pressure slip casting [6]. The lower apparent porosity might be ascribed to the higher filtration area in the pressure filtration cell compared to the small sprue of the used mold to cast the plates in a commercial pressure slip casting machine [6]. Thus, a higher relative filtration area may result in a higher compaction of the filter cakes and thus lower apparent porosity. Nevertheless, as indicated in Fig. 12 no effect was significantly active regarding the apparent porosity, although it seems that the binder filter aid (A) resulted in a slightly higher porosity. A longer filtration time of the first filter cake layer (D) also caused some increase in the apparent porosity of the graded filter cake layers. A higher filtration pressure of both filter cake layers (B and C), on the other hand, decreased the apparent porosity to some extent. In summary, the best overall combination of the factors regarding Y1 to Y3 was determined when the binder filter aid was applied (A), at low filtration pressure of the first filter cake layer (B), at a high filtration pressure of the second filter cake layer (C) and at a shorter filtration time of the first filter cake layer (D). The four samples, which attained the best overall results regarding Y1 to Y3 according to the full factorial experimental
−1.0
−0.5
Effects in % 0.0
0.5
ME
B A
D C
AC AB
AD BC
CD BD
ME
ABD BCD ABC ACD ABCD
Factors Fig. 12. Lenth plot of the effects regarding the apparent porosity of the graded filter cakes. Although no effect was significantly active, higher filtration pressures (B and C) resulted in rather lower porosities, whereas the binder filter aid (A) increased the porosity on average.
Fig. 14. Lateral view (2D) on the graded filter cake of run 2. Unfortunately it can be seen that the filter cake layers delaminated at the rim of the filter cake. Furthermore, the bonding interface was buckled to some extent.
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008
G Model JECS-10516; No. of Pages 9 8
ARTICLE IN PRESS S. Schafföner et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx
Fig. 15. Plan view (2D) on the bottom of the filter cake layer FC3 of the graded filter cake 6. A homogeneous distribution of the coarse grains was noticed. However, some cracks can be observed around the largest grains. These cracks were more pronounced at the bottom than on the top of the filter cake layer.
difference to conventional refractory castables, which often contain a large amount of large pores. Generally, the coarse grains were very homogeneously distributed and only some gradients of the coarse grains in the filter cake layer FC3 were observed. The concentration of the coarse grains was at times somewhat higher at the bottom and closer to the rim of the filter cakes. These small gradients of the coarse grains could be explained by possible sedimentation and higher friction close to the wall of the pressure filtration cell. However, in all investigated samples except in the filter cake run 1, some cracks around the coarse grains were observed in the lower filter cake layer FC3 (see Fig. 15). These cracks were more pronounced at the bottom of the filter cake layer FC3 than on the top of this filter cake layer. In addition, these kind of cracks were absent in case of the second filter cake layer FC1 (see Fig. 16). Possible sources of cracks could be a fast pressure release after filtration, an inhomogeneous particle size distribution due to possible sedimentation, or a too high pressure on the filter cakes during the demolding caused by the used circular die. These possible sources of errors will be investigated in future studies. Finally, Fig. 17 shows a representative digital microscope image of the interface between the two filter cake layers of the graded filter cake of run 1. A discrete transition between the two filter cake layers was observed. The digital microscope analyses confirmed the results of the XCT investigations because no cracks were detected at the interface between the two filter cake layers.
Fig. 17. Representative digital microscope image (cross section) of the graded filter cake of run 1. The FC1 filter cake layer is displayed on the top, while the FC3 filter cake layer is on the bottom. The different maximum grain sizes can be clearly distinguished. Furthermore, no cracks were found at the interface between the two filter cake layers.
4. Conclusions The present study investigated the production of graded functional composites for refractory applications with the pressure slip casting technique. For that purpose two filter cake layers build up from slips with 1 mm and 3 mm maximum grain size and a thickness of 20 mm were sequentially filtrated in a laboratory pressure filtration cell. The graded filter cakes were optimized using a full factorial experimental design. It was demonstrated that a lower filtration pressure for the first, i.e. coarser, slip in combination with a higher filtration pressure for the second, i.e. finer, slip was beneficial to improve the bonding between the two filter cake layers. Furthermore, the application of a thin prefiltered layer consisting of the same xanthan/guar gum solution which was used to stabilize the slips resulted in crack-free filter cakes and helped to remove the filter cakes from the PMMA filter medium. Finally, in order to achieve an improved joining between the filter cakes, the first filter cake layer should only be filtrated to the point that a soft layer still remains on the top. This soft layer improved the bonding between the filter cake layers significantly. The excellent joining between the filter cake layers and the lack of macroscopic pores and cracks was later demonstrated with X-ray computed tomography and digital microscopy. The pressure slip casting of graded refractories can provide a functionality to each filter cake layer. Thereby tailored compositions, microstructures, and thermomechanical properties can be achieved. Future work will thus concentrate on upscaling experiments and has the aim to provide a further functionality to the produced refractory components. Acknowledgments This work was kindly supported by the European Ceramic Society (ECerS) and by the JECS Trust in the “Frontiers of research” program for young scientists (contract 201242-1). References
Fig. 16. Plan view (2D) on the bottom of the filter cake layer FC1 of the graded filter cake 6. A very homogeneous distribution of the coarse grains was again noticed. No cracks and only a negligible amount of large pores was detected.
[1] E.G. Blanchard, Pressure casting improves productivity, Am. Ceram. Soc. Bull. 67 (1988) 1680–1683. [2] F.F. Lange, Powder processing science and technology for increased reliability, J. Am. Ceram. Soc. 72 (1989) 3–15. [3] O. Lyckfeldt, E. Lidén, M. Persson, R. Carlsson, P. Apell, Progress in the fabrication of Si3 N4 turbine rotors by pressure slip casting, J. Eur. Ceram. Soc. 14 (1994) 383–395. [4] R. Rice, Ceramic Fabrication Technology, Marcel Dekker, 2003. [5] A. Mattern, R. Oberacker, M.J. Hoffmann, Multi-phase ceramics by computercontrolled pressure filtration, J. Eur. Ceram. Soc. 24 (2004) 3219–3225. [6] S. Schafföner, C.G. Aneziris, Pressure slip casting of coarse grain oxide ceramics, Ceram. Int. 38 (2012) 417–422.
