Effect of solids concentration on particle size distribution of deagglomerated barium titanate in stirred media mills

Effect of solids concentration on particle size distribution of deagglomerated barium titanate in stirred media mills

ARTICLE IN PRESS CHERD-1550; No. of Pages 6 chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx Contents lists available at ScienceD...

1021KB Sizes 4 Downloads 63 Views

ARTICLE IN PRESS

CHERD-1550; No. of Pages 6

chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx

Contents lists available at ScienceDirect

Chemical Engineering Research and Design journal homepage: www.elsevier.com/locate/cherd

Effect of solids concentration on particle size distribution of deagglomerated barium titanate in stirred media mills A.B.G. Simpson a,∗ , J.A. Byrne a , J.A.D. McLaughlin a , M. Strawhorne b a

Nanotechnology and Integrated Bioengineering Centre (NIBEC), School of Engineering, University of Ulster, Jordanstown BT37 0QB, United Kingdom b AVX Ltd, Hillmans Way, Coleraine BT52 2DA, United Kingdom

a b s t r a c t The effect of solids concentration on particle size distribution of a suspension of barium titanate in toluene/ethanol solvent, deagglomerated in a re-circulating stirred media mill, has been examined by experimentation. The effect on the particle size distribution curves was recorded for solids concentrations ranging from 10% to 85% by weight and for impeller tip velocities of 4 m/s, 6 m/s and 8 m/s. Lower D50 (median) particle sizes were found at both low and high concentrations. Additionally the span of the distribution curves was found to be widest at both the lower and higher concentrations. At low concentrations high specific energies per pass of product through the grinding chamber were measured and this was as a result of high calculated stress numbers indicating high numbers of successful contact between particles and media. At high concentrations, high specific energies per pass were also measured and this was due to the increased density and viscosity of the suspension which had the effect of increasing residence time, reducing the number of passes. Thus an optimum range for solids concentration was determined which gave the narrowest spans of the particle size distribution curve, an important characteristic for deagglomeration of barium titanate for MLCC manufacture. © 2014 Published by Elsevier B.V. on behalf of The Institution of Chemical Engineers.

Keywords: Stirred media mill; Barium titanate; MLCC; Solids concentration; Particle size; Specific energy

1.

Introduction

Barium titanate (BaTiO3 ) is widely used as a dielectric in multilayer ceramic capacitors (MLCCs), the largest class of capacitor in terms of volume produced (Sato et al., 2011; Buchanan, 2004; Horiba Inc., 2008). MLCCs consist of multiple layers of mainly BaTiO3 dielectric separated by layers of mainly nickel metal electrodes and this type of construction gives a high capacitance density relative to the volume of the component. With advances in technology, the demand for miniaturization and higher capacitance has resulted in designs which increase the number of layers in a device, achieved by thinning of the dielectric and electrode layers. In addition, the demand for MLCCs to operate under harsher conditions (higher ambient temperatures and greater voltages) has increased. Well

deagglomerated and dispersed dielectric particles as indicated by reduced and narrower particle size distributions is critical to ensure a homogeneous microstructure and high packing density within the dielectric layers to reduce the occurrence of physical defect sites (Paik et al., 1998; Kim et al., 2003; Ying and Hsieh, 2007). A commonly used manufacturing technique for the building of MLCCs is the tape casting process. BaTiO3 is dispersed with organic additives in an organic solvent medium to form ceramic suspension (slip). This slip is fed through a precision controlled slot die onto polyester film which passes continuously underneath. The slip forms a continuous layer of precise thickness on the film. Electrode ink is screen printed onto the ceramic sheets to form the electrode pattern and the sheets are stacked to create the multilayer structure.



Corresponding author. Tel.: +44 028 70340523. E-mail address: [email protected] (A.B.G. Simpson). http://dx.doi.org/10.1016/j.cherd.2014.04.006 0263-8762/© 2014 Published by Elsevier B.V. on behalf of The Institution of Chemical Engineers.

