1 IMPROVING AEROSOL DISPERS ION THROUGH A FUNDAMENTAL UNDERSTANDING OF INTERPARTICLE FORCES By STEPHEN THOMAS TEDESCHI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2007
2 2007 Stephen Thomas Tedeschi
3 To my Grandfather, Frank Mone, who first taught me about engine ering and instilled in me the desire learn about how and why a ll things work, that is, the sc ience of everything. The time I spent helping you on those projects, although too fe w, started me down the path Im on today. Thank you.
4 ACKNOWLEDGMENTS I first thank my advisor, Dr. Hassan El-Sha ll, for all of his wisdom and guidance throughout my studies here at the University of Florida. His ex tensive knowledge and experience were invaluab le sources of information and encouragement. I acknowledge my committee members for their helpful discussions and suggestions including Dr. Rolf Hummel, Dr. Brij Moudgil, Dr. Arun Rande, Dr. Rajiv Singh, and Dr. Spyros Svoronos. Special recognition goes to Professor Br ian Scarlett. He will always be remembered for his unwavering belief in his students and enco uragement for the pursuit of knowledge and the understanding of science of all th ings. These values he lived for live on in all of us who knew him. I thank my family, especially my parents Fr ank and Raffele and my brother Daniel, for always believing in me and encouraging me to follow my dreams. E alla mia Nonna, grazie mille per tutto! Their support was esse ntial to accomplishing this goal. I especially want to thank my fiance Dr. Anna Fuller for her never ending patience and encouragement throughout our many years here. So many dinner convers ations over research helped us through so many of those challenges. It wont have b een possible without you. Many thanks go to Dr. Nate Stevens, Dr. Sc ott Brown and Dr. Yakov Rabinovich, for their help in several details of my research. I also thank Dr. Kevin Powers for his advice and debates with me over almost every point of the project, as well as Gill Brubaker and Gary Scheiffele for their invaluable assistance and Kerry Siebei n for all of her assist ance on the SEM and TEM. I also gratefully acknowledge my current group members including Dr Maria Palazuelos Jorganes, Dr. Saijit Ply Daosukho, Rhye Hamey, Yang-Yao Simon Lee, Kerri-Ann Hue, Milorad Djomlija, Julio Castro, Amit Singh, De bamitra Dutta, and Kalyan Gohkale; past members including Dr. Ayman El-M idany, Dr. Olesya Zhupanska, Dr. Ecivit Bilgili, Dr. Mario
5 Hubert, Dr. Dimitri Eskin, Dr. Nishanth Gopinath an, Dr. Madhavan Esayanur, Dr. Caner Yurteri, Dr. Dauntel Specht, Dr. Marco Verwijs, and Dr Kerry Johanson; and undergraduate students including Calvin Faucett for thei r help in carrying out my resear ch, and also for their support and encouragement. Dear collogues and friends, th ank you for making my time here in Gainesville interesting and memorable. I thank the staff of the Particle Engineeri ng Research Center in cluding Jamie Anders, Ashley Boller, Noeleen Brophy, Anne Donnelly John Henderson, Donna Jackson, Dr. Victor Jackson, Sophie Leone, James Newman, Greg Nort on, and Jo-Anne Standridge for all of their support and assistance. Lastly, I acknowledge the Particle Engineering Research Ce nter, the United States Army Environmental, Chemical, and Biol ogical Center, and our industrial partners for their financial support.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........9 LIST OF FIGURES................................................................................................................ .......10 ABSTRACT....................................................................................................................... ............14 CHAPTERS 1 INTRODUCTION..................................................................................................................16 Impact of Dry Powder Dispersion..........................................................................................16 Current Research............................................................................................................... .....18 2 BACKGROUND....................................................................................................................21 State-of-The-Art Dispersion Techniques................................................................................21 Mechanical Dispersion Devices and Methods................................................................21 Reducing Attractive Forces.............................................................................................26 Particle Charging to In crease Repulsive Forces..............................................................28 Powder Processing Techniques to Reduce Capillary Effects..........................................29 Summary of Current Research................................................................................................32 3 INTERPARTICLE FORCES IN AEROSOLS......................................................................36 Introduction................................................................................................................... ..........36 van der Waals Forces........................................................................................................... ...37 Microscopic Approach....................................................................................................38 Macroscopic Approach....................................................................................................39 Surface Energy Approach to Adhesion...........................................................................40 Capillary Forces............................................................................................................... .......41 Roughness...................................................................................................................... .........42 Electrostatic Forces........................................................................................................... ......44 4 MATERIALS AND METHODS...........................................................................................51 Material....................................................................................................................... ............51 Experimental Factors and Processing Methods......................................................................52 Drying Method................................................................................................................53 Spacer Material................................................................................................................54 Relative Humidity...........................................................................................................55 Surface Modification.......................................................................................................56 Dispersion Medium.........................................................................................................56
7 Characterization Methods.......................................................................................................57 Particle Size and Morphology.........................................................................................57 Material surface energy...................................................................................................60 Dispersion Characterization............................................................................................60 Particle size comparison...........................................................................................61 Settling measurements..............................................................................................61 Powder Mechanics...........................................................................................................62 Interparticle Forces..........................................................................................................64 Design of Experiments.......................................................................................................... .65 5 RESULTS AND DISCUSSION.............................................................................................73 Characterization............................................................................................................... .......73 Particle Size of Single Materials.....................................................................................73 Particle Size of Mixtures.................................................................................................75 Morphology.....................................................................................................................75 Drying Methods................................................................................................................. .....76 Aerosol Experiments............................................................................................................ ..78 Spacer Addition...............................................................................................................78 Dispersion Medium.........................................................................................................83 Surface Modification.......................................................................................................85 Design of Experiments....................................................................................................88 Interparticle Forces on th e Macro and Micro Scale................................................................89 Bulk Powder Strength......................................................................................................90 Experimental methods..............................................................................................90 Shear test results.......................................................................................................91 Shear test discussion.................................................................................................92 Interparticle Forces..........................................................................................................93 Experimental methods..............................................................................................94 Interparticle force re sults and discussion.................................................................96 Data Correlation............................................................................................................... .......99 6 CONCLUSIONS..................................................................................................................136 7 CASE STUDY: ARMY OBSCURANTS............................................................................141 Introduction................................................................................................................... ........141 Experimental................................................................................................................... ......143 Factors and Preparation:................................................................................................143 Flake Thickness.............................................................................................................143 Spacer material.......................................................................................................144 Characterization.............................................................................................................145 Particle size and morphology.................................................................................145 Aerosol performance..............................................................................................145 Powder performance..............................................................................................146 Results and Discussion..................................................................................................148 Conclusions...................................................................................................................150
8 8 SUGGESTIONS FOR FURTURE WORK..........................................................................157 LIST OF REFERENCES.............................................................................................................160 BIOGRAPHICAL SKETCH.......................................................................................................171
9 LIST OF TABLES Table page 3-1 van der Waals interaction forces an d energies for different geometries............................50 3-2 Interaction energy using the Derjaguin Approximation....................................................50 5-1 Summary of optimum su rface modified aluminum.........................................................135
10 LIST OF FIGURES Figure page 1-1 Representation of the relationshi p developed through this research.................................20 2-1 The Thermodynamics of drying methods..........................................................................35 3-1 Illustration of important interpartic le forces in powders and aerosols..............................48 3-2 Depiction of the three interacti ons known as van der Waals forces..................................48 3-3 Depiction of the calculation of fo rce using the Derjarguin Approximation......................49 3-4 Change in contact ar ea with increasing roughness............................................................49 4-1 A roadmap to dispersion: depiction of the direction for the presented research...............67 4-2 Thermodynamic path of dr ying methods investigated.......................................................68 4-3 Images of the Air dispersion chamber...............................................................................69 4-4 The Sci-Tec ShearScan TS12 shear tester and test cell.....................................................70 4-5 A typical result from the consolida tion and failure of a powder sample...........................70 4-6 Schematic of the AFM operation.......................................................................................71 4-7 Example of a force distance curve for the AFM................................................................72 5-1 Particle size distributi on of aluminum flakes...................................................................101 5-2 Particle size distribut ion of Valimet aluminum in isopropyl alcohol..............................101 5-3 SEM images of Sigma VPD Aluminum flakes................................................................102 5-4 TEM images of Sigma 1215 aluminum flakes................................................................102 5-5 Contact mode topographical im ages of aluminum flakes................................................103 5-6 Detailed topographical image of flake surface................................................................103 5-7 SEM Image of Valimet aluminum spheres......................................................................104 5-8 Phase diagram for carbon dioxide as related to supe rcritical drying...............................104 5-9 Particle size di stribution measured follo wing each drying method................................105 5-10 Contact area with increasing roughness...........................................................................105
11 5-11 Calculated van der Waals for ces of a smooth and rough surface....................................106 5-12 Illustration of the change in contact area as a result of the addition of fumed silica.......106 5-13 Calculated van der Waals forces with the addition of silica............................................107 5-14 SEM images of aluminum flakes with silica...................................................................108 5-15 Air Chamber transmission measurement.........................................................................108 5-16 Average and standard deviation of settling rate...............................................................109 5-17 Airborne concentration at selected times after partial vacuum dispersion......................109 5-18 Airborne concentration by vac uum dissemination with silica.........................................110 5-19 Dissemination pressure a nd airborne concentration........................................................110 5-20 Airborne concentration by high pres sure gas dissemination with silica..........................111 5-21 Illustration of the dispersion effect of liquid carbon dioxide...........................................111 5-22 Airborne concentration produced by liquid carbon dioxide dissemination.....................112 5-23 Airborne concentration fo r gas vs. liquid carbon dioxide................................................112 5-24 Airborne concentration usi ng liquid carbon dioxide and silica.......................................113 5-25 Summary of airborne concentr ation by carbon dioxide dispersion.................................113 5-26 Particle size of aluminum flakes at 20, 50, and 80% rela tive humidity...........................114 5-27 Digital image of water drops on al uminum as used for measurement.............................114 5-28 Contact angle on ODTMS surface modified aluminum..................................................115 5-29 Change in contact angle of TDFS surface modified aluminum.......................................115 5-30 Digital image of water drop on modified flake bed.........................................................116 5-31 Airborne concentration of surface modi fied aluminum flakes at 50% humidity.............116 5-32 Airborne concentration of bare al uminum at 20, 50, and 80% humidity. .....................117 5-33 Experimental design dispersion results. .........................................................................117 5-34 Unconfined yield strength vs major conso lidation stress at 20% relative humidity........118 5-35 Unconfined yield strength vs major conso lidation stress at 50% relative humidity........119
12 5-36 Unconfined yield strength vs major conso lidation stress at 80% relative humidity........120 5-37 Unconfined yield strength vs majo r principle stress for bare aluminum.........................121 5-38 Unconfined yield strength vs major prin ciple stress for TDFS modified aluminum.......122 5-39 Unconfined yield strength vs major pr inciple stress for ODTMS modified aluminum..123 5-40 Response plot for unconfined yield strength...................................................................124 5-41 Damaged aluminum co ating on cantilever tip.................................................................124 5-42 Bare aluminum particle probe..........................................................................................125 5-43 ODTMS modified Alum inum particle probe...................................................................125 5-44 Contact mode image of th e CVD bare aluminum surface...............................................126 5-45 Contact mode image of bare aluminum in 3D.................................................................126 5-46 Detailed contact mode image of bare CVD aluminum....................................................127 5-47 Contact mode image of bare aluminum in 3D.................................................................127 5-48 Contact mode image of ODTMS modified aluminum....................................................128 5-49 Contact mode image of ODTMS aluminum in 3D..........................................................128 5-50 Typical force distance curve............................................................................................129 5-51 Interparticle for ce at humidity for bare aluminum. ........................................................130 5-52 Interparticle force at humidity for ODTMS modified aluminum....................................131 5-53 Comparison of the effect of surface modification...........................................................132 5-54 Interaction energy and surface modification....................................................................133 5-56 Overlay of response plots.................................................................................................134 5-57 Overlay of contour on air borne concentration results......................................................134 7-1 Laser Diffraction particle size di stribution for aluminum flakes.....................................152 7-2 Typical Images of aluminum flakes and silica................................................................152 7-3 Packing density of flakes and FoM. ...............................................................................153 7-4 Figure of Merit for aluminum flakes of varying thickness..............................................154
13 7-5 Figure of Merit for aluminum flakes of varying thickness and spacer concentration.....154 7-6 Average FoM for aluminum flakes of va rying thickness and silica concentrations........155 7-7 Design Expert Surface Response plot for FoM...............................................................156
14 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy IMPROVING AEROSOL DISPERS ION THROUGH A FUNDAMENTAL UNDERSTANDING OF INTERPARTICLE FORCES By Stephen Thomas Tedeschi December 2007 Chair: Hassan El-Shall Major: Materials Science and Engineering Dispersion is a vital aspect of powder te chnology impacting traditional and emerging technologies in many diverse areas including healt h, industry, the environment, and the military. In such applications, the propert ies and quality of th e resultant product will be directly and noticeably affected by the degree to which the pa rticles are dispersed. Reduced dispersion, or increasing agglomeration, will lead to decreased e fficiency and ultimately result in a low yield and increased cost. As powder technology moves into the nano-age new, more complex challenges arise. Decreasing particle size into the submicron range significantly increases the effect of surface forces leading to an increase in the cohesive forces of the powder, thus, making an understanding of dispersion even more essential. In order to develop the know ledge base to understand and control aerosol dispersion, a fundamental study of the significant powder char acteristics and environmental conditions that effect dispersion has been completed and through this investigation the resulting change in interparticle forces has been m easured. Supercritical drying wa s shown to successfully produce a dry, dispersable powder though the supercritical phase and the evasion of capillary forces. A fumed silica spacer material was used to impr ove dispersion though a reduction in contact area
15 and in turn van der Waals forces. Surface modification was shown to control the surface energy of the aluminum surfaces, leading to a multifacet ed contribution to dispersion. First, the increased hydrophobicity reduced th e amount of adsorbed water and was successfully used to control capillary forces. In the dry system, the silane was shown to also reduce van der Waals forces through the change in surface energy and th e steric barrier produced. This research was guided using statistical experime ntal design in order to effi ciently study significant factors affecting flake dispersion in air. An improvement in the dispersion of an aluminum flake system was shown. However, more importantly, a corr elation between the processing parameters and the change interparticle forces that led to th e increase in dispersion was determined. It was shown that an understanding of the interparticle forces and the processing methods that affect them can be used to decrease adhesion and de velop novel dissemination methods that result in improved dispersion. Ultimately, this knowledge allows the application of this research to other powder systems.
16 CHAPTER 1 INTRODUCTION Powder dispersion is vital to traditional a nd emerging technologies in many diverse areas including health care, industry, the environment, and the military. In each application, the degree to which the powder is dispersed will ha ve direct and noticeable effects on the properties and quality of the resultant product. Reduced dispersion, or increasing agglomeration, will lead to defects and decreased process efficiency result ing in low yield and increased cost. As powder technology moves into the nano-age, new and more complex challenges arise due to decreasing particle size into the submicron range. This sign ificantly increases the surface forces leading to an increase in cohesiveness of the powder. Ther efore, understanding these forces and their role in powder dispersion is essential. Impact of Dry Powder Dispersion Health care is a major area where the dispersi on of drug powders greatly effects treatment. Inhaled drug aerosols are used to treat a number of lung disorders including chronic obstructive pulmonary disease (COPD), a group of lung disorders that includes as thma, and is attributed with a cost of over 14 billion dollars in the United States alone. The drug powder must be under a critical size as onl y drug particles 5 m in size or smaller are capable of deep lung delivery. Agglomeration of the drug leads to larger sizes and deposition on the tongue or throat rendering that portion of the dose ineffective. Industrial powder processes, such as ceramic processing, powder metallurgy, chemical reactors, powder coating applic ations, and aerosol technologies require controlle d dispersion of powders. Unwanted agglomeration during die filling or pressing will reduce the quality of the product by creating density gradients and voids. Chemical reactors, coal, and other powder burning operations require a uniform particle size distribution in order to maintain a constant
17 heating rate. Agglomeration of the powder will resu lt in an inconsistent particle size and lead to variations in the products. Powder coating ope rations rely on dispersed powder to produce uniform coatings on the substrate. Powder disp ersion is also an important factor in aerosol technologies. Separation and cl assification systems must be able to separate starting materials into the true primary particle sizes in order to obtain optimum results. Also, the evaluation of filters requires that the powder passing through th e filtration system be fully dispersed into a known and uniform size to measure the true filter efficiency. These are only a few industrial applications, but the great extent of the impact is readily acknowledged. Increased dispersion will also have an impact on the environment. Dry powders are used in agriculture as fertilizers to enrich the soil and as pesticides to increase crop yield. Reports have shown that fertilization has many benefi cial effects on soil including increased water holding capacity, porosity, infiltra tion capacity, hydraulic conductiv ity, water stable aggregation, and decreased surface crusting. In the long te rm, this will lead to higher quality soil and increased crop yield. Dry chemicals used in firefi ghting are vital when water is not available or when water cannot be used, i.e. electrical fires. In both cases by increasing the degree of dispersion, the higher surface area available for re action will increase the e ffectiveness of these chemicals and in turn a smaller amount of the chemical will be required for the same performance. Farmers will use less fertilizer lowering costs and leading to more organic, healthier foods and firefighters will be able extin guish fires more efficiently with less chemicals reducing the concern for envir onmental cleanup following a fire. Effective aerosolization is also important for defense. The United States Military Services rely on obscurant clouds to hide personnel a nd equipment from enemy acquisition. A program has been initiated to develop more efficient obscurant s by engineering the particle
18 which make up the cloud. In orde r to realize the full potential of the engi neered particulates, dispersion of the powder in the aerosol into prim ary particles must be ach ieved. In addition, the powder settling rate greatly depends on particle size and reducing the size will increase the amount of time the powder is airborne. Therefore, any agglomeration of the material, before or after dissemination, will result in decreased obscuration efficiency. Great improvements in all of these areas can be realized by contro lling the degree of powder dispersion. Benefits include decreased co sts and, simultaneously, increased efficiency. Towards these ends, researchers have undertaken the i nvestigation of powder dispersion for a variety of these applications. However, disp ersion is a very complex problem as powder properties are unique to the materi al of which they are made and can change as their environment changes. The majority of the published research is empirical in nature and focuses on optimizing the dispersion of a specific powder, for a partic ular application, under a given set of conditions. Little work has been done to correlate the interparticle forces involved in dispersion with the powder properties, effect of environmental conditi ons, and general trends se en in particle size, and degree of dispersion, measurements. A syst ematic study of the parameters involved in the particulate system, including correlating the effect microscopic (interparticle forces and surface forces) and macroscopic (bulk forces) interactions together with the in fluence of the surrounding medium on these intera ctions, is required. Current Research The first step towards the realization of an increased state of dispersion is an understanding of the interparticle forces of a given system a nd the factors that affect them. It should be recognized that the dominant forces are dependent on the physical and chem ical properties of the material and thus will vary between particle systems. However, if a general battery of tests is
19 developed along with the understanding of the rela tionship to interparticle forces, this knowledge can be applied to any powder. Therein lies the focus of this research. The goal of this research is to study the funda mental interactions be tween particles in air and their magnitude. In addition, the effect of significant factors on the particle and bulk scale forces and their subsequent influence on the pa rticle size distribution after aerosolization has been studied both experimentally and theoretical ly. This document describes a systematic study of the physical, chemical, and e nvironmental factors th at has been completed for an aluminum flake material in order to develop a fundamental understanding of the eff ect of these aspects on airborne concentration and partic le size through the correl ation of factor effe ct on single particle and bulk scale force measurements. The degree of dispersion in air is dependent on two things, the source and magnitude of the significant interparticle forces, and the ener gy supplied to the agglomerated powder by the disperser. An illustration of the interrelations hip developed through this research is given in Figure 1-1. For this research, relative humidity, powder surface modification, the effect of flow aids such as second phase particles, and disp ersion medium were st udied and an understanding of the fundamental interactions relevant to disp ersion were determined. Statistical experimental designs were used to systematically investigat e these factors to efficiently determine those significant to the aerosolized particle size di stribution. Additionall y, interparticle force measurements using the atomic force microscope (AFM) were made to de termine the effect of the significant factors on particle or micro-scale, and shear were made to see their effect on the bulk, or macro-scale. When these results we re compared, insight into the mechanism of adhesion for a given powder, and thus the be st method of dispersion, was determined.
20 Figure 1-1. Representation of th e relationship developed through this research. Through an understanding of the effect of signifi cant powder characteristics, processing parameters on aerosolized particle size and their correlation to bulk and interparticle forces the fundamental knowledge base requi red to improve aerosol dispersion can be developed. Bulk and Interparticle Forces Aerosol Dispersion Powder Characteristics Improved Dispersion Bulk and Interparticle Forces Aerosol Dispersion Powder Characteristics Improved Dispersion
21 CHAPTER 2 BACKGROUND A great deal of work has been put forth to improve dispersion in several diverse fields. The research covers a large number of material s in many different areas and technologies. However, since powder propertie s are commonly unique to a speci fic material, one dispersion method may not be effective for more than one powder[5-14]. In order for powder to be dispersed, the external force applied to the clus ter of powder as a bulk or agglomerate must be larger than the attractive forces holding the pow der together. The powder particles themselves can also be modified to reduce the attractive fo rces between them and th us reduce the amount of energy needed to disperse them. Many examples are seen for the dispersion of pharmaceutical drug powders, which is a major area of research due to their direct eff ect on the treatment of health problems and accounts for a majority of th e published research. Efforts have focused on four main areas: developing mechanical dispersi on devices, reducing attractive forces between particles, increasing repulsive forces, and impr oving processing and drying methods. Each will be discussed in more detail. State-of-The-Art Dispersion Techniques Mechanical Dispersion Devices and Methods In general, there are a number of differen t types of dispersion devices that can be characterized by the method in which the aerosol is disseminated. Many variations of dispersion devises can also be found in literature[15-22]. For some devices, the powd er is dispersed in a dilute suspension with a volatile liquid. It is then nebulized into small droplets containing particles. The liquid then volat ilizes leaving the powder suspende d in an aerosol. If, however, more than one particle is present in each dr op, capillary forces caused by a decrease in liquid volume of the meniscus between them can cause the particles to become agglomerated. Dry
22 powders are also dispersed in fl uidized beds where larger particles can be added to aid in dispersion. Smaller particle are entrained in the air stream flow ing through the bed and carried out of the device. Dry powders can also be di sseminated by high speed air jets that entrain and mix the powders. These dispersion systems rely on airflow to impart the energy required to disperse the powder through rapid acceleration a nd deceleration along with rapid compression or expansion of the carrier gas. Venturi type nozzles, ejectors, and by annular nozzles, which can operate at subsonic, sonic, or supe rsonic velocities, are examples of this type of de vice. Internal propellers have also been added to aid the mi xing of different powders and in dispersion. A number of commercial and expe rimental devices for dry powder dispersion, such as classifier nozzles and dust generators, have been reviewed in literature by Ranade and Calabrese as have a number of dry powder inhalers (DPIs) for the administration of pharmaceutical powders by Hickey and coworkers. Improved performance requires a better unde rstanding of how these devices work. A dispersion device has two functions: (1) entrain th e powder in air from the bulk reservoir and (2) carry the particles out of the device and into the environment to produ ce an aerosol cloud. During one or both of these func tions, additional steps are require d to disperse the powder and maintain that state of dispersion. Often mechan ical dispersion, or dea gglomeration, is achieved by shear and impaction. Shear can be produced by acceleration in ai r and is a common dispersion method employed by many nozzle-type dispersion devi ces including venturies, ejecto rs or sub-sonic, sonic, and super-sonic nozzles discussed previously. Th e rapid acceleration or deceleration of the air around the powder and by rapid volume change through expansion or compression of the air causes shearing and normal forces that act to sepa rate particles. The particles suspended in a
23 fluid third phase, liquid or gas, will experience forces from the movement of medium in the form of these pressure and velocity gradients. As the agglomerate enters the flow field, it will experience a shear force produced by the sudden acceler ation, which, if larger than the attractive forces between the particles, will pull them apar t. The magnitude of the airflow velocity and pressure change through the nozzle can be cont rolled by design and will affect the amount of energy supplied to the agglomerate. Unstable airflow leads to large velocity gradients in the form of small turbulent eddies. The agglomerate, possessing a sufficiently large size and inertia in comparison to that of the eddy, will not be able to follow the abnormal path of the airflow. The resulting stresses experienced by the agglomerate as it travels past th ese eddies act to disperse it. Agglomerates smaller in scale than the eddies wi ll, however, be able to follow th e fluid flow of the turbulent air more closely and not experience the large stresse s needed for dispersion. Thus, conditions for a lower limit to the effectiveness of this met hod are recognized. In their study, Ranade and Calabrese found that an annular geometry dispersion device configured such that the particles enter the air stream at the point of maximum energy, i.e. maximum turbulence created by an acceleration gradients, led to the best dispersion. Turbulence created by devise configuration was experimentally studied for a number of dry powde r inhalers (DPIs) by Chew[24-26] for pharmaceutical powders. It was shown that the airflow was highly dependent on the device configuration but varied with the drug powder used such that no one configuration worked well of all powders. Turbulence has also been experimentally studie d with latex particles in conjunction with a numerical simulation developed by Endo. The simulation was co mpared to the experimental results and used to calculate the adhesion force of spherical particles. A value of approximately
24 1 MPa was calculated for two 1um latex particles. In this work, the effect on the particle size distribution was not included. Ch anges in particle size distribu tion will change the powder packing and thus the number of particle contacts in the bulk powder. This will influence the powder strength and the force needed to disperse it. Particle size distribu tion has been shown to change the number of contacts between partic les and thus the total adhesion of the bulk powder[28-31]. It was acknowledg ed that increasing the total number of contacts between particles in a bulk powder increased the points of adhesion and in turn led to a stronger powder. Straying from a spherical particle shape introduces errors that have not yet been included in the model. One reason for this is that the particle shape also affects the number of contacts as well as the total area in contact. High aspect ratio pa rticles such as flakes may have a smaller number of contacts but have a larger total contact area wh ich will also lead to an increase in powder strength. This has been experime ntally shown to affect the degree of dispersion in a number of studies[9, 32, 33]. Turbulent flow through any system will also lead to impaction. Turbulence will cause particles to collide with each other and the wall as they move though the turbulent air[34, 35]. The force of the collisions will be dependent on the particle size and inertia, the gas velocity, and the velocity gradients determined by the eddy size Y. He studied the effect of collision on dispersion and showed that collisions of sufficien t force were enough to separate agglomerates of some powders. The force of the airflow comb ined with impaction has to be larger than the interparticle forces of the agglomerate. Galk and coworkers study this technique in a classifier. It was shown th at by adding an impeller in the di spersion step of a industrial classification system, improvement s in dispersion were achieved.