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008
G Model JECS-10516; No. of Pages 9
ARTICLE IN PRESS S. Schafföner et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx
[7] J.S. Reed, Principles of Ceramic Processing, 2 ed., Wiley & Sons, 1995. [8] J.M.F. Ferreira, H.M.M. Diz, Pressure slip casting of bimodal silicon carbide powder suspensions, Ceram. Int. 25 (1999) 491–495. [9] P. Greil, U. Gruber, N. Travitsky, M. Kulig, Pressure filtration of silicon nitride suspensions with constant filtration rate, Mater. Sci. Eng. A 151 (1992) 247–254. [10] S. Maleksaeedi, M.H. Paydar, S. Saadat, H. Ahmadi, In situ vibration enhanced pressure slip casting of submicrometer alumina powders, J. Eur. Ceram. Soc. 28 (2008) 3059–3064. [11] J. Requena, R. Moreno, J.S. Moya, Alumina and alumina/zirconia multilayer composites obtained by slip casting, J. Am. Ceram. Soc. 72 (1989) 1511–1513. [12] J.S. Moya, A.J. Sánchez-Herencia, J. Requena, R. Moreno, Functionally gradient ceramics by sequential slip casting, Mater. Lett. 14 (1992) 333–335. [13] B.R. Marple, J. Boulanger, Graded casting of materials with continuous gradients, J. Am. Ceram. Soc. 77 (1994) 2747–2750. [14] L. Zhang, V.D. Krstic, High toughness silicon carbide/graphite laminar composite by slip casting, Theor. Appl. Fract. Mech. 24 (1995) 13–19. [15] C.-H. Yeh, M.-H. Hon, The Si3 N4 and Si3 N4 /TiC layered composites by slip casting, Ceram. Int. 23 (1997) 361–366. [16] M. Kato, Y. Goto, T. Fukasaw, Rheological characterization of ˇ-sialon and Si3 N4 slips and fabrication of multilayered composites by slip casting, J. Mater. Sci. Lett. 16 (1997) 2055–2057. [17] Y.-G. Jung, S.-C. Choi, Fabrication of 3Y-TZP/SUS304 functionally graded materials by slip casting; application of porous alumina molds, Mater. Lett. 37 (1998) 312–319. [18] H. Ishibashi, H. Tobimatsu, K. Hayashi, T. Matsumoto, A. Tomsia, E. Saiz, Characterization of Mo-SiO2 functionally graded materials, Metall. Mater. Trans. A 31 (2000) 299–308.
9
[19] B. Kieback, A. Neubrand, H. Riedel, Processing techniques for functionally graded materials, Mater. Sci. Eng. A 362 (2003) 81–106. [20] M. Yan, X.L. Peng, T.Y. Ma, Microstructures of Ni-ZrO2 functionally graded materials fabricated via slip casting under gradient magnetic fields, J. Alloy. Compd. 479 (2009) 750–754. [21] M. Dröschel, M.J. Hoffmann, R. Oberacker, H. Both, W. Schaller, Y.Y. Yang, D. Munz, SiC-ceramics with tailored porosity gradients for combustion chambers, Key Engineering Materials, vol. 175, Trans Tech Publ, 1999, pp. 149–162. [22] U. Klippel, C.G. Aneziris, A.J. Metzger, Shaped coarse grained refractories by pressure slip casting, Adv. Eng. Mater. 13 (2011) 68–76. [23] J.A. Casas, A.F. Mohedano, F. García-Ochoa, Viscosity of guar gum and xanthan/guar gum mixture solutions, J. Sci. Food Agric. 80 (2000) 1722–1727. [24] R.J. Wakeman, Vacuum dewatering and residual saturation of incompressible filter cakes, Int. J. Miner. Process. 3 (1976) 193–206. [25] C¸. Hos¸ten, O. S¸an, Reassessment of correlations for the dewatering characteristics of filter cakes, Miner. Eng. 15 (2002) 347–353. [26] D.C. Montgomery, Design and Analysis of Experiments, John Wiley & Sons, 2008. [27] R Core Team, R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna, Austria, 2015. [28] R. Wakeman, S. Tarleton, Solid/Liquid Separation: Principles of Industrial Filtration, Elsevier, 2005. [29] S. Schafföner, C.G. Aneziris, H. Berek, J. Hubálková, A. Priese, Fused calcium zirconate for refractory applications, J. Eur. Ceram. Soc. 33 (2013) 3411–3418. [30] R.V. Lenth, Quick and easy analysis of unreplicated factorials, Technometrics 31 (1989) 469–473.
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008
ARTICLE IN PRESS
JECS-10516; No. of Pages 9
Journal of the European Ceramic Society xxx (2016) xxx–xxx
Contents lists available at www.sciencedirect.com
Journal of the European Ceramic Society journal homepage: www.elsevier.com/locate/jeurceramsoc
Functional composites based on refractories produced by pressure slip casting Stefan Schafföner ∗ , Lisa Freitag, Jana Hubálková, Christos G. Aneziris Institute of Ceramic, Glass and Construction Materials, TU Bergakademie Freiberg, Agricolastrasse 17, 09599 Freiberg, Germany
a r t i c l e
i n f o
Article history: Received 19 December 2015 Accepted 3 February 2016 Available online xxx Keywords: Refractories Composites Pressure slip casting X-ray computed tomography Spinels
a b s t r a c t The present study investigates the sequential pressure filtration of two filter cake layers with a maximum grain size of 3 mm and 1 mm, respectively, using a water soluble additive system consisting of xanthan and guar gum. The pressure slip casting of the resulting composites was optimized with a factorial experimental design. The best combination was a lower filtration pressure for the first filter cake layer and a higher filtration pressure for the second one. Furthermore, a shorter filtration time of the first filter cake layer together with the application of a filter aid of a xanthan/guar gum solution resulted in an easy demolding of crack-free graded composites with an excellent bonding. The results of the full factorial experimental design were verified by X-ray computed tomography. The demonstrated pressure slip casting of graded refractories offers the possibility to produce functional composites with tailored compositions, microstructure and thus functionality. © 2016 Elsevier Ltd. All rights reserved.
1. Introduction Pressure slip casting is state of the art for the production of silicate ceramics with complex shapes such as whiteware and sanitary ware [1–6]. Compared with conventional slip casting, the process time for pressure slip casting is much shorter. The durability of the used polymer molds is also much longer than plaster of Paris molds and they do not need any drying between the casting cycles. Furthermore, pressure slip casting can be automated to a large extent. Thus, pressure slip casting offers a significantly higher efficiency compared to conventional slip casting [1,3–5,7,8,]. Besides, technical ceramics produced by pressure slip casting also exhibit an improved microstructure with less agglomerates and superior mechanical properties [2–4,9,10]. Conventional and pressure slip casting was also repeatedly investigated to produce fine graded composites [5,11–19,20]. By changing the slip composition discrete or continuous gradients of the filter cakes can be achieved. The investigated compositions ranged from oxide ceramics including alumina–zirconia mixtures [5,11–13] over non-oxide ceramics [14–16] to ceramicmetal composites [17,18,20]. Besides changing the ceramic phase composition along the gradient in the filter cakes it is also possible to tailor other characteristics such as the porosity using layered pressure slip casting [21].