Please cite this article in press as: Simpson, A.B.G., et al., Effect of solids concentration on particle size distribution of deagglomerated barium titanate in stirred media mills. Chem. Eng. Res. Des. (2014), http://dx.doi.org/10.1016/j.cherd.2014.04.006

ARTICLE IN PRESS

CHERD-1550; No. of Pages 6

2

chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx

Table 1 – Particle size of barium titanate feed. Particle size distribution

␮m

D90 D50 (median) D10

5.33 0.87 0.47

One of the key elements in the dispersion of BaTiO3 particles is deagglomeration where agglomerates of the particles are broken up into discrete, primary particles (Parfitt, 1981; TTP, 1993). A widely used technique for the deagglomeration of ultrafine materials is high energy media milling in stirred media mills, where the grinding chamber is fixed and the grinding media is stirred by an agitator (Snow et al., 1984; Kwade, 1999). The level of deagglomeration is determined by the number of stress events per unit time and unit volume and the stress intensity at these events. Using grinding media with reduced diameter beads with reduced mass allows operation at high rpm therefore inducing a high number of stress events at reduced stress intensities. In this paper, the effect of solids concentration of the BaTiO3 suspension on the particle size distribution achieved in a stirred media mill over the same milling time period is examined.

2.

Materials and methods

To reflect current MLCC manufacturing techniques, BaTiO3 particles dispersed in a blend of organic solvents (toluene and ethanol) were examined. BaTiO3 powder (BT03, Sakai Chemicals) was used. The BaTiO3 feed particles are approximately spherical and are agglomerated. Particle size distribution of the feed particles (Table 1) was determined by laser light scattering (Horiba LA-950). A solvent mix of toluene, min 99.5% purity, and ethanol, 97.6–97.9% alcohol content, (Brenntag Group) at a ratio of 1:1 by weight was used as the liquid medium. Blends of these two solvents are used extensively in barium titanate slurry manufacture (Cho et al., 2003). The dispersant used was a commercially available dispersant (NOF Corporation, Japan) composed of a maleic anhydride/styrene copolymer with polyoxy alkylene monalkyl ether side chains which adsorbs onto the surface of the BaTiO3 providing a steric hindrance effect (NCJ, 2010). The dispersant was used at a quantity of 25 mg dispersant per 1 g of powder. Varying solids concentrations from 10 wt% to 85 wt% (Table 2) were made up by firstly pre-mixing the solvents and dissolving the dispersant and then wetting the powder using a bench-top mixer with a toothed, dispersion-type blade (IKA Eurostar® Model Euro-ST). The 85 wt% solids concentration was determined to be the maximum practical concentration achievable as higher concentrations could not be successfully wetted to form a fluid suspension. The wetted suspension was then milled for 1 h in a laboratory horizontal stirred media mill (Dispermat® model SL-12C) where the grinding media was contained in 125 ml volume, horizontal chamber and spun by a disc impeller. The grinding chamber was water cooled. The experiments were carried out at three different impeller speeds, 4 m/s, 6 m/s and 8 m/s, measured as the speed of the outer tip of the discs. The suspensions were re-circulated through the grinding chamber by integrated electric pump/stirrer mounted within the product vessel and the pump speed setting was kept constant at 6000 rpm. The grinding media used was 0.3 mm

Fig. 1 – Representation of Dispermat® SL-12C laboratory mill (WMA, 2013). diameter yttrium stabilized zirconia beads and occupied 80% of the volume of the grinding chamber (100 ml of media). The total volume of each suspension was consistent at approximately 240 ml and therefore the ratio of suspension to media was 2.4/1 (Fig. 1). Net mechanical power consumption determines the energy transmitted by the impeller via the media to the product and is the total mill power consumption minus the idling power (power consumption under no product load conditions). The net power consumption relative to the duration of milling is recorded as Wh and was measured by mill instrumentation. Particle size distribution (PSD) measurements of the milled suspension were determined by laser light scattering (Horiba LA-950). Measurements were made of the D50 (median particle size), D90 (the particle size where 90% of the particles are below that size) and D10 (the particle size where 10% of the particles are below that size). In addition the span of the PSD curve was determined by the formula (D90-D10)/D50. A lower value of PSD span indicates a tighter distribution which implies a more even and consistent milling process (Horiba Inc., 2008; Mizuta et al., 1984). Viscosity measurements of the suspensions were made with a Brookfield model LV viscometer using a cylindrical spindle (#34) at a shear rate of 56 1/sec (200 rpm). Measurements were taken within a temperature stabilized sample chamber (25 ◦ ).

