Classification of Particles Dispersed by Bead Milling with Electrophoresis †

The purpose of the present study is to develop a new particle classification technique and apparatus to replace centrifugal separation method, where continuous operation for classification is impossible because of the deposition of the coarse particles on the wall of the separator. In our recent research, it was clear that the zeta potential of the silica particles dispersed by a bead mill had size dependency. Hence, the particles were classified using an electrical field flow fractionation (EFFF) system under the condition that the zeta potential of the smaller particles was more negative than that of larger particles. Previous EFFF apparatus utilized horizontal field flow; however, the apparatus designed in this study had vertical field flow and cylindrical channel with length of 350 mm and radius width of 6 mm to reduce the running cost and operation time for classification of particles. About the classification performance of this apparatus, it was found that the silica particles with the size from 50 nm to 400 nm were classified using a low applied voltage. This method prevented deposition of particles on the wall of the apparatus and allowed continuous operation. Results of theoretical calculations supported qualitatively the experimental results obtained in this research.


Introduction
One of the current industrial requirements is technique of separation and classification for nanoparticles because their applications become widespread.So far, we have developed particle-classification techniques using electrical field flow fractionation [1][2][3] and beads mill dispersion. Bad mills have been widely used for the dispersion of particles in suspension 4) .In recent studies, a bead mill capable of nanoparticle dispersion was developed 5) , the surface potential of silica particles treated with a bead mill demonstrated size dependency ─ smaller par ticles had a more negative zeta potential.It was found that the mechanism for size dependency was related with the collision and frictions between particles and beads in the beads mill 6,7) .This technique effectively classifies nano-sized particles using vertical or horizontal electrophoresis [8][9][10] . Th purpose of the present study is to develop compact classification apparatus with electrophoresis and to evaluate its classification performance.

Measurement of particle size distribution
and zeta potential Particle size distribution of the slurries was measured using dynamic light scattering (DLS; LB-550, HORIBA Co., Ltd.).For the DLS measurements, the particle concentration of the slurry was adjusted so that the intensity of the scattered light would be detectable using the DLS apparatus.Because the DLS measurement of nanoparticle distribution included error, the parameters of the DLS measurements were modified to reduce the difference between the size distributions obtained using DLS and transmission electron microscopy.Modification of the DLS parameters allowed for a nearly accurate measurement of the nanoparticle size distribution 11) .† Accepted: July 26 th , 2011  Zeta potentials of the slurries were measured using a ZETASIZER (ZETASIZER 2000, MALVARAN Co., Ltd.).

Properties of the particle dispersed by a bead mill
Silica powder (DENKA FUSED SILICA, Denki Kagaku Kogyo, Co., Ltd.) with a specific surface area of 22.4-24.1 m 2 /g was used as the test powder in the present study.Glass beads (Plasma Beads, Neturen Co., Ltd.) ranging in size from 100 μm to 150 μm were selected for use in the bead mill (UAM-015, Kotobuki Industries Co., Ltd.), because their composition was almost identical to that of the test powder.
A bead mill was used to disperse the silica powder in water that had been deionized using an ionexchanged resin.The resulting slurr y was subsequently used for nanoparticle classification by EFFF.The bead mill was operated using the following procedures and conditions.First, the silica powder was added to 1.25 L of deionized water with gentle stirring at a final concentration of 2 wt% of the slurry mass.Then, the slurr y was circulated from the slurry tank to the bead mill at a flow rate of 3.33×10 -3 L/s using a pump.The peripheral speed of the bead mill rotor was 6.24 m/s and the glass beads occupied 60% of the chamber volume of the bead mill.After the milling operation for 30 min, the dispersed slurries were prepared.The size distribution of the tested silica particles was shown in Fig. 1a.Each size of particle obtained as a result of bead milling was characterized.The dispersed slurry prepared using the bead mill was separated into different sized particles using a centrifugal separator (Kansai Centrifugal Separator Mfg.Co., Ltd.), and then the zeta potential of the each sized particles were measured by the ZETASIZER.The relationship between zeta potential and median size was shown in Fig. 1b, which indicated that the zeta potential of the slurry had size dependency; the zeta potential of the smaller particles was more negative than that of the larger particles.The mechanism of which has been investigated elsewhere 5) .The relationship between the zeta potential and the particle diameter is shown in Eq. (1).
Also, the sedimentation balance method 12) was used to clarify the relationship between zeta potential and the specific size of the particle dispersed by the beads mill instead of the centrifugal classification.
Here, the conventional sedimentation balance apparatus was equipped with vertical electrophoresis.And then, the result was shown in the following equation calculating differences of the sedimentation curves before and after applying voltage at the electrical potential.This equation was applied to the silica particles with the size from 30 nm to 500 nm.

