Particle Formation

Introduction

This interim report summarises work carried out between April and December, 1989. The project is scheduled for completion in mid-1990. The investigation consists largely of a case study of granulation using three commercial binders; here we describe the context of the work, summarise the relevant studies which have recently been carried out at the University of Surrey and give preliminary results of the case study.

Summary

During the first two years of this project, we have developed a scheme for quantitatively measuring the bond strength distribution of agglomerates via the use of a calibrated ultrasonic field and developed a synthesis technique for producing model agglomerates with monosize primary particles and narrow strength distributions which can be varied over a wide strength range. Also, the use of this calibrated ultrasonic field approach as a diagnostic tool was demonstrated during titania processing to show at what points in the process, hard agglomerate formation occurs and identify possible processing changes to eliminate hard agglomerate formation.

Introduction

In many studies of precipitation, the length of the nucleation period is commonly invoked as the primary control parameter for forming uniform particles in a reaction involving homogeneous nucleation and growth (l-3). In the model originally proposed by LaMer and Dinegar (14) for the mechanism of formation of sulfur sols, uniform size distributions result if all the particles are formed in a short burst of nucleation, and then particle growth occurs by a mechanism where large particles increase in diameter slower than smaller particles (as occurs when growth is limited by diffusion to the particle surface). Despite the influence this model has had on studies of inorganic particle precipitation chemistry, the model has seen little corroboration and indeed has been brought into question for the sulfur sol for which it was developed (15). While the LaMer mechanism would undoubtedly result in uniform particles, finding systems meeting the conditions required for the model to hold has been elusive.

In recent years studies on the formation of polymer latex particles have provided an alternative mechanism whereby uniform particles can result from a homogeneous nucleation, precipitation reaction. In emulsion polymerization, an insoluble monomer is mixed with water and a water soluble free radical initiator added. Final particle size depends on reaction temperature, reagent concentration and parameters controlling the colloidal stability of the growing particles (i.e., ionic strength and pH). The initial locus of the reaction is in the aqueous phase. Oligimers grow to a size where they become insoluble and undergo a sol to gel transition. Due to their relatively low concentration, this transition involves only few polymer molecules and the gel phase grows by aggregation. The charge on the primary particles is small but as the aggregation process proceeds, the charge per particle grows. As a consequence, aggregation rates between particles of equal size decrease but the rate of aggregation of particles of dramatically different size increases. The result is a bimodal particle (or gel phase) size distribution with one peak located at the primary particle size and the second peak containing particles that are stable to mutual coagulation and grow by scavenging smaller particles. Upon aggregation, the gel phase particles coalesce and thus the particles are able to retain a spherical shape. As the reaction proceeds the growing particles reach a constant number density. These colloidally stable particles swell with monomer and the locus of the reaction is transferred from the aqueous phase to the inside of the particle phase. Uniformity is achieved through control of the colloidal stability of the primary particles and aggregates of these particles (6-8).

A second organic analogy to the precipitation of inorganic particles occurs in dispersion polymerization. Here a solvent is chosen where the monomer is soluble but the polymer is not. The reaction is initiated and proceeds through polymerization until the oligimers grow to a size where they undergo a phase transition and precipitate to form polymer and solvent rich phases. Uniform particles are achieved through the addition of steric stabilizers that control the aggregation of the growing gel phase (9).

The major distinction between the model of kMer and that developed for uniform latex particles lies in the incorporation of colloidal stability of small particles. The LaMer model assumes that each nucleus is colloidally stable and survives at the end of the reaction at the center of a particle. The aggregation models argue that while stabilizing such small particles is difficult, aggregation does not necessarily result in a broad particle size distribution. In developing schemes for control of particle size distribution, the result of accepting that colloidal stability can play an important role is that rather than focussing attention on the length of the nucleation period, one becomes very concerned about the colloidal properties of the growing particles.