25 In many devices, an impeller, a rotating disk or internal propeller, was added to increase turbulence. Voss and Finley investigated the effect of tu rbulence by comparing a dispersion rig they developed against a common pharmaceuti cal inhaler that uses this technology, the Diskhaler. They concluded that turbulence wa s not the only method of deagglomeration but that impaction was also involve d. Agglomerates that were to o large to follow the turbulent eddies were impacted by the impeller which ai ded in the dispersion. A mixer-type dispersion device, which uses an internal propeller to di sperse the powder, was studied by Masuda and Gotoh. They found that pa rticle size was inversely propo rtional to the impeller rate regardless of the powder used. The lowest opera tional speed was also found to be dependent on device geometry. This shows that dispersion is known to be dependent on the powder properties but the device design also signi ficantly contributes to disper sion and operational parameters. Particles in a stable air str eam can also be impinged upon a flat plate, or other such obstacle, set perpendicular to thei r trajectory. The force of the agglomerates collision with the obstacle will impart on it energy needed for fract ure. After the collision the particles and agglomerates that remain will be carried past the obstacle by the air stream into a turbulent region. An experimental study by Wang using a common DPI showed that by this method an increase in dispersion of 2-3 times was achieve d as compared to acceleration in air alone. Gomes et al have shown experimentally that the inertia of the particle is related to its residence time in the wake region of the obstacl e where turbulence is hi gh. Longer residence time in the wake region leads to more particle-par ticle collisions and a smaller resultant particle size. Impaction has been experimentally show n to improve dispersion in the SPIROS and rotohaler, devices.
26 Shear, turbulence, and impaction are dispersi on mechanisms that are relatively easy to generate, and can be created simp ly by controlling device design. Howe ver, it is seen that alone these mechanisms are not entirely sufficient and further research is re quired, especially for particle under 1 m in size. It is therefore necessary to also study the powder itself and add processing steps during powder prod uction to reduce cohesion belo w the force of the dispersion device. A better understanding of both interparticl e forces and energy ava ilable applied by these dispersion mechanisms will l ead to increased dispersion. Reducing Attractive Forces If the attractive forces betw een particles are reduced, accordin gly is the energy needed to separate the particles. The majo r contribution to attract ion is attributed to van der Waals forces or capillary forces, in the presence of a liquid. The magnitude will be due to the number of contacts within the powder and th e contact area between particles. These forces can be reduced by changing the Hamaker constant or surface ener gy of the interacting surfaces. This reduction has been accomplished in a number of ways. A coating of a low surface energy material such as Teflon, or chemical surface modification will reduce the attractive forces at the particle surface. Chemical surface modifications and physical coatings on metal pa rticles were studied experimentally by Ranade and coworkers[11, 12, 40]. Five methods were inve stigated in an attempt to reduce the cohesion between metal powders including the addition of a spacer, chemical surface alteration, monolayer coatings of surfactants, particle encapsulation by a silane, and altering particle geometry. They found that the growth of a th in oxide layer on the metal powder significantly increased the dispersion. In or der to further understand the effect on dispersion, they studied the effect of surface modification on the tensile streng th of the powders. It was determined that
27 liquid-phase deposition of silica and a conti nuous oxide layer coating by steam treatment improved dispersion. The use of adsorbed films has been studied theoretically by Vold and is shown to reduce particle cohesion thereby increasing dispersion. Particles of a different material can also be added in an attempt to alter the interparticle forces. Second phase particles have been inve stigated for use in pharmaceutical powders for DPIs. Zeng observed an increase in dispersion wh en fine lactose particles were added to a drug powder[13, 30, 31]. These particles had a lower in terparticle attraction and thus reduced the attraction between drug pa rticles. Louey and coworkers showed an increase in the fine powder fraction (FPF), the fraction of powder below 5 m, by mixing two non-active spacer particles with the drug. They theorized that the ca rrier particles have sites of varying degrees of adhesion. The last powder to be added will attach to the site of lowest adhesion and will detach easier increasing the final FPR of drug particles released. The size distribution and polydisper sity of dry powders has also been identified to affect dispersion by increasing the packing and numbe r of contacts. Safatov and coworkers determined that the powder size distribution affected the amount of energy required for dispersion. In their work, they found that a la rge size distribution lead to agglomeration and attempts to disperse the powder proved more difficult than for a more monodispersed powder. Chew and Chan  concluded that dispersion of a smaller size distribution powder generated more fine particles when using a typical DPI. It was determined that large particles and agglomerates were lost to impaction inside th e dispersion device and were not aerosolized. An increase in the width of the size distribution will also affect th e particle packing, and in turn affect the number of cont act points for cohesion.
28 Particle Charging to Increase Repulsive Forces Dispersion can also be affected by charging the particles. Im parting a similar charge to powder particles will increase the repulsive forc es by the magnitude of the charge and will counter act the attraction due to van der Waals forces. Partic les can obtain a charge from a number of different methods[15, 44-46]. Commonly, charge can be transferred to particles upon contact with another surface, as in tribocharging, irradiation of the powder, and collisions with electrons, ions, or ion clusters in the surrounding environment. Ren and coworkers developed a disperser for the electrostatic stabil ization of fine, dry calcium carbonate and talc materials. The para meters used were in agreement with their theoretical calculations based on first principle equations. It was show n through experimentation that electrostatic stabilization of CaCO2 was achieved for a maximum of 48 hours by corona charging. The process was highl y dependent on humidity and ad sorbed moisture due to the nullification of charge by water. In th e study, it was concluded that the method was ineffective for particles less than 2 m in size. An expression for the summation of interparticle forces was developed and the force of elect rostatic repulsion need ed for dispersion was determined. Simple assumptions were made and the effect of roughness was not included, a factor that is known to have a significant effect on interp article forces[7, 8, 32, 49-55]. In subsequent research, the particle surfaces were first coated with a polymer then subjected to corona charging. The polymer a ddition was expected to decrease the attractive forces between particles as compared to the uncoated CaCO3 particles and hold a better charge, increasing the ability of this technique to disper se the powder yet the resu lts showed only slight improvements. Due to these limiting factors, including humidity[15, 45, 47], the charge density
29 on the particles, particle conc entration, and those presente d in the work by Ren, it can be seen that charging is effective but only under limited conditions. Powder Processing Techniques to Reduce Capillary Effects Many powder processing methods require the powd er to be suspended in a liquid and thus an additional drying step is needed. Conventiona l drying methods, such as simple evaporation or even heating under vacuum, result in a hard, dry cake. Often the agglomerates that form cannot be re-dispersed. If the powder is manufactur ed already containing ag glomerates, the production of a dispersed aerosol becomes that much hard er. Therefore, drying is a crucial step in developing a material that can be disseminated as a dispersed dry powder aerosol and will now be described in detail. As water is removed from a powder, a liqui d bridge, or meniscus, forms between the particles. These liquid bridges will be detrim ental to the powder dispersion. The liquid can dissolve some of the particle surface and redeposit the material in the neck region. If the material is soluble in water, excess material can be deposited at the particle contacts as the powder is dried and form a solid bridge holding th e particles together. If no dissolution occurs the liquid bridges still lead to the formation of capillary forces between the particles which will bring them into intimate contact after which va n der Waals forces hold the particles together. The phase change across the gas liquid phase bo undary followed by evaporation can be seen by the red line in Figure 2-1, a ge neral gas-liquid-solid phase diag ram depicting the thermodynamic phase change path of each drying method. The pr esence of water in the air is the source of another capillary attracti on since only a very small amount is n eeded to form liquid bridges in the neck regions between particles. It can then be understood that drying can also be an important process for powders which are not completely immersed in liquid. A number of processing methods have been developed to overcom e these difficulties including freeze-drying,
30 supercritical fluid processing, and a boiling met hod. The phase changes that occur during these drying methods can be seen in Figure 2-1 and will be explained in more detail for each process. Freeze-drying, as the name implies, is a method during which the powder is initially exposed to temperatures low enough to freeze the liqui d in the system after which the pressure is lowered by placing the sample under a high vacuum to sublime away the liquid. Using this method, capillary forces due to water bridges are avoided. The path thro ugh the phase diagram for freeze drying is depicted by the blue line in Figure 2-1. However, during freezing the liquid will crystallize and stresses develop that, if sufficiently high, may be damaging to the powders. Supercritical (SC) Fluid processing is becomi ng a popular focus of research because of the many unique possibilities. During Supercritical Fluid Processing, or crit ical point drying, the liquid is heated beyond the critical point of the liquid directly to a supe rcritical fluid, avoiding the formation of a liquid / gas interface completel y. The critical point is a critical temperature and pressure at which the dens ities of the gas and liquid ar e equal and the phase boundary between the two phases ceases to ex ist. Above the critic al point the system is in a gaseous state and cannot be liquefied by applying pressure. SC processing is carri ed out in a pressure vessel in order to maintain a constant proc ess volume. The fluid is added as a liquid and heat is applied causing the pressure to increase. The system is brought above the critical point of the fluid, that is, a temperature above which the increased pressure cannot maintain the liquid phase. This is a unique phase having qualities of both gas and liquid. As the pressure vessel is vented, the SC fluid escapes as a gas without forming liquid brid ges. The process as it relates to the phase diagram is shown by the green line in Figure 2-1.
31 In order to carry out this process, the powder must be immersed in the medium to be taken to the supercritical point. This substance is chosen by its chemical properties, toxicity, abundance, etc., and the ease of goi ng supercritical. The most commonly used medium for this type of processing is CO2 mostly due to its abundance, envi ronmental inertne ss, relatively low Critical point, and low toxi city. This may require an intermedia te step to insure that the medium the powder is first dispersed in will be miscible with the substance. Powders dispersed in water that are not miscible in CO2, for example, are commonly solven t exchanged with alcohol, which is miscible in CO2. The solubility and dispersion of powder in SC-CO2 mediums have been improved by the use of surfactants in a similar method to comm on aqueous detergents. Through this research, DeSimone and coworkers have developed surfactants that will adsorb onto the particle surface allowing particles that are normally insol uble to readily disperse in the SC-CO2 medium[59-62]. Shekunov and coworkers supercritically processed a common pharmaceutical powder used in DPIs in order to evaluate the effect on di spersion. During supe rcritical processing, the powder formed a very loosely packed, soft agglom erate structure. There are several advantages to this loosely agglomerated structure. It yiel ds a larger aerodynamic shape factor and low bulk density, which increases the volume diameter available to the airf low, providing rapid acceleration. In addition, the resulting agglomerates do not have a high inherent strength since no capillary forces were present to pull part icles together limiti ng strong van der Waals attraction. The soft agglomerates that are formed can easily be broken up by the turbulent flow through an inhaler. It was shown that this processing method result ed in a factor of two increase in the drug fine particle fraction (FPF) after aerosolization as compared to other processing methods. However, this increase in the fine powder fraction was direc tly after production and it
32 did not appear to prevent agglomeration during st orage due to the re-adsorption of water from the air. Sievers and coworkers developed a met hod in which high concentration aqueous suspensions of therapeutic proteins we re aerosolized usi ng a supercritical CO2-assisted method[64, 65]. The suspension was mixed with in a low dead volume tee creating an emulsion. The CO2 was brought to its supercriti cal state and then passed th rough a capillary restrictor resulting in the aerosolization of the aqueous solution into sm all droplets. Many powders and solvents have been successfully investigated using this method. It was found that, the powder must be mixed with a solvent which then evaporat es after aerosolization. The final particle size is thus controlled by the size of the droplet formed after aerosolization since capillary forces will hold the particles within the droplet together. Recently, a new method of aerosol formation ha s been invented in which powder dispersed in water was passed through a heat ed capillary tube. Upon heati ng the water boils and generates an aerosol. This process is depicted by the orange ling in Figure 2-1. It was reported that high concentration slurries were aerosolized with high fine vol ume fractions by this method. This process is limited by the need for a large am ount of equipment limiting its ability to be used in a portable device. Summary of Current Research To date, the evaluation and op timization of the mechanisms and efficiency of dispersion methods in systems such as pharmaceutical powders for inhalers have typica lly been empirical in nature. That is, the effect of a single variab le on the measured fine powder fraction (FPF), the weight percent of powder with a size less than 5 m, in the resultant aerosol is measured. The combination that yields the highest FPF is then chosen. A very limited scientific understanding
33 pertaining to the reasons for a dhesion or magnitude of the force in the powder system is available. Through these investigations researcher s have determined that the particle dispersion of the aerosolized powder was affected by devi ce configuration, chamber size, impeller speed, primary particle size and mo rphology, powder formation, crysta llinity, and particle size distribution. Estimations of the energy input n eeded for high fine partic le fractions of powder have been developed using simple assumptions fo r inhaler type dispersers Complete dispersion into primary particles, however, was not ach ieved by any method. These studies have shown that the generation of an aerosol from a bulk f eed powder results in a maximum fine particle fraction of only 50 wt %. Examination of all of the work completed yields, at best, a vague relationship between all of the relevant variables for only a limited number of materials. None of the work found in literature to date considers the powder system as a whole. Included in the current research, is the investigat ion of dispersion factors such as: the effect of relative humidity, surface modifica tion, the effect of flow aids, and dispersion medium on the particle size and settling of the pow der. In addition, interparticle forces were measured using the Atomic Force Microscope (AFM), and bulk powde r properties were determined using a shear cell to understand the effect of th ese properties on the micro and ma cro scale. Statistical Design of Experiments was used to insure efficient and statistical analysis of the data. It should be recognized that the dominant forces are dependent on the physical and chem ical properties of the material and thus will vary be tween particle systems. Once these forces are understood, the dispersion method can be optimized to overcome th e forces of attraction between particles. In Chapter 3, a more detailed expression of the significant interparticle forces found in bulk powders and aerosols as well as their source s is developed. Chapter 4 will then outline the materials and experimental procedure used in th is investigation. Experimental results will be
34 presented in Chapter 5 and the discussion will be given in Chapter 6. The conclusions drawn from this research are explained in Chapter 7. Th e application of these co nclusions is shown in a case study in Chapter 8.
35 Figure 2-1. The Thermodynamics of drying met hods. A depiction of the thermodynamic path across phase boundaries followed by each of th e three drying processes investigated in this research. Evaporation crosses the gas-liquid phase boundary leading to capillary forces. During freeze drying, low temperature and low pressure are used and the liquid is first frozen then remove d by sublimation. If the pressure is not sufficiently low, condensation is possible a nd will lead to capi llary forces. The boiling method used high temperature and takes the liquid thro ugh the boiling point. Expansion of the liquid into a gas is used to overcome capillary forces. Supercritical drying takes a path to high temperature a nd high pressure comp letely avoiding the formation of capillary forces through the entire process. Supercritical Drying Evaporation Freeze Drying Capillary Forces Boiling Supercritical Drying Evaporation Freeze Drying Capillary Forces Boiling
36 CHAPTER 3 INTERPARTICLE FORCES IN AEROSOLS Introduction An aerosol is a dispersion of particles in air. Bulk powder is commonly disseminated to form the aerosol. The state of dispersion of the aerosol is governed by the conditions under which it is disseminated, the environment into which it is distributed, and the characteristics of the bulk powder from which it is taken. Particle properties are de termined by the inherent nature of the powder, the interaction be tween particles on both the macroscale and micro-scale, and the medium surrounding them. In order to improve aer osol dispersion, it is important to recognize all of the significant interact ions that determine these prope rties and understand the correlation between them on the microand macro-scale. The particle interactions consist of the su m of both the attractive (or cohesive) forces holding the particles together a nd the repulsive forces acting to separate them. The forces common to dry powder systems in clude attraction by van der Waal s forces, cohesion by capillary forces in the presence of liquid bridges, and repu lsion by electrostatic forc es. In addition, for the bulk state, the powder will also experience fric tion when in contact with other particles and shear forces as they move past each other. An illustration depicting these forces can be seen in Figure 3-1. Here Fad is the force of adhesion due to van der Waals and capillary bridge forces, Fes is the electrostatic force, FF and FS are the friction and shear forces respectively acting at the particle level, and F and S are the overall normal and shear forces acting on the ensemble of particles. Each of these forces will now be ex amined in more detail in order to develop an understanding of their interactio ns in dry powder systems.
37 van der Waals Forces The most common interparticle forces for powde rs in air arise from the electromagnetic interaction, coulombic attracti on, or dipoles created by the se paration of the positive and negative portions of atoms or mo lecules[68, 69]. These dipoles from a number of mechanisms including: the creation of fluc tuating induced dipoles by vibrational motion, permanent dipoles, or polar molecules. There are three main cont ributions collectively called van der Waals forces which include: Keesom interactions, or dipole-di pole interactions; Debye interactions, or dipole induced dipole interactions; a nd London dispersion interactions, or instantaneous, fluctuating dipole induced dipole interactions An illustration of these forces can be seen in Figure 3-2. London dispersion forces are considered to be the predominant part of van der Waals force between colloidal particles[70, 71]. The theory behind these forces was derived fr om the ideal gas law, which predicted the behavior of monatomic and simple diatomic gas. To explain the deviation of more complex molecules for the ideal gas law, Johannes D. van der Waals suggested a modification to the ideal gas law to include intermolecular interactions and it became known as the van der Waals equation of state, Equation 3-1[71, 72]. RT V V P 2 (3-1) Here P is pressure, V is volume, R is the gas constant, T is temperature, and and are constants specific to a partic ular gas. The constant describes the finite volume of the molecules comprising the gas, and the constant takes into account the attractive forces between the molecules. After some time, a paired poten tial form of the equation for interaction included, for the first time, both an attractive and repulsive term. This would later become known as the commonly used LennardJones pot ential, Equation 3-2 .
38 12 6r B r C Wa a (3-2) Here the net potential en ergy between two atoms, Wa/a, separated by a distance, r, is described by the competition of an attractive London dispersion force , C, related to the synchronization of instantaneous dipoles created when the energy fields of neighboring atoms overlap , and a Born repulsion term, B, arising fr om the overlap of electron clouds. Microscopic Approach In 1937 Hamaker expanded the concept of the van der Waals forces from single atoms and molecules to all atoms in solid bodies[70, 76]. He assumed that ea ch atom in one body interacts with all atoms in another body and added these forces by a method known as pair wise summation. Bradley , Derjarguin , and  all followed Hamaker to improve the understanding of the forces acti ng between bodies. According to their approach, the interaction energy, W, of two spheres of equal size can be ca lculated by Equation 3-3. If the radius of the particles is much greater than the separation distance, the results can be simplified to the more common equation for the van der Waals energy of interac tion between two spheres shown in Equation 3-4. 222 / 22224 ln1 64 22sphsphARRR W HRH RHRH (3-3) H AD Wsph sph12/ (3-4) Here A is the Hamaker constant, R is the particle radius, H is the separation distance between the particle surfaces, and D is the particle diameter. One important aspect of this work is that the van der Waals forces can now be considered by th e geometry of the body and an interaction term that is dependent only on the material propertie s, not shape. A compilation of the interaction forces and energies for various geometries can be seen in Table 3-1.
39 Macroscopic Approach This microscopic approach does have one dr awback; it does not take into account the effect of the interaction of dipoles within one body on other dipoles within the same body. In an attempt to include these interactions, Lifshitz  modified the Hamaker theory in 1956. Lifshitz ignored the atoms completely, assuming each material was a single body. He developed a calculation for van der Waals forces using the di fference in dielectric pr operties. Equation 3-5 is the final expression Lifshitz derived for the Hamaker constant. A132 3 2 kBT1in3in1in3in n 0 2in 3in 2in3in (3-5) Here the subscripts 1, 2, and 3 refer to material 1 and 2 interacting in medium 3; kBT is Boltzmanns constant times temperature; and (v) is the dielectric permittivity of the material at a frequency, v. There is some complexity involved in employing this equation as it is quite difficult to measure the dielectric response of materials over the wide range of frequencies required for the calculation. Work has been done by Ninham and Parsegian  and Hough and White  to simplify the measurement to specifi c characteristic frequencies in the ultraviolet (UV) range. The result is Equation 36, the Tabor-Wintert on approximation . A132 3 4 kBT1313 1313 3 he82 n1 2 n3 2 n2 2 n3 2 n1 2 n3 212 n2 2 n3 212 n1 2 n3 212 n2 2 n3 212 (3-6) Again, the subscripts 1, 2, and 3 refer to material 1 and 2 interacting in medium 3; kBT is Boltzmanns constant times temperature; and (v) is the dielectric permittivity of the material at a frequency, v; ni is the refractive index of the respective ma terial; and h is Planks constant. It is interesting to note that where the medium and the solids have the same index of refraction (n1=n2
40 and 1= 2) the Hamaker constant A132 goes to zero. This method is known as index matching and can be used to minimize or eliminate the effect of the Hamaker constant. Up to this point, the relationships for inter action have used intera ction energies, not the interaction force. For molecules, this is reasona ble but as larger bodies ar e studied, force is of more interest. To overcome the challenges pr esented by the geometry of different surfaces, Derjaguin developed an approxima tion to relate the force of an interaction between a body of some geometry to the energy of in teraction per unit area of two para llel flat surfaces . To do so, the force is integrated over infinitely small, flat concentric rings of radius x and thickness dx on each body as the separation distance between the bodies, H, goes from H to infinity as depicted in Figure 3-3. The relationship betwee n the interaction force for bodies of different geometries and the interaction of planar surf aces can then be extracted making it possible to compare the interaction force of bodies of diffe rent sizes and geometries. For this reason, interaction force measurements ar e often reported as force normali zed by radius. A list of factors that can be used to calculate the force of interaction from the energy of interaction of two plates of different geometries ca n be seen in Table 3-2. Surface Energy Approach to Adhesion In the previous section, the interaction betw een surfaces is determin ed as the surfaces are at some distance apart. However, when the surf aces are in contact, as is the case for powder in the bulk state, a different approach can also be considered using the su rface energies of the surfaces described by the work of adhesion. Work of Adhesion is defined as the energy needed to separate two surfaces from contact to an infinite separation distance and takes into consideration the surface ener gy of the bodies in contact as seen in Equation 3-7 . Wa = 12 + 23 (3-7)
41 Here Wa is the work of adhesion, 12 is the surface energy of one new surface 1 in medium 2 and 23 is the surface energy of new surface 3 in medium 2. By combining Equation 3-7 with Equation 3-8, Youngs equation, Equation 3-9 is obtained for the work of adhesion based on contact angle. 12cos = 2313 (3-8) Here 12 is the surface energy at the liquid and vapor interface, 23 is the surface energy at the solid and vapor interface, 13 is the surface energy at the so lid and liquid interface, and cos is the angle between 12 and 13 through the liquid phase. Wa = 12 (1+cos ) (3-9) From this equation, the work of adhesion can be calculated by measuring the contact angle of a liquid, with a known surface energy, on the materi al of interest. By modifying the surface energies of the bodies through coatings and chem ical surface modification, the work needed to separate them can also be modified. Doing so in a controlled manner can lead to the lowering of the cohesion between par ticles and thus the energy needed for dispersion. Capillary Forces Another important source of cohesion arises from capillary forces that form in the liquid pendulums between particles in the presence of liquid and is known to have a significant effect on the mechanical properties, transport, and di spersion of powder. The first attempt to understand this cohesion in soil was made by Hain es and Fisher[3, 84] and an expression for the force of cohesion due to the formation of aque ous liquid bridges was developed. Condensation of water vapor, present as relative humidity in the air, is sufficient to create liquid bridges[85, 86]. Since, most powder processing and transp ort occurs in ambient conditions, and in the presence of humidity, an understand ing of the effect of capillary fo rces on dispersion is vital.