In recent years pressure slip casting was also studied as a new forming technique for coarse grained oxide ceramics such as refractories [6,22]. By combining coarse and fine grain fractions the thermal shock, corrosion, and erosion resistance of refractories is optimized. Furthermore, the sintering shrinkage is also reduced to a minimum, which allows the production of large sized components. Klippel et al. [22] focused especially on the rheology of suitable slips and demonstrated the casting of coarse grained ceramics using a pressure filtration cell. In a further study Schafföner and Aneziris [6] reported the production of large plate shaped oxide ceramics with only minimal gradients caused by sedimentation. The addition of larger particle sizes had the effect of a reduced Young’s modulus, which is an often observed feature of refractories. However, only the production of coarse grained oxide ceramics from a single slip composition was reported and the production of coarse grained composites by pressure slip casting remains unstudied. The purpose of this study is to describe and examine the pressure slip casting of functional composites based on refractories. The approach in this study is to use sequential pressure slip casting of magnesium aluminate spinel slips. Thereby the overall aim was to produce refractories with a high quality bonding of filter cakes with different particle size distributions.
2. Experimental ∗ Corresponding author. E-mail address: [email protected] (S. Schafföner).
In the present study graded composites made of filter cakes of the same material but different particle size distributions were
http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008 0955-2219/© 2016 Elsevier Ltd. All rights reserved.
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008
G Model
ARTICLE IN PRESS
JECS-10516; No. of Pages 9
S. Schafföner et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx
2 Table 1 Batches of the slips. Product name
Raw material
Almatis AR 78
MgAl2 O4
Martoxid MR 70
Al2 O3
Water for binder Axilat RH 50 MD Food grade guar gum Additional water
H2 O Xanthan gum Guar gum H2 O
Grain size fraction
Slip S3
Slip S1
1–3 mm 0.5–1 mm 0–0.5 mm 0–0.045 mm –
15 wt.% 10 wt.% 30 wt .% 10 wt.% 20 wt .%
0 wt.% 11.76 wt.% 35.3 wt .% 11.76 wt .% 23.53 wt .% 12 wt .% 0.024 wt .% 0.024 wt .%
Mass relative to dry mass
produced by pressure slip casting. The casting of graded composites was optimized using a full factorial experimental design. In all experiments slip compositions similar to the ones presented by Schafföner and Aneziris [6] were applied. The used raw materials were alumina rich magnesium aluminate spinel (AR 78, Almatis GmbH, Germany) and reactive alumina (Martoxid MR 70, Martinswerke, Germany) for the finest particle fraction. For the experiments two different slip compositions with a maximum grain size of 1 mm (S1 ) and 3 mm (S3 ), respectively, were prepared. The batches are summarized in Table 1. In order to prevent a sedimentation or segregation of the coarse grain fraction, dissolved xanthan (Axilat RH 50 MD, C.H. Erbslöh GmbH & Co. KG, Germany) and food grade guar gum were used as stabilizers because these two hydrocolloids show a strong positive interaction regarding their viscosity [23]. As a rule the added amount was 0.024 wt .% for both stabilizers regarding the raw material dry mass. The xanthan/guar gum solution did not only act as a stabilizer to prevent a possible sedimentation, but also acted as a binder because the viscosity of this hydrocolloid solution depends strongly on the water content. During the filtration of the slips, the water content decreases which hence increases the viscosity of the binder/water solution. Thereby a sufficient green strength of the filter cakes can be attained. Prior to mixing the slip, the binder solution was prepared using 12 wt .% water regarding the dry mass by carefully adding the binders separately to half of the water using a high shearing laboratory mixer (RZR 2102 control, Heidolph Instruments GmbH & Co. KG, Germany). Afterwards, the two separate binder solutions were mixed together. In all experiments 3 kg of dry mass were used for each slip. For the slip preparation the dry mass was first mixed for 4 min using a standard concrete mixer (ToniMIX, Toni Technik Baustoffprüfsysteme GmbH, Germany). The slips were in general thixotropic, therefore the mixer was kept at the lowest power level to avoid a strong stirring and hence decrease in viscosity of the slips. After the dry mixing, the binder solution was added in three steps to ensure a homogeneous mixture. After each adding the slip was stirred for 3 min followed by a careful scrapping of the mixer walls to remove any lumps. Finally an amount of 3 wt .% and 4 wt .% of distilled water was added to the S1 and S3 slips, respectively. The slips were then stirred for 3 min for a last time. Shortly after preparing the slips, the water content was controlled using a thermobalance (MA 30, Sartorius AG, Germany) in order to ensure a comparable slip composition in the experiments. For all the filtration experiments a laboratory pressure filtration cell was applied. The pressure filtration cell operated similar to a pressure casting machine using pressurized air as the pressure medium. It allows the testing of a wide range of factors before performing upscaling experiments. The filter medium of the pressure filtration cell was poly(methyl methacrylate) (PMMA), which was identical to the usual filter medium material used in commercial pressure slip casting machines. The filter medium had a filtration
3 wt .%
4 wt .%