In this paper we review our studies on the formation of uniform precipitates through the hydrolysis and condensation of titanium and silicon alkoxides. Our work has been aimed at elucidating whether uniformity is the result of the details of the chemistry or if particle size distributions can be controlled by physico-chetnical means. In particular we focus on experiments probing the length of the nucleation period in these systems and the effects of parameters controlling particle interaction potentials. We find that for the systems studied, the nucleation period appears to be a substantial fraction of the entire reaction period and that particle size is largely controlled by parameters related to particle interaction potentials. These results suggest that there are strong links between the physical chemistry underlying the formation of uniform latex particles and that controlling particle size distributions of hydrous metal oxide precipitates (10-14).

EXECUTIVE SUMMARY

The primary objective of the Caltech program supported by IFPRI is to understand the dynamics of particles which do not coalesce immediately upon coagulation and, through that understanding, to guide the development of processes for production of particles with particular properties. Theoretical and experimental investigations of the dynamics of aggregate aerosols have been undertaken. The factors that must be taken into account in such models were first explored theoretically. Particles are represented as fractal agglomerates. The mobilities and cross sections of such aggregates were examined, and upper and lower bound estimates of the collision frequency function are developed. Predictions of the dynamics of a coagulating aerosol indicate that aggregates coagulate more rapidly than do dense spheres.

In the initial year of the program we have focussed our experimental program on describing the mobility of agglomerates. Agglomerate particles were classified using electrical mobility techniques. The classified particles were collected on electron microscope grids for image analysis of the photographs taken from those samples to estimate fractal structure parameters. The particles studied include titanium dioxide particles produced by pyrolysis of titanium tetraisopropoxide and elemental silicon particles produced by pyrolysis of silane. The former particles are representative of particles of considerable industrial interest. The latter particles are a convenient model system that has been used for initial studies of coalescence of agglomerate particles.

The mobility equivalent size of the free molecular and transition regime aggregates studied to date has been found to correlate well with the projected area of the particles. During the past year we also had an opportunity to study the mass transfer to agglomerate particles using an instrument which was brought to Caltech by Urs Baltensperger, a visitor from ETH in Zurich. The instrument, called an epiphaniometer, is a device for measuring the mass transfer to particles by attachment of radioactive lead atoms. The experiments have been performed using mobility classified particles and, as expected, we find that the mass transfer rate scales directly with the mobility for drag equivalent diameter of the particles. That is, spherical particles and agglomerate particles with the same migration velocity in an electric field exhibit the same mass transfer rates.

Because electrical mobility measurements play a key role in our studies of the dynamics of agglomerate particles, we have also conducted a series of experiments aimed at measuring the charge uptake by these agglomerate particles. These experiments have been conducted using bipolar chargers, that is, radioactive sources that generate both positive and negative ions in the gas that then attach to the particles. The experiments involve taking a neutral aerosol from which all charge particles have been removed with an electrostatic precipitator and exposing it to bipolar ions. The number of neutral particles remaining after exposure to a fixed ion concentration is measured. The neutral fraction scales, once again, directly with the mobility equivalent diameter.

Thus, from each of our studies to date, we find that the mobility equivalent diameter is a very convenient quantity for characterizing the agglomerate particles. During the past year we have also advanced our efforts somewhat in the modeling of the agglomeration process itself. This requires an understanding not only of the mobility or diffusivities of the particles, but also of their collision cross sections. To date, we do not have direct measurements of the collision cross section, so we have been forced to make estimates to provide upper and lower bounds on those cross sections and, hence, on the coagulation rates. Still, the estimates that we have made do provide some insights into the differences between the agglomeration process and coagulation of rapidly coalescing particles.

Executive Summary

The overall goal of this feasibility study is to produce sub-50 um droplets when spraying highly viscous (up to 100,000 cP) non-Newtonian fluids at process level throughputs (up to 1 kg/s). Work was conducted using a new type of nozzle, developed during the last three years at Purdue, because it is the only candidate likely to meet the sub-50 um criterion.

Work was carried out using fluids comprised of glycerin/water/polymer and coal-water slurry/glycerine/polymer mixtures. Fluids with consistency indices ns high as 41.500 and flow behavior indices as low as 0.27 were employed. All spray data is reported as mean particle size, in terms of the Sauter mean diameter. Mean drop sizes as low as 28 pm have been achieved with air-liquid ratio values of less than 0.20.