42 Liquid present in a particle system will collect at contact points betw een particles and form liquid bridges. Capillary forces in the liqui d bridge will increase the cohesion between the particles at these points. For this reason, the de gree of powder packing is an important factor of the system. Close packed structures will ha ve more particle contac ts leading to a higher number of points for adhesion resulting in an incr ease in the powder strength. On the other hand, more open structures will have fewer contacts for adhesion lead ing to a weaker powder with a higher flowability. It can be readily seen that this phenomenon is very important for both the dispersion of dry powder from a reservoir, as in inhalers, and in the pr eservation of the powder dispersion after aerosolization. The equation for the attractive forces due to the presence of liquid bridges between two sphere s is given in Equation 3-10. cos 2 R Fcap (3-10) Here R is the radius of the capillary, is the surface tension of the liquid phase, is the contact angle of the liquid at the part icle surface, and the sign of th e equation represents the force direction, which is cohesive in nature. The a dhesion force due to the presence of a liquid bridge is calculated by the Laplace equation, Equation 3-11. ) 1 1 (2 1r r P (3-11) Here, P is the pressure difference across the liquid interface, is the surface tension of the liquid and r1 and r2 are the radii of the particles. Decr easing particle size will therefore have a large effect on the adhesion force when liquid bridges are present. Roughness Up to this point, only ideal, smooth surfaces have been considered. However, this is unlikely for real surfaces. At the atomic level there will be roughness due to the size of atoms
43 and their spacing within the crystal structure. These perturbations on the particle surface will influence the effective interparticle spacing, th e consequences of which are multifold since van der Waals forces are sensitive to the area of the surfaces in contact and separation distance between particles. The classical model for th e effect of roughness on van der Waals adhesion was created by Rumpf [88, 89] and shown in Equation 3-12. 2 21 6o o adH r R R r rR H A F (3-12) Here A is the Hamaker constant, H is the interpar ticle separation distance, R is the radius of the particle, and r is the radius of the asperity on the surface. This model, however, significantly underestimated the adhesion force. The model ha s since been modified by others and research by Rabinovich and coworkers[53, 54, 90] has further refined the model to better predict experimental results, and in its final form is s een as Equation 3-13. Their work indicates that surface roughness on the nanometer scale (<2nm) great ly increases the force of adhesion, and as the scale of the roughness increases there is a subsequent decrease in adhesion forces. 2 1 2 1 21 1 32 1 1 6o o adH rms k rms Rk H AR F (3-13) The additional terms in this equation include k1, the maximum peak to peak height between asperities; rms, the root mean s quared roughness of the surface; and the average peak to peak distance between asperities. The first term accounts for the pa rticle contact with the asperity
44 surface and the second term represents the inte raction between the par ticle and the average surface plane. Surface roughness will also decrease the minimu m amount of water needed to form liquid bridges at particle contact point s. Rabinovich and coworkers have investigated the effect of nanoscale surface roughness on the critical humidity re quired for the onset of capillary forces in such systems[49, 52, 55, 90, 91]. The presence of roughness on surfaces on a nanoscale has been shown to have a profound effect on the critical hu midity required to indu ce capillary adhesion. Their study also validates a simp le analytical expression devel oped to predict the force of adhesion as a function of partic le size, humidity, and roughness, which is most relevant for practical systems. A visualization of the effect of surface roughness can be seen in Figure 3-4. Esayanur  furthered this research by in corporating the friction coefficient, surface roughness, and the force of adhesion due to van der Waals forces into a single adhesion model. As a result, Equation 3-14 was developed: 13 22 31 2 224cos()() (2cos) cos12 ()cos(3)rlcsolrl H c hliqknpCnknnp A f RHK kR (3-14) Here, the additional variables are: AH, the Hamaker constant, H, the distance of closest approach, K, the roughness factor, and the friction coefficient. The first term on the right side of Equation 3-14 represents the contribution of the liquid bridge (capillary) forces and the second term represents the adhesion force in the dry state. Electrostatic Forces Electrostatic force is also considered impor tant for aerosol dispersion and has been the focus of research[9, 43, 46-48, 56, 58, 93-102]. St atic electricity arises when charges on the surface of a material are not free to move to neutralize each othe r or do so slowly. These
45 charges remain on the surface wher e they can interact with thei r surroundings. Particles with a similar net surface charge, greater in magnitude than the attractive forces, will experience coulombic repulsion, a force named after Charle s Coulomb who was the first to measure these forces between glass spheres using a torsion ba lance he invented. Equation 3-15 shows the general equation for the elec trostatic repulsive force. 2 2 14 r e q q Fo e (3-15) Here q1 and q2 are the charge on each particle, eo is the unit charge of an electron, and r is the radius of the particle. Particles suspended in an aer osol continually gain, lose, and transfer charge between themselves and the surrounding environment. Ther e are four basic ways by which charge can be imparted to the particle: (1) direct ionization, (2) ionization of part icles by electromagnetic radiation, (3) static electrification, and (4) collis ions with ions or ion clusters. Due to the low ratio of particle to air mass, the first mechanism is not considered signific ant[44, 45]. The latter two mechanisms are the most common and will be discussed in more detail. Static electrification occurs when part icles acquire charge from the surrounding environment. This can happen by any one or co mbination of the following five methods. (1) Electrolytic charging occurs when liquids of high dielectric constants exchange ions with the surfaces when they are in contact with particles. When the liquid is removed, a charge is left behind. This is not a significant source of charge for this reported research. (2) Electrons can also migrate from a clean, dry surface to anot her, dissimilar surface with a lower work function through contact electrification. This method is strictly electronic in nature and requi res that there be no impurities between the surfaces making this unlik ely to be important to real aerosols. (3) When particles come into contact with other pa rticles or surfaces, charge can be imparted to
46 them because of this contact by tr ibo electrifcation, or frictional electrification. (4) In some cases the powders can be aerosolized as liquid suspensions and during evaporation the surface forces will increase the concentrat ion of charges at the surface. (5) The last method is flame ionization where particles are ionized as they pass through the reaction zone of a flame. Electromagnetic radiation such as ultraviolet or vi sible light can also be used to create charge by ionizing the particles instead of a flame. Particles can also gain charge by collisions with electrons, ions, or ion clusters that can be produced from a number of different processes and types of electrical discharge. When an ion strikes a particle, it attaches to the surface[103, 104]. As charge s accumulate on a particle an electric field will develop around the particle an d prevent ions or electrons from reaching the particle. Therefore the accumulation will decrease as the charge on the particle increases and a point is reached where no furthe r charging happens, this is called the saturation charge. Ions can move by thermal motion alone requiring no external force and as a result of diffusion charging. In order to make the process more efficient, an external electric field is applied to control the movement of ions[105-107]. Ofte n, it is the combined occurren ce of diffusion and field charging that leads to particle charging. The most common method of i on production is by corona disc harge. Two electrodes are arranged such that the field strength between them is constant, e.g. with a wire/tube or point/plane construction. The pot ential is increased until there is electrical breakdown of the gas around the point or wire and ions are forced to leave. These ions collide with gas molecules leading to electrical breakdown in the gas. Oppositely charges i ons are drawn towards the wire and produce additional ions upon impact. This pro cess creates a rapidly growing number of ions
47 with the same charge moving towards the electro de between which passing aerosol particles can be charged by collisions with these ions. The amount of charge a particle can obtain by any of these met hods is material dependent. The method is also ineffective above 50 percent relative humidity since the presence of water will nullify the charge[45, 47]. The presence of water does, however, lead to cohesion due to capillary forces, as discussed above. Now that an understanding of the fundame ntal forces underlying adhesion between particles has been developed, their significance to the dispersion of powder in air will be presented through the experimental studies outlined in the following chapters.
48 Figure 3-1. Illustration of impor tant interparticle forces in powders and aerosols. Here Fad is the force of adhesion due to van der W aals and capillary bridge forces, Fes is the electrostatic force, FF and FS are the friction and shear fo rces respectively acting at the particle level, and F and S are the ove rall normal and shear forces acting on the ensemble of particles. Figure 3-2. A depiction of the th ree interactions that are collectively known as van der Waals forces. van der Waals forces include: Keesom interactions, or dipole-dipole interactions; Debye interactions, or dipol e induced dipole in teractions; and London dispersion interactions, or instantaneous fluctuating dipole induced dipole interactions. Fad S FFCapillary Bond FS F Fes + -+ -+ + + + + + +Fad S FFCapillary Bond FS F Fes + -+ -+ + + + + + +KeesomInteraction Dipole Dipole Debye Interaction Dipole Induced Dipole London Interaction Fluctuating Dipole Induced Dipole KeesomInteraction Dipole Dipole Debye Interaction Dipole Induced Dipole London Interaction Fluctuating Dipole Induced Dipole
49 Figure 3-3. Depiction of the calculation of interparticle force using the Derjarguin Approximation. An approximation developed to relate the force of an interaction between a body of some geometry to the energy of interaction per unit area of two parallel flat surfaces. The fo rce is integrated over infini tely small, flat concentric rings of radius x and thickness dx on each body as the separation distance between the bodies, H, goes from H to infinity. Figure 3-4. Change in cont act area with increasing roughness. The illustration shows the decrease in contact area that occurs with incr easing surface roughness and is attributed to the difference in interparticl e forces between smooth and rough surfaces. Rumpf [19, 20] initially derived the equation for the contribution of surface asperities to the decrease in van der Waals forces. This work was later expanded by Rabinovich et al to more closely define surface roughness[53, 54] and include the effect of roughness on capillary forces[49, 55, 91, 108]. H H R2R1dx x x H H R2R1dx x x Decreasing contact area with roughness Decreasing contact area with roughness
50 Table 3-1. van der Waals interaction forces and energies for different geometries. The theoretical equations for comparing inte raction force and energy where A is the Hamaker Constant, H is surface separation distance, and R1 and R2 are the plate or sphere radii. Interaction Geometry Interact ion Force Interaction Energy Plate Plate 212H A 36H A Sphere Plate H AR 6 26 H AR Sphere Sphere ) ( 62 1 2 1R R R R H A ) ( 62 1 2 1 2R R R R H A Table 3-2. Interaction energy us ing the Derjaguin Approximation. A list of factors used to calculate interaction force from the interaction energy of two flat plates as determined by the Derjaguin Approximation. The inte raction force, F, between different geometries can be calculated by multiplying the energy of interaction between two plates, Wplt/plt, by a geometry factor. Interaction Geometries Force Energy Plate Sphere Fplt/sph (2R)Wplt/plt Sphere Sphere Fsph/sph (R)Wplt/plt
51 CHAPTER 4 MATERIALS AND METHODS The first step towards realizing an increase in aerosol dispersion is a complete understanding of all interparticle fo rces significant to a given system This requires a systematic study of single particle and bul k interaction measurements to determine the dominant forces between the powder particles on both the micro and macro scale. By measuring and understanding the dominant interpartic le forces and how they change with the different factors, processing methods can be developed to reduce th eir magnitude so that the energy needed for dispersion is within the range of the current dissemination met hods. Statistical experimental design was used to systematically investigate th ese factors to establish those significant to the aerosol dispersion as determined by particle size and settling measurem ents; bulk properties through shear cell experiments; and the magnitude of significant in terparticle forces using the atomic force microscope. Figure 4-1 is a depiction of the direc tion of this research and is explained in more detail below. Material In this study, a variety of materials were used to investigate interpar ticle forces. To study the interactions between flakes a vapor phase deposited (VPD) aluminum flake material was used, Sigma 1215, provided by Sigma Technologies, In c. The material is vapor deposited onto a Mylar sheet using a continuous proc ess until the desired thickness is achieved. The VPD process uses a laser to sputter molecules from a bulk of the pure material that then adsorb onto the substrate surface. The metal layer is then remove d from the Mylar and milled into flakes in ethyl alcohol to dissolve any residual polymer and inhibit oxidation of the aluminum. Through this process the flake thickness can be carefully cont rolled and is designated by the numeric value following the name. For example Sigma 1215 signifi es that those flakes are 12-15 nanometers
52 thick. Unlike flakes produced by milling, which requires milling aids such as stearic acid that change the surface chemistry of the particles, the VPD result is a ba re aluminum flake with a thin controllable native oxide laye r suspended in alcohol. Flakes are not suitable for compaction a nd shear tests, however, because they are susceptible to mechanical deformation leading to physical interlocking under those conditions. Once the flakes are mechanically connected, th e investigation of interparticle forces is impossible since the particles can no longer slide past each other, a condition required for bulk powder testing. For this reason, atomized al uminum spherical powder from Valimet, Inc was used for the shear measurements. To create a surface for contact angle and inte rparticle force measurements, aluminum was also deposited onto a silicon wafer using a chem ical vapor deposition (CVD) process. By this process, the precursor is vaporized and the gase ous molecules flow into the reactor. At the reaction zone, they adsorb onto the substrate su rface after which thermal energy is added to aid in their decomposition and formation of a solid f ilm. This produces a smooth, near-ideal surface that enables the investigation of surface energy and how it change d with surface modification. Experimental Factors and Processing Methods A number of factors are invol ved in powder dispersion and will vary between systems. They include powder drying method, relative humid ity, surface modification, flow aids such as second phase particles, and disper sion medium. These parameters were systematically varied and the resulting powder was characterized in order to determine the degree of dispersion. First, the material itself was characterized before and after processing to determine the particle size and morphology of the materials. The surface energy was determined for the ideal surfaces using contact angle measurements. Next, the aerosol dispersion was characterized experimentally through particle size, airborne c oncentration and settling. The fi rst means used to characterize
53 the degree of dispersion of the aer osol was by a comparing the partic le size distribution in a welldispersed state; i.e. measurement of the primary pa rticle size, to that measured after processing in the dry state. The settling rate and dissemina ted concentration of the powders in an Air Dispersion Chamber fabricated in house was al so employed to study the aerosol by using laser attenuation and the Beer-Lambert Law as will be explained in more detail below. Dispersion is complex and the change in the significant forces involved and their magnitude must also be investigated. Therefor e dispersion was also studied theoretically on the macro scale by treating the powder as a conti nuum using shear cell measurements and on the micro scale by investigating the interparticle forces between individua l particles using the Atomic Force Microscope (AFM). These results coupled with the aerosolized particle size distribution data under different conditions, give insight into the mechanism of adhesion and the processing methods that can be used to su ccessfully lower adhesi on and thereby improve dispersion. As seen in Figure 41, the experimental factors chosen for this research were drying methods, surface modification, humidity, spacer mate rials, and dispersion medium. Each effect was determined by two sets of experiments. Firs t, experiments were performed to investigate the change in particle size, aerosolization, and, set tling. Second, those results were compared to the change in micro and macro scale force measuremen ts made using the shear cell and atomic force microscope (AFM) measurements. Drying Method Liquid, whether water due to humidity or pr ocessing methods, must be removed from the powder in a way that does not le ad to the formation of liquid bridges and capillary forces. Common drying methods were describe d in detail in Chapter 2. Fo r this investigation, material was dried using evaporation under va cuum, freeze drying, and a liquid CO2 supercritical fluid processing method for powders developed during th is research, then the particle size of the
54 resulting powder was measured using laser diffr action. A depiction of each of these drying methods as they relate to a general thermodynamic phase diagra m is given in Figure 4-2. For evaporation and freeze drying, the effect of liquid surface tension was investigated by dispersing the powder in water and isopropyl alcohol. It was found th at water is not miscible in liquid CO2 thus only isopropyl alcohol was used. Fo r the evaporation process, the dispersed suspensions were placed in a vacuum oven at 40oC and -20psi for 24 hours. The resulting cake was broken up with a spatula followed by particle size analysis. The process for freeze drying included placing the water suspensions in the fr eezer overnight then tr ansferring them to the freeze drying vacuum chamber. Suspensions in alcohol were frozen by mixing the suspension in liquid nitrogen then transferring the frozen samp le to the freeze drying vacuum chamber. The supercritical drying method developed for the research involved a slightly different process than the conventional methods and is show n by the green line in Figure 4-2. To avoid the inherent challenges of mixing the powder directly in liquid CO2, i.e. the laws of thermodynamics that made CO2 a liquid only above 800psi, the powder was first dispersed in isopropyl alcohol then solven t exchanged with liquid CO2 in a pressure reactor using a batch solvent exchange process. Once no alcohol was left, the pressure reactor was heated above 32oC at which point the pressure in the reacto r was above 1060psi, the critical point of CO2. Above the critical point, the surface tension of CO2 is negligible and capillary forces cease to exist. The supercritical fluid is vented from the chamber as a gas and as long as the temperature is above the critical temperature, Tc, it cannot re-condense back to a liquid and capillary forces are eliminated. Spacer Material High aspect ratio particles have a very high mass to surface area ratio. That is, because the thickness of the flake is an order of magnitude sm aller that the diameter the contact area between
55 the flakes is very high while the mass to whic h momentum can be applied is very low. Associated with the high surface area are also high attractive surface forces, especially van der Waals forces. For these reasons, once the surfa ces come into contact, dispersion can become impossible. If the contact area between flakes is decreased, the adhesion between them will also be decreased and dispersion can be improved, as described in Chapter 3. The spacer material used was a 10nm mean diameter superhydrophobic fumed silica, Degussa Aerosil LE2. The addition of the silica can also ha ve a negative effect: since the silica is not the material of interest, the addition of silica pa st the point of improving disper sion will decrease the amount of desired material available to be dispersed. Ther efore, the spacer was a dded at 0, 5, 10 weight percent. Relative Humidity Powders can be subjected to a wide range of humidity levels during processing and aerosolization. Adsorption of wa ter onto the particle due to humidity leads to particle adhesion by the formation of liquid bridges an d capillary adhesion. It was s hown in Chapter 3 that there is a critical humidity at which the liquid bridge fo rms. Since the effect of humidity has such a profound effect on adhesion, it is important to understand at what point it becomes significant (critical humidity level) and how this factor will change the pow der system. Therefore, humidity levels in the range of 10-90 percen t have been selected. The relative humidity will be increased in 40 percent intervals to allow the full range of possible environments to be studied. Each powder was conditioned at the gi ven humidity for 48 hours to allow for equilibration prior to experimentation. To do so, the samples were placed into the chamber and humid air was pumped through the chamber for 48 hours prior to measurements to allo w the powders time to equilibrate. Humidity was c ontrolled by pumping dry, compre ssed air through a water trap and by controlling the temperature of the water. Humidity levels were controlled to within %.
56 Surface Modification Capillary forces are determined by the amount of water adsorbed onto the particle surfaces at their point of contact. The ability of wa ter to condense on the surface of a particle is determined by the ability of the water to wet th e particle surface and controlled by surface energy of the water and of the partic le surface. By changing the surface energy of the powder through surface modification, the powder can be made hydrophobic and thus water cannot wet the surface. This can be done by attaching monome rs to the surface with hydrocarbon chain tails. Water molecules no longer can see the aluminum surface, but instead see a surface with which it is not thermodynamically favorable for them to interact. In order to investigate this, the aluminum surface was modified with two types of silanes. The first, Octydecyltrimethoxysilane (ODTMS), is a tri-functional monomer. That is the silicon has three bonds that are used to create a two dimensional structure on the alum inum. One bond attaches to the surface and the other two attach to the neighboring monomers. The second silane used was (TDFS). In addition to creating a hydrophobic surface, the Teflon aspect of this particular silane is thought to also decrease the surface energy between the particle s much like it can between eggs and a Teflon coated frying pan. The reaction time between the silane and the surface was also investigated to determine the optimum processing time needed to achieve the highe st hydrophobicity. Dispersion Medium Liquid Carbon Dioxide Assisted Aerosolization: As part of the experimental research, a novel dispersion method has also been developed. This dispersion method utilizes liquid CO2 as a propellant and source of energy for dispersion. Powder particles to be aerosolized are suspended in Liquid CO2. The suspension is then released su ch that the liquid instantaneously undergoes a phase change to a gas, the therm odynamically stable phase at standard room temperature and pressure. The phase change causes an expansion in the volume of the CO2 and
57 in turn creates added stresses from within the agglomerate pushing outward significantly increasing the degree of dispersion. This dispersion method has advantages over ot her dispersion methods regardless of the dominant interparticle forces. Liquid CO2 at room temperature reside s at a pressure of 800psi. This pressure will supply a large amount of en ergy to the agglomerates in the form of high velocities leading to tremendous sh ear, turbulence, and impaction. Pressure changes can also be increased further with the use of nozzles. The instantaneous phase change and expansion of the CO2 will add to the energy of dispersion by supplying dispersion force in an outward direction. The state of dispersion of the powder in the liquid CO2 is also important as an increased state of dispersion in the li quid phase is expected to lead to higher dispersion upon aerosolization. Technology for the dispersion of particles in liquid CO2 has been developed and implemented in the dry cleaning industry alread y[59, 60, 109, 110]. Surfactants have been developed with a CO2-philic end and a CO2-phobic. This surfactant is used to disperse particles in a similar way to surfactants in aque ous mediums such as detergents. This technology can be leveraged to aid in this dispersion method. Characterization Methods Particle Size and Morphology All materials underwent the same rigorous characterization protoc ol upon recite. For particle size, this included using the appropriate li ght scattering instrument for the size range of the powder, optical image analysis, and scanning electron microscopy (SEM). Laser Light Scattering The light scattering instrument used for particle between 40nm and 2mm was the Beckman Coulter LS13320. It is based on the principle that particles scatter and diffract light at cert ain angles based on their size, shap e, and optical pr operties. A 750nm diode laser is used for analysis of particles in the size range of 0.4 m to 2000 m. The beam
58 passes through filters as well as projection and Fourier lenses and is spatially recorded on 126 photodiode detectors. The Polariza tion Intensity Differential Scatte ring (PIDS) assembly is used to measure particles from 40 nm to 400 nm a nd improves resolution in the 400 nm to 800 nm range. PIDS uses a tungsten-halogen lamp and thr ee sets of vertically an d horizontally polarized color filters at 450, 600, and 900 nm as the light source. The tec hnique is based on the principle that at high scattering angles (~90 degrees) the difference in s cattering intensity of the two polarizations is a function of the pa rticle size to wavelength ratio. The advantages of this technique include eas e of operation, large range of detectable particle sizes, and accuracy in the micron and submicron range. The calculations assume the scattering pattern is due to singl e scattering events by spherical particles of the same refractive index. This raised some questions since this study involved high aspect ratio particles, i.e. flakes, and mixtures. For flakes moving in turbulen ce past that laser, it is not the actual diameter being measured but the average random orientat ion of the flake. However, when using a differential volume distribution, the particle size is weighted towards large particles, which in this case is the maximum diameter. Mixtures on the other hand, could not be measured using this instrument as there is no way to simulta neously determine the scattering pattern from two materials of a different refractive index and other methods were used. The LS13320 has two sample modules which made it ideal for this re search: a fluid module a nd a dry powder module. This allows for the measurement of the powders in a wet, well dispersed state and again after processing and drying using the same instrument. All samples in liquid were placed in a sonic bath for twenty minutes before measurement to insure complete dispersion. While all dry materials were sampled according to the appropr iate methods before the measurements. The results presented in this work are an average of at least three measurements.
59 Dynamic Light Scattering Powders with a mean diameter less than 10 nm, particularly the spacer materials, were measured usi ng the Microtrac Nanotrac NPA250 dynamic light scattering instrument. It is capable of measuring particles between 0.8nm and 6.5 m based on dynamic light scattering by meas urement of the velocities of particles undergoing Brownian motion in a liquid. This motion is detected by measuring the spectrum of Doppler-shifted photons from light scattered by th e particles. Photons scattere d from particles moving away from the scattering photon have their frequency lowered. A photon scattering from a particle moving towards it has its scattered frequency incr eased. The instrument compares the frequency shifted photons to unshifted photons, recording the di fference or beat frequency of the interfering photons. The Brownian velocity of particles is inversely proportional to particle size and becomes quite significant if the particle dimensions are less than 1 m. By measuring the spectrum of Doppler shifted frequenc ies, the particle velocity distri bution and, hence, the particle size distribution can be determined. All mate rials in a liquid medium were placed in an ultrasonic bath for a minimum of twenty minutes prior to measurement to insure dispersion. Small and more agglomerated materials required up to one hour of ultrason ication. All results presented are an average of at least three measurements. Scanning Electron Microscopy SEM images of the particles were taken using the JEOL 6330, a cold field emission scanning electron micr oscope. Cold field emission scopes have the advantage of high brightness and high resoluti on even at low accelera ting voltages to allow imaging of a wide variety of ma terials including soft polymeric materials without causing sample damage. Resolution of the instrument is around 15 Angstroms depending on the sample. The JEOL 6330 is equipped with an energy dispersi ve X-ray spectrometer (EDS) for elemental analysis and a backscattered electron dete ctor that allows compositional analysis.
60 Optical Microscopy and Image Analysis The Olympus BX60 optical microscope is equipped with a SPOT Insight Digital Camera. Th e SPOT RT is a "three pass" charge coupled device (CCD) digital color camera suitable for taking digital images of a particulate sample. The SPOT RT is cooled to 37C below room temperature to reduce thermal noise inherent in CCD devices. It is equipped with dual light sources for both transmitted and reflectance imaging. It can be operated in the bright field, dark field, and differential interferen ce contrast (DIC) modes. Five objectives from 5 100 times magnification are available. Image Pro v. 6.2 Optical Analysis Software analyzes digital pictures taken of particles. Pa rticle properties such as size, shape, roundness, aspect ratio, and percent imag e coverage can be measured. All particle size distributions were made from images of at least 10,000 particles to insure statistical significance. Material Surface Energy Ram-Hart Contact Angle Goniometer In order to understand th e effect of surface modification on the material, advancing and reced ing contact angle measurements were made before and after surface modifi cation of the material with a Rame-Hart Contact Angle Goniometer. This instrument consists of a st age for droplet formation on a material surface, a digital camera for image acquisition, and software for data analysis. Ten to fifteen images of the advancing and receding contact angle of a wate r drop on each surface were taken and the average of the measurement made from these images is reported. The signifi cance of the different contact angles is presented in Chapter 3. Dispersion Characterization To determine the effect of processing me thods on improving dispersion, the degree of dispersion, or dispersability, of the aerosol wa s measured by two methods For single materials the comparing the particle size measured by lase r light scattering was used. By comparing the
61 primary particle size measured in a wet, well dispersed state, to that of the powder after processing and drying, the degree of dispersion can be seen by the shift in mean particle size. For mixtures with different indices of refrac tion, airborne concentr ation and settling rate measurements of the powders were made in a chamber fabricated for the FTIR. Particle size comparison The first method used to determine the degree of dispersion was by measuring the particle size using the Beckman/Coulter LS13 320. Th e LS13320 has two sample modules that made it ideal for this research: a fluid module and a dry powder module. Because of this, powders could be measured in a wet, well dispersed state th en processed and dried and measured as a dry powder in the same particle size instrument. A comparison of the two par ticle size distributions yields a measurement of the degree of disper sion. Agglomeration can be seen by both an increase in the mean particle si ze, and an increase in the width of the particle size distribution. Complete dispersion is achiev ed if both particle size m easurements are equal. Settling measurements Air Dispersion Chamber for the Fourier Transform Infrared Spectrometer: In order to characterize dispersion as aerosols, a lab scale air dispersion chamber was fabricated that fits into the Thermo Electron Magna 760 FTIR Micros cope. The chamber, with a volume of five liters, fits into the FTIR such that the IR laser can pass thr ough the windows of the chamber to the detector. A picture of the FTIR Air Cham ber in place can be seen in Figure 4-3A. The design of the chamber allows for the induction of powder by low vacuum, -12.5psi, or a variety of positive pressures produced by the use of an N2 cylinder, 5-90psi, or a CO2 cylinder, 800psi gas or liquid. The IR laser passes through the di sseminated aerosol cloud and the intensity is measured by the FTIR detector and reported at pe rcent transmission. A schematic of an aerosol measurement in the chamber after dissemination can be seen in Figure 4-3B Transmission (T) is
62 measured over time and graphed as the negative log of one divided by transmission. The graph of time versus the log of one divided by transmi ssion is indicative of se ttling rate. A slower settling rate would result from a smaller particle size by increased dispersion. If a constant mass of material is used, from which a constant concentration is assu med immediately after dissemination, the mass extinction coefficient of the Beer-Lambert Law, Equation 4-1, can be calculated and used to determine th e airborne concentration over time. Beer-Lambert Law: -Log(1/T) = Me C PL (4-1) Here T is transmission, Me is the mass extinction coefficient, C is concentration, and PL is the path length of the laser through th e aerosol cloud and assumed to be the diameter of the chamber. Powder Mechanics Shear Cell There are many methods available to m easure the cohesive forces of bulk powders including measurement of the shear stre ngth or tensile strength of a powder compact, angle of repose, and powder flow behavior. In this study the bul k properties was measured using the Sci-Tec ShearScan bulk powder tester. A diag ram of the Sci-Tec ShearS can is seen in Figure 4-4. This method offers the most simplicity and several models have been developed to relate the tensile strength to interparticle force. For a typical measurement, powder is loaded in to the cell, which has an annular geometry and covered by a lid. A load is applied to th e lid by the overhead piston pushing down onto the powder causing consolidation. Th e lid of the cell is then rota ted by the piston and the torque applied to the lid by the powder is measured with a load cell connected to the base of the test cell. The shear stress versus distance sheared da ta recorded at each normal load is used to generate the Yield Locus. Mohrs circles are drawn tangent to the Yield Locus, one of which will pass through the origin. The Mohr circle is a two dimensi onal representation of all the equivalent states of stress that de scribes the Unconfined Yield Stress ( fc ) in terms of a shear
63 stress () and a normal stress () that correspond to the major principal stresses (1, 3) acting on the system. A typical direct shear test of a powder is pe rformed by consolidating the sample at a given normal load then lowering the normal load and determining the shear stress required to initiate flow (failure) in the sample. From a graph of the stress vers us distance sheared consolidation state of a powder and the subsequent failure stress that leads to in cipient flow in the system can be determined, as shown in Figure 4-5. The yiel d locus is the line tangent to the failure Mohr circles, and the slope of the yiel d locus is called the angle of inte rnal friction of the powder. The unconfined yield strength is define d as the yield stress of a bulk sample at a given consolidation load where one of the major principal stresses eq ual to zero. A zero stress on one of the axes represents a powder not confined in the direction of the stress which is the case during powder flow of powder. A set of yield loci is generated at cons olidation loads of 2. 5kpa, 5kpa, 10kpa, 20kpa, and 30kpa. The unconfined yield strength verses the major principle stress, 1, is plotted at each consolidation load to generate a flow function. The flow functi on defines the change in strength of the powder, measured as fc as a function of the applied conso lidation load. A flow function is generated for each powder at each relative hum idity level to determine the change in fc Each flow function reported in this research is an aver age of at least four repeats of a given relative humidity and consolidation load. Apart from the unconfined yield strength, the M ohr circle also enable s the determination of the cohesion (C) and the angle of internal friction (angle of repose, ) both of which are bulk parameters characterizing the flow characteristic s of the powder. The unconfined yield strength (fc) and principle stresses (1,2) can then be determined as shown in Figure 4-5.