3. Energy considerations in stirred media mills Energy considerations in stirred media mills are discussed by Kwade et al. (Kwade, 1999; Kwade and Schwedes, 2002)

3.1.

Specific energy

Specific energy, SE (kJ/kg), is the amount of energy directly transferred into the mill chamber related to the quantity of

Please cite this article in press as: Simpson, A.B.G., et al., Effect of solids concentration on particle size distribution of deagglomerated barium titanate in stirred media mills. Chem. Eng. Res. Des. (2014), http://dx.doi.org/10.1016/j.cherd.2014.04.006

ARTICLE IN PRESS

CHERD-1550; No. of Pages 6

3

chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx

Table 2 – Experimental solids weight concentrations and volume fractions of BaTiO3 suspensions. 85 wt% 0.45

80 wt% 0.36

70 wt% 0.25

60 wt% 0.17

product and it is proportional to the product of stress number, SN, and stress intensity, SI. At constant stress intensities the specific energy is directly proportional to the stress number. Energy usage of the mill is measured by instrumentation and is recorded in the units Wh. From this, the specific energy is derived as; ((wh × 3600)/1000) m

(3)

where GM (kg/m3 ) is the density of the grinding media.

4.

Results and discussion

4.1.

Particle sizes and particle size distribution

4 ms

1

0.6 0.4

0%

20%

40%

60%

80%

100%

Solids concentration (wt%)

Fig. 2 – D90 particle sizes at different impeller tip velocities with increasing solids concentration. 0.6

8 ms

0.55

6 ms

0.5

4 ms

0.45 0.4 0.35 0.3 0.25 0.2

0%

20%

(2)

where ϕGM is the filling ratio of the grinding media, ε is the porosity of the bulk of grinding media beads, CV (–) is the volumetric solids fraction, n (s−1 ) is the number of impeller revolutions per second, t (s) is the milling time and dGM (m) is the diameter of the grinding media beads. The stress intensity at each of the stress events may be said to be proportional to the kinetic energy of the media beads (Kwade, 1999). It is assumed that the tangential velocity of the media beads, vt is equivalent to the tip speed of the impeller discs. Thus; SI ∝ d3GM GM v2t

10 wt% 0.02

6 ms

1.2

Stress number and stress intensity

ϕGM (1 − ε) nt (1 − ϕGM (1 − ε)cV ) dGM

20 wt% 0.03

0.8

Stress number, SN, is essentially how often each agglomerate is stressed. Kwade (1999) describes how the average number of stress events, SN, in a batch grinding process is a function of the number of media contacts, the probability that the agglomerate is caught and sufficiently stressed, and by the number of product agglomerate particles inside the mill. SN for deagglomeration may be calculated by; SN ∝

30 wt% 0.06

8 ms

1.4

(1)

where m (kg) is the mass of product deagglomerated. The specific energy value characterizes the process of deagglomeration as it describes the influence of mill size, circumferential speed of the impeller, solids concentration of the suspension and the density and size of the grinding media beads (Blecher et al., 1996).

3.2.

40 wt% 0.08

1.6

um

SE =

50 wt% 0.12

um

BaTiO3 wt% BaTiO3 vol. fraction

40%

60%

80%

100%

Solids concentration (wt%)

Fig. 3 – D50 particle sizes at different impeller tip velocities with increasing solids concentration.

The calculated spans of the PSD curves (Fig. 5) showed higher values, i.e. wider distribution curves both at lower solids concentrations and at higher solids concentrations with the lowest values being achieved in the range of 30–70 wt%. The spans for the 6 m/s and 4 m/s speeds are generally reduced compared to the higher 8 m/s speed. When viewed as PSD curves (Fig. 6 for 8 m/s velocity), the 85 wt% concentration shows a high proportion of larger particle sizes, the 80 wt% and 10 wt% concentrations show the highest proportion of finer particles. The curve with the lowest span is at the 50 wt% concentration.