The goal of this feasibility study was achieved. In fact, mean drop sizes as low as 38 um were measured at an air-liquid mass ratio of 0.2 and a nominal throughput of 1 kg/s by using an effervescent atomizer. In addition, nozzle performance was shown to improve with throughput, in contrast to the behavior of conventional twin-fluid injectors. Finally, the addition of polymer to either single- or multiphase fluids was shown to increase mean drop size, although the extent of the increase left SMD below the target value of SO um. An explanation for this increase is being pursued.

Work during the next contract year will be focused in three areas:

• An investigation into how the tightness of the particle size distribution varies with fluid properties and throughput.
• An investigation into why polymer addition increases mean drop size.
• Extension of the mathematical model for effervescent atomization developed by the principal investigator and his colleagues to fluids and conditions of interest to IFPRI members.

The model will then be used to determine the minimum mean drop size achievable with an effervescent nozzle, given a particular fluid, throughput rind nozzle geometry, and to identify the physical mechanisms responsible for performance barriers associated with effervescent atomizer operation.

Executive Summary

The goal of this one year study was to demonstrate that sub-50 pun mcxl drop size sprays can be produced when using an effervescent atomizer to spray high viscosity, non-Newtonian fluids at mass flow rates up to 1 kg/s. That goal was met. A secondary goal was to determine the spatial structure of the spray, in terms of how mean drop size and the width of the drop size distribution varied with axial and radial position within the spray. That objective was also met.

1991- 1992 begins the first year of our three year study into the fundamental mechanisms responsible for effervescent atomization. This year, we will be focusing on the transition region where the two-phase supersonic flow that exits the nozzle as discrete gas bubbles in a continuous liquid is transformed into a continuous gas stream containing discrete liquid drops. Single and multiple-pulse holography will be used to obtain the data necessary to achieve that goal. In particular, single-pulse holography will be used to determine the mechanisms of effervescent atomization while multiple-pulse holography will be used to determine drop size distribution and droplet velocities. A qualitative explanation of the mechanisms responsible for effervescent atomization should be available by the 1992 annual meeting with the first quantitative results available later in 1992.

Executive Summary

The primary objective of the Caltech program supported by IFPRI is to understand the dynamics of particles which do not coalesce immediately upon coagulation and, through that understanding, to guide the development of processes for production of particles with particular properties. Theoretical and experimental investigations of the dynamics of aggregate aerosols have been undertaken with IFPRI support. Following on our studies of the properties of agglomerate particles, we have developed models to describe the kinetics of agglomeration. The kinetics of agglomeration are determined by the mobilities of the agglomerate particles and by their collision cross sections. The lower density of agglomerates has competing effects on the coagulation kinetics as compared with dense particles of equal mass: (i) the aerodynamic drag on the particles is increased due to the larger size of the particle of the same mass; (ii) the low density tends to increase the collision cross section of the agglomerate over that of the dense sphere. We have used fractal scaling concepts to evaluate the relationship between these measures of particle size and the particle structure.

In this report, we first examine the scaling that determines the asymptotic form of the size distribution, the so-called self-presenting particle size distribution. A dynamic exponent is defined that describes the rate of growth of the mean particle size. The dynamic exponent is shown to pass through a minimum in the transition regime, behavior that has not previously been described. Essential features of this asymptotic solution are observed experimentally, although direct comparison between experiment and theory is complicated by the transition between coalescing coagulation and agglomeration that took place in our experiments at about the same particle size as the transition between the free molecular and continuum size regimes.

A more complete description of aerosol agglomeration was made in numerical solutions to the coagulation equation using a modified sectional code. Comparison of the collision frequency functions of spheres and agglomerate particles reveals that the predominant effect is the increased collision cross section of the agglomerate, leading to a dramatic increase in the collision frequency of agglomerates as compared to dense particles of the same mass. Simulations that assume that particles coalesce completely up to a limiting size and do not coalesce at all beyond that size exhibit a broadening of the size distribution at that transitional size. Experimental evidence of that broadening is provided. The comparisons at this point can be considered only provisional in as much as direct experimental validation of the collision frequency function is still lacking. This experimental evaluation is the key objective of the research under this program for the coming year.