64 Interparticle Forces Atomic Force Microscope (AFM) The properties of the powder as a bulk give insight about the strength and flowability of the powder but not all of the information needed to understand powder cohesion and how it relates to dispersion. In or der to determine the forces on the particle level, an Asylum Research AFM was used. A humidity cell accessory was utilized to measure and control the humidity in the sample chamber. With this instrument, the significant interparticle forces and their ma gnitude can be measured for each of the surfaces at the relative humidity values. This is done to determine the effect of the surface coa tings and humidity on the interparticle forces that are dominant at different conditions and methods that can be used to help overcome them. Interparticle force measurements were ma de on each surface with an AFM cantilever according to the methods described by Ducker et al and Rabinovich[111, 112]. For the measurement, the cantilever was positioned close to the flat substrate. Then, as the AFM tip assembly is moved towards and away from th e surface by a piezoelectric column, the deflection of the cantilever is monitored by a diode laser th at reflected off the top of the cantilever onto a position sensitive photodiode. As the cantilever is attracted to or repelled from the surface, depending on the situation, the amount of defl ection is measured by the position of the laser on the photodiode, as shown in Figure 4-6A. By this procedure, the force between the two surfaces as a function of separation distance and can tilever spring constant was obtained, and representation of this can be s een in Figure 4-6B. The approach ing portion of the force curve is characterized by a jump-to point as the tip is attracted to the surface and can be seen by the black solid line in Figure 4-7. Once the tip is in co ntact with the surface, the piezoelectric column raises the assembly away from the surface. The cantilever will bend until the force holding the cantilever tip to the surface is grea ter than its spring constant at wh ich time it springs back to the
65 normal position. The retraction portion of the for ce/distance curve can is represented by the red dotted line in Figure 4-7. The force/distance profiles were then norma lized by dividing the measured force by the radius of the sphere thereby allowing the data to be presented in terms of energy per unit area of a flat surface. This enables th e determination of interaction for ce for different geometries as long as the range of the forces is much less than the ra dius of curvature of the particles interacting. Force measurements were taken at four differe nt spots on each surface and an average force of fifteen measurements at each spot was reported. Design of Experiments The order of experiments performed was dete rmined by developing a statistical design using the Design Expert 6 software package. Do ing so has two advantages. First, it allows evaluation of the error associated both within th e experiments of one c ondition and that between the different conditions. This will help separate experimental error from the results allowing a more accurate evaluation of the factors under inves tigation. Second, the software will provide an equation base on the model developed to fit that da ta and the factors chosen to be significant. From this equation, the optimum result within th e experimental design space can be reached. In addition, it will also aid in determining what di rection to move the experimental space if the highest value has not been reached. A design was developed for the airborne concentration experiments, shear cell measurements, and interp article force measurements. A comparison of the results was used to develop the correlation be tween the change in processing, powder forces, and dispersion. The results to the work outlined in Chapter 4 will be presented in Chapter 5. The results for the particle size and settling and the bulk an d interparticle force measurements will be
66 discussed as well as a correlation of the data. The conclusions developed during this research will be explained in Chapter 6 and applied in a case study described in Chapter 7.
67 Figure 4-1. A roadmap to disper sion: depiction of the directio n for the presented research. Design of experiments was employed to dete rmine the magnitude of the effect of factors that have been shown in literature to effect disper sion in air. Experimental results are compared to theory and micr o and macro scale force measurements to develop an understanding of their sign ificance, magnitude, and change with processing. Surface Modification Spacer Determine significant powder characteristics* Effect on dispersion: Aerosolization Particle Size Understand Cohesion Drying Method Effect on powder properties/ interparticle forces Air Chamber Shear Cell AFM Humidity Improve Powder Dispersion *Different for each powder system Surface Modification Spacer Determine significant powder characteristics* Effect on dispersion: Aerosolization Particle Size Understand Cohesion Drying Method Effect on powder properties/ interparticle forces Air Chamber Shear Cell AFM Humidity Improve Powder Dispersion Surface Modification Spacer Determine significant powder characteristics* Effect on dispersion: Aerosolization Particle Size Understand Cohesion Drying Method Effect on powder properties/ interparticle forces Air Chamber Shear Cell AFM Humidity Improve Powder Dispersion *Different for each powder system
68 Figure 4-2. Thermodynamic path of drying me thods investigated. A depiction of the thermodynamic path across phase boundaries followed by each of the three drying processes investigated in this research. Evaporation crosses the gas-liquid phase boundary leading to capillary forces and agglomeration. Freeze drying takes the liquid to low pressure and temperature as it is first frozen then removed by sublimation. If the pressure is not sufficiently low, conde nsation is possible and will lead to capillary forces. Supercritical dr ying takes a path to high temperature and pressure completely avoiding the formati on of capillary forces through the entire process. The best results were f ound using the supercritical process. Supercritical Drying Evaporation Freeze Drying Capillary Forces Supercritical Drying Evaporation Freeze Drying Capillary Forces
69 A Liquid CO2 Bottle Valves Powder in Liquid CO2 Sapphire crystal windows Liquid CO2 Bottle Valves Powder in Liquid CO2 Sapphire crystal windows B Figure 4-3. Images of the Air dispersion chamber. A) is a photograph of the air dispersion chamber when positioned in the FTIR during an experiment. B) is a schematic of the chamber showing the laser path through an aerosol cloud pr oduced in the chamber. The aerosol can be produce by high pr essure gas or a volatile medium.
70 A B Figure 4-4. The Sci-Tec ShearScan TS12 sh ear tester and test cell. A) shows a photograph of the Sci-Tec ShearScan inst rument. B) depicts the compression cell used for the measurements. Powder is loaded into the cell and a consolidation load is app lied by the piston from above as it rotates. The torque applied is measured by the load cell attached to the base. Figure 4-5. A typical result from the consolidation and failure of a powder sample. The corresponding points on the Mohr circle are shown to illustrate the procedure of generating the Yield Locus, the major prin ciple stress, sigma one, and the unconfined yield strength, fc Yield Locus 00.511.522.533.54 0 0.5 1 1.5 2 Normal Stress (KPa)Shear Stress (KPa) Continual Deformation Failure fc Normal Stress, (kPa) Shear Stress, (kPa) Yield Locus 00.511.522.533.54 0 0.5 1 1.5 2 Normal Stress (KPa)Shear Stress (KPa) Continual Deformation Failure fc Normal Stress, (kPa) Shear Stress, (kPa)
71 Figure 4-6. Schematic of how the AFM works. A) A laser is reflected off the end of the cantilever onto a position sens itive photodiode. The cantile ver is brought close to the sample, on a moveable stage, and the inter action between them causes the cantilever to move. The movement is measured by th e change in position of the laser on the photodiode. B) The measured cantilever m ovement and its spring constant allow calculation of the interaction forces. Cantilevers of varying compositions and spring constants allow for a large ra nge of measurement options. Z X Y Position Sensitive PhotodiodeL a s e rXYZ Scanner XYZ Scanner Z X Y Position Sensitive PhotodiodeL a s e rXYZ Scanner XYZ Scanner A B
72 -1.00E-08 -8.00E-09 -6.00E-09 -4.00E-09 -2.00E-09 0.00E+00 2.00E-09 4.00E-09 6.00E-09 8.00E-09 1.00E-08 0510152025303540 Separtion Distance (nm)Force (N) Approach Retraction Figure 4-7. Example of a force distance curve for the AFM. The cantilever approach is depicted by line 1 and characterized by the jump to su rface point at 2.5nm. After this point, the piezoelectric increases the separation di stance and the adhesion force is calculated using the height of the retr action curve shown by line 2. 2 1
73 CHAPTER 5 RESULTS AND DISCUSSION Characterization Particle Size of Single Materials All materials were characterized prior to e xperimentation. Particle size was measured using laser diffraction and compared to image analysis and SEM images. The non-spherical nature of the aluminum material and the mixture of materials with a diffe rent index of refraction presented new challenges for common characteriza tion techniques such as laser diffraction. Laser diffraction uses the scatteri ng pattern produced by the laser interaction with the particles and the index of refraction of the material to calculate a particle size. The first assumption required to make accurate measurements is that th e particles are spherical and the second is that the scattering pattern is produced by only one index of refraction. Laser light scattering is an efficient and repeatable sizing method; therefore the use of this method for flakes was further investigated. To do so, both assumpti ons must be further understood. The orientation of the particles as they m ove past the laser is random and cannot be controlled. A spherical particle does not have an effect due to orientation, a flake, however, does. Although flakes can be considered spherica l in two dimensions, they cannot be considered spherical in the third, their thickness. Two impor tant concepts must also be taken into account. First, the differential volume di stribution produced by li ght scattering is weighted towards the larger particle sizes. This is because only a few 10 m particles will have a much larger volume than an even higher number of sub micron particle s. Second, statistica lly the instrument will measure the average random orient ation of the particles being measured due to the fact that the flow field past the laser is tur bulent and therefore the orientation of the par ticles will constantly be changing. For these reasons it can be expect ed that light scattering will measure the average
74 random orientation of the flake and, using a volume distribution, the size will be weighted towards the diameter of the particles. To test this hypothesis, pa rticle size distribution measurements made of pure aluminum flak es suspended in isopr opyl alcohol using the Beckman/Coulter LS 13320 laser diffraction instrume nt were compared to those made using image analysis as describe in Chapter 4. The re sults can be seen in Figure 5-1, which shows the graph of particle size distribu tion produced by both laser diffrac tion and image analysis. The difference can be explained by the method of particle size measurement used for each instrument. Image analysis measures the maximum particle diameter of the flakes as they lay flat on the microscope slide, whereas lase r diffraction is measuring the average random orientation of the flakes as they move pass th e laser in a liquid suspen sion; the average random orientation always being smaller than the maxi mum diameter. However, the difference between them is consistent. It was therefore determined that laser diffraction is suitable to measure the size of flakes if it is acknowledged that the repor ted particle size is typi cally 25 percent smaller than the maximum dimension. The laser diffraction pattern used to calc ulate the particle size is obtained by a simultaneous measurement of the laser intensity on a number of detectors placed at different angles from the normal laser position. It is impossible to decipher whic h particles produce the pattern. For this reason the particle size measure of a sample made up of a mixture of materials each with a different index of refraction is inaccura te. This led to the use of other measurement methods to determine the degree of dispersion of the samples to which spacer materials were added and will be described below.
75 The spherical aluminum material from Valimet Inc used for the shear cell measurements was also characterized using laser diffraction. Figure 5-2 shows the particle size distribution with a mean diameter of 4 m and a log normal distribution from 1-10 m. Particle Size of Mixtures The characterization of mixtures and the degree of dispersion posed new challenges. Due to each material having a unique index of refract ion, laser diffraction could not be used to characterize dispersion. A new method was developed using the at tenuation of light in a settling chamber built for the FTIR. The powder can be disseminated by pulling a partial vacuum on the chamber or by pressure using a compressed ga s or liquid cylinder. Once the powder is disseminated, measurements of the laser transm ission through the partic le cloud are made over time. These transmission measurements can then be used to calculate airborne concentration using the Beer-Lambert Law, Equation 4-1, and effective particle size using the Stokes settling equation explained in more deta il below[113, 114]. From these the degree of dispersion can be determined and with it the effectiveness of the processing and dissemination methods for that sample. Morphology To obtain a better understanding of particle mor phology, images of all of the particles were taken using the scanning electron microscope (S EM). Figure 5-3 shows SEM images of the VPD aluminum flakes. A face of one flake is sh own in Figure 5-3A. The irregular shape of the flake shown in this image is indicative of the milling process which was used to produce it after vapor deposition onto the My lar sheet. Edges of a number of flakes can be seen in Figure 5-3B and the large diameter to thickness ratio can be read ily appreciated. It is interesting to note in Figure 5-3B, that the flakes are so thin that some are becoming transparent to the electron beam and the stack of flakes can be seen under th e flakes laying on top. Transmission electron
76 microscope pictures were also ta ken of the flakes to study the fo rmation of the oxide coating at the flake edges. The images were taken with the assistance of Kerry Siebein at the Major Analytical Instrumentation Center Department of Materials Scien ce and Engineering, University of Florida. As seen in Figure 5-4 a unifor m and continuous oxide layer is present and the aluminum below the oxide layer is preserved. The atomic force microscope (AFM) was also used to characterize the flake morphology as well as surface roughness. A standard silicon nitr ide cantilever was used to image the flakes deposited onto a silica surface. Th e image of two flakes can be seen in Figure 5-5. The irregular edges of the flakes can be seen as well as some large scale roughness which is expected to be attritted pieces of aluminum generated during the milling process. Higher magnification images of the flake surface can be seen in Figure 5-6 an d show that the flake itself has a very smooth surface. The root mean squared (rms) surface roughness of the flakes was measured, using the AFM, to be 4nm. SEM images of the Valimet spherical aluminum we re also taken and seen in Figure 5-7. It is apparent from the image that the material has a broad particle size as observed in the particle size measurements, Figure 5-2. Drying Methods For drying, each sample was mixed in water or isopropyl alcohol (IP A) and sonicated for 30 minutes to insure the powder was dispersed. For oven drying, the suspension was then placed in a vacuum oven at 60oC and -25 inches Hg for 24 hours to dry. For freeze drying, the samples were pipetted into liquid nitrogen to freeze them and then placed in the freeze dryer. For SC fluid drying with liquid CO2, the powders were dispersed in IPA, as water is not highly miscible in CO2. As this is a new process for drying powde rs, it will be looked at in more detail.
77 A custom designed Parr Instruments pressure re actor was used for the supercritical drying. Solvent exchanges were performed to complete ly replace the isopropyl alcohol medium with liquid CO2. After this was completed, the chamber was heated to a temperature above the supercritical state of liquid CO2, held to allow for the temperat ure throughout pressure vessel to reach equilibrium, and then slowly releas ed. This process is depicted by the CO2 phase diagram in Figure 5-8. The process starts at room temperature where the liquid CO2 is at 800psi inside the pressurized chamber, point A in Figure 5-8. The chamber is heated until a temperature above 31Celsius at which point a pressure above 1072 psi is reached, the critical point of carbon dioxide, and held for at least thir ty minutes to insure the chamber and its contents equilibrate to this temperature, seen as point B in Figure 5-8. A pin valve on the top of the chamber is then opened to release the supercritical fluid as a gas at a rate no greater than three liters per minute until the pressure in the cham ber returned to 0 psi. The results for drying can be seen in Figure 5-9, the particle size distribution of each material made after drying. Evaporation is represented by the red curve. As the liquid evaporates, capillary forces draw the particles together, so that attractive van der Waals forces cause the particles to agglomerate. The agglomera tion can be seen in the particle size distribution which has a large mean particle size and wide distribution. Duri ng freeze drying, the inability to maintain sufficiently low pressure during the length of the drying process leads to the condensation of water vapor and the formation of liquid bridges and capilla ry forces. This was determined to be due to the limitations of the vacu um pump to maintain the pressure required to sublime the frozen liquid. The resulting materi al produced by supercri tical drying using liquid CO2 is shown to be the most disper sible in Figure 5-9. This mate rial has the smallest particle size and narrowest distribution of all the processes investigated due to the fact that supercritical
78 drying completely avoids the gas-liquid phase bounda ry throughout all steps of the process. The primary particle size distribution of the material measured by dispersion in isopropyl alcohol is shown for a comparison. Despite the improve ment provided by superc ritical processing, completely removing the capillary forces dur ing drying was not alone sufficient enough to produce a completely dispersible powder. Alth ough drying was shown to have a significant effect on powder dispersion, othe r factors that influence disper sion must also be taken into account in order to achieve a fully dispersed powde r. Aside from capillary forces, van der Waals forces are always present between particles and are very significant when particles are in close proximity. Reducing the van der Waals attraction will increase the dispersability of the powder and will now be discussed. Aerosol Experiments Spacer Addition van der Waals forces are generally accepted to be significant only at short ranges, often considered less than 50nm. Therefore, by mainta ining some separation between the particles the van der Waals attraction will decrease. Surface roughness also impacts van der Waals forces in two ways. First, the asperities on the surface lead to a less intimate contact and a lower total contact area. Second, distance between the averag e surface plane is increased. This can be seen in Figure 5-10, an illustration of the effect of roughness on two flat surfaces. The surfaces to the left are smooth and are seen to be in intimate contact. The surfaces in the center and to the right illustrate increasing roughness. The contact area between the flakes can be seen to decrease and the distance between the averag e surface plains increase with increasing roughness. This was first shown by Rumpf [88, 89] w ho derived an equation for the effect of roughness on van der Waals forces to explain the difference between the calculated and measured adhesion force of two surfaces as seen by Equation 3-12.
79 Recently, Rabinovich et al. furt her expanded this equation to better account for the true geometry of the surface taking into account not only asperity hei ght but asperity width and peak to peak distance as well[53, 54] and shown in Equation 3-13. The effect roughness has on van der Waals force can be seen by comparing the fo rce distance curves the smooth rough surfaces as shown in Figure 5-11. Here the van der Waals fo rce of attraction is pl otted versus separation distance. Curve 1 shows the attractive forces for two smooth surfaces using the plate-plate equation in Table 3-1 and curve 2 shows the attractive forces for th e same size flakes with a 4nm RMS surface roughness as calculated by the PERC Mode l, Equation 3-13. As seen in Figure 511, this increase in surface roughness has a dram atic effect of decrea sing surface forces. By applying this idea towards particles to be dissemi nated, decreasing the attractive forces will lead to an improvement in dispersion. The surface roughness of the aluminum flakes cannot be altered easily. However, by adding a spacer particle between th e flakes, roughness can be in eff ect simulated as illustrated in Figure 5-12. The fumed silica is nearly the sa me size as the flake thickness, 10nm, and has a lower Hamaker constant than aluminum 6.55x10-20 J as opposed to 1.5x10-19 J respectively. The combination of these two effects will redu ce the interparticle forces. The significant decrease in the magnitude of the forces can be seen in Figure 5-13, a comparison of the calculated force distance curve for smooth aluminum flakes 12 m in diameter and 12nm thick alone; with the addition of five weight percent fumed silica; and with the addition of ten weight percent fumed silica. There is a slight increase in interaction for ces at ten weight percent silica because of the increasing number of contacts with increasing silica addition. Based on these calculations, experiments were c onducted to investigate the tr ue effect of adding Degussa Aerosil LE2 fumed silica to the flakes.
80 The silica was added to the aluminum flak es suspended in isopropyl alcohol before supercritical drying. SEM images of the flakes alone and mixtures with silica can be seen in Figure 5-14 A-C. Figure 5-14 A shows the flakes themselves, their smooth surface can easily be seen. Figure 5-14 B is an image of the mixture of flakes with five weight percent silica added. The silica is seen to be attached to the flake surfaces thereby in creasing the interparticle spacing. Figure 5-14 C shows the mixt ure with ten weight percent silica added. The silica is distributed evenly throughout the sample; how ever large agglomerates of th e silica can be seen indicating that dispersion of the spacer material is also impor tant. It was determined from this information that ten weight percent silica was in excess of what was needed therefore experiments were carried out with mixtures of five weight percent Degussa Aerosil LE2 fumed silica added to the aluminum flakes. For each experiment, a sample of 3.850.05mg of material was disseminated into the FTIR Air Chamber and measurements were taken as the powder settled. The data was plotted as the log of inverse transmission over time as seen by the blue points conn ected by a line in Figure 5-15. The slope of this line is indicative of th e settling rate since the transmission is decreasing due to the powder settling out of suspension in the air. A trend line is added to the data and the trend line equation displayed on the graph, wh ich is shown in Figure 5-15. Repeats of each sample were run and the trend line equations used to determine the average and standard deviation for each condition. This was used to insure statistical signif icance between results. The graph in Figure 5-16 shows a typical standard deviation for a given set of conditions. The average is depicted by the solid line and the stan dard deviation is given by the dashed lines. Settling times of interest were chosen to be one second, taken to be the maximum initial dispersion; ten seconds, a time of interest for both pharmaceutical powders and other aerosol
81 applications; and one hundred seconds, a length of time considered to be in excess of most applications but interesting for the point of view of total aerosol lifetime. Using the average settling rate, as shown in Figur e 5-16, along with knowledge of th e initial material mass and the Beer Lambert Law, Equation 4-1, the airborne c oncentration can be calculated. An example of the results can be shown in Figure 5-17, the airb orne concentration of aluminum flakes produced by 12psig vacuum induction into the chamber. From the graph, it is seen the 66% of the powder was introduced into the aerosol. Using this type of graph, the maximum airborne concentration of powder produced and the effectiv e settling rate can be compared for the different conditions. These measurements are made by laser attenuation, thus increasing airbor ne concentration is accomplished by increasing the number of flakes scattering the light and thus increasing dispersion. Very large agglomerates will not be aerosolized and fall direc tly to the bottom of the chamber. Smaller agglomerates settle out very quickly and will be seen in a rapid settling rate. A well dispersed powder will have a high initial airborne concentration and maintain a higher concentration over the selected measurement time. Samples containing five weight percent silica were th en disseminated using the same partial vacuum pressure. As determined above, th e silica is expected to reduce the interparticle forces and this is, in fact, confirmed by the se ttling results. Figure 518 shows a comparison of the airborne concentration for the vacuum disseminated aluminum and aluminum containing the silica. The addition of the silica increased disp ersion and lead to a seven percent increase in airborne concentration. The settling rate was co mparable to the aluminum alone and the airborne concentration has room for furt her improvement. The partial v acuum induction method uses a relatively low pressure and so the effect of in creasing dissemination pressure was subsequently investigated.
82 To study the effect of pressure a compresse d gas cylinder was attached to the induction tube after the sample was loaded. Using co mpressed nitrogen pressures from 5-90psig were safely attainable. However, experimentation at a higher pressure was also desired and for this a compressed gas carbon dioxide cylinder was empl oyed for a test pressure of 800psig. The results are seen in Figure 5-19, a plot of th e airborne concentrati on over time of aluminum disseminated using the varying pres sure. As expected, increasing pressure resulted in a higher airborne concentration and a slower settling rate. The addition of silica was also investigated at 800psi. Figure 5-20 shows the airborne concentr ation for samples disseminated with and without silica. The sample with silica disseminated by 800psig gas carbon dioxide showed a decrease in airborne concentration. In none of these expe riments, did the airborne concentration reach 100 percent, thus inviting further consid eration of the dispersion method. As discussed in Chapter 3, turbulent flow is a ssociated with fluctuations in velocity called eddies which impart energy to break up agglom erates, and a minimum eddy size determined by pipe diameter and air flow velocity. Particles smaller that the minimum eddy size would be able to follow the velocity gradient and not be di spersed. Using the maximum pressure, 800psig, the minimum eddy size for this system was determined to be 1.2 m. Taking into consideration that the flakes are 12 m in diameter but only 12nm thick and that the force needs to be applied along the thickness such that the fl akes can be separated, it was determined that even at high pressures, the minimum eddy size was larger than that needed to separate flakes from intimate contact. To do so would require higher pressures or the application of the pressure in a different way. Instead of applying the dispersion for ce only by linear methods such as turbulence and drag, perhaps adding a radial component from inside the agglomerate acting outward would increase dispersion.
83 Dispersion Medium Turbulence, although effective for larger partic les, has a minimum turbulent eddy size of 1.2 m for this setup that limits the smallest aggl omerate that can experience dispersion forces during dissemination. In order to overcome this challenge, turbulen ce was coupled with a volatile medium, liquid carbon dioxide. By this me thod, not only will there be the high pressure, 800psi, available to create turbul ence for dispersion, in addition there will be a radial outward force that will develop as the liquid carbon diox ide is released and unde rgoes a phase change from liquid to gas. An illustration of this id ea is provided in Figure 521. Taking into account the change in density, the change in volume associated with the phase change under ideal conditions and neglecting any temperature change due to Joule cooling, an increase in volume of approximately 500 times is calculated. Taking the sample chamber volume to be 3cc, this is calculated to be a force of 3 N per flake. Considering that th e van der Waals force calculated by the PERC Model and shown in Figure 5-11 is 132 N this force is two orders of magnitude less than that required for dispersion. If the flake and silica mixture is considered, an interparticle force of 16 N is calculated. For this case, neglecting all other forces acting on the particles, the expansion force of the liquid CO2 alone is near the force needed for dispersion. This additional force, in combination with the 800psi dissemination pressure is expected to further improve dispersion. For these experiments, after the sample was loaded, a liquid carbon di oxide cylinder with a siphon tube was connected and the sample chamber filled. The measurement protocol described above was used. The aluminum flakes were di sseminated using the liquid and the results are seen in Figure 5-22, a plot of airborne con centration at select times. The maximum initial airborne concentration was calcula ted to be only 65%, a value less th an that measured previously
84 by 800psi gas dispersion as seen in Figure 5-23. Fr om this result and further consideration of the phase change process, the following two conclusi ons were drawn. First, the volume of liquid actually between the flakes must be considered. The flakes are smooth and once in contact, very little liquid will be able to penetr ate the space between them. If this happens, the liquid does not expand from between the flakes and no extra dispersion force is achie ved. Second, it is considered that the liquid to gas phase change was not, in actuality, instantaneous. Because of this time delay, a meniscus of liquid carbon di oxide can form between the flakes. The low surface tension of liquid carbon dioxide would wet th e particles and capillary forces would draw the particles closer together incr easing the adhesion force. The addition of a spacer in this case would increase the interparticl e spacing, increasing the amount of liquid carbon dioxide between the flakes and decrease the area over which capillary forces can act. A simple evaluation of the interparticle vol ume for two flakes with an rms roughness of 4nm is calculated to be 0.6nm3, while adding five weight percen t spacer increases that volume by a factor of three to 1.5nm3. Experimentally the effect can be seen by the increase in airborne concentration in Figure 5-24. The addition of the sp acer shows a twenty percent increase in airborne concentration over that of liquid alone. In addition, the settling rate of the mixture is shown to be slower a second indication of impr oved dispersion. The same conclusion is drawn when the liquid carbon dioxide results are compared to those of the gas, Figure 5-25; the mixture disseminated using liquid carbon dioxid e provides the best results. Perfect dispersion has not yet been achieved a nd other factors must al so be considered. Due to the readiness of moisture from the envi ronment to adsorb onto the particles, and the significant increase in adhesion created by capillar y forces, surface modification to neutralize the effect of humidity will next be investigated.