area of 33.7 cm2 and was supported by a porous metal plate with a relative open area of 40.2%. In all cases the slips were first filled into the pressure filtration cell, then sealed before the pressure was abruptly increased using a lever valve. The pressure was held constant during the whole filtration time. In preliminary experiments the filtration properties of both slips were first separately investigated. For that purpose the slips were filled with a height of 20 mm into the pressure filtration cell and were then filtrated with a pressure of 5 bar and 7 bar, respectively, until the dewatering of the filter cake started. The dewatering starts at the point when the cake filtration of the slip ceases, i.e. the filter cake is completely built up and no slip remains on the filter cake surface. With the beginning of the dewatering the remaining water of the filter cake is then displaced by air [22,24,25]. The start of the dewatering was clearly perceivable by the time that a fizzling sound became evident, which was caused by the breakthrough of pressurized air. The start of the filter cake dewatering was defined as the termination criterion for the filtration and the necessary time to reach the dewatering was thus recorded for later experiments at each pressure. The following experiments regarding the graded filter cake composites were mainly conducted using a 24 full factorial experimental design because factorial designs have several advantages. They allow the simultaneous study of several factors, i.e. independent variables, and also of so called interactions, i.e. the influence of one factor on the other, with only a minimum of experiments [26]. The overall aim was to achieve a composite of high quality which can be easily demolded at the same time. The factorial experimental design is summarized in Table 2 in conjunction with Table 3. All of the following statistical analyses were conducted with the software package R [27]. The general procedure for the casting of graded filter cakes was always to filtrate the slip S1 (dmax = 1 mm) on top of a filter cake of the S3 slip (dmax = 3 mm). In the following the chosen factors of the factorial experimental design will be described. The first factor (A) was the use (+1) or nonuse (−1) of a so called binder filter aid. This binder filter aid had the aim to improve the demolding of the filter cake and to prevent the penetration and thus possible clogging of the filter medium
Table 2 Factors of the 24 full factorial experimental design. Factor
Identifier
Lower level (−1)
Higher level (+1)
Binder filter Filtration pressure of first (coarse) slip S3 Filtration pressure of second (finer) slip S1 Filtration time relative to beginning of filter cake dewatering
A B
Without 5 bar
With 7 bar
C
5 bar
7 bar
D
75%
100%
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008
G Model
ARTICLE IN PRESS
JECS-10516; No. of Pages 9
S. Schafföner et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx
3
Table 3 24 Full factorial experimental design for the pressure slip casting of graded filter cakes including results for the investigated responses, the filter medium and filter cake resistances, and the apparent porosity. Run
Order
Factors A
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
14 16 12 10 2 8 4 6 13 15 11 9 1 7 3 5
−1 +1 −1 +1 −1 +1 −1 +1 −1 +1 −1 +1 −1 +1 −1 +1
˛av,3
Rm,3
˛av,1
Rm,1
Y1 (1–12)
Y2 (1–12)
Y3 (1–9)
Apparent porosity
(m kg−1 )
(m−1 )
(m kg−1 )
(m−1 )
(%)
10 12 5 8.5 4.5 12 7 7 6.5 10 5 8 6 12 5 8
11 11 6 11 10 11 10 9 11 9 10 11 10 10 9 8
7 8 7 5 9 8 8 9 4 4 4 3 6 6 6 6
1.34 × 1013 8.82 × 1012 2.22 × 1013 1.35 × 1013 1.41 × 1013 7.81 × 1012 1.01 × 1013 9.48 × 1012 1.34 × 1013 9.81 × 1012 1.87 × 1013 1.01 × 1013 1.24 × 1013 1.19 × 1013 1.82 × 1013 7.78 × 1012
2.16 × 1014 2.29 × 1014 5.32 × 1013 1.74 × 1014 1.73 × 1014 1.53 × 1014 2.16 × 1014 1.54 × 1014 1.99 × 1014 3.16 × 1014 2.18 × 1014 2.25 × 1014 2.50 × 1014 1.24 × 1014 1.65 × 1014 1.99 × 1014
2.09 × 1013 1.11 × 1013 2.34 × 1013 1.56 × 1013 1.74 × 1013 1.11 × 1013 1.67 × 1013 1.23 × 1013 1.79 × 1013 1.40 × 1013 1.67 × 1013 1.20 × 1013 1.82 × 1013 8.20 × 1012 1.82 × 1013 6.32 × 1012
4.40 × 1014 6.40 × 1014 6.93 × 1014 5.87 × 1014 5.34 × 1014 4.35 × 1014 3.95 × 1014 4.93 × 1014 7.28 × 1014 6.45 × 1014 8.70 × 1014 8.70 × 1014 7.50 × 1014 5.82 × 1014 5.82 × 1014 7.39 × 1014
16.9 17.4 16.0 16.3 15.4 17.2 15.6 16.0 17.4 17.9 16.6 17.2 15.8 17.8 16.3 16.5
Responses B −1 −1 +1 +1 −1 −1 +1 +1 −1 −1 +1 +1 −1 −1 +1 +1
C −1 −1 −1 −1 +1 +1 +1 +1 −1 −1 −1 −1 +1 +1 +1 +1
D −1 −1 −1 −1 −1 −1 −1 −1 +1 +1 +1 +1 +1 +1 +1 +1
by the fine particle fraction of the slips. For this purpose a small amount of the xanthan/guar gum binder solution (10 ml) was first filtrated with 1 bar before filtrating the S3 slip. Thus, a thin skin of the filtrated binder solution was formed on the filter medium. The second investigated factor (B) was the filtration pressure of the first slip, i.e. the coarse grained slip S3 , because during filtration the filter cakes might exhibit some degree of compressibility. In case of compressible filter cakes a higher filtration pressure does not necessarily significantly increase the filtration speed because highly compacted filter cakes increase the filtration resistance [28]. Therefore, the filtration pressure was varied between 5 bar (−1) and 7 bar (+1). The third investigated factor (C) was accordingly the filtration pressure of the second layer, i.e. the S1 slip. The filtration pressure was again varied between 5 bar (−1) and 7 bar (+1). The fourth and last factor D was the filtration time of the first slip layer S3 . At the lower level (−1) the filtration time was only 75% of the time needed until the filter cake dewatering started, therefore a slip layer was still present on the top of the slip. At the higher level (100%, +1) the slip was filtrated until the filter cake dewatering stage was reached. The necessary filtration time as a function of the applied pressure and the usage of the binder filter aid was determined in the preliminary experiments described above. In order to determine the filtration properties of the slips, the filtrate mass was always registered with a scale below the pressure filtration cell and recorded with a computer in an interval of 1 s. To reduce the time offset until the first filtrate was registered, the filter medium was always immersed into distilled water before mounting. Thus, the filtrate always displaced other water in the filter medium. The data were later used to determine the filtration properties of the filter cakes using the conventional cake filtration theory (see also Eq. (1)) [28]. After filtration the filter cakes were carefully demolded using a circular die which fitted exactly in the tubular pressure filtration cell. Not once this demolding caused any cracking. The filter cakes were first evaluated regarding the bottom surface quality and the effort to remove the filter cakes from the filter medium (Y1 ). Furthermore, the filter cakes were examined on the overall quality of the filter cake, henceforth called response Y2 . As a third response (Y3 ) the joining of the composite, i.e. the connection of the top (FC1 ) and bottom layer (FC3 ) of the filter cake, was analyzed. In all cases an ordinal scale was applied to evaluated each response. The ordinal scale ranged from 1 (worst) to 12 (best) for Y1 and Y2 and 1 (worst) to 9 (best) for Y3 .
