Fluidization of nanoparticle and nanoparticle agglomerates with supercritical carbon dioxide

Transcription

1 Fluidization of nanoparticle and nanoparticle agglomerates with supercritical carbon dioxide Victor Martín, Soraya Rodriguez-Rojo*, Marta Salgado, Maria José Cocero High Pressure Processes Grp, Dept Chem Engn & Environm Technol, Universidad de Valladolid, Dr Mergelina s/n Valladolid, Spain * sorayarr@iq.uva.es Abstract Agglomerates of aluminum oxide nanoparticles of 13 nm have been chosen as model particles. The fluidization is assessed through experimental determination of minimum fluidization velocity. The measurement is based in the stabilization of pressure drop across the bed. The study has been carried out in a pressure range from 9.7 to 11.7 MPa and temperatures between 35 to 55 C, that covers a range of carbon dioxide density values between 490 to 710 kg/m 3. The minimum fluidization velocity at these operating conditions varies from to cm/s. Introduction Fluidized bed technology is widely used in different industrial processes for reactions, drying, coating and combustion as it provides good mixing conditions, homogeneity in temperature and concentration. Nowadays, the interest in methods on handling and fluidization of particle of submicronic (100 nm 1 μm) and nanometric (< 100 nm) size is increasing. The fluidization of submicro- and nano-particles is challenging due to their strong interparticle forces (London - Van der Waals forces, electrostatic interactions and liquid bridges). Currently, the investigated methods to avoid channeling and improve the fluidization of nanoparticles involved the use mechanical systems (stirring, centrifugal devices, ) and field of forces (magnetic, acoustic, ultrasonic, electric) [1]. The aim of this study is to investigate the fluidization of nanoparticles with supercritical carbon dioxide, as high pressure has been probed to improve fluidization of microparticles [2,3,4] through increasing fluid density, and fluid particle interactions [2] and avoiding the formation of liquid bridges, as supercritical fluids have no surface tension. Coupling the fluidization technology with the use of supercritical fluids (SCFs), namely supercritical carbon dioxide (SC-CO 2 ), has been proved to be valuable in coating of microparticles [5, 6] as SCFs make possible the solubility control of the coating agent by small changes in pressure and temperature, and reduce interfacial forces between fluids decreasing the agglomeration of particles. It could be also interesting in other processes such as reaction due to the absence of mass transfer limitations. The most common method to assess the fluidized state of a bed of particles is to determine the minimum fluidization velocity (u mf ) via the measure of the pressure drop experimented by a fluid circulating upwards through the bed [7]. This pressure drop is generated by the resistance of the particles to the fluid flow. When the fluidized bed is established, the pressure drop keeps a constant value even if the fluid flow is increased. This means that the gravitational force on the particles and the drag frictional force are compensated by the buoyancy or force exert by the fluid. In this steady state, the pressure drop value is equal to the effective weight of bed per unit cross sectional area in pressure units (W eff ):

2 W eff m p g p S (equation 1) Where, mp is the mass of particles, g is the acceleration of gravity, ρ p is the density of the solid particles, ρ is the density of the fluid and S is the cross-sectional area of the bed. The fluidization can be follow by the fluidization index that is defined as the dimensionless ratio between the experimental pressure drop (ΔP) and the W eff. When fully fluidized, all particles are suspended by the fluid so the fluidization index is equal to unity. Experimental Aluminium oxide (Al 2 O 3 ) nanoparticles (Aeroxide AluC, Evonik) have been used as model powder. According to supplier data, the mean particle size is 13nm, the bulk density is 40 g/l and the tapped density is 58 g/l, in contrast to the aluminium oxide density of g/cm 3. Technical grade (99.5%) carbon dioxide (Carburos Metálicos S.A.) was used as fluidizing media. In Figure 2, the schematic flow diagram of the equipment is displayed: Carbon dioxide, stored in a gas cylinder (a) at MPa, is pumped in liquid state (T<5ºC) by a membrane pump (b,eh-m-510v1, Lewa) until the operating pressure is reached. This pressure is controlled at the end of the line by a backpressure valve (c, BP-66, GO Inc). After the pump, there is an equalizing reservoir (c) in order to avoid pressure and flow oscillations due to the pulsating flow of the pump. This vessel has a volume of 1.2L, the same as the vessel for the fluidization bed (e). A stainless steel basket (H = mm; D= 41 mm), with stainless steel sintered plates at bottom and top, is used to introduce the particles inside the vessel and to make their handling easier. The bottom plate acts as fluid distributor. At the outlet of the fluidized bed vessel, there is a filter to avoid that any particle could be recirculated with the fluid. The fluidized state was assured by means of pressure drop across the bed measurements that were carried out with a (j) differential pressure meter (f, 3051CD, Fisher-Rosemunt) by means of two concentrical tubes of 1/8 and 1/4 in. inside the bed. The equipment was calibrated in the range of -1mbar to 3 mbar. In a previous work of microparticles fluidization with SC-CO 2 [4], it was tested that this device has no influence on the fluidization performance. In the first step of the study, the flow rate was controlled with a manual metering valve (2232 HOKE, Cv = ) in the recirculation line. However, it was not possible to work with flow rates below 500 g/h and the oscillations were also quite big ( g/h). Consequently, a flow controller (d, Mini Cori-Flow M14- RGD-220S, Bronkhorst) was purchased. It allows controlling the flow rate with a mean incertitude of 0.5% of the measured value for values above 500g/h and a maximum of 3% incertitude for values between 100 and 500g/h. Additionally, the temperature in both vessels is controlled by water circulating through the jacket of each vessel from two external thermostatic baths.