85 Surface Modification Capillary forces between particles caused by the condensation of water from the atmosphere can lead to a significant increase in particle adhesion and in turn decrease the dispersability of the powder. This can readily be seen by the change in particle size distribution of aluminum flakes in Figure 5-26 after conditioning at 20, 50, and 80% relative humidity for 48 hours. The particle size is seen to increase with relative humidity as capillary forces increase the adhesion between the particles. The primary particle size is shown for a comparison. The surface hydrophobicity of aluminum wa s changed by surface modification using noctadecyltrimethoxysilane (ODTMS) and tridecafl uoro-1,1,2,2-tetrahydro octyl-1-triethoxysilane (TDFS). For surface modification, 2% (by volum e) solutions of ODTMS and TDFS were made as follows. 2 mL of silane and 0.5 mL of wa ter was placed into 100 mL of ethanol which was adjusted to a pH of 5 usi ng glacial acetic acid. Measuring the macroscopic contact angle of a drop on powders is challenging due to particle surface roughness and the porosity of a packed bed of powder and due to the small particle diameter, traditional particle contact an gle methods were not employed. Thus, in order to determine the optimum reac tion time to produce the most hydrophobic surface, surfaces were produced by chemical vapor deposition of alum inum onto polished silicon wafers were used which is more conducive to contact angle measurem ents. 1 cm x 1 cm square samples were cut from larger wafers and the samples were reac ted with ODTMS and TDFS as described above. Samples were stored in a clean and dry desiccator and prior to contact angle measurement, samples were rinsed with methanol and wa ter and then dried in a clean drying oven. High purity water (Barnstead Nanopure Infinity, 18 M .cm-1) was used to measure advancing and receding contact angl es with a Ram Hart contact angle goniometer as described previously and contact angle imag es were captured with a digital camera. Contact angles were
86 determined from digital images using the ImageJ software package (Version 1.36b). All contact angle measurements were performed at 22C, atmospheric pressure and 50% RH. Figure 5-27 shows examples of digital images used for cont act angle measurements of bare aluminum, A, and aluminum surface modified with ODT MS and TDFS, B and C respectively. The results for reaction time can be seen in Fi gure 5-28, a plot of the change in advancing and receding contact angle of water on the ODT MS modified aluminum surface at varying reaction times. The contact angle on the bare al uminum surface is represented at time zero with a contact angle of 673o. The advancing contact angle is represented by the solid line and measurement points are indicated by the closed diamonds. As the aluminum surface was mixed in the solution, an increase in cont act angle was seen and a maximum of 1141o obtained at one hour. After this point, a plateau is reached and the c ontact angle remained constant up to 12 hours of mixing after which there was a decrease. This is expected, as with time a bi-layer or other surface complex begins to form causing pola r end groups to be expos ed to the environment as opposed to the hydrophobic non-polar ends. The receding contact angle is represented by the dashed line and the correspondi ng measurement points are indicat ed by the open circles. The receding contact angle followed a similar trend to that observed by the advancing measurements. The change in the contact angle of water on surfaces modified w ith TDFS is shown in Figure 5-29. Again, the bare alum inum, with a contact angle of 673o is seen at time zero. The advancing contact angle is shown by the solid line and closed circ les to rapidly increase to a maximum of 1113o after one hour. As with the ODTMS, a plateau is reached between 1 and 12 hours after which a decrease in contact angle is seen. The reced ing contact angle, represented by the dashed line and open circles, also show s similar trends to that of the advancing measurements.
87 Contact angle measurements were also used to calculate the work of adhesion between the surface and a water droplet, shown in Table 51. The effect of the hydrophobic silane surface coating on the work of adhesion is most pronounced in the advancing measurements where on an untreated surface, the work of adhesion was 101 mJ/m2, and in comparison, the work of adhesion on the treated aluminum surfaces was 43 and 44 mJ/m2 for TDFS and ODTMS respectively. Similar decreases were observed in the wo rk of adhesion calculated from receding measurements. The advancing and receding work of adhesion ar e both important to consider with respect to the agglomeration and separa tion between two surfaces where c ondensed liquids are present. Hydrophobic surfaces with low contact angle hyst eresis exhibit lower values of work of adhesion. Thus two hydrophobic surfaces joined by a liquid bridge would be less likely to agglomerate by capillary forces and therefore shou ld also be easier to separate. Hydrophobic particles under humid conditions should therefore display lower values for bulk powder strength compared with hydrophilic particles and also be easier to disperse back to their primary particle size. Based on the preceding results, the aluminum flakes were surface modified with ODTMS for one hour, rinsed by dilution and centrifugatio n, and supercritical dr ied. Figure 5-30A-C shows examples of digital imag es taken of a water drop on a la yer of the aluminum flakes deposited on a glass slide. Figure 5-30A show s the contact angle of a water drop on a bed of bare aluminum flakes, Figure 5-30B shows a wate r drop on a bed of aluminum flakes modified with ODTMS, and Figure 5-30C shows a water drop on TDFS modified flakes. It is important to note that due to the particle surface roughness an d flake bed porosity, these measurements do not
88 represent the contact angle of water on the flake but instead are the effect ive contact angle of a similarly treated flat surface. After drying, all three samples of the flakes were conditi oned at 50% relative humidity for 48 hours prior to dissemination in the air chamber. Liqui d carbon dioxide was used to disseminate the flakes for measurement. The airborne concentration of the bare and ODTMS surface modified flakes at 50% relati ve humidity disseminated using liquid CO2 can be seen in Figure 5-31. No significant change in airborne concentration is obs erved. To further investigate this, bare aluminum flakes were conditioned at 20, 50, and 80% relative humidity for 48 hrs then disseminated into the air chamber using liquid CO2. As seen in Figure 5-32 no significant change in airborne concentration or settling rate can be seen w ith increasing humidity. This led to the conclusion that using liquid carbon dioxi de to disseminate the powder overwhelmed any effect of prior powder conditioning on dispersion. Design of Experiments The order of the airborne concentration e xperiments performed above was determined using Design Expert 6 software package. Th e results can be seen in Figure 5-33, a 3D representation of the design space. Each side of the cube represents one experimental factor with the corners signifying the minimum and maximum levels. The airborne concentration obtained from each level of each set of th e three factors is designated by the number at each corner inside the cube. From these results, the best results can be seen using liquid CO2, 5 weight % silica, and using surface modification. Based on the data, a model described by Equation 5-1 was developed by the software. Airborne Concentration = 73.55 +0.70A (Medium) -1.80B(Surface Modification) +3.95C (Spacer) +6.80AC +1.80BC (5-1)
89 In Equation 5-1, the factors that have a signifi cant effect on airborne concentration are shown along with the magnitude of their e ffect by the factor coefficient. The standard deviation for the coefficients was 0.6. Half the coefficient value is the f actor effect. It is seen that the spacer and the interaction between the spacer and the me dium have the largest effect on dispersion. Using this model, a maximum airborne concentra tion can be calculated based on different levels of these parameters. Within this design space, a maximum air borne concentration of 85% is predicted and experimentally obtained. Up to this point experiments have been perf ormed to investigate th e effect of important parameters on the particle size a nd airborne concentration. The effect of chosen parameters on interparticle forces has been calculated and compared to the e xperimental results. In order to obtain a better understanding of the effect of these factors on the interparticle forces measurements of the macro and micro scale must be performed. To do so, bulk properties were studied using the shear cell and interparticle force measurements using the atomic force microscope. A comparison of the change in magnitude of the interparticle forces as they relate to these factors results in a co rrelation between the significant interparticle forces and their relationship to dispersion is di scussed in the next section. Interparticle Forces on th e Macro and Micro Scale In order to develop the knowledge base about dispersion that will apply, in general, to many different powder systems, an understanding of the changes in a powder system that occur following processing is required. The above re search investigated important factors of dispersion, from which an expres sion of their effect on disper sion was developed. Now, the interparticle forces on both the macro-scale, the forces of powder as a continuum, and on the micro-scale, forces between individual particles, will be investigated. From this work,
90 knowledge of the change in interparticle forces w ill be developed. The correlation of the results of these two aspects will lead to an understanding of why processing parameters have an effect on dispersion, from which an extension to other powder systems will be possible. Bulk Powder Strength Most aerosol applications begi n with the powder being drawn fr om a reservoir in the bulk state before forming the aerosol. Knowledge of powder as a continuum, specifically macro-scale powder properties such as strength and flow, is therefore important to optimizing dispersion. In order to better understand the cohesi ve behavior of powder due to th e effect of capillary forces, a monodispersed spherical aluminum powder was test ed in a bulk powder shear tester, the Sci-Tec ShearScan, with and without hydr ophobic surface modification at varying rela tive humidity values. Experimental methods The aluminum powder was divided into three samples. Two of the powder samples were modified by ODTMS and TDFS using the paramete rs determined previously, after which both powder samples were supercritical dried. The third sample was left in the as received state with a minimal native aluminum oxide surface. The powder was conditioned for 48 hours at a constant relative humidity after which the powder st rength was measured using the shear tester. Preconditioning of aluminum powders for shear testing was performed inside a humidifying chamber. Samples were placed into the chamber and humid air was pumped through the chamber for 48 hours prior to meas urements to allow the powders time to equilibrate. Humidity was c ontrolled by pumping dry, compre ssed air through a water trap and by controlling the temperature of the water. Humidity levels were controlled to within %. Approximately 30 cm3 of conditioned powder was loaded in to the annular shear cell inside the humidity chamber and then mounted on the shear tester. Samples were sheared under normal
91 consolidation loads of 2.5, 5, 10, 20 and 30 kPa. The linear rotational speed of the shearing lid was maintained at 5 mm/min. Outlying data poin ts, greater than one stan dard deviation, were removed from the yield locus and the unconfined yield strength and maximum principal stress recorded and plotted as a powder flow function. Shear test results The results for unconfined yield strength, fc, versus maximum principal stress, 1, at the different relative humidity levels have been plo tted in Figure 5-34, Figur e 5-35, and Figure 5-36 In each graph the bare aluminum is represented by squares, the ODTMS modified aluminum is represented by circles, and that of TDFS by triangles. Measurem ent points are designated by the markers. Figure 5-34 shows the results for fc after conditioning at 20 pe rcent relative humidity. At this condition, all three powders have the same unconfi ned yield strengt h, a maximum fc of 12kPa at the highest normal load of 30kPa. The cohesion in this case is dominated only by van der Waals attractive forces and frictional forces between the pow der particles. Figure 5-35 shows the powder flow function measured for the aluminum powders treated at 50 percent humidity. It can be seen there that the strength of the bare aluminum increased at the higher consolidation loads, 37%, to a maximum of 18kPa, while the surface modified aluminum powders, which each had the same strength, increa sed 20% to a maximum of 15kPa. The results of fc vs. 1 for the three powders at 80 percent humid ity are seen in Figu re 5-36. The powder strength of the bare aluminum again increa ses an additional 25 percent to 26kPa while no significant increase in the strength was seen for the surface modified powders. In order to better understand the effect of relative humidity on each of the powders themselves, the unconfin ed yield strength, fc, versus major consolidation stress, 1, for each powder at the three relative humidity levels were compared. The shear testing results on bare aluminum powder is shown in Figure 5-37. Th e powder strength at the three measured humidity
92 levels is within experimental error at principal stresses below about 20kPa. Principal stresses above 20 kPa show differences in powder behavior with respect to relative humidity. There is a 40 percent increase in strength wh ich is a result of increasing the relative humidity from 20 percent to 50 percent humidity. An additional 25 percent increase is observed as the relative humidity was increased to 80 percent. Hence it is shown that the powder strength is directly affected by relative humidity. The mechanism fo r the increase in particle strength is the formation of pendular liquid bonds between particles that cause capillary forces adhering particles together. As a result of the increase in interparticl e forces, a corresponding increase in cohesion is seen. Shear test discussion The results for the aluminum powder modified with TDFS are seen in Figure 5-38. There is a 25% increase in strength as humidity is increased from 20 to 50 percent; however the subsequent increase in relative humidity does not result in a signi ficant increase of strength. The initial increase is significantly lower than that observed for the bare aluminum samples hence the hydrophobic coating has decreased the powder strength due to le ss adsorbed moisture and therefore less capillary forces acting on the particles. Similar conclusions were drawn for the ODTMS modified aluminum powder results seen in Figure 5-39. As humidity is increased there is a trend toward s slightly higher bulk strength, however the increase is not signi ficant when compared to the experimental error of 10% for these measurements. This indicates that humid ity did not significantly change the unconfined yield strength of this material and th e capillary forces were controlled. Comparing the powder flow functions of bare aluminum, Figure 5-40, to that of the TDFS and ODTMS surface modified aluminum, Figure 541 and Figure 5-42 respectively, it is seen that at 20% relative humidity, the modified and un modified powders have very similar behavior.
93 When the relative humidity is increased to 50% the unconfined yield strength for modified and unmodified aluminum, 19 kPa and 14.5 kPa respectiv ely, begins to deviate. A similar difference is seen for measurements at 80% relative humidit y where the unconfined yield stress for the bare aluminum increases while there is not signifi cant change in the surf ace modified powders. These results indicate that the hydrophobic surface coatings directly influenced the mechanical strength measured on bulk powders by reducing adsorbed moisture on the surface of the particles. From the data presented above, it can be seen th at the use of surface modification can be successfully used to maintain the unc onfined yield strength of a spherical aluminum powder in an environment of high relative humidity. The powder modified with ODTMS showed no significant change in strength unde r any conditions. Even at the high relative humidity levels a significant amount of water did not adsorb onto the hydrophobic surface. The powder modified with TDFS showed a slight incr ease in strength from 20 to 50 percent relative humidity but no further change in strength from 50 to 80 percent relative humidity. This implies that the powder is not completely hydrophobic, ye t humidity can still affect the bulk strength. This could be due to incomplete modificat ion of the surface or because of the powder morphology, roughness, or partic le size distribution. After adding these results into the Design Expert software, a model is fit to the data. From the model, a contour plot of the results with in the design space can be made. The plot for the results of the shear tests are s hown in Figure 5-43. From the pl ot it is seen that surface modification reduces the adhesion force at higher humidity levels, which verifies the results described for the shear cell above. Interparticle Forces The adhesion force measurements are now ta ken one step further from the macro-scale down to the micro-scale to inves tigate the magnitude of the change in interparticle forces as a
94 function of surface modification and relative humid ity. This set of experiments involved the measurement of particle adhesion force and fo llowed the methods described by Ducker  and Rabinovich . Experimental methods The first attempt to measure interparticle forc es was by using silicon cantilevers that were coated with aluminum by chemical vapor depos ition (CVD). The cantilevers themselves were then surface modified such that each combination of tip and surface could be investigated. Using the CVD process, however, the tip diameters co uld not be controlled so that the force measurements would be difficult to interpret. In addition, after inspection of the tips after the first set of measurements usi ng the transmission electron micr oscope, it was found that the aluminum coating was soft and easily damaged du ring the experiments. The force exerted on the tip during the measurements was enough to actua lly remove the aluminum coating and expose the underlying silicon nitride tip. This can be seen in Figure 5-44, a TEM image of one such cantilever tip. Therefore, for remainder of the measurements, particle probes were employed. Each particle probe was made by mounting eith er a modified or ba re Valimet aluminum particle, as used for the shear tests, to a tip less silicon cantilever employing a standard quick set epoxy. Figure 5-45 shows an SEM image of a bare aluminum particle m ounted to the cantilever and Figure 5-46 shows the same for an ODTMS surface modified aluminum sphere. These particles, however, are near the same size as the flakes. During force measurement there was a good chance they would lift the flakes off the substrate and hinder force measurement. Therefore, the CVD aluminum surfaces made dur ing the surface modification experiments were used. Contact mode topographical images of the bare aluminum surface can be seen in Figure 547 through Figure 6-15. Images of the ODTMS surface made through the same method can be seen in Figure 5-48 and Figure 5-49. Using the AFM, the root mean squared (rms) surface
95 roughness for the bare aluminum and ODTMS modi fied aluminum was measured to be 1.6nm and 1.8nm respectively. For each force measurement, the sphere attach ed to a cantilever of known spring constant, was positioned close to the flat substrate. Th en, as the substrate and particle were moved towards and away from each other by a piezoel ectric, the deflection of the cantilever as it interacted with the surface was monitored by a lase r that reflects off the top of the cantilever onto a position sensitive photodiode. In this manner, th e force between the two surfaces as a function of separation distance was obtai ned. The force-distance data presented can be normalized by dividing the measured force by th e radius of the sphere. The result is energy per unit surface area and thereby allows a straight forward comparison of the sphere -surface interaction forces to different geometries such as flakes. A typical force-distance plot is given in Figure 5-50. As the particle and surface approach each other, the onset of interac tion is shown by the jump-to-surface point. Once the particle is in contact with the surface, the cantilever bends slightly as seen by the upward slope of the line. After a maximum deflection equal to 5nN, the so ftware triggers the piezoelectric to change direction and the particle and surf ace move apart. During retraction, the point of particle release is seen by the jump-off-surface point. From the deflection distance and cantilever spring constant the force of inte raction is calculated. The environment around the sample was controlle d through the use of a humidity chamber, also by Asylum Research, which has a sensor to monitor the internal temperature and relative humidity. The standard room hu midity varied between 45 and 50% as measured by the device. The humidity was decreased below this point by fl owing dry compressed nitrogen at a controlled rate through the chamber. To raise the humidity, the nitrogen wa s first bubbled through water
96 then to the chamber. By this method, the hum idity could easily be cont rolled between 5% and 85% and within 0.5% relative humidity of the target valu e. The change in interparticle forces over a range of humidity levels from 10% to 80% were measured for modified and bare aluminum. Interparticle force results and discussion For the bare aluminum particle and bare aluminum surface, the measured forces over changing humidity are seen in Figure 5-51. In a dry state, between 10% and 20% relative humidity, the interparticle fo rces are seen to be 83560nN and 80060nN respectively. This value can then be compared to that calculated using the general van de r Waals equation for the force of interaction between a smooth sphere and plate from Table 3-1, wh ich yields a force of interaction of 888nN. It should be noted that the contact angle was measured on the CVD aluminum surface with 1.6nm rms surface roughness. Taking into account that the real material has surface roughness and other imperfections that w ill lower the adhesion force as discussed in Chapter 3, the values are favorably comparable. After 20% relative humidity, the cr itical humidity is reached. It is at this point that enough water is adsorbed onto the surfaces that a me niscus forms between them and the onset of capillary forces between the surf ace and particle can be seen. To verify the data, the capillary forces can be calculated using Equation 3-10, the average particle size of 5 m, and the receding contact angle of 7.8o measured on the CVD aluminum surface. The result is an interaction force of 1800nN which again is extremely close to the measured value of 1920nN at 30% relative humidity for the onset of capillary forces. Here again, the difference can be attributed to the variation between the real and ideal surface. Th e volume of water in the meniscus increases due
97 to the increase relative humidity from 20% to 40% relative humidity and accordingly so does the capillary force. These results are in agreement with the work completed by Ra binovich et al.[55, 91] that there is a critical humidity to the onset of capillary forces a nd this value is proportional to roughness. Below this critical point there is no change in in terparticle forces, which are dominated by van der Waals attraction. At the critical humidity, enough water is adsorbed onto the surface that a meniscus can form and along with it capillary forces. The forces will increase with the volume of the meniscus. Measurements above 40% relative humidity were not possible because the capillary forces exce eded the spring constant of th e cantilever. Upon retraction, excess cantilever deflection leads to saturation of the photodiode and an incomplete force measurement. Despite this complication, the ra nge of the van der Waals forces, the critical humidity required for the formati on of a meniscus, and capillary fo rces have been calculated and verified by the AFM measurements. Next, the effect of surface modification on in terparticle forces was investigated. Since both silane coupling agents showed similar result s, only ODTMS was used for the experiments. The results can be seen in Figure 5-52, a plot of adhesion force betw een an ODTMS modified particle and ODTMS modified surface. At low relative humidity, the adhesion force is significantly lower than that seen for the bare aluminum surfaces in Figure 5-53, 14nN for ODTMS as opposed to 800nN for the bare aluminum despite a similar particle size. The surface energy of the silane modified material is much lo wer than that of the bare aluminum as seen by the contact angle measurements made previous ly. Bare aluminum had an advancing water contact angle of 67o whereas the ODTMS modified surface has an advancing contact angle of 114o. These values were used to calculate a work of adhesion, WA, of 101 J/m2 for the bare
98 aluminum surface versus 43 J/m2 of the modified surface. A d ecrease in the work of adhesion means less energy is needed to separate the materi als and indicants a change in the interparticle forces. This can be due to two different reasons. Fi rst, the ODTMS has a lower work of adhesion and therefore will decrease the value of the effective Hamaker constant below that of bare aluminum and in turn lead to a de crease in interparticle forces. A second effect of the coating is creating an increased separation distance between the aluminum particles themselves, and using the new, increased separation distance in the va n der Waals equation from Table 3-1, a value of 24nN is calculated. This value is quite compar able to the measured value of 15nN using the AFM. From this it can be concluded that it is both of these effects of surface modification that contribute to the decrease in interparticle forces. Upon increasing the relative humidit y, an increase in adhesion force is seen. This trend is also comparable to that observed in the c ontact angle and shear cell measurements made previously. The contact angle measurements show that the ODTMS does not create a completely hydrophobic coating and a slight increas e in bulk strength was seen in the bulk measurements. To check the validity of thes e values, Equation 3-10 can again be employed. Using the receding contact a ngle reported above of 68.5o and the average particle size of 5 m, an interparticle force of 415 nN is calculated. This value is in agreement with the onset of capillary forces measured here to be 460nN at 40% rela tive humidity. The maximum increase in adhesion force for ODTMS of 770nN at 80% relative humidity is less that the observed value for the bare aluminum or 3700nN at the same relative humidit y. This demonstrat es that the surface modification changes the material hydrophobicity as well as control interparti cle forces in the presence of humidity.
99 By dividing by the radius of the particle, the interaction energy per unit area can be calculated. This was done and repo rted in the graph in Figure 5-54, a plot of the interaction force divided by particle radium versus relative humidit y. From this information, the energy of interaction of surfaces of different sized can be compared. This data can now be used to show that even for flakes, surface modification will reduce the van der Waals forces and improve dispersion. Again the data was added into Design Expert and the ensuing contour plot developed also confirms these results. As seen in Figure 555, the minimum adhesion force at all relative humidity levels are found when the su rface modified materials are used. Data Correlation The final step of this research was to take the experimental results for the dispersion experiments and correlate them to the change in ma gnitude of the interparticle forces. To aid in this, Design Expert 6, a statistical design software package, was used. During the analysis of the results from each set of experiments, contour plot s of the results were made. By overlaying these plots the optimum levels of the factors for a set of experiments can be compared to those of the other experiments. Figure 5-56 shows the overlay for the shear cell and AFM results. From here it can be seen that for both the interparticle fo rces and the bulk forces the optimum conditions, lowest adhesion, require surface modification. This overlay can also be combined with the experimental cube developed a bove. By aligning the surface m odification axis, thereby taking into account the optimum bulk forces and interpar ticle forces, the optimum results to satisfy all conditions will lie in the upper half of the cube which is shaded in Figure 5-57. The optimum results for airborne concentra tion, is found in the upper righ t corner of the design space corresponding to a surface modi fied powder, with 5 weight percent spacer, using liquid CO2 dissemination. It was determined, based on the shear cell measurements, that these conditions
100 will have the lowest bulk strength; based on the AF M to have the lowest interparticle forces; and thus as seen in the airborne concentra tion, to yield the hi ghest dispersion.
101 Figure 5-1. Particle size distribution of alum inum flakes. The graph of volume weighted equivalent spherical diameter, particle si ze, as measured by laser light scattering, shown a mean particle size of 10microns, represented by the curve labeled 1. The volume weighted particle diameter meas ured by image analysis shows a mean particle size of 17microns and is represented by curve 2. Figure 5-2. Particle size distribution of Valimet aluminum in isopropyl alcohol. The mean particle size is 4 microns. 0 1 2 3 4 5 6 0.1110100Equivalent Spherical Diameter (um)Differential Volume % Valimet Al in IPA 0 2 4 6 8 10 110100 Equivalent Spherical Diameter (um)Differential Volume % Image Analysis Laser Diffraction Equivalent Spherical Diameter ( m) Equivalent Spherical Diameter ( m) 1 2
102 Figure 5-3. SEM images of Sigma VPD Aluminum flakes. A) is an image of the Sigma 1215 flake face. B) shows the edges of a stack of Sigma 1215 flakes. The 15nm flakes are so thin that they are partia lly transparent to the SEM as seen by the faint flake faces visible in B. Figure 5-4. TEM images of Sigma 1215 alumi num flakes. Transmission electron microscope images taken by Kerry Siebein at the Major Analytical Instrument ation Center at the University of Florida. A)shows that the al uminum is polycrystalline. In B) the thin oxide layer can be seen. A B A B A B A B
103 Figure 5-5. Contact mode topograp hical images of aluminum flak es. The contact mode of the AFM was used to make topographical images of the aluminum flakes deposited on a silicon substrate. The irregular shape of the flakes and infrequent large scale roughness are indicative of the milling process. Figure 5-6. Detailed topographical image of flake surface. These contact mode topographical AFM images show the flakes themselves to have a very uniform and smooth surface. The roughness was measured to be 4nm rms.