Later the filter cakes were carefully dried at 40 ◦ C, 60 ◦ C, 80 ◦ C, and 110 ◦ C for 6 h at each temperature. The drying process was not optimized, however it was known that a carefully drying is necessary to prevent eventual cracking. After drying the filter cakes were fired in air for 4 h at a temperature of 1650 ◦ C with a heating and cooling rate of 3 K min−1 . After firing all filter cakes were investigated on their apparent porosity and apparent density using the Archimedes method according to DIN EN 993-1 with water as the immersion fluid. In order to evaluate the inner structure, the four filter cakes which achieved the best overall values regarding Y1 to Y3 were examined with a micro focus X-ray computed tomograph (XCT) (CT-ALPHA, ProCon X-Ray GmbH, Germany) similar to Schafföner et al. [29]. The XCT was equipped with a transmission X-ray tube (YXLON International Feinfocus GmbH, Germany) and a detector (Dexela, PerkinElmer Inc., USA) with a total area of 120 × 150 mm2 and a pixel size of 75 m. The whole samples were scanned with a voltage of 150 kV and a current of 90 A; the target power was 11.5 W. The edge length of the voxels after reconstruction was about 98 m. The reconstructed XCT data were processed and visualized with the software package VGStudio Max 2.2 (Volume Graphics GmbH, Germany). Although beam hardening artifacts occurred, no regions of interest (ROI) were extracted in order to obtain a full image of the graded filter cakes. Finally, to investigate the interface between the two coarse grained filter cake layers, cut samples were investigated with a digital microscope (VHX-2000, Keyence Corp., Japan). 3. Results and discussion The aim of the present study is to produce graded functional composites based on refractories using the pressure slip casting technique. For pressure slip casting the filtration properties can be explained by t/V vs. V plots. Fig. 1, for example, presents the t/V vs. V values of S3 slips as a function of the applied pressure. The plotted graphs express the filtration properties according to the conventional cake filtration equation at constant pressure: (˛av cV + ARm ) dt = dV A2 · p
(1)
where t is the filtration time, V the filtrate volume, A the filter area, the dynamic filtrate viscosity, ˛av the average specific resistance of the filter cake, c the effective concentration of solids in the
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008
G Model
ARTICLE IN PRESS
JECS-10516; No. of Pages 9
S. Schafföner et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx
1e+13
2.5e+08 1.0e+08
High pressure (7 bar) Cake filtration
0e+00
2e−06
4e−06
6e−06
8e−06
Fig. 1. Comparison of the filtration of the slips S3 of the runs 13 (low filtration pressure, 5 bar) and 15 (high filtration pressure, 7 bar). Both slips were filtrated without using the binder filtering aid and until the cake dewatering was reached.
slip, p the applied filtration pressure, Rm the resistance of the filter medium. Integrating Eq. (1) with the initial conditions t = V = 0 results in a linear equation: K1 t = V + K2 V 2 where ˛av c K1 = 2 A p
(2)
(3)
and Rm Ap
(4)
Eq. (2) in conjunction with the Eqs. (3) and (4) can then be used to determine ˛av and Rm of each filter cake by approximating the t/V vs. V data with a linear equation. The necessary effective solid concentration in the slip can be calculated according to the following equation: c=
l s 1 − ms
ME
A
B
D
AB
AD
BD
ABD
Factors
Filtrate volume in m3
K2 =
5e+12 −1e+13
5.0e+07
Filter cake dewatering
0e+00
Filter cake dewatering
Effects in m kg
2.0e+08 1.5e+08
Cake filtration
0.0e+00
Time per filtrate volume in s m3
Low pressure (5 bar)
ME
−5e+12
4
(5)
where l is the liquid of the dispersing medium of the slip, i.e. water, s the mass fraction of solids in the slip, and m the ratio of the mass of the wet cake to the mass of the dry cake. To determine m of the graded filter cakes, the same value was applied for both cake layers because only the wet and dry mass of the graded filter cakes was determined. Due to the large filtration resistance of the filter cakes and due to the setup of the filtration pressure cell, the time offset between the time of pressure increase to start the filtration and the registration of the first filtrate volume was often quite long. Therefore the calculations according to Eq. (2) took only the time after registrating the first filtrate into account and ignored the time offset. The steady-state cake filtration in Fig. 1 can be approximated by a line expressed in Eq. (2). Only values that fell in a 95% confidence interval around the approximated line were included to approximate the line. The coefficients of the line, i.e. slop and intercept, were thus determined by iteration. All determined values regarding the filter medium resistance Rm , the average filter cake resistance ˛av , and the concentration of the slips are listed in Table 3. In the following the results of the full factorial experimental design for these responses will be evaluated. Fig. 2 presents the effects of the full factorial experimental design for the average filter cake resistance of the filter cakes made up of the S3 slips (˛av,3 ). The effects are plotted in a so called
Fig. 2. Lenth plot of the effects regarding the average filter cake resistance made up of the S3 slips (˛av,3 ). Only effects that fall above or below the line of the margin of error (ME) are considered as possibly active. In this case, however, not a single effect was identified as active.
Lenth plot [30]. Lenth plots have the advantage that the results of an unreplicated factorial experimental design can be effectively presented. Furthermore, active and inactive effects can be easily distinguished. For the analysis of the resistance of the filter medium (Rm,3 ) and of the average filter cake resistance of the composition S3 (˛av,3 ) the filtration pressure of the second slip, i.e. the factor C, was not taken into account because the filtration of the second slip took always place after the measurements of the filtration of the S3 slips. On the other hand, excluding the factor C for the analysis of Rm,3 and ˛av,3 increases the degrees of freedom for the statistical tests and thus their power. The overall mean of ˛av,3 was 1.26 × 1013 m kg−1 . Such a specific filtration resistance is considered as very high [28] and is most probably caused by the broad particle size distribution and by the considerable fine particle content of the slips. As can be seen in Fig. 2 no factor was identified as significantly active regarding ˛av,3 . Especially, since the effect of the filtration pressure (B) was determined as inactive in the full factorial design on the chosen pressure levels, the compressibility of the filter cakes of S3 composition (FC3 ) can be regarded as negligible. Filter cakes of a high compressibility, on the other hand, would have been highly influenced by the filtration pressure. A low compressibility of the filter cakes FC3 can be regarded as an advantage because filter cakes with a low compressibility have generally a more homogeneous microstructure [28]. Fig. 3 provides the Lenth plot for the filter medium resistance of S3 composition (Rm,3 ). Similar to ˛av,3 no effect was identified as significant. This result can be again attributed to the lack of compressibility of the filter cake layer FC3 . On the other hand, in case of compressible filter cakes the pressure drop over the filter cake and filter medium are influenced by the filtration pressure. Thus, compressible filter cakes often exhibit a changing filter medium resistance as a function of the filtration pressure [28]. Furthermore, the analysis also revealed that the binder filter aid (A) did not influence Rm,3 . Fig. 4 presents the Lenth plot of the average filtration resistance (˛av,1 ) of the filter cake layers of the S1 composition (FC1 ). The binder filter aid (A) and the filtration pressure of the FC1 filter cake layer (C) had a negative effect on the filtration resistance of the second filter cake layer. Thus, the binder filter aid and a higher filtration pressure of the second filter cake layers helped to reduce the average filtration resistance of the second layer of the graded filter cake.
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008
G Model
ARTICLE IN PRESS
JECS-10516; No. of Pages 9
S. Schafföner et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx
−5e+13
Effects in m−1 0e+00
5e+13
ME
ME
A
B
D
AB
AD
BD
ABD
Factors Fig. 3. Lenth plot of the effects regarding Rm,3 of the first filter cake layer FC3 . Again not a single effect was identified as active, either.