0.4 0.3 0.25 um

Figs. 2–4 show the D90, D50 and D10 particle sizes respectively after 1 h milling time for the three impeller speeds. With increasing solids concentration, little difference was seen with the D90 except for very high solids concentrations where higher D90s were observed suggesting the presence of more agglomerates. For both the D50 and D10, reduced values were observed both at lower solids concentrations (<30 wt%) and at higher solids concentrations (>50 wt%) suggesting the presence of finer material. At very high solids concentrations (>80 wt%), however, the values observed showed larger increases.

0.35

0.2 0.15

8 ms

0.1

6 ms

0.05 0

4 ms

0%

20%

40%

60%

80%

100%

Solids concentration (wt%)

Fig. 4 – D10 particle sizes at different impeller tip velocities with increasing solids concentration.

Please cite this article in press as: Simpson, A.B.G., et al., Effect of solids concentration on particle size distribution of deagglomerated barium titanate in stirred media mills. Chem. Eng. Res. Des. (2014), http://dx.doi.org/10.1016/j.cherd.2014.04.006

ARTICLE IN PRESS

CHERD-1550; No. of Pages 6

4

chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx

2

6E+10

6 ms

1.6 um

Stress Number SN (-)

8 ms

1.8

4 ms

1.4 1.2 1 0.8

0%

20%

40%

60%

80%

6 ms 4 ms

4E+10 3E+10 2E+10 1E+10 0

100%

Solids concentration (wt%)

0%

20%

40%

60%

80%

100%

Solids concentration (wt%)

Fig. 5 – PSD curve spans at different impeller tip velocities with increasing solids concentration.

Fig. 8 – Theoretical stress number with increasing solids concentration.

18

1.20E-05

16

80% 50% 10%

12

1.00E-05

Stress intensity SI (Nm)

85%

14 Frequency (%)

8 ms

5E+10

10 8

8 ms 6 ms

8.00E-06

4 ms

6.00E-06 4.00E-06 2.00E-06

6 0.00E+00

4

0%

20%

40%

60%

80%

100%

Solids concentration (wt%)

2 0 0.01

0.1

1

10

Fig. 9 – Theoretical stress intensity with increasing solids concentration.

Particle size (um)

Fig. 6 – PSD curves for selected solids concentrations at 8 m/s impeller tip speed.

4.2.

Energy considerations

Specific energy, SE, derived from the measured values for net power input to the mill, shows a decrease with increasing solids concentration (Fig. 7). Calculated theoretical values of stress number, SN, and stress intensity, SI, indicate that the SN decreases exponentially with solids concentration while the SI remains constant with solids concentration as it is assumed that the media velocity is equivalent to the impeller tip speed (Figs. 8 and 9). Thus SE would be directly proportional to SN. This would tend to suggest that, with increasing solids concentration, larger values for particle size should be observed after similar milling times. From the graphs in Figs. 2–4,

140

6 ms

120

4 ms

4000

Suspension characteristics

To explain this observation, the physical characteristics of the suspensions must be considered. With increasing solids concentration there is an exponential increase in the viscosity of the suspensions due to the increased density and increased particle–particle interactions (Torsten, 2011) (Fig. 10). At concentrations greater than 80 wt% the viscosity increased dramatically. For this experiment, a value for viscosity at the concentration of 85 wt% had to be estimated at >5000 cPs due to excess torque experienced by the viscometer as a result of the high suspension density.

8 ms

5000

100

3000 2000

80 60 40

1000 0

4.3.

cP

Specific energy SE (kJ/kg)

6000

however, it was observed that at solids concentrations between 50 wt% and 80 wt%, the D50 and D10 values were reduced suggesting an increased milling effect. For the highest concentration, 85 wt%, higher particle sizes were achieved.

20

0%

20%

40%

60%

80%

Solids concentration (wt%)

Fig. 7 – Specific energy with increasing solids concentration.

100%

0

0%

20%

40%

60%

80%

100%

Solids concentration (wt%)

Fig. 10 – Suspension viscosity (cP) with increasing solids concentration.