104 Figure 5-7. SEM Image of Valimet aluminum spheres. Atomization produces a spherical powder; however the particle size di stribution is seen to be large. Figure 5-8. Phase diagram for carbon dioxide as related to supercritical drying. The thermodynamic path taken during supercritical drying is depicted by the lettered path. The process starts with CO2 as a liquid (A). It is heat ed above the critical point (B) and then released as a gas (C). For co mparison, standard room temperature and pressure is represented by the dot. ValimetAl ValimetAl P (psi)Solid Liquid GasSupercritical FluidT (oC ) A B C 1072 31 15 20 75 -56 800 P (psi)Solid Liquid GasSupercritical FluidT (oC ) A B C 1072 31 15 20 75 -56 800
105 Figure 5-9. Particle si ze distribution measured following each drying method. The curve labeled 4 shows the results for evaporation, with the largest amount of agglomeration due to capillary forces that pull th e particles together as the liquid evaporates. Freeze drying is represented by curve 3 and is the s econd most agglomerated powder. During freeze drying, capillary forces can also form if the pressure in not sufficiently low and the ice starts to melt before sublimation. Supercritical drying, shown by curve 2, completely avoids the gas-liquid phase boundary consistently producing the best results in the research. The primary particle size distribution of th e material is shown for comparison and represented by curve 1. Figure 5-10. Contact area w ith increasing roughness. Roughness has a large effect on interparticle forces due to its connection to co ntact area. As seen in this schematic, as the roughness of the surfaces in crease, the contact area between them decrease. van der Waals forces are also decreased becau se with increasing roughness, the average surface plain of each body moves further away. Decreasing contact area with roughness Decreasing contact area with roughness 0 1 2 3 4 5 6 7 8 9 0.11101001000 Equivalent Spherical Diameter (um)Differential Volume % As Rec SCDrying Freeze Dried IPA Oven 1 2 3 4
106 Figure 5-11. Calculated van de r Waals forces of a smooth and rough surface. Surface roughness has a large effect on interparti cle forces. The line 1 show s the calculation of van der Waals attraction for two smooth surfaces. Line 2 shows the effect of surface roughness incorporated by the PERC model[ 53, 54] which not only decreases the contact area, but also separa tes the average surf ace plane of each body. A significant decrease in the calculated interaction force is seen. Figure 5-12. Illustration of the change in contact area as a result of the addition of fumed silica. The particle surface cannot be changed to reduce contact area; however, fumed silica can be added to simulate the effect. 0 1 2 3 4 5 6 7 8 9 10 0102030405060 Separtation Distance (nm)Interaction Force (mN) Infinite Flake PERC Model 0 1 2 3 4 5 6 7 8 9 10 0102030405060 Separtation Distance (nm)Interaction Force (mN) Infinite Flake PERC Model 1 2
107 Figure 5-13. Calculated van der Waals forces with the addition of silica. A comparison of the calculated van der Waals forces for two smooth flakes is seen to be much higher than for two smooth flakes separated by sili ca spacer material. The line 1 shows the calculated van der Waals force for two smoot h disks and line 2 shows the calculation for two disks separated by five weight pe rcent added 10nm fumed silica particles. Line 3 shows the calculation when ten weight percent silica is added. The spacer has the same effect as increa sing surface roughness, but done w ithout having to alter the flakes themselves. The Hamaker constant of silica is also lower than that of aluminum which also cont ributes to a decrease in van der Waals forces. 0 1 2 3 4 5 6 012345678910 Separation Distance (nm)Interaction Force (mN) Finite Flake Flake plus 5wt% SiO2 Flake plus 10wt% SiO2 0 1 2 3 4 5 6 012345678910 Separation Distance (nm)Interaction Force (mN) Finite Flake Flake plus 5wt% SiO2 Flake plus 10wt% SiO2 1 2 3
108 Figure 5-14. SEM images of aluminum flakes with silica. In A) the edge of one flake, shows the smooth surface obtained from Physical Vapor Deposition. B) The aluminum/silica mixture at low silica concentrations. This silica can be seen around each flake. C) The aluminum/silica mixture at high silica con centrations. Note that the distribution of silica between the flakes incr easing the interparticle spacing. Figure 5-15. Air Chamber transmission measurement. In order to determine the settling rate of the materials, the log of inverse transmi ssion is plotted over time on a log scale for each sample. A trend line is then added. y = -0.0516Ln(x) + 0.3743 0 0.05 0.1 0.15 0.2 0.25 0.3 110100100010000 Time (s)Log(1/T) AL Flake Settling Log. (AL Flake Settling) 1 3 2 A C B
109 Figure 5-16. Average and standard deviation of settling rate. The trend line equation for each sample is averaged and the standard deviat ion is calculated in order to determine the statistical significance between factors. From this transmission data and the concentration of powder disseminated, the Beer Lambert Law, Equation 4-1, can be used to calculate the airborne concen tration and its change over time. Figure 5-17. Airborne concentratio n at selected times after partia l vacuum dispersion. Using the Beer Lambert Law, Equation 4-1, and the mass of powder disseminated, and the transmission at chosen times, the airborne concentration can be calculated and used to compare factors and their effect on disper sion. Transmission decreases due to an increased number of particles present to scatter the light; therefore, increasing airborne concentration is attri buted to increased dispersion. 0.00 0.05 0.10 0.15 0.20 0.25 0.30 110100100010000 Time (sec)log(1/T) Average Al settling +SD -SD 0 10 20 30 40 50 60 70 80 90 110100 Time (sec)Airborne Concentration (%) AL Concentration
110 Figure 5-18. Airborne concentration by vacuum dissemination with sili ca. The addition of 5 weight percent silica to the aluminum is seen to increase dispersi on ten percent. This increase is due to the decrease in contact area between flakes resulting in lower adhesion forces. Figure 5-19. Dissemination pressure and airborne concentration. Increasing pressure leads to an increase in airborne conc entration because of the larger force applied to the agglomerated and the reduction in the size of the turbulent eddy size. 0 10 20 30 40 50 60 70 80 90 110100 Time (sec)Airborne Concentration (%) Al by Vac Al+SiO2 by Vac 0 10 20 30 40 50 60 70 80 90 100 110100 Time (sec)Airborne Concentration (%) +5psi +20psi +50psi +70psi +90psi +800psi
111 Figure 5-20. Airborne concentration by high pr essure gas dissemination with silica. The addition of 5 weight percent silica to the aluminum decreases the airborne concentration when 800psi high pr essure gas is used to disseminate the material. This is expected to be due to the inability of the gas to effectively impart enough force on the agglomerates. Figure 5-21. Illustration of th e dispersion effect of liquid ca rbon dioxide. Upon release, the liquid carbon dioxide will unde rgo a phase transformation from a liquid to a gas resulting in an increase in volume of appr oximately 250 times. The high pressure of liquid carbon dioxide and th e radial expansion of th e liquid during the phase transformation will increase the amount of force applied to the agglomerate. Drag force + Force caused by radial pressure gradient 0 10 20 30 40 50 60 70 80 90 110100 Time (sec)Airborne Concentration (%) Al by Vac Al+5% SiO2by Vac 0 10 20 30 40 50 60 70 80 90 110100 Time (sec)Airborne Concentration (%) Al by Vac Al+5% SiO2by Vac
112 Figure 5-22. Airborne concentration produced by liquid carbon dioxi de dissemination. Figure 5-23. Airborne concentration fo r gas vs. liquid carbon dioxide. Gas CO2 dispersion is seen to be higher than liquid CO2 dispersion. This is attributed to the non instantaneous phase change from liquid to gas CO2 resulting in the formation of capillary forces between the flakes. 0 10 20 30 40 50 60 70 80 90 100 110100 Time (s)Airborne Concentration (%) Liquid CO2Dispersion 0 10 20 30 40 50 60 70 80 90 100 110100 Time (s)Airborne Concentration (%) Liquid CO2Dispersion 0 10 20 30 40 50 60 70 80 90 100 110100 Time (s)Airborne Concentration (%) Liquid CO2Dispersion Gas CO2Dispersoin 0 10 20 30 40 50 60 70 80 90 100 110100 Time (s)Airborne Concentration (%) Liquid CO2Dispersion Gas CO2Dispersoin
113 Figure 5-24. Airborne concentra tion using liquid carbon dioxide and silica. The addition of five weight percent silica has previ ously been shown to reduce the van der Waals forces between flakes by increasing the interpar ticle spacing. This also incr eases the volume of liquid carbon dioxide between the flakes and in turn the amount of pressure that can be generated upon dissemination. Figure 5-25. Summary of airborne concentration by carbon dioxide dispersion. The graph shows the airborne concentration produced by liquid and gas CO2 and the addition of five weight percent silica a dded as a spacer. Liquid CO2 with the five weight percent spacer yields the highest airborne concentrati on of eight five percent and thus the best dispersion. 0 10 20 30 40 50 60 70 80 90 100 110100 Time (s)Airborne Concentration (%) Al in Liquid CO2 Al+5wt% silica in Liquid CO2 0 10 20 30 40 50 60 70 80 90 100 110100 Time (s)Airborne Concentration (%) Al in Liquid CO2 Al+5wt% silica in Liquid CO2 0 10 20 30 40 50 60 70 80 90 100 110100 Time (s)Airborne Concentration (%) Gas CO2Dispersion 5wt% silica + Gas CO2 Liquid CO2Dispersion 5wt% silica + Liquid CO2 0 10 20 30 40 50 60 70 80 90 100 110100 Time (s)Airborne Concentration (%) Gas CO2Dispersion 5wt% silica + Gas CO2 Liquid CO2Dispersion 5wt% silica + Liquid CO2
114 Figure 5-26. Particle size of aluminum flakes at 20, 50, and 80% relati ve humidity. The graph represents change in particle size of the aluminum flakes after supercritical drying and conditioning for 48 hours at 20% relative humidity, line 2, 50% relative humidity, line 3, and 80% relative humidity, line 4. Line 1 represents the primary particle size measured in isopropyl alcohol and provided as a comparison. The standard deviation for the measurement is 3 m. The change in particle size and distribution due to changes in the relative humidity can readily been seen. Figure 5-27. Digital image of water drops on alum inum as used for measurement. Contact angle was measured by taking digital images of water drops on the surfaces at varying reaction times. A) A water drop on a bare aluminum surface at time equal to zero with a contact angle of 67o. B) A water drop on an aluminum surface modified by ODTMS for 1hr and C) A water drop on an aluminum surface modifiec by TDFS for 1 hr. 0 1 2 3 4 5 6 7 8 9 0.11101001000 Equivalent Spherical Diameter (um)Differential Volume % In IPA 20% 50% 80% Al + TDFS 110o Al + TDFS 110o Al + ODTMS 114o Al + ODTMS 114o Bare Al 67o Bare Al 67oA C B Al + TDFS 110o Al + TDFS 110o Al + ODTMS 114o Al + ODTMS 114o Bare Al 67o Bare Al 67o Al + TDFS 110o Al + TDFS 110o Al + ODTMS 114o Al + ODTMS 114o Bare Al 67o Bare Al 67oA C B 1 2 3 4
115 Figure 5-28. Contact angle on ODTMS surface modi fied aluminum. The change in the contact angle of water on an alumin um surface modified with n-octadecyltrimethoxysilane (ODTMS) after varyi ng reaction time. Based on thes e results, one hour was chosen as the optimum reaction time. Figure 5-29. Change in contact angle of TDFS surface modified aluminum. The change in the contact angle of water on an aluminum surface modified tridecafluoro-1,1,2,2tetrahydrooctyl-1-triethoxysilane (TDFS) w ith varying reaction time. Based on these results, one hour was chosen as the optimum reaction time. 0 20 40 60 80 100 120 02004006008001000120014001600 Time (min)Contact Angle (theta) TDFS Adv TDFS Rec 0 20 40 60 80 100 120 02004006008001000120014001600 Time (min)Contact Angle (theta) ODTMS Adv ODTMS Rec
116 Figure 5-30. Digital image of water drop on modi fied flake bed. Contact angle was measured on the surface modified alum inum flakes by taking digital images of water drops on packed bed of flakes at varying reaction times. A) A water drop on a bed of bare aluminum flakes with a contact angle of 62o. B) A water drop on an bed of aluminum flakes modified by ODTMS for 1hr with a contact angle of 109o and C) A water drop on an aluminum surface modified by TDFS for 1 hr with a contact angle of 73o. Figure 5-31. Airborne concentration of surface m odified aluminum flakes at 50% humidity. The plot shows the airborne c oncentration of bare alumin um flakes, blue bars, and ODTMS surface modified aluminum flakes, red bars. Surface modification does not improve dispersion over liquid CO2 dissemination alone. 62o109o73oBare AlAl + ODTMSAl + TDFSA C B 62o109o73o 62o109o73oBare AlAl + ODTMSAl + TDFSA C B 0 10 20 30 40 50 60 70 80 90 100 110100 Time (s)Airborne Concentration (%) Bare Al Al+ODTMSLiquid CO2Dissemination 0 10 20 30 40 50 60 70 80 90 100 110100 Time (s)Airborne Concentration (%) 0 10 20 30 40 50 60 70 80 90 100 110100 Time (s)Airborne Concentration (%) Bare Al Al+ODTMS Bare Al Al+ODTMSLiquid CO2Dissemination
117 Figure 5-32. Airborne concentration of bare aluminum at 20, 50, and 80% humidity. Using liquid CO2 dissemination, the usual effect of increasing relative humidity on increasing agglomeration is no longer a factor due to the improved dispersion method. Figure 5-33. Experimental design dispersion results. The figure shows the results for airborne concentration for all of the experiments used to investigate the effect of humidity, spacer, and dispersion medium. The condition s for optimum dispersion can be seen to result from 5 wt% spacer and liquid CO2. Dispersion MediumHumiditySpacer Gas Liquid 20% 80% 0 wt% 5wt% 79 71 77 68 65 85 65 85 Dispersion MediumHumiditySpacer Gas Liquid 20% 80% 0 wt% 5wt% 79 71 77 68 65 85 65 85 0 10 20 30 40 50 60 70 80 90 100 110100 Time (sec)Airborne Concentration (%) 20% RH 50% RH 80% RH 0 10 20 30 40 50 60 70 80 90 100 110100 Time (sec)Airborne Concentration (%) 20% RH 50% RH 80% RH
118 Figure 5-34. Unconfined yield strength versus major consolidation stress at 20% relative humidity. The plot shows the unconfined yield strength, fc, versus major consolidation strength, 1, for bare aluminum powder, diamonds, and aluminum powder surface modified with OTDMS, circles, and TDFS, squares, at 20 percent relative humidity. van der Waals and frictional forces dominate the interparticle forces and no significant difference in fc is seen. 0 5 10 15 20 25 30 0102030405060 Major Consolidation Stress, 1(kPa)Unconfined Yield Strength, fc (kPa) Al Al+TDFS Al+ODTMS 0 5 10 15 20 25 30 0102030405060 Major Consolidation Stress, 1(kPa)Unconfined Yield Strength, fc (kPa) Al Al+TDFS Al+ODTMS
119 Figure 5-35. Unconfined yield strength versus major consolidation stress at 50% relative humidity. The plot shows the unconfined yield strength, fc, versus major consolidation strength, 1, for bare aluminum powder, squares, and aluminum powder surface modified with OTDMS, circles, and TDFS, triangles, at 50 percent relative humidity. A significant increase in strength of 37% is seen in the bare aluminum, while a smaller increase of 17% is seen for the surface modified powders. Capillary forces in the liquid pendulums between pa rticles increase the adhesion force and in turn the unconfined yield strength. 0 5 10 15 20 25 30 010203040506070 Major Consolidation Stress, 1 (kpa)Unconfined Yield Strenght, fc (kpa) Al Al+TDFS Al+ODTMS
120 Figure 5-36. Unconfined yield strength versus major consolidation stress at 80% relative humidity. The plot shows the unconfined yield strength, fc, versus major consolidation strength, 1, for bare aluminum powder, squares, and aluminum powder surface modified with OTDMS, circles, and TDFS, triangles, at 80 percent relative humidity. A further increase in strength of 25% is seen in the bare aluminum, while no significant increase is seen for the surf ace modified powders. Capillary forces in the liquid pendulums between particles increase in the bare material as more water is adsorbed. There is no significant increas e in adsorbed water on the modified materials as seen by the consta nt unconfined yield strength. 0 5 10 15 20 25 30 01020304050607080 Major Consolidation Stress, 1(kPa)Unconfined Yield Strength, fc (kpa) Al Al+TDFS AL+ODTMS 0 5 10 15 20 25 30 01020304050607080 Major Consolidation Stress, 1(kPa)Unconfined Yield Strength, fc (kpa) Al Al+TDFS AL+ODTMS
121 Figure 5-37. Unconfined yield strength versus major consolidation stress for bare aluminum. The plot shows the uncon fined yield strength, fc, versus major consolidation strength, 1, for bare aluminum powder at 20%, 50%, and 80% humidity. An increase in strength can be clearly seen as the relativ e humidity increases. This is due to the additional capillary adhesion fo rce between the particles due to liquid adsorbed from the increased relative humidity. 0 5 10 15 20 25 30 010203040506070 Major Consolidation Stress, 1(kPa)Unconfined Yield Strength, fc (kPa) 20% Al 50% Al 80% Al 0 5 10 15 20 25 30 010203040506070 Major Consolidation Stress, 1(kPa)Unconfined Yield Strength, fc (kPa) 20% Al 50% Al 80% Al
122 Figure 5-38. Unconfined yield strength versus major consolidation st ress for TDFS modified aluminum. The plot shows th e unconfined yield strength, fc, versus major consolidation strength, 1, for aluminum powder surface modified with TDFS at 20%, 50%, and 80% humidity. A small increase in strength can be seen as the relative humidity increase from 20 -50%, however no significant increase in seen as the relative humidity is further increased to 80% These results indicate that the powder is not completely hydrophobic and due to some adsorbed water additional capillary adhesion force between the particles lead to the initial increase in strength at 50% relative humidity. Since there is not signifi cant increase in strength at 80% relative humidity, it is concluded that no further adsorption occurs. 0 5 10 15 20 25 30 010203040506070 Major Consolidation Stress, 1(kPa)Unconfined Yield Strength, fc (kPa) 20% Al+TDFS 50% Al+TDFS 80% Al+TDFS 0 5 10 15 20 25 30 010203040506070 Major Consolidation Stress, 1(kPa)Unconfined Yield Strength, fc (kPa) 20% Al+TDFS 50% Al+TDFS 80% Al+TDFS
123 Figure 5-39. Unconfined yield strength versus major consolidation stre ss for ODTMS modified aluminum. The plot shows th e unconfined yield strength, fc, versus major consolidation strength, 1, for aluminum powder surface modified with TDFS at 20%, 50%, and 80% humidity. A trend towards an increase in strength can be seen as the relative humidity increase from 20 -80%, how ever no significant increase in seen. These results indicate that the powder is hydrophobic and no significant adsorption of water occurs. By controlling capillary for ces the powder strength can be controlled. 0 5 10 15 20 25 30 010203040506070 Major Consolidation Stress, 1(kPa)Unconfined Yield Strength, fc (kPa) 20% Al+ODTMS 50% Al+ODTMS 80% Al+ODTMS 0 5 10 15 20 25 30 010203040506070 Major Consolidation Stress, 1(kPa)Unconfined Yield Strength, fc (kPa) 20% Al+ODTMS 50% Al+ODTMS 80% Al+ODTMS
124 Figure 5-40. Response plot for unc onfined yield strength. The c ontour plot from Design Expert shows the effect of humidity and su rface modification on the unconfined yield strength of the bulk powder. Surface modification can be used to control the effect of humidity. Figure 5-41. Damaged aluminum coating on cantil ever tip. The CVP alum inum coating was not hard enough for force measurements a nd was damaged after contact with the aluminum substrate. This led to th e use of aluminum particle probes. HumiditySurface Modification 20 50 80 Bare TDFS ODTMS 7.0 7.6 8.2 8.8 9.4HumiditySurface Modification 20 50 80 Bare TDFS ODTMS 7.0 7.6 8.2 8.8 9.4
125 Figure 5-42. Bare aluminum particle probe. A bare spherical aluminum particle attached to tipless silicon cantilevers can be seen in th is SEM image. A particle probe such as this one was used in order to measure interparticle forces at varying relative humidity levers. The spheres diameters were measur ed using a line analys is tool in the SEM software. Figure 5-43. ODTMS modified Al uminum particle probe. SEM image of an ODTMS modified aluminum particle attached to a tipless canti lever, one of the part icle probes used to measure the interparticle fo rces between surface modified aluminum at varying relative humidity.
126 Figure 5-44. Contact mode image of the CVD bare aluminum surface. Using a 1 m x 1 m 2D contact mode image of bare CVD alumin um the rms surface roughness was measured to be 1.6nm. Figure 5-45. Contact mode image of bare alum inum in 3D. The surface roughness of the bare CVD aluminum surface can be seen in this 1 m x 1 m 3D contact mode. 1.0 0.8 0.6 0.4 0.2 0.0m 1.0 0.8 0.6 0.4 0.2 0.0 m -1.5 -1.0 -0.5 0.0 0.5 1.0nm
127 Figure 5-46. Detailed contact mode image of bare CVD aluminum. The surface roughness of the aluminum can be seen in this 500nm x 500nm 2D c ontact mode image of bare CVD aluminum. Figure 5-47. Contact mode image of bare alum inum in 3D. The surface roughness can be better appreciated in this 500nm x 500nm 3D contact mode imag e of bare CVD aluminum. 500 400 300 200 100 0nm 500 400 300 200 100 0 nm 4 2 0 -2 -4nm
128 Figure 5-48. Contact mode image of ODTMS modified aluminum. A 1 m x 1 m 3D contact mode image of a ODTMS modified CVD aluminum surface. The average rms surface roughness for these samples was measured to be 1.8nm. Figure 5-49. Contact mode image of ODTMS alumin um in 3D. To obtain a better sense of the surface roughness, a 500nm x 500nm 3D cont act mode image of ODTMS modified CVD aluminum.
129 Figure 5-50. Typical force distance curve. As the particle and surface approach each other, represented by line 1, the onset of interac tion is shown by the jump-to-surface point. Once the particle is in contact with the su rface, the cantilever bends slightly as seen by the upward slope of the line. After a maximum defection equal to 5nN, the software triggers the piezoelectric cha nges direction and the particle and surface move apart. During retraction, represented by line 2, the poi nt of particle release is seen by the jump-off-surface point. From the deflection distance and cantilever spring constant the force of in teraction is calculated. -40 -30 -20 -10 0Interaction Force (nN) -100 -50 0 Distance (nm) -40 -30 -20 -10 0Interaction Force (nN) -100 -50 0 Distance (nm) Approach Retraction -40 -30 -20 -10 0Interaction Force (nN) -100 -50 0 Distance (nm) -40 -30 -20 -10 0Interaction Force (nN) -100 -50 0 Distance (nm) -40 -30 -20 -10 0Interaction Force (nN) -100 -50 0 Distance (nm) -40 -30 -20 -10 0Interaction Force (nN) -100 -50 0 Distance (nm) Approach Retraction1 2
130 Figure 5-51. Interparticle fo rce at humidity for bare aluminum. The graph shows the interparticle adhesion force between a bare al uminum particle and substrate. At low relative humidity, van der Waals forces dominate and no change in force is seen. As humidity increases, the amount of water adso rbed on the surface al so increases. Once enough water is present to from a meniscus at the contact point between the particle and surface, capillary forces arise leading to an increase in adhesion force. 0 500 1000 1500 2000 2500 3000 3500 051015202530354045 Relative Humidity (%)Interparticle Force (nN) Al Bare
131 Figure 5-52. Interparticle force at humidity for ODTMS modified aluminum. This graph shows the interparticle adhesion force between an ODTMS modified aluminum particle and substrate. At low relative humidity, van der Waals forces dominate. As humidity increases, the amount of water adsorbed on the surface also increases, and capillary forces arise leading to an increase in adhesion force. Because the hydrophobic surfaces created by the ODTMS modificati on are not completely hydrophobic (the contact angle was 110o much less than 180o) some water can adso rb onto the surface as seen by the increased adhesion. However, the effect of this partial hy drophobicity can be seen by only a minimal increase in adhesion forces. 0 100 200 300 400 500 600 700 800 900 0102030405060708090 Relative Humidity (%)Adhesion Force (nN) ODTMS Modified Al
132 Figure 5-53. Comparison of the effect of surf ace modification. Insp ection of this graph comparing the adhesion force for that of bare aluminum and of surface modified aluminum, the extent to which the surf ace modification decreases the effect of capillary forces can be seen. Even t hough the ODTMS does not create a perfectly hydrophobic surface, contact angle of 180o, there is still a very significant effect on interparticle forces and a much lo wer adhesion force is measured. 0 500 1000 1500 2000 2500 3000 3500 0102030405060708090 Relative Humidity (%)Adhesion Force (nN) Al Bare AL+ODTMS
133 Figure 5-54. Interaction energy and surface modifi cation. Dividing the interaction force by the radius of the particle on the probe yields th e energy of interaction per unit area. This allows the comparison of this data for spheres to that of other geometries of interest such as flakes. Figure 5-55. Response plot for a dhesion force. The contour plot from Design Expert shows the result of changing relative humidity and surf ace modification on the adhesion force between a particle and surface. Surface modification is show n to lower adhesion force even at high relative humidity. 0 0.5 1 1.5 2 2.5 3 3.5 4 0102030405060708090 Relative Humidity (%)Adhesion Force/Particle Radius (N/m^2) Al Bare Al+ODTMS Relative Humidity (%)Surface Modification205080 Bare Al TDFS ODTMS 653.7 955.6 1257.5 1559.4 1861.3 Relative Humidity (%)Surface Modification205080 Bare Al TDFS ODTMS 653.7 955.6 1257.5 1559.4 1861.3
134 Figure 5-56. Overlay of response plots. By ove rlaying the response cont our plots from both the shear cell and AFM experiments the optimum results for all conditi ons can be seen. This plot can now be compared to the experimental results also. Figure 5-57. Overlay of contour on airborne concentration results From the overlay shear cell and AFM results, optimum performance requires surface modification, limiting the conditions to those in the upper portion of th e cube. Thus the optimum conditions are seen to be Surface modifica tion, 5% spacer, and liquid CO2 dissemination. Unconfined Yield Strength, fc(kPa)Relative Humidity (%)Surface Modification 20 50 80 Bare Al TDFS ODTMS Adhesion Force +Unconfined Yield Strength, fc(kPa)Relative Humidity (%)Surface Modification 20 50 80 Bare Al TDFS ODTMS Unconfined Yield Strength, fc(kPa)Relative Humidity (%)Surface Modification 20 50 80 Bare Al TDFS ODTMS Adhesion Force + Adhesion Force + Airborne Concentration MediumSurface ModificationSpacer GasLiquid Bare Al ODTMS 0 wt% 5 wt% 79.3 70 72.1 70 67.1 85 59.9 85 TDFS Airborne Concentration MediumSurface ModificationSpacer GasLiquid Bare Al ODTMS 0 wt% 5 wt% 79.3 70 72.1 70 67.1 85 59.9 85 TDFS
135 Table 5-1. Summary of optimum surface modifi ed aluminum. The table shows the contact angle measurements of water for bare alum inum and aluminum surface modified with ODTMS and TDFS as well as the calculated wo rk of adhesion. These results are for a reaction time of one hour, which was determined to be the optimum time. AluminumODTMSTDFS 67.2113.5110.5 7.868.543.0 AluminumODTMSTDFS 101.943.847.3 144.999.5126.0 Advancing Receding Contact Angle W or k o f Adh es i on W a (J/ m 2) Advancing Receding
136 CHAPTER 6 CONCLUSIONS Improving aerosol dispersion requi res knowledge of the significan t interparticle forces of the system and the method to control them, an aspect that has been previously by and large neglected. Through this research, a systematic inve stigation of the effect of significant factors on powder dispersion in air ha s been conducted. An understanding of the effect of these factors on dispersion, the magnitude of their effect, and the subsequent change in in terparticle forces has been developed. Using this appr oach, the interparticle forces of any system can be determined allowing the determination of op timum processing methods to signi ficantly alter t hose forces and ultimately improve dispersion. In the first aspect of this research, factors that were determined to be significant to dispersion were investigated to determine not only their effect, but also magnitude, on dispersion. These factors includ ed the physical factors of dryi ng methods and contact area, as well as the chemical factors of capillary forces and surface modification. Dissemination methods were also investigated. The s econd part of this research invo lved developing an understanding of the reason for the observed effect. The change in dominant interparticle forces were measured and correlated to the processing parameters. Each will now be discussed in more detail. Drying was shown to have a significant im pact on producing a dry dispersible powder. Ineffective drying methods, such as evaporati on and freeze drying, allowed the presence of a liquid gas phase boundary and were shown to lead to strong interpartic le adhesion due to the capillary forces found in the liquid bridges at particle contacts. The capillary forces drew particles together which allowed van der Waals adhesion forces to become significant. An increase in mean particle size of near 10 100 times the primary particle size as a result of capillary force driven agglomeration was seen. Drying the Sigma aluminum flake powder with a
137 12m mean particle size by evapor ation and freeze drying led to a dry powder mean particle size of 600m and 300m respectively. The third drying method using s upercritical liquid CO2 was shown to overcome this challenge and effectivel y eliminate capillary forces by the complete avoidance of a liquid-gas interface through the use of the supercritical phase. The unique properties of supercritical fluids, in particular the lack of surface tens ion due to no definable liquid vapor phase boundary, result ed in the elimination capillary forces. This method was further developed and optimized for the produc tion of a dry, dispersable powder. A dry aluminum flake powder with a mean particle size of 18 m was produced, a significant improvement over any other drying method. Because of its superior results, supercritical drying became the standard drying process for all further research. Interparticle adhesion, specifica lly by van der Waals forces, is known to be affected by the contact area between the surfaces Through the use of spacer part icles, the contact area between the flake surfaces was controlled and in turn the adhesion forces were decreased. By calculation, a theoretical reduction in va n der Waals attraction of 1 107 N was found by including spacer particles in the aluminum matrix. Empirically, it was demonstrated that the addition of 5 weight percent of spacer particles resulted in a 10% incr ease in airborne partic le concentration. Dissemination methods using varying pressure and medium phase were also investigated in order to optimize dispersion. The airborne concentration was shown to be related to dissemination pressure and improve by increa sing dissemination pre ssure to 800psi. A maximum airborne concentration of 80% was seen under these c onditions. Calculation of the minimum turbulent eddy size s howed that despite the high pr essure, a minimum eddy of only 1.2m was possible. However, the thickness to which the force needed to be applied was only 12nm, well below the minimum eddy size.