The overall mean of ˛av,1 was 1.50 × 1013 m kg−1 and thus on average 21.5% higher than the overall mean of ˛av,3 . A paired Student’s t-test also indicated a significant difference in the means of the two average filtration resistances (p < 0.01). The results of the full factorial design regarding ˛av,1 seem to contradict the above presented result regarding ˛av,3 because no effect was identified as significant for ˛av,3 . Yet, comparing the Lenth plot of the effects for both filter cake layers in Figs. 2 and 4 it can be observed that for the average filtration resistance of both filtration cake layers the effect of the binder filter aid (A) was quite pronounced. But, for ˛av,3 the so called margin of error (ME) was much larger than for ˛av,1 . According to Lenth [30] ME is calculated from the 0.975th quantile of the t distribution on d degrees of freedom (t0.975;d ) and the pseudo standard error (PSE): ME = t0.975;d · PSE
(6)
The PSE, on the other hand, is calculated from the estimates of the contrasts of interest cj : PSE = 1.5 · median |cj |
(7)
|cj |<2.5s0
whereas s0 is defined as s0 = 1.5 · median|cj |
(8)
j
5
Examining Eqs. (7) and (8) it becomes clear that ME becomes inflated in case of several possibly active effects, or in case of a larger variance in the response values. Thus, whether an effect is active depends on the size of the estimate of the effect, i.e. cj , and on the value of ME. This also explains the difference in Figs. 2 and 4 because for ˛av,1 ME was much smaller compared to ˛av,3 (2.76 × 1012 m kg−1 and 9.40 × 1012 m kg−1 , respectively). A possible reason for the smaller ME value for ˛av,1 could be a smaller variance in the cake buildup because the FC1 filter cake layers were filtrated on top of the first filter cake layers FC3 . What is more, all active effects regarding ˛av,1 reduced the specific filtration resistance. Especially the negative effect of the filtration pressure of the second slip (C) is somewhat unusual because in case of a compressible filter cake layer FC1 an increasing filtration pressure would be expected on a higher level of the factor C, i.e. on a higher filtration pressure. The activity of the effects A and C regarding ˛av,1 can be interpreted in view of the fact that the filtrate from S1 slips has to flow through the already consolidated filter cake layers FC3 . Thus, according to the conventional filtration theory (Eq. (1)) these results imply that a higher specific filtrate volume is registered at the higher levels of the factors A and C. Fig. 5 presents the filter medium resistance of the filter cakes FC1 (Rm,1 ). In case of Rm,1 C and D are possibly active. A higher filtration pressure of the S1 slips, i.e. a higher level of the factor C, causes a decreasing Rm,1 . A higher pressure for the filtration of the second filter cake layer, as present at the higher level of the factor C, might contribute to overcome the filter medium resistance provided by the first filter cake layer FC3 . By contrast, a higher consolidation of the filter cake layer FC3 , as present at the higher level of the factor D, acts as a higher filter medium resistance for the filtration of the second filter cake layer FC1 . The overall mean of Rm,1 (6.24 × 1014 m−1 ) was 3.79 times higher than the overall mean of Rm,3 (1.92 × 1014 m−1 ). The much higher values for Rm,1 than for Rm,3 are caused by the sequential filtration of the two slips, whereas the filter cake layer FC3 acts as the filter medium for the filtration of the slip S1 . In the following the results of the full factorial experimental design on the properties of the graded filter cakes will be evaluated. The result for each run is presented in Table 3. As can be seen from Fig. 6 the surface quality of the demolded filter cakes and the effort to remove the filter cakes from the filter medium improved with the application of the binder filter aid (see Fig. 7). By contrast, the surface quality deteriorated and the effort to
1.5e+14
B A
D C
AC AB
AD BC
CD BD
ABD ABC
BCD
ACD
ABCD
Factors
ME
Effects in m−1 −5.0e+13 5.0e+13
SME
−1.5e+14
−2e+12
ME
−6e+12
Effects in m kg
2e+12
ME
B A
D C
AC AB
AD BC
CD BD
ME
ABD BCD ABC ACD ABCD
Factors Fig. 4. Lenth plot of the effects regarding average filtration resistance (˛av,1 ) of the filter cake layer made up by the S1 slip (FC1 ). The application of the binder filter aid prior to filtrating the slip S3 (A) and the filtration pressure of the second filter cake layer (C) were identified as probably active.
Fig. 5. Lenth plot of the effects regarding the resistance of the filter medium of the second filter cake layer (Rm,1 ). An increasing pressure of the second filter cake layer (C) reduced Rm,1 , whereas a longer filtration time of FC3 (D) increased Rm,1 .
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008
G Model
ARTICLE IN PRESS
JECS-10516; No. of Pages 9
S. Schafföner et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx
3
6
3
ME
−2 B A
D C
AC AB
AD BC
CD BD
ME
ABD BCD ABC ACD ABCD
−3
−3
−2
−1
−1
Effects 0
Effects 0 1
1
2
2
ME
B
D
A
C
AC AB
AD BC
Factors Fig. 6. The surface of the filter cakes and the effort to remove the filter cakes from the filter medium improved substantially with the application of the binder filter aid (A). On the other hand, a higher filtration pressure of the first filter cake layer FC3 (B) frequently impaired the surface quality of the filter cakes.
CD BD
ME
ABD BCD ABC ACD ABCD
Factors Fig. 8. No effect was determined as active regarding the integrity of the graded filter cakes.
Fig. 7. Plan view on the bottom of the graded filter cake no. 14. The binder filter aid (factor A) together with a low filtration pressure of the first filter cake layer (factor B) resulted in an easy separation of the filter cake from the filter medium and in an excellent surface at the bottom of the filter cake.
Effects
−1
0
1
ME
−2
ME
−3
remove the graded filter cakes increased significantly with a higher filtration pressure of the first filter cake layer FC3 . When the first filter cake layer FC3 was filtrated with a higher filtration pressure and without the binder filter aid, the filter cake layers frequently stuck on the filter medium which at times caused some damage of the graded filter cakes. By contrast, a lower filtration pressure of the first filter cake layer (lower level of factor B) together with the binder filter aid generally allowed the demolding of filter cakes with a superior surface. According to Fig. 8 no effect was active regarding the integrity of the graded filter cakes as the filter cakes were generally of high quality. The integrity was evaluated regarding the uniformity of the filter cakes, the occurrence of cracks or other flaws, and the quality of the rim. The evaluation regarding the integrity of the filter cakes did not include the bottom surface adjacent to the filter medium and the interface between the two filter cake layers. The Lenth plot in Fig. 9 contains the effects regarding the composite quality of the two filter cake layers. The bonding between the two filter cake layers was especially evaluated regarding the difference in diameter of the FC3 and FC1 filter cake layers after demolding, drying, and firing. Furthermore, a possible delamination of the two filter cake layers was evaluated after each process
B A
D C
AC AB
AD BC
CD BD
SME ABD BCD ABC ACD ABCD
Factors Fig. 9. Lenth plot of the effects regarding the composite and bonding quality of the two filter cake layers. Especially a higher filtration pressure of the second filter cake layer FC1 (C) and a shorter filtration time of the first filter cake (D) improved the bonding of the graded filter cakes.
step. As a rule the two filter cake layers had the same diameter and did not exhibit any delamination in green or fired state. However, it was often possible to clearly distinguish the filter cake layers (see Fig. 10). Yet, this clearly visible interface between the two filter cake layers became almost negligible in case of a higher filtration pressure of the second filter cake (C) combined with shorter filtration time of the first filter cake layer FC3 (D), see Fig. 11. The activity of these effects can be attributed to a higher
Fig. 10. Lateral view on the graded filter cake no. 2. The filter cake layer FC1 containing a dmax = 1 mm is on the top, while the filter cake layer FC3 with a dmax = 3 mm is at the bottom. Unfortunately, the bonding was less than perfect and the two filter cake layers can be clearly distinguished.