Please cite this article in press as: Simpson, A.B.G., et al., Effect of solids concentration on particle size distribution of deagglomerated barium titanate in stirred media mills. Chem. Eng. Res. Des. (2014), http://dx.doi.org/10.1016/j.cherd.2014.04.006

ARTICLE IN PRESS

CHERD-1550; No. of Pages 6

chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx

45.0 Residence time tR (s)

40.0

8 ms

35.0

6 ms

30.0

4 ms

25.0 20.0 15.0 10.0 5.0 0.0

0%

20%

40%

60%

80%

100%

Solids concentration (wt%)

Fig. 11 – Residence time (s) with increasing solids concentration. 800

8 ms

No. of passes (-)

700

6 ms

600

4 ms

500 400 300 200 100 0

0%

20%

40%

60%

80%

100%

Solids concentration (wt%)

Fig. 12 – Number of passes through mill grinding chamber with increasing solids concentration. The increase in viscosity has the effect of reducing the volumetric flowrate of the suspension through the mill chamber which in turn increases the residence time tR (s) in the mill chamber (Fig. 11). Residence time is calculated by dividing the active volume of the mill chamber VGCA (m3 ) by the volumetric flowrate Q (m/s). tR =

VGC (1 − ϕGM ) + VGC ϕGM ε VGCA = Q Q

(8)

From this residence time, the number of product passes through the milling chamber over the total milling time t (s) can be calculated. This shows the number of passes reducing as the solids concentration increases (Fig. 12). Passes =

t tR

(9)

When the specific energy is related to the number of passes (Fig. 13) it is found that the lowest specific energy per pass is seen around 50 wt% concentration with the values rising as

Specific energy per pass (kJ/kg)

9.00

8 ms

8.00

6 ms

7.00

4 ms

6.00 5.00 4.00 3.00 2.00 1.00 0.00

0%

20%

40%

60%

80%

100%

5

the solids concentration both decreases and increases from that point. A steep rise in specific energy per pass is observed at both less than 30 wt% and greater than 70 wt% solids concentrations, equating to the reduced particle sizes after 60 minutes milling time. At less than 30 wt% there is a sharp increase in the calculated stress number and therefore a sharp increase in the proportion of successful media–particle interactions (i.e. those which produce a deagglomeration event). At greater than 70 wt% there is a sharp increase in the residence times due to the increasing suspension viscosity. This has the effect of reducing the number of passes through the grinding chamber effectively increasing the proportion of successful media–particle interactions. Between 30 wt% and 70 wt% the value for specific energy per pass is essentially consistent as there is both a reduction in the stress number and an increase in residence time (and therefore a reduction in the number of passes) and there is a balancing effect between the number of media–particle interactions and the time spent in the grinding chamber. The exception is at very high concentration (>80 wt%) where the resultant high suspension density and subsequent increased viscosity has a dampening effect on the grinding media beads, effectively reducing the velocity and therefore the stress intensity of the media (Knieke et al., 2010). The deagglomeration action becomes ineffective and this is reflected in the increased particle sizes after milling and specifically the high number of larger particles detected in the particle size distribution (Fig. 6).

5.

Conclusions

Solids concentration of the feed suspension has a strong influence on the outcome of the de-agglomeration process. Reduced particle size may be achieved within the same milling time at both lower concentrations (<30 wt%) and higher concentrations (>50 wt%) with the exception of very high concentrations (>80 wt%). Measurements of specific energy per pass through the mill indicate that increased specific energy is experienced by the suspension at both lower and higher concentrations. At lower concentrations this is as a result of increased number of successful stress events (i.e. those stress events resulting in a successful de-agglomeration event). At higher concentrations the increased specific energy per pass is as a result of the increasing residence times due to the increased suspension density and viscosity. At very high concentrations, despite a high specific energy per pass being measured, a higher particle size and an increase in frequency of larger particles is achieved. This is due to the high suspension viscosity as a result of high particle density causing a dampening effect on the media beads. For the deagglomeration of BaTiO3 for MLCC dielectric slip manufacture it is critical to achieve a reduced particle size distribution curve span which indicates reduced proportions of both agglomerates and finer particles. For this parameter, the lowest values of span are achieved in the range of 30–70 wt%. With optimization of impeller speed, this value can be reduced further.

Acknowledgments

Solids concentration (wt%)

Fig. 13 – Specific energy per pass (kJ/kg) with increasing solids concentration.

The authors would like to acknowledge the contributions from Mr. M. Conway and Mr. J. McCarry of AVX Ltd.