138 To overcome this challenge, a novel method was develope d using liquid CO2 as the dissemination medium. Through this method, the particles were dispersed in a liquid that expanded in volume during disse mination because of its phase change to a gas, the thermodynamically favored state at standard temperatur e and pressure, once released from the devise. The expansion added a radi al force, on the order of magnit ude of the interparticle forces, in addition to the shear normally produced throu gh dispersion. The volume between bare flakes, calculated based on their surface roughness, however, was not sufficiently large for enough liquid CO2 to reside to aid in dispersion. Howeve r, once the spacer was a dded, the significant increase in volume between the flakes, and therefore of liquid CO2, led to a significant increase in dispersion. A 20% increase in ai rborne concentration when liquid CO2 dissemination was employed, and the maximum airborne c oncentration of 85% was produced. Capillary forces that develop due to moisture in the form of relative humidity were shown to result in agglomeration and a five fold increase in dry particle size from 10 to 50m as relative humidity was varied from 20 to 80%. The cr eation of a hydrophobic surface through the use of surface modification was employed to control aggl omeration caused by humidity. Two silanes, n-octadecyltrimethoxysilane (ODTMS) and tr idecafluoro-1,1,2,2tetrahydrooctyl-1triethoxysilane (TDFS), were used. ODTMS wa s shown to create a hydrophobic surface with an advancing water contact angle change fr om bare aluminum measured to be 673o to that of ODTMS modified surface of 1141o. TDFS was show to produ ce a slightly less hydrophobic surface with a measured contact angle of 1113o under the same conditions. The CO2 dissemination method was again used to i nvestigate the ability of this method to disperse powder that was precond itioned at the varying humidity as described for the particle size measurements above. It was shown to be equally effective regardless of the powder
139 conditioning prior to diss emination using liquid CO2. This was the case even for the bare aluminum, which was susceptible to the most si gnificant agglomeration due to capillary forces. Despite conditioning for 48hr at high humidity, the increased agglomeration seen in the particle size measurements was not seen in liquid CO2 dissemination. No si gnificant decrease in airborne concentration was seen as the relative humidity was increased from 20-80%. With the use of statistical design, these results were evaluated to determine both their significance and the magnitude of their effect. It was shown that the interaction between the spacer addition and dispersion medi um that had the largest effect on airborne concentration. From this data, a relationship was also develope d to determine airborne dispersion based on the factor levels, namely Equation 5-1. The optimum levels of each factor were also determined. They were surface modification, 5 weight % spacer, and liquid CO2 dissemination. Next, the change in interparticle forces associ ated with these factors was investigated. It has already been shown that co ntact area reduced van der Waals forces and that the liquid CO2 dissemination, given a sufficient volume, will add a radial force on the same order of magnitude as the van der Walls adhesion. In addition, th e bulk powder properties a nd interparticle forces were also investigated. Since in most aerosol applications the pow der is drawn from a reservoir, bulk powder forces are important. It was s hown that at low humidity levels only van der Waals and frictional forces contributed to adhesion, however, at higher humidity capillary forces dominated. Through the use of surface modification, the adhesi on force at high humidity could be controlled. With ODTMS surface modification, unconfined yield strength could be controlled and no significant increase in adhesion was s een at any humidity level.
140 To understand the effect at the particle leve l, the atomic force microscope was used. Interparticle forces we re measured between bare and surf ace modified aluminum spheres and substrates. The bare aluminum at low humidity had an adhesion force of 850nN which is in close agreement with the calculated value. A critical humidity was reached at 20% relative humidity at which point a large enough amount of water adsorbed to the surface that a meniscus formed at the particle/surface contact and an increase in adhesion force due to the additional capillary contribution was seen. As the relative humidity increased, so did the volume of liquid in the meniscus and therefore the magnitude of the capillary forces. The ODTMS modified surfaces showed an initial adhesion force much lo wer than that of the bare aluminum surface. This was due to the lower surface energy of the modified surface a nd the steric ba rrier created by the silane. Taking this into account, the adhesi on force could be compared to the calculated value. As humidity was increased, the adhesion force also increased slig htly. The surfaces are therefore not completely hydrophobi c and some water can adsorb. The maximum adhesion force at 80% humidity was an order of magnitude lower for the surface modified materials as compared to the bare aluminum surfaces. This result shows that the surface modification does work to control both the interpartic le forces and capillary forces. The correlation of the two aspects of this rese arch shows not only the factors that effect dispersion as seen by an increase d airborne concentration, but al so the change in interparticle forces that are responsible for the change. From this understanding, information of what interparticle forces ar e involved in adhesion is enough to determine what processing parameters should be changed thereby allowing the application of this knowledge to other powder systems.
141 CHAPTER 7 CASE STUDY: ARMY OBSCURANTS Introduction Obscurant smokes and clouds are commonly used by the U.S. Army as a defense against spectral targeting and have been for centuries. Advancements in technology have led to communication and detection systems now functioni ng in the infrared a nd microwave region of the electromagnetic spectrum. For this reason ne w obscurant materials ar e under investigation to meet these requirements. An initiative has b een undertaken by the US Army to improve the current infrared obscurant grenades by a factor of four. Modeling by Army and independent civilian scientists has shown that high aspect ratio metal particles would result in the maximum performance[115-120]. Despite the extent of commercially availa ble materials, none meet the specifications to achie ve these goals. A large amount of work has been put towards fa bricating a flake materi al with the desired characteristics. This is, however, only the first ch allenge. The Infrared attenuation efficiency of a particulate cloud is dependent on the particle char acteristic and the amount of material in the cloud available to attenuat e the light. Particle dispersion is th erefore essential to the optimization of an obscurant particulate cloud. In order to maximize the numbe r of particles in the cloud all of the material discharged must disperse into primary particles. The focus of this research was to apply the knowledge gained th rough this dissertation research to develop an understa nding of the particle-particle interaction between the chosen particulate systems, determine the properties th at control dispersion, and develop methods to controllably manipulate these pr operties through processing in or der to develop an obscurant material that meets the performance criteria of a four times improvement over the current material across the selected region of the infrared spectrum.
142 Performance parameters have been selected by the Army to characterize a materials ability to obscure in the wavele ngths of interest[116, 121, 122]. The ratio of aerosol mass to transported precursor material ma ss, or yield factor, is the firs t performance parameter. The second is the electromagnetic extin ction cross section per mass of aerosol. This refers to the extinction cross section, or sh adow, created by the aerosol cloud. The third performance parameter is deposition velocity which is a meas ure of the settling rate of the cloud as it is transported downwind. Each of these factors will be infl uenced by the degree of powder dispersion. Particle shape also plays a role in the overall aerosol attenuation effici ency. Particles that have been found to produce the greatest extincti on over the selected wavelengths are small absorbing particles, flakes, hollow particles, a nd particles producing an optical resonance of some kind. A wide variety of commercially available incl uding highly absorbing materials like carbon black and graphite as well as ma n-made graphite flakes have also been investigated[123, 124] however a material genera ting greater extinction is desired. Through the use of fully dispersed aerosols of engineered high aspect ratio particles improvements in efficiency of up to an order of magni tude are theoretically possible. Improvements have been shown by taking advant age of the nanodimension. By using high aspect ratio particles, such as flakes, the eff ective scattering area per unit volume of the material in the aerosol cloud will be dram atically increased. There ar e some challenges, however, to dispersing particles of this shape and size range. High aspect ratio partic les have much larger contact area and decreasing the flake thickness into the submicr on range significantly increases the surface to mass ratio. In tu rn the effect of surface forces is much higher leading to an increase in the cohesiveness of the powder. The fundamentals associated w ith these interactions
143 must first be understood and then processing met hods developed to control dispersion in order for the maximum obscurant pe rformance to be realized. Experimental It is readily acknowledged that dispersion in air is a complex multifaceted problem and requires a systematic approach in order to determine the significant factors. For this reason statistical design of experiments is used to guide the research. Drying has already been shown to be important and supercritical drying was used exclusively. The first two factors under investigation were flake thickness and the addition of a spacer material. Suspensions of different thickness aluminum flake powders with and without fumed silica were characterized and then dried and the effects of these fact ors on performance, and dispersion, were evaluated. Finally the dissemination method was also considered and a novel liquid carbon dioxide method was investigated Factors and Preparation: Flake thickness The number of particles in the aerosol is c onstrained by their size and the volume of the container from which they are dispersed. By decr easing the particle thickness, while maintaining the diameter, the volume of each particle is de creased without changing the performance of the individual flake. As a result, a higher number of partic les can be packed into the same volume. As can be imagined, there will be a point of diminishing return. Once the flake thickness is decreased beyond a point, two problems will arise. First, the flakes will become transparent as there is not enough material to interact with the infrared ra diation. In addition, the mechanical strength of the flake decreases and they are no longer flat and ridged, but instead bend or curl on themselves.
144 Commercially available aluminum flake mate rials of three different thicknesses were received. Vapor phase deposited material of 15nm and 35nm mean thicknesses were received from Sigma Technologies, Inc and a milled materi al with a 60nm mean thickness was received from Silberline. All of the powders were rece ived in paste form and solvent exchanged three times with isopropyl alcohol (IPA) to remove any other solvents or contaminants. The non-polar nature of IPA resulted in a well dispersed Al su spension. The final aluminum IPA suspensions were between 10 and 20 weight percen t solids for supercritical drying. Spacer material High aspect ratio particles have a very high surface area to ma ss ratio. That is, because the thickness of the flake is an order of magnitude sm aller that the diameter the contact area between the flakes is very high while the mass to which mo mentum can be applied is very low. For this reason, once these high aspect ratio flakes come in to contact, dispersion ca n become impossible. If the contact area between flakes is decreased, the adhesion between them will also be decreased and dispersion can be improved. To achieve this goal, the flakes were mixed with fumed silica particles that act as spacers and prohibit the fl akes from coming into intimate contact. The spacer material used was a 10nm mean diam eter superhydrophobic fumed silica, Degussa Aerosil LE2. The addition of the silica can also have a harmful effect: since the silica does not attenuate in the infrared, the addition of s ilica past the point of improving dispersion will decrease the amount of aluminum material per mass and decrease the aerosol performance. Therefore, the spacer was added at 0, 5, 10 weight percent. All suspensions were then dr ied using the supercritical CO2 drying processing developed in this research. The dry powde rs were then dispersed in a Galai vacuum dispersion chamber onto (1) a glass microscope slide and (2) a Dual Pass Transmission (DPT) slide for evaluation in the FTIR.
145 Characterization Particle size and morphology As discussed in Chapter 4, although high aspe ct ratio particles do not completely agree with the assumptions needed for accurate light scattering measurements, size analysis on a volume basis has been shown to reliably describe th e particle diameter as has been confirmed by Image Analysis. The particle size distribution of each material, as received, was measured in a well dispersed suspension with isopropyl alcohol. The particle size was also measured after processing, before drying, to ve rify the processing did not cha nge the particle size. Scanning electron microscope (SEM) images and transmi ssion electron microscope (TEM) images were also taken to understand the flake morphol ogy, thickness, and the oxide layer. Aerosol performance Obscurant performance was assessed on two levels. Ultimately the entire system, a grenade, will be evaluated. Fo r this, a figure of merit, or FoMgrenade shown in equation 8-1, was developed and takes into account all aspects of the system including material performance, amount of material in the grenade, and the amount of material that is dispersed into the aerosol. Y FoMFill Grenade (8-1) Here the refers to the extinction cross section of the aerosol by mass; Fill is the fill density, or the density of the material multiplied by the volum e of material in the grenade; and Y is the dissemination yield, or ma ss fraction of powder aerosol ized in the cloud. The material performance is based on pr oduction methods and material properties including conductivity, ductility, and density. Th e fill density is determined by the packing density of the material in the container. For high aspect ratio flakes, the packing density limit has been shown to be much lower that that for spheres and only minimal pressure can be applied
146 to increase packing or the flakes will deform and mechanical interlocking will prevent future dispersion. It is in the Y fact or that the powder dispersion is ta ken into account. Agglomerated material will not contribute to the aerosol but instead fall to the ground. Incr easing the powder dispersion will result in more ma terial in the aerosol cloud, a longer cloud life, and therefore improved performance. The latter, dispersion, is the major focus of this case study. The effect of packing was also briefly investigated. Di fferent packing and dryi ng methods were used followed by FoM measurements of samples taken for evaluation. Grenades were not developed though this research, instead the study centered on improving the material dispersion through processing, drying, and dissemination techniques. To evaluate the material performance, instead of maki ng a grenade, a figure of merit of the material, FoMmaterial, was developed using the Beer-Lambert La w as shown in Equation 8-2 and will be discussed in detail below. c T100 log FoMmaterial (8-2) Here T is the transmission and c is the concentr ation in of the material obscuring the light. Powder performance Duel pass transmission (DPT) measurement In order to have a quick and effective screening process and to overcom e the challenges presented by mixt ures of materials, different characterization methods were developed and ba sed on the scattering and absorption of the IR laser of the FTIR. The theoretical absorption of light is proportional to the number of particles present. By dispersing the powder onto DPT slid es and measuring the transmission of IR light through the slide measured in a reflection m ode the degree of dispersion and material performance can be evaluated. This method is esp ecially useful for powders with silica added as
147 a flow aid; the metallic particles will attenuate light in the IR while the silica is transparent. This method follows the hypothesis that if the powder is more dispersed after pro cessing, there will be a decrease in FTIR transmission as there will be a higher number of pa rticles on the slide to attenuate with the laser. The mass of powde r on the slide will directly effect the FTIR transmission also. For this reason, the tran smission was normalized by the mass of powder on the slide. The amount of powder used was within the measurement error of lab balance so for a higher degree of accuracy Inducti vely Coupled Plasma Spectroscopy (ICP) was used and is described in detail below. This allowed the accurate measurement of masses well below those seen for these experiments. The material pe rformance was determined based on the material Figure of Merit (FoM), or the improvement of the new materials over a standard material supplied by the US Army. FoM values at chosen IR wavelengths were calculated and plotted for each material processing techniqu e. If dispersion of a constant mass of powder increased, the transmission would decrease and result in a higher FoM. On the other hand, powder agglomeration would be portrayed by a decrease in the FoM due to less laser attenuation. Decreasing the thickness of the flake by half will result in twice as many flakes per unit mass and theoretically half the transmission. The standard procedure used for FoM measurem ents is as follows. The dry powders were dispersed in a Galai vacuum dispersion chambe r onto (1) a glass micros cope slide and (2) a DPT slide. The powder was allowed to settle for forty five minutes after which the DPT slide was placed in the FTIR for measurement. The gl ass microscope slide was placed in a Nalgene bottle with a dilute acid soluti on to dissolve the metal in or der to accurately measure the concentration using the ICP. Mass is calculate d by measuring the concentration of Al in
148 solution, corrected for the presence of the oxide layer, and normalized by the area of the glass slide. Inductively Coupled Plasma Spectroscopy (ICP) The Perkin-Elmer Plasma 3200 Inductively Coupled Plasma Sp ectroscopy (ICP) system is equi pped with two monochromators covering the spectral range of 165-785 nm with a grated ruling of 3600 lines/mm. The ICP operates on the principle of atomic emission by at oms ionized in the argon plasma. Light of specific wavelengths is emitted as electrons return to the ground state of the ionized elements, quantitatively identifying the species present. Th e system is capable of analyzing materials in both organic and aqueous matrices with a detection limit range of less than one part per million. This detection limit is well below the concentrat ions used in these experiments and therefore allowed accurate measurement of the material mass disseminated. Results and Discussion The particle size of each material is shown in Figure 8-1. The mean particle diameter of both the 12nm and 35nm thick material was 10 2 microns. The 60nm thick material was slightly smaller with a mean particle size of 6 microns. Scanning electron microscope images of the aluminum flakes with varying concentrati ons of silica can be seen in Figure 8-2 A-C. Figure 8-2A shows an image of the 35nm thick fl akes without silica adde d. Figure 8-2B and 82C show images of the same ma terial with 5 and 10 weight pe rcent silica added respectively. The results for packing can be seen in Figure 8-3. The flake packing can be increased but at the expense of the FoM. Optimum packi ng is shown to be between 15 and 20 percent by volume and achieved by vibrational packing of the superc ritical dried powder. Packing by slurry evaporation uses the capillary for ces present to pull the powder toge ther and in turn increases the amount of agglomeration. Centrifuging can lead to mechanical interlocking by physical deformation of the flakes.
149 Supercritical Fluid processing in liquid CO2 resulted is a very l oosely packed powder of soft agglomerates with a low bulk density whic h was shown to effectiv ely produce a dispersible dry powder. During SCF processing the forma tion of the of the gas-liquid interface is completely avoided as are the capillary forces limiting the strong van der Waals forces present found after conventional drying. These dry, disp ersible powders were then disseminated in the Galai vacuum dispersion chambe r onto a glass microscope slide and a DPT for evaluation. The effect of thickness can be seen in Figure 8-4. For thinner flakes, the FoM continuously increased. In these experiments, decreasing the aluminum thickness from the 60nm flakes to 35nm flakes resulted in an increase in the average FoM from 3.75 to 6. A further reduction in thickness using the 15nm flakes resulte d in an increase in the average FoM to 10. The increase in FoM for these samples is lower than theory due to the presents of agglomerates. The addition of the fumed silica as a spacer materi al is shown to have a maximum effect as seen in Figure 8-5. When silica was added there was an increase in the FoM. It should be noted that the addition of five weight perc ent silica to the 35nm aluminum flake results in an increase in FoM from 6 to 7.9, close to the theoretical value suggesting that th e powder is better dispersed. However, at the highest concentr ation of silica, ten weight per cent, there was a decrease in the FoM below that of the aluminum without silica. Unlike aluminum, silica is tran sparent through most of the IR and does not attenuate any IR light. As silica is added to the mixture, it has been shown to reside between the aluminum flakes and increases the interpar ticle spacing, seen in Figure 8-2. In turn, the strong but short ranged van der Waals forces are reduced and the powder disperses to a hi gher degree as seen by the increased FoM. The silica is added on a weight basis thus as silica is added, the amount of aluminum in the mixture decreases. When the sili ca concentration is in ex cess of that needed to
150 aid in dispersion, the reduced amount of aluminum results in less material to attenuate the IR and as it follows, a decrease in the FoM. A simila r trend can be seen at all flake thicknesses; however, the effect is not as pr ofound for the 60nm flakes, which disperse equally at all silica concentrations. A summary of the re sults can be seen in Figure 8-6. These results were also analyzed using the De sign Expert 6 statistical design software. Three repeats of each point were run to take into consideration the sample-to-sample variability. The surface response plot is below in Figure 8-7. The standard deviation of the data is 0.5, which means that the values on the graph are signif icantly different. From this output it can been seen that the FoM increases with decreasing th ickness, similarly to the data from the FTIR results above, and that there is a maximum in the spacer concentration at f our weight percent. Conclusions The drying process is crucial to the production of a dispersible, dry powder. By using Supercritical Fluid processing in liquid CO2 the production of a dispersible dry powder has been demonstrated. This is due to the ability of this processing technique to remove liquid without the formation of the gas/liquid interface and thus capill ary forces that lead to agglomeration. The FoM of the aluminum powder was shown to increas e as the thickness of the flakes decreased as long as SCF processing was leveraged to produce the dry material As flake thickness decreased, a higher number of flakes per unit mass were ava ilable to interact with the laser, lowering the transmission. Fumed silica was shown to incr ease the dispersion of the aluminum flakes by increasing the interparticle spac ing and thus reducing the van der Waals forces between the flakes. The silica was shown to have a maximum in effectiveness at four weight percent. Although the silica increased the in terparticle spacing to improve dispersion, it was added on a weight basis, so the further addi tion reduced the FoM. Since silica is transparent in the IR, thus after four weight percent sili ca was added, the number of aluminum particles available to
151 attenuate with the laser was d ecreased causing the decrease in FoM. Statistical design of experiments as used to insure that the results were obtained efficiently and were significant.
152 Figure 7-1. Laser Diffraction pa rticle size distribution for aluminum flakes. A graph of the particle size distribution of the aluminum flakes as meas ured by laser light scattering in the Bachman/Counter 13320. The blue cu rve depicts the partic le size for the 12nm thick flakes, the red cured shows that of the 35nm thick flakes, and the green curve show the particle size of the 60 nm thick flakes. Figure 7-2. Typical Images of aluminum flakes and silica. A) shows the edge of one flake, B) shows the aluminum/silica mixture at low silica concentrations, and C) shows the aluminum/silica mixture at high silica concen trations. Note the distribution of silica between the flakes increasing the interparticle spacing. 0 1 2 3 4 5 6 7 8 9 0.11101001000 Equivalent Spherical Diameter (um)Differential Volume (%) 12nm 35nm 60nm 1 3 2 A C B
153 Figure 7-3. Packing density of flakes and FoM. By using diffe rent consolidation methods the maximum packing density is increased but at the sacrifice of FoM. Slurry evaprotation leads to agglomeration due to capillary forces and centrifuging can lead to mechanical deformation and physical interlocking of the flakes. The optimum is shown to be between 15-20 volume percent. 0 5 10 15 20 25 30 PourVibrateSlurry evaporation Centrifuge Packing MethodsFill % VolumeFoM:5.1FoM:2.8 FoM:3 FoM:5 0 5 10 15 20 25 30 PourVibrateSlurry evaporation Centrifuge Packing MethodsFill % VolumeFoM:5.1FoM:2.8 FoM:3 FoM:5
154 Figure 7-4. Figure of Merit fo r aluminum flakes of varying thickness. As thickness is decreased, the number of flakes present per unit mass incareases resulting in an increase in the FoM. Figure 7-5. Figure of Merit for aluminum flakes of varying thickness and spacer concentration. 0 2 4 6 8 10 12 24681012141618WavelengthFoM 60nm 40nm 15nm Wavelength (um) 0 2 4 6 8 10 12 24681012141618WavelengthFoM 60nm 40nm 15nm Wavelength (um) 0 2 4 6 8 10 12 24681012141618Wavelength (um)FoM 15nm 15nm + 5wtLE2 15nm + 10wtLE2 40nm 40nm + 5wtLE2 40nm + 10wtLE2 60nm 60nm + 5wtLE2 60nm + 10wtLE2
155 Figure 7-6. Average FoM across all wavelengths for aluminum fl akes of varying thickness and silica concentrations. The addition of 5wt% silica increases the interparticle spacing, aiding in dispersion as seen in the FoM. W ith the further addition of silica above the optimun, excess silica decreases the amount of aluminum resulting in a decrease in FoM. Similar trends were seen for all materials. 0 2 4 6 8 10 12 15nm 15nm + 5wt silica 15nm + 10wt silica 40nm40nm + 5 wt silica 40nm + 10wt Silica 60nm60nm + 5wt silica 60nm + 10wt silicaMaterialAverage FoM
156 Figure 7-7. Design Expert Surf ace Response plot for FoM. Fr om the plot the effect of decreasing flake thickness and increasing spacer concentration can be seen. Decreasing flake thickness increases FoM due to the higher number of flakes available to attenuate the light. The spacer decreases the contact area between flakes thereby improving dispersion. A maximum at four weight percent spacer can be seen as additional spacer decreases the amount of aluminum present. The best results are found with a 12nm thick flake and 5 weight percent spacer yielding a FoM value of 11.