3 d) h) g) e) a) b) f) c) Figure 1. Schematic flow diagram of the experimental device Results and discussion The minimum fluidization velocity (u mf ) was determined through the fluiddynamic curve using a mass of solid of 8.5g. The first set up with the manual control of flow came out inefficient: the control of flow rates was difficult, as mention before, so the change from fixed bed to fluidized bed was no clear and not reproducible, as shown in Figure 2. 0,9 0,8 0,7 0,6 - P (mbar) 0,5 0,4 0,3 0,2 0, ,01 0,02 0,03 0,04 0,05 0,06 0,07 u mf (cm/s) Figure 2. Fluid-dynamic curve for a mean SC-CO2 density of 650 ± 20 kg/m 3 (Effective Bed weight; W eff = ± mbar): Black symbols (, & ) different experiments with manual flow control. Open symbols ( ) experiment with automatic flow control. In Figure 3, the experimental minimum fluidization velocity determined with the automatic flow control system in the range of temperature from 35 to 55ºC and pressure between 9.7 to 11.7 MPa is shown. As expected, the u mf decreases when increasing the fluid density [2-4].

4 0,10 0,08 u mf (cm/s) 0,06 0,04 0,02 0, (kg/m 3 ) Figure 3. Experimental u mf values of Aeroxide Alu C ( ) as a function of fluid density. Correlation of u mf according to Wen Yu equation for different agglomerated sizes considering the density of the solid alumina (3900 kg/m 3 ): ( ) 17 μm; (---) 31μm and ( ) 40μm. These results are quite promising regarding the fluidization of nanoparticles without external field forces or mechanical devices. In a recent work, S. Kaliyaperumal [8] and collaborators have fluidized the same type of particles with air assisted by acoustic vibration. The minimum fluidization velocity of alumina nanoparticles at a sound pressure level of 120 db and frequency 120 Hz, was 0.75 cm/s. This value is tenfold higher, indicating the existence of higher agglomerates and interparticle forces. Moreover, the fluidization index never achieved a value equal to unity due to the unfeasibility of completely fluidize nanoparticles, as it was pointed out by the authors. In this work, the fluidization index exceeds this value and then decrease to unity, as inter-particle forces are overcome by the up-ward fluid [7]. There are several correlations, derived from the Ergum equation for particulate beds in terms of the Reynols number for particles (Re mf ) at the umf and the Arquimides number (Ar) to predict the umf. The general formula is Re mf = (A 2 + B Ar) ½ - A. One of the most well-known is the Wen-Yu equation (A = 33.7 and B = ), that has been already succesfully used to correlate umf for microparticles fluidized with sc-co 2 [3,4]. The first drawback to test this correlation for the fluidization of nanoparticles is the particle size of the agglomerates that remains unknown, expresed in the Re mf as mean Sauter Diameter or Surface Weighted Mean D[3,2]. The size of nanoparticle agglomerates during fluidization is mainly determined by non-invasive techniques such as laser-based planar imaging that requires transparent bed walls. Another method that has been used to obtain information about agglomerate size is to fit bed expansion data to the Richardson Zaki empirical equation [9], which are usually get from direct observation of the bed, as well. The vessel of the used experimental setup (Figure 2) does not allow direct observation of bed, thus, this parameter could not be experimentally determined.