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008
G Model
ARTICLE IN PRESS
JECS-10516; No. of Pages 9
S. Schafföner et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx
7
Fig. 11. Lateral view on the graded filter cake no. 6. In this case the bonding between the two filter cake layers was considerably better and the transition between the filter cake layer was much less evident.
Fig. 13. Lateral XCT view (2D) on the filter cake of run 1 with the filter cake layer FC3 on the left and the filter cake layer FC1 on the right. Some beam hardening was noticed. Nevertheless, a very homogeneous distribution of the coarse grains can be observed in both filter cake layers. Both filter cakes contained only a negligible amount of large pores. The bonding between both filter cake layers was excellent and without cracks.
design, i.e. the samples of the runs 1, 2, 6, 14, were further investigated using XCT. Fig. 13 shows a lateral slice of the graded filter cake of run 1, which had the best bonding between the FC3 and FC1 filter cake layers. Meanwhile, for all other three investigated graded filter cakes the area of the bonding can be still clearly distinguished (see Fig. 14). That is, the bonding was less than ideal. Furthermore, the XCT analyses also revealed some curvature of the bonding, which was likely caused by friction at the pressure filtration cell walls (see Fig. 14). In the XCT analyses it was also observed that both filter cake layers were very dense and contained only a negligible amount of large sized pores caused by entrapped air. This is a significant
1.0
compression in case of a higher filtration pressure of the second filter cake layer FC1 . A shorter filtration time of the first filter cake layer, on the other hand, resulted in a softer top layer of the FC3 filter cake layer, which consequently improved the bonding between the two filter cake layers. Finally, Fig. 12 presents the Lenth plot regarding the apparent porosity. The overall mean of the apparent porosity was 16.66% and thus significantly lower than reported previously for large plates produced by pressure slip casting [6]. The lower apparent porosity might be ascribed to the higher filtration area in the pressure filtration cell compared to the small sprue of the used mold to cast the plates in a commercial pressure slip casting machine [6]. Thus, a higher relative filtration area may result in a higher compaction of the filter cakes and thus lower apparent porosity. Nevertheless, as indicated in Fig. 12 no effect was significantly active regarding the apparent porosity, although it seems that the binder filter aid (A) resulted in a slightly higher porosity. A longer filtration time of the first filter cake layer (D) also caused some increase in the apparent porosity of the graded filter cake layers. A higher filtration pressure of both filter cake layers (B and C), on the other hand, decreased the apparent porosity to some extent. In summary, the best overall combination of the factors regarding Y1 to Y3 was determined when the binder filter aid was applied (A), at low filtration pressure of the first filter cake layer (B), at a high filtration pressure of the second filter cake layer (C) and at a shorter filtration time of the first filter cake layer (D). The four samples, which attained the best overall results regarding Y1 to Y3 according to the full factorial experimental
−1.0
−0.5
Effects in % 0.0
0.5
ME
B A
D C
AC AB
AD BC
CD BD
ME
ABD BCD ABC ACD ABCD
Factors Fig. 12. Lenth plot of the effects regarding the apparent porosity of the graded filter cakes. Although no effect was significantly active, higher filtration pressures (B and C) resulted in rather lower porosities, whereas the binder filter aid (A) increased the porosity on average.
Fig. 14. Lateral view (2D) on the graded filter cake of run 2. Unfortunately it can be seen that the filter cake layers delaminated at the rim of the filter cake. Furthermore, the bonding interface was buckled to some extent.
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008
G Model JECS-10516; No. of Pages 9 8
ARTICLE IN PRESS S. Schafföner et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx
Fig. 15. Plan view (2D) on the bottom of the filter cake layer FC3 of the graded filter cake 6. A homogeneous distribution of the coarse grains was noticed. However, some cracks can be observed around the largest grains. These cracks were more pronounced at the bottom than on the top of the filter cake layer.
difference to conventional refractory castables, which often contain a large amount of large pores. Generally, the coarse grains were very homogeneously distributed and only some gradients of the coarse grains in the filter cake layer FC3 were observed. The concentration of the coarse grains was at times somewhat higher at the bottom and closer to the rim of the filter cakes. These small gradients of the coarse grains could be explained by possible sedimentation and higher friction close to the wall of the pressure filtration cell. However, in all investigated samples except in the filter cake run 1, some cracks around the coarse grains were observed in the lower filter cake layer FC3 (see Fig. 15). These cracks were more pronounced at the bottom of the filter cake layer FC3 than on the top of this filter cake layer. In addition, these kind of cracks were absent in case of the second filter cake layer FC1 (see Fig. 16). Possible sources of cracks could be a fast pressure release after filtration, an inhomogeneous particle size distribution due to possible sedimentation, or a too high pressure on the filter cakes during the demolding caused by the used circular die. These possible sources of errors will be investigated in future studies. Finally, Fig. 17 shows a representative digital microscope image of the interface between the two filter cake layers of the graded filter cake of run 1. A discrete transition between the two filter cake layers was observed. The digital microscope analyses confirmed the results of the XCT investigations because no cracks were detected at the interface between the two filter cake layers.
Fig. 17. Representative digital microscope image (cross section) of the graded filter cake of run 1. The FC1 filter cake layer is displayed on the top, while the FC3 filter cake layer is on the bottom. The different maximum grain sizes can be clearly distinguished. Furthermore, no cracks were found at the interface between the two filter cake layers.