157 CHAPTER 8 SUGGESTIONS FOR FURTURE WORK Through the research presented in this dissert ation, an understanding of the factors that affect dispersion and the changes in interparti cle forces associated with those factors was developed. Because of this fundamental unders tanding, the conclusions developed through this work can be applied towards other powder sy stems. Despite these achievements, ideal dispersion still was not achieved. To this end, further research is now suggested. The first step to understanding dispersion is its measurement. In this research a settling chamber was built to measure transmission de cay of the settling aerosol. From these measurements and the Beer-Lambert Law, the ai rborne concentration of powder was calculated and used to compare the effectiveness of the processing and dissemination methods. Transmission measurements over time also result in a line whose slope is proportional to settling rate. Specifically, it is equal to the change in concentration divide d by the height of the chamber. From the settling rate and the equation for Stokes settling, a particle size for the powder can be calculated. This would be a va luable undertaking that would comp lement airborne concentration measurements. With this disseminated particle size, a comparison can be made to the actual primary particle size and result in the development of a dispersion in dex. This would also give a more accurate description of the state of dispersion. For this, improvements to the dispersion chamber are required. The current measurements made in the disper sion chamber result in a large variation in transmission at the initial times due to the high tu rbulence associated dissemination. One factor that can affect this variation is chamber volume, specifically height. Incr easing the height of the chamber could allow the initial turb ulence in the system to stabili ze before reaching the height of the laser. Less turbulence at the level of the measuring laser beam would lead to more consistent
158 measurements and more accurate calculation of th e particle size both of which are required to use this measurement method to develop a dispersion index. In this research, methods were developed to measure interparticle fo rces at the macro and micro scale. These measurements were then us ed to develop a better unde rstanding of the effect of the processing parameters on th e change in interparticle forces and a subsequent correlation to dispersion. However, the direct correlation between interparticle forces and dispersion was not made. In continuation of the above discussion of the effective part icle size, or dispersion index, this could be related directly to interparticle forces. From this a predictive model could be developed to relate interparticle fo rces directly to particle size, a nd thus the degree of dispersion. A model based on the empirical results, as that produced by statistical experimental design, would also serve in such a way if the disseminated particle size was used as a response. This would expand the model to also include the analysis of error within the e xperiments. The source of error, and those areas where it should be re duced, can also be used to determine optimum processing factors and levels thereof. Another aspect of this resear ch that had proven to be chal lenging is the application of a dispersion force to the agglomerates. As show n in this research, th e effect of increasing pressure, and turbulence, is limited to the minimum turbulent eddy size. The addition of a radial dispersion force acting from within the agglom erate, through a volatile medium, was shown to improve dispersion. This concept could also be accompanied by the use of nozzles. Through nozzle design, rapid changes in ve locity, volume, and pressure can be created. Using liquid CO2 to create a large increase in volume combined wi th a standard dispersion nozzle geometry, such as a converging diverging nozzle, large gradients can be created. This could resu lt in a reduction of the minimum turbulent eddy size and the app lication of a large disp ersion force to smaller
159 agglomerates. This idea can be taken to the ex treme case. With the correct configuration, a sufficient change in conditions can result in s upersonic velocities and associated with that supersonic transition is a shockwave which may a dd enough force to greatly improve dispersion.
160 LIST OF REFERENCES 1. Byron, P.R., Prediction of Drug Residence Time in Regions of the Human Respitory Track Following Aerosol Inhalation. Journal of Pharmaceutical Sciences, 1986. 75(5): p. 433-438. 2. Reed, J.S., Principles of Ceramic Processing. 2nd ed. 1995, New York: John Wiley and Sons. 3. Haynes, R.J. and R. Naidu, Influence of Lime, Fertilizer and Manure Applications on Soil Organic Matter Content and Soil Physical Conditions: A Review. Nutrient Cycling in Agroecosystems, 1998. 51(2): p. 123-137. 4. Embury, J., Optimum smoke Nanoparticles that Maxi mize Attenuation at Infradred and Radar Wavelengths, US Army Edgewood Chemi cal and Biologi cal Center. 5. de Boer, A.H., et al., Air Classifier Technology (ACT) in Dry Powder Inhalation Part 2. The Effect of Lactose Carrier Surface Prope rties on the Drug-to-Carrier Interaction in Adhesive Mixtures for Inhalation. International Journal of Pharmaceutics, 2003. 260(2): p. 201. 6. French, D.L., D.A. Edwards, and R.W. Niven, The Influence of Formulation on Emission, Deaggregation and Deposition of Dry Powders for Inhalation. Journal of Aerosol Science, 1996. 27(5): p. 769-783. 7. Louey, M.D., P. Mulvaney, and P.J. Stewart, Characterization of Adhesional Properties of Lactose Carriers Using Atomic Force Microscopy. Journal of Pharmaceutical and Biomedical Analysis, 2001. 25(3-4): p. 559-567. 8. Louey, M.D. and P.J. Stewart, Particle Interactions Involved in Aerosol Dispersion of Ternary Interactive Mixtures. Pharmaceutical Research, 2002. 19(10): p. 1524-1531. 9. Mullins, M.E., et al., Effect of Geometry on Particle Adhesion. Aerosol Science and Technology, 1992. 17(2): p. 105-118. 10. Musante, C.J., et al., Factors Affecting the Deposition of Inhaled Porous Drug Particles. Journal of Pharmaceutical Sciences, 2002. 91(7): p. 1590-1600.
161 11. Ranade, M.B., Research on Coatings or Surface Treatment of Metal Powders, A.F.C.R. Laboratory, Editor. 1975. 12. Ranade, M.B., Surface Treatment on Metal Powd ers for Dispersion Improvement, C.R.a.D. Center, Editor. 1984. 13. Zeng, X.M., et al., The Role of Fine Particle Lactose on the Dispersion and Deaggregation of Salbutamol Sulphat e in an Air Stream in Vitro. International Journal of Pharmaceutics, 1998. 176(1): p. 99-110. 14. Zijlstra, G.S., et al., The Role of Particle Engineering in Relation to Formulation and DeAgglomeration Principle in the Developm ent of a Dry Powder Formulation for Inhalation of Cetrorelix. European Journal of Pharmaceutical Sciences, 2004. 23(2): p. 139. 15. Masuda, H., Higashitani, K., Yoshida, H., ed. Powder Technology Handbook. Third ed. 2006, Taylor and Francis Gr oup, LLC: Boca Ratone, Fl. 16. Chan, H.-K. and N.Y.K. Chew, Novel Alternative Methods for the Delivery of Drugs for the Treatment of Asthma. Advanced Drug Delivery Reviews, 2003. 55(7): p. 793. 17. Galk, J., W. Peukert, and J. Krahnen, Industrial Classification in a New Impeller Wheel Classifier. Powder Technology, 1999. 105(1-3): p. 186. 18. Hickey, A., A New Millenium for Inhaler Technology. Pharmaceutical Technology, 1997. 21(6): p. 116-125. 19. Holdich, R., Fundamentals of Particle Technology. 2002, Shepshed, U.K.: Midland Information and Technology Publishing. 20. Masuda, H. and K. Gotoh, Dry Dispersion of Fine Particles. Colloids and Surfaces A: Physicochemical and Engi neering Aspects, 1996. 109: p. 29. 21. Sweeney, L.G., et al., Spray-Freeze-Dried Liposomal Cipr ofloxacin Powder for Inhaled aerosol drug delivery. International Journal of Pharmaceutics, 2005. 305(1-2): p. 180. 22. Voss, A. and W. Finley, Deagglomeration of Dry Powder Pharmaceutical Aerosols. International Journal of Pharmaceutics, 2002. 248.
162 23. Calabrese, R.V. and M.B. Ranade, Pneumatic Nozzle Dissemination of Powders into Air, U.S.A.R. Office, Editor. 1993. 24. Chew, N.Y.K., D.F. Bagster, and H.K. Chan, Effect of Particle Size, Air Flow and Inhaler Device on the Aerosolization of Disodium Cromoglycate Powders. International Journal of Pharmaceutics, 2000. 206(1-2): p. 75-83. 25. Chew, N.Y.K. and H.K. Chan, Influence of Particle Size, Air Flow, and Inhaler Device on the Dispersion of Mannitol Powders as Aerosols. Pharmaceutical Research, 1999. 16(7): p. 1098-1103. 26. Chew, N.Y.K., et al., Characterization of Pharmaceutica l Powder Inhalers: Estimation of Energy Input for Powder Dispersion and Effect of Capsule Device Configuration. Journal of Aerosol Science, 2002. 33(7): p. 999-1008. 27. Endo, Y., S. Hasebe, and Y. Kousaka, Dispersion of Aggregates of Fine Powder by Acceleration in an Air Stream and its Applicat ion to the Evaluation of Adhesion Between Particles. Powder Technology, 1997. 91(1): p. 25-30. 28. Chew, N.Y.K. and H.K. Chan, Effect of Powder Polydispersity on Aerosol Generation. Journal of Pharmacy and Pharmaceutical Sciences, 2002. 5(2): p. 162-168. 29. Raula, J., et al., Study of the Dispersion Behaviour of l-leucine Containing Microparticles Synthesized with an Aerosol Flow Reactor Method. Powder Technology, 2007. 177(3): p. 125. 30. Zeng, R.J. and B. Rand, Particle Size Distribution, Powder Agglomerates and their Effects on Sinterability of Ultrafine Alumina Powders. Journal of Materials Science & Technology, 2000. 16(4): p. 393-396. 31. Zeng, X.M., et al., Effects of Particle Size and Addi ng Sequence of Fine Lactose on the Deposition of Salbutamol Sulphate fr om a Dry Powder Formulation. International Journal of Pharmaceutics, 1999. 182(2): p. 133-144. 32. Mizes, H., et al., Small Particle Adhesion: Measurement and Control. Colloids and Surfaces a-Physicochemical and Engineering Aspects, 2000. 165(1-3): p. 11-23.
163 33. Crowder, T.M., et al., Fundamental Effects of Particle Morphol ogy on Lung Delivery: Predictions of Stokes' Law and the Partic ular Relevance to Dry Powder Inhaler Formulation and Development. Pharmaceutical Research, 2002. 19(3): p. 239-245. 34. Ho, C.A. and M. Sommerfeld, Modelling of Micro-Particle Agglomeration in Turbulent Flows. Chemical Engineering Science, 2002. 57(15): p. 3073-3084. 35. Lightstone, M.F. and G.D. Raithby, A Stochastic Model of Particle Dispersion in a Turbulent Gaseous Environment. Combustion and Flame, 1998. 113(3): p. 424-441. 36. He, Y.H., A Criterion for Particle Agglomeration by Collision. Powder Technology, 1999. 103(2): p. 189-193. 37. Han, R., G. Papadopoulos, and B.J. Greenspan, Investigation of Powder Dispersion Inside a SPIROS(R) Dry Powder Inhaler using Particle Image Velocimetry. Powder Technology, 2002. 125(2-3): p. 266. 38. Wang, Z.L., C.F. Lange, and W.H. Finlay, Use of an Impinging Jet for Dispersion of Dry Powder Inhalation Aerosols. International Journal of Pharmaceutics, 2004. 275(1-2): p. 123-131. 39. Gomes, M. and e. al, The Effect of Inertial on the Dispersion of Particles in the Flow Around a Two-Dimensional Flate Plate. Chemical Engineering Science, 2002. 57: p. 1319-1329. 40. Mullins, M.E. and M.B. Ranade, Force of Adhesion and the Energy Required to Disperse Test Dust in Air, U.S.A. Chemical Research and Development Center, Editor. 1984. 41. Vold, M.J., The Effect of Adsorption on the van der Waals Interaction of Spherical Colloidal Particles. Journal of Colloid Science, 1961. 16(1). 42. Safatov, A.S., et al., Variations in Disperse Composition of Dry Powders According to Energy of their Dispersion. Powder Technology, 1998. 97(3): p. 227-232. 43. Feng, J.Q. and D.A. Hays, Relative Importance of Electrostatic Forces on Powder Particles. Powder Technology, 2003. 135: p. 65-75. 44. Reist, P.C., Intoduction to Aerosol Science. 1984: p. 299.
164 45. Reist, P.C., Aerosol Science and Technology. Second ed. 1993, New York: McGraw-Hill, Inc. 379. 46. Yurteri, C.U., et al., Electrostatic Effects on Dispersi on, Transport, and Deposition of Fine Pharmaceutical Powders: Development of an Experime ntal Method for Quantitative Analysis. Particulate Science and Technology, 2002. 20(1): p. 59-79. 47. Ren, J., et al., Electrostatic Dispersion of Fine Particles in the Air. Powder Technology, 2001. 120(3): p. 187. 48. Ren, J., S.C. Lu, and F.Q. Tang, The Voltage-Current Charac teristics and Dispersion Effect of RLW Electrostatic Disperser. Powder Technology, 2003. 135: p. 261-265. 49. Ata, A., Y.I. Rabi novich, and R.K. Singh, Role of Surface Roughness in Capillary Adhesion. Journal of Adhesion Science and Technology, 2002. 16(4): p. 337-346. 50. Hooton, J.C., et al., An Atomic Force Microscopy Study of the Effect of Nanoscale Contact Geometry and Surface Chemistry on the Adhesion of Pharmaceutical Particles. Pharmaceutical Research, 2004. 21(6): p. 953-961. 51. Hooton, J.C., et al., Characterization of Particle-I nteractions by Atomic Force Microscopy: Effect of contact Area. Pharmaceutical Research, 2003. 20(3): p. 508-514. 52. Nguyen, A.V., et al., Attraction between Hydrophobic Surfa ces Studied by Atomic Force Microscopy. International Journal of Mineral Processing, 2003. 72(1-4): p. 215-225. 53. Rabinovich, Y.I., et al., Adhesion between Nanoscale Rough Sur faces I. Role of Asperity Geometry. Journal of Colloid and Interface Science, 2000. 232(1): p. 10-16. 54. Rabinovich, Y.I., et al., Adhesion between Nanoscale R ough Surfaces II. Measurement and Comparison with Theory. Journal of Colloid and Interface Science, 2000. 232(1): p. 17-24. 55. Rabinovich, Y.I., et al., Capillary Forces between Sur faces with Nanoscale Roughness. Advances in Colloid and Interface Science, 2002. 96(1-3): p. 213-230. 56. Ren, J., et al., Research on the Composite Dispersion of Ultra Fine Powder in the Air. Materials Chemistry and Physics, 2001. 69(1-3): p. 204.
165 57. Tammet, H., Productivity of an Electric Mobility Classifier. Journal of Aerosol Science, 1996. 27(Supplement 1): p. S219. 58. Gupta, R., D. Gidaspow, and D.T. Wasan, Electrostatic Separation of Powder Mixtures Based on the Work-Functions of Its Constituents. Powder Technology, 1993. 75(1): p. 79-87. 59. DeSimone, J.M. and J.S. Keiper, Surfactants and self-assembly in carbon dioxide. Current Opinion in Solid Stat e & Materials Science, 2001. 5(4): p. 333-341. 60. Maury, E.E., et al., Graft Copolymer Surfactants fo r Supercritical Carbon-Dioxide Applications. Abstracts of Papers of the American Chemical Society, 1993. 206: p. 383POLY. 61. DeSimone, J.M., et al., Self assembly of small mol ecules and macromolecules in compressible liquid and supercritical Co-2. Abstracts of Papers of the American Chemical Society, 2002. 224: p. U313-U314. 62. Desimone, J.M., et al., Dispersion Polymerizations in Supercritical Carbon-Dioxide. Science, 1994. 265(5170): p. 356-359. 63. Shekunov, B.Y., et al., Aerosolization behaviour of micronised and supercriticallyprocessed powders. Journal of Aerosol Science, 2003. 34(5): p. 553-568. 64. Sellers, S.P., et al., Dry powders of stable protein form ulations from aqueous solutions prepared using supercritical CO2-assisted aerosolization. Journal of Pharmaceutical Sciences, 2001. 90(6): p. 785-797. 65. Sievers, R.E., et al., Supercritical fluid carbon dioxide technologies for fine particle formation for pulm onary drug delivery. Abstracts of Papers of the American Chemical Society, 2000. 219: p. U33-U34. 66. Kousada, Y. and e. al, Generation of an Aerosol by the Boiling of Suspensions. Aerosol Science and Technology, 1994. 8(21): p. 236. 67. Zimon, A.D., Adhesion of Dust and Powder. 1969, New York: Plenum Press.
166 68. Callister, W.D., Materials Science and Engi neering, An Introduction. 1997, New York: Wiley and Sons, Inc. 69. London, F., Eisenschitz, E., The relation between the van der Weals forces and the homeopolar valence forces. Zeits. f. Physik, 1930. 60(491). 70. Hamaker, H.C., The London van der Waals attrac tion between sphe rical particles. Physica IV, 1937. 10(1048). 71. van der Waals, J.D., The Equation of State for Gases and Liquids, in Nobel Lecture. 1910. 72. van der Waals, J.D., Over de Continui'teit van de GasEn Vloeistoftoestand (On the Continuity of the Ga seous and Liquid States). 1873, Leiden University, Netherlands. 73. Lennard-Jones, J.E.D., B.M., Cohesion at a Crystal Surface. Transactions of the Faraday Society, 1928. 24(92). 74. London, F., The General Theroy of Molecular Forces. Transactions of the Faraday Society, 1937. 33(8). 75. Derjaguin, B.V., Kolloid Zeitschrift, 1934. 69(155). 76. Hamaker, H.C., Physica, 1937. 4(1058). 77. Bradley, R.S., The Cohesive Force Between Solid Surfaces and the Surface Energy of Solids. Philosophical Magazine, 1932. 13(86): p. 853-862. 78. de Boer, J.H., Transactions of the Faraday Society, 1936. 32: p. 10-18. 79. Lifshitz, E.M., The Theory of Molecular Attractive Forces Between Solids. Soviet Physics JETP, 1956. 2(73). 80. Ninham, B.W. and V.A. Parsegian, van der Waals Forces Special Characteristics in Lipid-Water Systems and a Ge neral Method of Calculation Based on Lifshitz Theory. Biophysical Journal, 1970. 10(646).
167 81. Hough, D.B. and L.R. White, The Calculation of Hamaker Constandts from Lifshitz Theroy with Applications to Weddting Phenomena. Advances in Colloid and Interface Science, 1980. 14(1): p. 3-41. 82. Tabor, D. and H. Winterton, Driect Measurement of Normal and Retarded van der Waals Forces. Proceedings fo the Royal Society of Lo ndon Series a-Mathematical and Physical Sciences, 1969. 312(1511). 83. Israelachvili, J., Intermolecular & Surface Forces. 2nd Edition ed. 1992, San Diego: Academic Press Inc. 84. Fisher, R.A., On the Capillary Forces in an Ideal Soil; Correction of Formulae given by W. B. Haines. Journal of Agriculture Science, 1926. 16: p. 492. 85. Coelho, M.C., Upper Humidity Limit of the Pendular State. Powder Technology, 1979. 23(2): p. 203-207. 86. Coelho, M.C. and N. Harnby, Effect of Humidity on Form of Water-Retention in a Powder. Powder Technology, 1978. 20(2): p. 197-200. 87. Coelho, M.C. and N. Harnby, Moisture Bonding in Powders. Powder Technology, 1978. 20(2): p. 201-205. 88. Rumpf, H., Die Wissenschaft des Agglomerierens. Chem.-Ing.-Tech, 1974. 46: p. 1-11. 89. Rumpf, H., K. So mmer, and K. Steier, Mechanismen der Haftkraftverstrkung bei der Partikelhaftung durch plastisches Verforme n, Sintern und viskoelastisches Flieben. Chem. Eng. Tech, 1976. 48: p. 300-307. 90. Adler, J., Interaction of Non-Ideal Surf aces in Particulate Systems. 2001, PhD Dissertation, University of Florida: Gainesville. 91. Rabinovich, Y.I., M.S. Es ayanur, and B.M. Moudgil, Capillary Forces between Two Spheres with a Fixed Volume Liqui d Bridge: Theory and Experiment. Langmuir, 2005. 21(24): p. 10992-10997.
168 92. Esayanur, M.S., Interparticle Force Based Methodol ogy for Prediction of Cohesive Powder Flow Properties, in Materials Science and Engineering. 2004, PhD Dissertation, University of Florida: Gainesville. 93. Alonso, M. and F.J. Alguacil, Penetration of aerosol underg oing combined electrostatic dispersion and diffusion in a cylindrical tube. Journal of Aerosol Science, 2007. 38(5): p. 481. 94. Stotzel, H., et al., Adhesion measurements for elect rostatic powder coatings. Journal of Electrostatics, 1997. 40-1: p. 253-258. 95. Rimai, D.S., D.J. Quesnel, and A.A. Busnaina, The adhesion of dry particles in the nanometer to micrometer-size range. Colloids and Surfaces a-Physicochemical and Engineering Aspects, 2000. 165(1-3): p. 3-10. 96. Kunkel, W.B., Comment on Static Electrification of Dust Particles on Dispersion into a Cloud. Journal of Applied Physics, 1951. 22(1): p. 103-104. 97. Kunkel, W.B., The Static Electrificati on of Dust Particles on Dispersion into a Cloud. Journal of Applied Physics, 1950. 21(8): p. 820-832. 98. Nakajima, Y. and T. Sato, Electrostatic collection of submicron particles with the aid of electrostatic agglomeration prom oted by particle vibration. Powder Technology, 2003. 135: p. 266-284. 99. Watanabe, T., et al., Submicron Particle Agglomera tion by an Electrostatic Agglomerator. Journal of Electrostatics, 1995. 34(4): p. 367-383. 100. Laitinen, A., et al., Bipolar charged aerosol agglomeration with alternating electric field in laminar gas flow. Journal of Electrostatics, 1996. 38(4): p. 303-315. 101. Law, S.E., S.A. Thompson, and W. Balachandran, Electroclamping forc es for controlling bulk particulate flow: Charge relaxation effects. Journal of Electrostatics, 1996. 37(1-2): p. 79-93. 102. Crowder, T. and A. Hickey, Powder specific active disp ersion for generation of pharmaceutical aerosols. International Journal of Pharmaceutics, 2006. 327(1-2): p. 6572.
169 103. Boisdron, Y. and J.R. Brock, On Stochastic Nature of Acquisition of Electrical Charge and Radioactivity by Aerosol Particles. Atmospheric Environment, 1970. 4(1): p. 35-&. 104. Natanson, G.L., On the Theory of the Charging of Amicroscopic Aerosol Particles as a Result of Capture of Gas Ions. Soviet Physics-Tec hnical Physics, 1960. 5(5): p. 538-551. 105. Pauthenier and M. Moreau-Hanot, Spherical particle charge s in an ionized field. Journal De Physique Et Le Radium, 1932. 3: p. 590-613. 106. Pauthenier, M. and R. Guillien, Direct electrometric study of the limited charge of a conductor sphere in an ionised electric field. Comptes Rendus Hebdomadaires Des Seances De L Academie Des Sciences, 1932. 195: p. 115-116. 107. Pauthenier, M., M. Moreau-Hanot, and R. Guillien, Charge of small dielectric spheres in an ionised electric field. Comptes Rendus Hebdomadaires Des Seances De L Academie Des Sciences, 1932. 195: p. 213-215. 108. Yoon, R.-H., D.H. Flinn, and Y.I. Rabinovich, Hydrophobic Interactions between Dissimilar Surfaces. Journal of Colloid and Interface Science, 1997. 185(2): p. 363. 109. Keiper, J.S., et al., New phosphate fluorosurfactants for carbon dioxide. Journal of the American Chemical Society, 2002. 124(9): p. 1834-1835. 110. Ye, W.J., J.S. Keiper, and J.M. DeSimone, Polymeric nanoparticles from supercritical CO2 microemulsion polymerization. Chinese Journal of Polymer Science, 2006. 24(1): p. 95-101. 111. Ducker, W.A., T.J. Senden, and R.M. Pashley, Measurement of Forces in Liquids Using a Force Microscope. Langmuir, 1992. 8: p. 1831. 112. Rabinovich, Y.I. and R.H. Yoon, Use of Atomic Force Microscope for the Measurements of Hydrophobic Forces between Silanated Silica Plate and Glass Sphere. Langmuir, 1994. 10(6): p. 1903-1909. 113. Cadle, R.D., The Measurement of Airborne Particles. 1975, New York: John Wiley and Sons, Inc.
170 114. Fuks, N.A., The mechanics of aerosols. Rev. and enl. ed. 1964, Oxford, New York: Pergamon Press; [distributed in the West ern Hemisphere by Macmillan. xiv, 408 p. 115. Embury, J., Particles Producing Strong Ex tinction in the Infrared. Technical Report, 1994. US. Chemical Systems Laboratory. 116. Embury, J., Maximizing Infrared Extinction Coef ficients for Metal Discs, Rods, and Spheres. Technical Report, 2002. US Army Edge wood Chemical and Biological Center. 117. Appleyard, P.G., Infrared extinction performanc e of high aspect ratio carbon nanoparticles. Journal of Optics a-Pure and Applied Optics, 2006. 8(2): p. 101-113. 118. Appleyard, P.G., Modelled infrared extinction and attenuation performance of atmospherically disseminated high aspect ratio metal nanoparticles. Journal of Optics aPure and Applied Optics, 2007. 9(3): p. 278-300. 119. Appleyard, P.G. and N. Davies, Modelling infrared extinction of high aspect ratio, highly conducting small particles. Journal of Optics a-Pure and Applied Optics, 2004. 6(10): p. 977-990. 120. Appleyard, P.G. and N. Davies, Calculation and measurement of infrared mass extinction coefficients of selected i onic and partially ionic insulators and semiconductors: a guide for infrared obscuration applications. Optical Engineering, 2004. 43(2): p. 376-386. 121. Embury, J., Particles Producing Strong Ex tinction in the Infrared. Technical Report, 1995. US Army Edgewood Chemi cal and Biological Center. 122. Embury, J., Optimum Smoke Nanoparticles that Maximize Attenuat ion at Infrared and Radar Wavelengths. Technical Report, 1998. US Army Edgewood Chemical and Biological Center. 123. Embury, J., Extinction of Laser Radiation by Ma n-Made Graphite Flake Aerosols. Technical Report, 1993. US Army Edge wood Chemical and Biological Center. 124. Embury, J., Screening Smokes Performance of Commercially Available Powders III: Infrared and Visible Secreening by Carbon Black. Technical Report, 1994. US Army Edgewood Chemical an d Biological Center.
171 BIOGRAPHICAL SKETCH Stephen Tedeschi was born in Schenectady, New York. He graduated from Notre Dame Bishop Gibbons High School in May 1997 and enroll ed in The New York State School of Ceramics at Alfred University in Alfred, New Yo rk that fall. Stephen earned a Bachelors of Science in Ceramic Engineering upon graduation in May 2002. Stephen want ed to continue this education and in August of 2002 St ephen began his studies at the University of Florida. He received a Masters of Science in Materials Science and Engi neering in December 2004 and a Doctorate of Philosophy in December 2007. After co mpletion of his PhD, Stephen hopes to take the knowledge he gained here at th e University of Florida and conti nue to do research in the area of Particle Science and Technology as part of an industrial research and de velopment facility.