Classification of particles by the channel with electrophoresis
As shown in Figs. 2, an EFFF system was used to separate fine particles from the dispersed slurry produced by bead milling.A double cylindrical channel with length of 350 mm and a radius width of 6 mm stood vertically, as shown in Fig. 2b.The slurry was stirred in the slurry tank where the temperature and concentration of the slurry were controlled to be 30 ℃ and 0.2 wt%, respectively.There were three inlets in the top of the EFFF channel.One inlet was located at the nearest center of the apparatus and was used to supply the dispersed slurry from the slurry tank, and the others were used to supply the deionized water from the water tank.The flow rates of the slurry and deionized water could be controlled at their respective inlets.The linear velocities of the slurr y and deionized water were equivalent, and the total flow rate was from 200 to 500 mL/min to create laminar flow in the channel.The outside wall of the channel was the positive electrode and the inside cylinder served as the negative electrode.A DC electric field was generated perpendicular to the direction of fluid flow in the channel.Therefore, negatively charged silica particles moved towards the wall of the channel.A thin plate was placed at the bottom outlet for separation of particles in the outlet flow.Partial separation efficiency, Δη, which is defined by Eq. ( 3), was used to evaluate classification performance of the present apparatus.The mass of the dried powders from the fine silica and coarse particle slurries were measured.Sample of each slurry was collected by sample bottle, and the size distribution of each sample was determined by DLS.
In the above equation, mc and mf represent the mass of the coarse silica powder and the separated fine powder, while fc and f f represent the size distributions of the coarse silica powder and the separated fine powder, respectively.

Theoretical Calculations
vg and ve represent the Stokes sedimentation velocity and the velocity of electrophoresis, which are determined by Eqs. ( 4) and ( 5), respectively 13) . ) The velocity distribution of channel flow was calculated using the control volume method, the r-z two-dimensional grid portion of the channel shown in Fig. 3 on the assumption of axial symmetry, the equation of continuity, Navier-Stokes equations, and diffusion equation, as shown Eqs. ( 6) - (9).

The influence of applied voltage on partial separation efficiency cur ve Figs. 4 show the influence of applied voltage on
partial separation efficiency at the total flow rate of 300 mL/min by the experiments and theoretical calculations, where it was assumed that zeta potential, ζ, in Eqs (5) was represented by Eq. ( 1).The experimental and theoretical results were qualitatively consistent with each other.As the applied voltage decreased, the partial separation efficiency curve shifted towards small particles.This result indicated that the larger particles also moved to the fine outlet side, which was near the outside wall, by higher applied voltage according to the relationship between zeta potential and median size in Fig. 1(b).For example, in the case of 300-nm particles, all of them cannot be collected by the fine side at the applied voltage 4 V, however, all of the particles were collected the fine side when the applied voltage was 7 V as shown in Figs. 5. Thus, partial separation efficiency curve was controlled by changing the applied voltage; however, collection ratio as defined by Eq. ( 11) was decreased with the applied voltage increasing as shown in Fig. 6.
The reason of this result is the amount of particles collected by the wall increased.For example, in the case of the par ticles with the size of 120 nm, the   1).When the total flow rate was increased, the residence time for par ticles in the apparatus became short.Thus, partial separation efficiency curve was shifted towards smaller particles  because of larger influence on the smaller particle by electrophoresis.However, the difference between experimental and calculated values was large probably because Eq.( 1) did not show the relationship between zeta potential and particle size accurately.Then, numerical calculations were carried out under the condition of Eq.( 2) on the basis of the sedimentation balance method.The differences were remained because the slurry and water might not flow in the apparatus as simulated, but the calculated cur ves approached the experimental cur ves as shown in Fig. 9. Hence, it was possible to think that Eq.( 2) was more precise to express size dependency of zeta potential than Eq.( 1).The sedimentation method was effective for investigation of the relationship between zeta potential and particle size.creased because the residence time became shorter so that the particles were not easy to be collected by the outside wall of the apparatus.And also, the collection ratio was decreased with the applied voltage increasing because high collection of particles by the outside wall resulting from larger electrophoresis mobility as described in the section of 4.1.Hence, to keep high collection ratio for classification, the operating conditions of flow rate and applied voltage must be selected carefully.

Conclusions
The purpose of the present study is to develop a compact nanoparticle-classification apparatus with easy operation using EFFF.The channel of the apparatus was double cylinder with the length and diam-eter of 350 mm and 59 mm, respectively.The outside wall and inside cylinder of the channel ser ved as positive and negative electrodes, respectively.Using a low-applied potential, the silica particles with the size from 50 to 400 nm were classified.This method prevented deposition of negatively charged silica particles with low applied voltage and high flow rate, and allowed both low-energy and continuous operation.Results of theoretical calculations qualitatively supported the experimental results obtained in this study.

Fig. 1 Fig. 2
Fig. 1 Properties of the tested silica particle: (a) size distribution; (b) zeta potential with median size.

4. 2
The influence of flow rate on partial separation efficiency cur ve Fig. 8 (a) and (b) show the influence of flow rate on partial separation efficiency at the applied voltage of 15 V and the results of theoretical calculation according to Eq. (

Fig. 5
Fig. 5 Calculated 300 nm-particle trajectories in the apparatus at the flow rate of 300 mL/min and the applied voltage of: (a) 4 V; (b) 7 V.

Fig. 6
Fig. 6 Relationship between collection ratio and applied voltage at the total flow rate of 300 mL/min.

Fig. 10
represents the relationship between collection ratio and flow rate at various applied voltages.As flow rate increased, the collection ratio was in-

Fig. 7
Fig. 7 Calculated 120 nm-particle trajectories in the apparatus at the flow rate of 300 mL/min and the applied voltage of: (a) 7 V; (b)15V.

Fig. 8
Fig. 8 Effect of flow rate on partial separation efficiency curves at the applied voltage of 15 V by: (a) experiment; (b) numerical simulation according to Eq.(1).

Fig. 9
Fig.9 Effect of flow rate on partial separation efficiency curves at the applied voltage of 15 V calculated according to Eq.(2).

Fig. 10 Relationship
Fig. 10 Relationship between collection ratio and applied voltage.