5 In other to get a rough estimation of the size of the agglomerates, particle size was measured with a laser diffractometer (Mastersizer 2000, Malvern) equipped with a dry powder feeding system (Sirocco, Malvern). During the measurement, the sample powder (1 2 g) was shaken from a vibrating sample tray where the sample was then blown by compressed air, (1 bar and 2 bar), against and around a bend then across the lens were the particles passed through two focused laser beams (red light: helium-neon laser emitting at 633 nm, and blue light: solid-state light source at 466 nm). Also, particles were analyzed in water using a manual dispersion unit (Hydro MU) that allows using ultrasounds to break agglomerated. The measurements, summarized in Figure 4, show a bimodal agglomerate distribution: the higher volume proportion of the particles forms agglomerates around 10 μm and small volume produces agglomerates between μm. Only the ultrasonicated sample presents a unimodal distribution centered at 9 μm. As expected, the sample analyzed in water presents more agglomerates due to the formation of liquid bridges between the particles. Regarding the measurement in pressurized air, the size of the agglomerates is reduced as the pressure increases. Taking into account that the density of sc-co 2 at operating conditions is liquid-like ( kg/m 3 ) and viscosity is gas-like ( Pa s), and the samples present a quite similar distribution, a Surface Weighted Mean D[3,2] value of 17 μm, corresponding to the analysis in water, would be used to test the Wen-Yu equation. a) d) c) b) c) b) a) Figure 4. Volume agglomerate size distribution of the Aeroxide AluC nanoparticles under different measurement conditions: a) air at 1 bar ( ); b) air at 2 bar ( ); c) water ( ) ; d) water with ultrasounds ( ) As shown in Figure 3, the Wen-Yu equation highly under-predicts the minimum fluidization velocity. However, due to the impossibility to measure the real size of the agglomerates during the fluidization, its applicability cannot be excluded. The second uncertainty regarding the correlation is the value of solid phase density used to calculate the Ar number. Some authors [9], in fluidization experiments with air or N2 at ambient pressure, assume a value equal to the tapped density of the solid. In this case this assumption is not valid, as it would provide negative values of Ar since Ar α (ρ s ρ f ). Conclusions and Outlook In this preliminary work, nanoparticles of aluminum oxide (13 nm) have been fluidized for the first time, up to the best knowledge of the authors, with supercritical carbon dioxide. The minimum fluidization velocity was measured in a pressure range

6 from 9.7 to 11.7 MPa and temperature range between 35 to 55 C. The minimum fluidization velocity at these operating conditions varies from to cm/s. Nevertheless, the capability of supercritical carbon dioxide to fluidize nanoparticles should be assessed for nanoparticles of different characteristic (surface charge, bulk density, ) and also a broader range of operating conditions should be tested. Moreover, strategies should be developed to determine study of the bed expansion and to estimated/predict the characteristics (size and density) of the agglomerates formed during the fluidization, despite the fact that the bed of particles is not visually accessible. Acknowledgments The authors would like to thank the financial support from the Spanish Ministerio de Ciencia e Innovación through the project CTQ 2010/ S.Rodríguez-Rojo thanks the Spanish Ministry of Education for her postdoctoral grant. References [1] J.R. van Ommen and R. Pfeffer, Fluidization of nanopowders: Experiments, modeling, and applications, 13th International Conference on Fluidization, May (2010) Gyeong-ju, Korea [2] J. Li and J.A.M. Kuipers, Effect of pressure on gas solid flow behavior in dense gas-fluidized beds: a discrete particle simulation study, Powder Technology 127 (2002) [3] C. Vogt, R. Schreiber, G. Brunner and J. Werther, Fluid dynamics of the supercritical fluidized bed, Powder Technology 158 (2005) [4] S. Rodríguez-Rojo, N. López-Valdezate and M.J. Cocero, Residence time distribution studies of high pressure fluidized bed of microparticles, J. Supercritical Fluids 44 (2008) [5] R. Schreiber, C. Vogt, J. Werther and G. Brunner, Fluidized bed coating at supercritical fluid conditions, J. Supercrit. Fluids 24 (2002) [6] S. Rodríguez-Rojo, J. Marienfeld and M.J. Cocero, RESS Process in Coating Aplications in a High Pressure Fluidized bed Environment: Bottom and Top Spray Experiments, Chemical Engineering Journal 144 (2008) [7] J.R. Howard, Fluidized bed technology principles and applications. Ed. A. Hilger (Bristol), [8] S. Kaliyaperumal, S. Barghi, J. Zhu, L. Briens, S. Rohani, Effects of acoustic vibration on nano and sub-micron powders fluidization, Powder Technology 210 (2011) [9] J.M. Valverde, A. Castellanos, Fluidization of nanoparticles: A simple equation for estimating the size of agglomerates, Chemical Engineering Journal 140 (2008)