Please cite this article in press as: Simpson, A.B.G., et al., Effect of solids concentration on particle size distribution of deagglomerated barium titanate in stirred media mills. Chem. Eng. Res. Des. (2014), http://dx.doi.org/10.1016/j.cherd.2014.04.006

CHERD-1550; No. of Pages 6

6

ARTICLE IN PRESS chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx

References Blecher, L., Kwade, A., Schwedes, J., 1996. Motion and stress intensity of grinding beads in a stirred media mill. Part 1: Energy density distribution and motion of single grinding beads. Powder Technol. 86, 59–68. Buchanan, R.C., 2004. Ceramic Materials for Electronics, second ed. Marcel Dekker Inc., New York. Cho, C., Cho, Y., Yeo, J., Kim, J., Paik, U., 2003. Effects of PVB on the gelation behavior of BaTiO3 -based dielectric particles and glass suspension. J. Eur. Ceram. Soc. 23, 2315–2322. Horiba Inc., 2008. Applications Note, Particle Size of Electronic Ceramic Powders. Horiba Instruments Inc., Irvine. Kim, D.H., Yeo, J.G., Jung, Y.G., Choi, S.C., Paik, U., 2003. Suspension stability and consolidation behaviour of ultrafine BaTiO3 particles in nonazeotropic solvent system. Mater. Chem. Phys. 82, 181–187. Knieke, C., Steinborn, C., Romeis, S., Peukert, W., Breitung-Faes, S., Kwade, A., 2010. Nanoparticle production with stirred-media mills: opportunities and limits. Chem. Eng. Technol. 33, 1401–1411. Kwade, A., 1999. Wet comminution in stirred media mills – research and its practical application. Powder Technol. 105, 14–20. Kwade, A., Schwedes, J., 2002. Breaking characteristics of different materials and their effect on stress intensity and stress number in stirred media mills. Powder Technol. 122, 109–121. Mizuta, S., Parish, M., Bowen, H.K., 1984. Dispersion of BaTiO3 powders (Part II). Ceram. Int. 10, 83–86.

Nof Corporation, Japan, Oleo and Speciality Chemicals, Petrochemical Products, Product Information (Online), available from: http://www.nof.co.jp/english/business/oleo/product04.html (accessed March 2010). Paik, U., Hackley, V.A., Choi, S.C., Jung, Y.G., 1998. The effect of electrostatic repulsive forces on the stability of BaTiO3 particles suspended in non-aqueous media. Colloids Surf. A – Physicochem. Eng. Aspects 135, 77–88. Parfitt, G.D., 1981. Dispersion of Powders in Liquids, third ed. Elsevier Applied Science Publishers, London. Sato, S., Isozaki, H., Suk Shin, Y., 2011. Electronic Components. Asia Insight: MLCC. Future Market Dynamics. Morgan Stanley MUFG Research, New York. Snow, R.H., Kaye, B.H., Capes, C.E., Sresty, G.C., 1984. Size reduction and size enlargement. In: Perry, R.H., Green, D.W. (Eds.), Perry’s Chemical Engineer’s Handbook. , sixth ed. McGraw-Hill Inc., London, 8-1–8-72. Torsten, T., November 2011. Nano seminar: production and processing of nanoparticles by top-down processes. In: Rheology of Dispersions. Netzsch GmbH, Malvern. 1993. Tioxide Titanium Pigments, Dispersion of Titanium Dioxide Pigments, General Principles. Tioxide Group, London. WMA-Getzmann GMBH, Germany, Dispersion and Milling Systems, Bead mill SL (Online), available from: http:// www.vma-getzmann.com/english/dispersion&millingsystems (accessed July 2013). Ying, K.L., Hsieh, T.E., 2007. Dispersion of nanoscale BaTiO3 suspensions by a combination of chemical and mechanical grinding/mixing processes. J. Appl. Polym. Sci. 106, 1550–1556.

Please cite this article in press as: Simpson, A.B.G., et al., Effect of solids concentration on particle size distribution of deagglomerated barium titanate in stirred media mills. Chem. Eng. Res. Des. (2014), http://dx.doi.org/10.1016/j.cherd.2014.04.006