Executive Summary

This IFPRI project investigates the formation mechanisms of uniform, submicron inorganic particles. Due to poor understanding of the physical chemistry governing particle nucleation and growth, this project has undertaken to determine if there are underlying physical or chemical themes that give rise to uniform precipitates. Previous work has focussed on spheres produced from the hydrolysis and condensation of silicon alkoxides. Here the necessity of achieving colloidal stability among the growing particles was hypothcsized as an essential step in the formation of process, Studies on the silicon alkoxide system indicated that particle growth occurred through the agglomeration of primary particles produced at a rate independent of the presence of larger particles and that unifomlity was the result of size dependent rates of aggregation.

In this annual report we review studies on a second alkoxide system. Here the formation of titanium hydrous oxide particles are discussed. The results of these studies demonstrate that a mechanism similar to that observed for silica is operable. In addition the necessity of maintaining colIoidal stability among the large particles is emphasized. This is demonstrated through experiments showing that early in precipitation reaction a constant number of colloidalIy stable particles is formed. If the particle surface potential is less than 12mV for the ionic conditions investigated, late in the reaction agglomerates are formed and become fused. If the surface potential is greater than 12mV and the reacting solution is not sheared, uniform particles are formed. However, a critical shear rate exists where for particles with surface potentials greater than 12mV, agglomerates are formed from uniform particles of a particular size. These results are interpreted in terms of particle interactions consisting of van der Waals, electrostatic and short range repulsive interactions, Agglomeration of uniform particles is hypothcsized to be the result of the combined action of the formation of a shallow minimum in pair interaction potential for particles of a partiuar size, and the continued precipitation of reactive material.

EXECUTIVE SUMMARY

Agglomeration is an inherent problem in almost all industrially relevant powder processing techniques. Successful dispersion of powders often requires the total elimination of agglomerates. To achieve this, it is important to understand the nature of, and to ascertain the properties of, these agglomerates. Of particular importance is the strength of the interparticle bonds, (between primary particle units forming the agglomerate), in relation to the powder processing procedure used for their generation.

In this IFPRI-funded project, a characterization technique has been developed to quantitatively determine agglomerate strength distributions in bulk powders by exposing a sample dispersion to a calibrated ultrasonic field and following the ensuing changes via particle size analysis. This technique may be used throughout a powder processing procedure to determine the source (onset) of agglomeration, as well as an aid in the elimination of these agglomerates by helping the user to understand the nature of the interparticle bonds.

The primary advantage of this technique (over other agglomerate strength determination techniques) is that it classifies agglomerate strength with a distribution rather than a single value, giving the user a more complete understanding of the system under consideration, as single value agglomerate strengths describe only the magnitude of agglomeration and not the extent. Furthermore, this is a ‘wet’ testing technique that allows agglomerate strengths to be determined under conditions similar to process conditions, which can be of great significance as powders often display different physical and/or chemical properties in and out of a liquid. Also, the simplicity of the testing procedure lends itself to automated, in-situ analysis (currently under development), whereby agglomeration in a system can be continuously monitored. Thus changes in the extent/magnitude of agglomeration (in response to modifications in the processing procedure) can be observed, and also monitored in a time dependent fashion.

The validity of this approach has been demonstrated for model silica agglomerates of known bond strength distribution and primary particle size prepared by heat treating ordered, sub-micron silica spheres at various temperatures. The use of this approach to find at what step agglomerates form in industrial processing is illustrated by measuring agglomerate strength during various titania processing schemes. As an example, the type of washing (ethanol versus water) and the drying conditions (temperature) caused large differences in the quantity and strength of hard agglomerates formed from the same titania precursor powder.

A numerical scheme has been developed for analyzing the change in the particle size distribution of a powder dispersion during ultrasonic breakdown and determining the actual mechanism of particle breakdown (erosion vs. fracture). This type of analysis also allows a specific particle size group (and thus particle strength group) to be followed throughout the degradation process. When applied to industrial processes, this approach should provide design information to minimize agglomerate formation during processing and for selecting successful dispersion and size reduction strategies.