4. Conclusions The present study investigated the production of graded functional composites for refractory applications with the pressure slip casting technique. For that purpose two filter cake layers build up from slips with 1 mm and 3 mm maximum grain size and a thickness of 20 mm were sequentially filtrated in a laboratory pressure filtration cell. The graded filter cakes were optimized using a full factorial experimental design. It was demonstrated that a lower filtration pressure for the first, i.e. coarser, slip in combination with a higher filtration pressure for the second, i.e. finer, slip was beneficial to improve the bonding between the two filter cake layers. Furthermore, the application of a thin prefiltered layer consisting of the same xanthan/guar gum solution which was used to stabilize the slips resulted in crack-free filter cakes and helped to remove the filter cakes from the PMMA filter medium. Finally, in order to achieve an improved joining between the filter cakes, the first filter cake layer should only be filtrated to the point that a soft layer still remains on the top. This soft layer improved the bonding between the filter cake layers significantly. The excellent joining between the filter cake layers and the lack of macroscopic pores and cracks was later demonstrated with X-ray computed tomography and digital microscopy. The pressure slip casting of graded refractories can provide a functionality to each filter cake layer. Thereby tailored compositions, microstructures, and thermomechanical properties can be achieved. Future work will thus concentrate on upscaling experiments and has the aim to provide a further functionality to the produced refractory components. Acknowledgments This work was kindly supported by the European Ceramic Society (ECerS) and by the JECS Trust in the “Frontiers of research” program for young scientists (contract 201242-1). References
Fig. 16. Plan view (2D) on the bottom of the filter cake layer FC1 of the graded filter cake 6. A very homogeneous distribution of the coarse grains was again noticed. No cracks and only a negligible amount of large pores was detected.
[1] E.G. Blanchard, Pressure casting improves productivity, Am. Ceram. Soc. Bull. 67 (1988) 1680–1683. [2] F.F. Lange, Powder processing science and technology for increased reliability, J. Am. Ceram. Soc. 72 (1989) 3–15. [3] O. Lyckfeldt, E. Lidén, M. Persson, R. Carlsson, P. Apell, Progress in the fabrication of Si3 N4 turbine rotors by pressure slip casting, J. Eur. Ceram. Soc. 14 (1994) 383–395. [4] R. Rice, Ceramic Fabrication Technology, Marcel Dekker, 2003. [5] A. Mattern, R. Oberacker, M.J. Hoffmann, Multi-phase ceramics by computercontrolled pressure filtration, J. Eur. Ceram. Soc. 24 (2004) 3219–3225. [6] S. Schafföner, C.G. Aneziris, Pressure slip casting of coarse grain oxide ceramics, Ceram. Int. 38 (2012) 417–422.
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008
G Model JECS-10516; No. of Pages 9
ARTICLE IN PRESS S. Schafföner et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx
[7] J.S. Reed, Principles of Ceramic Processing, 2 ed., Wiley & Sons, 1995. [8] J.M.F. Ferreira, H.M.M. Diz, Pressure slip casting of bimodal silicon carbide powder suspensions, Ceram. Int. 25 (1999) 491–495. [9] P. Greil, U. Gruber, N. Travitsky, M. Kulig, Pressure filtration of silicon nitride suspensions with constant filtration rate, Mater. Sci. Eng. A 151 (1992) 247–254. [10] S. Maleksaeedi, M.H. Paydar, S. Saadat, H. Ahmadi, In situ vibration enhanced pressure slip casting of submicrometer alumina powders, J. Eur. Ceram. Soc. 28 (2008) 3059–3064. [11] J. Requena, R. Moreno, J.S. Moya, Alumina and alumina/zirconia multilayer composites obtained by slip casting, J. Am. Ceram. Soc. 72 (1989) 1511–1513. [12] J.S. Moya, A.J. Sánchez-Herencia, J. Requena, R. Moreno, Functionally gradient ceramics by sequential slip casting, Mater. Lett. 14 (1992) 333–335. [13] B.R. Marple, J. Boulanger, Graded casting of materials with continuous gradients, J. Am. Ceram. Soc. 77 (1994) 2747–2750. [14] L. Zhang, V.D. Krstic, High toughness silicon carbide/graphite laminar composite by slip casting, Theor. Appl. Fract. Mech. 24 (1995) 13–19. [15] C.-H. Yeh, M.-H. Hon, The Si3 N4 and Si3 N4 /TiC layered composites by slip casting, Ceram. Int. 23 (1997) 361–366. [16] M. Kato, Y. Goto, T. Fukasaw, Rheological characterization of ˇ-sialon and Si3 N4 slips and fabrication of multilayered composites by slip casting, J. Mater. Sci. Lett. 16 (1997) 2055–2057. [17] Y.-G. Jung, S.-C. Choi, Fabrication of 3Y-TZP/SUS304 functionally graded materials by slip casting; application of porous alumina molds, Mater. Lett. 37 (1998) 312–319. [18] H. Ishibashi, H. Tobimatsu, K. Hayashi, T. Matsumoto, A. Tomsia, E. Saiz, Characterization of Mo-SiO2 functionally graded materials, Metall. Mater. Trans. A 31 (2000) 299–308.
9
[19] B. Kieback, A. Neubrand, H. Riedel, Processing techniques for functionally graded materials, Mater. Sci. Eng. A 362 (2003) 81–106. [20] M. Yan, X.L. Peng, T.Y. Ma, Microstructures of Ni-ZrO2 functionally graded materials fabricated via slip casting under gradient magnetic fields, J. Alloy. Compd. 479 (2009) 750–754. [21] M. Dröschel, M.J. Hoffmann, R. Oberacker, H. Both, W. Schaller, Y.Y. Yang, D. Munz, SiC-ceramics with tailored porosity gradients for combustion chambers, Key Engineering Materials, vol. 175, Trans Tech Publ, 1999, pp. 149–162. [22] U. Klippel, C.G. Aneziris, A.J. Metzger, Shaped coarse grained refractories by pressure slip casting, Adv. Eng. Mater. 13 (2011) 68–76. [23] J.A. Casas, A.F. Mohedano, F. García-Ochoa, Viscosity of guar gum and xanthan/guar gum mixture solutions, J. Sci. Food Agric. 80 (2000) 1722–1727. [24] R.J. Wakeman, Vacuum dewatering and residual saturation of incompressible filter cakes, Int. J. Miner. Process. 3 (1976) 193–206. [25] C¸. Hos¸ten, O. S¸an, Reassessment of correlations for the dewatering characteristics of filter cakes, Miner. Eng. 15 (2002) 347–353. [26] D.C. Montgomery, Design and Analysis of Experiments, John Wiley & Sons, 2008. [27] R Core Team, R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna, Austria, 2015. [28] R. Wakeman, S. Tarleton, Solid/Liquid Separation: Principles of Industrial Filtration, Elsevier, 2005. [29] S. Schafföner, C.G. Aneziris, H. Berek, J. Hubálková, A. Priese, Fused calcium zirconate for refractory applications, J. Eur. Ceram. Soc. 33 (2013) 3411–3418. [30] R.V. Lenth, Quick and easy analysis of unreplicated factorials, Technometrics 31 (1989) 469–473.
Please cite this article in press as: S. Schafföner, et al., Functional composites based on refractories produced by pressure slip casting, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.02.008