Improving Aerosol Dispersion through Processing and Dissemination Techniques †

The state of dispersion is a vital aspect of powder technology impacting both traditional and emerging technologies in many diverse areas of interest including health care, industry, the environment, and the military. In such applications, the properties and quality of the resultant product will be directly and noticeably af fected by the degree to which the particles are dispersed. In any operation where satisfactory particle dispersion is imperative, a reduction in dispersion, or increased agglomeration, leads to poor efficiency and ultimately results in lower yields and increased cost. As powder technology moves into the nano-age, the challenges of maintaining and improving particle dispersion become increasingly important. Decreasing particle size into the submicron range can 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 knowledge base to understand and control aerosol dispersion, a fundamental study into significant powder characteristics and environmental conditions that influence dispersion has been conducted. This paper discusses the results of testing various techniques employed to improve the dispersion of sub-micron aluminum flakes such as using a fumed silica spacer to reduce interparticle contact area and, in turn, attractive van der Waals forces, as well as examining the ef fect of dissemination pressure on the state of dispersion of the particles. This research was conducted using statistical experimental design in order to efficiently study significant


Introduction
A great deal of work has been put forth to improve dispersion in several diverse fields.Although the research covers a large number of materials in many different areas and technologies, the efforts have focused on one or more main areas: improving drying methods [1][2][3][4] , reducing attractive forces between particles [5][6][7][8][9][10] , increasing repulsive forces 11,12) , and developing mechanical dispersion devices [13][14][15][16] .Due to the complexity of the problem, however, most work has been empirical in nature and only incremental im-provements have been seen 17) . The frst step towards realizing a significant increase in aerosol dispersion is an understanding of the dominant causes of particle agglomeration and methods that can be used to effectively overcome them. This reuires a systematic study of these factors and their effect on the dispersion of the resultant aerosol.In previous work, the efficacies of different dr ying methods were investigated and two novel CO2 based drying methods developed 18) as was the use of surface modification to control capillary adhesion 19) . Duringthis research, the ef fect of humidity and hydrophobicity, flow aids and interparticle contact area, and dispersion methods have been studied and the results analyzed through the guidance of a statistical experimental design.
Much research effort has been directed at understanding powder dispersion through aerosolization.Ranade and Calabrese 14) found that an annular geom-etry 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 gradient, led to the best dispersion.Chew [20][21][22] investigated dry powder inhalers for pharmaceutical powders and found that the airflow was highly dependent on the device configuration but varied with the drug powder used such that no one configuration worked well with all powders due to differences in powder strength. Tbulence has also been experimentally studied with latex particles in conjunction with a numerical simulation developed by Endo 23) .The simulation was compared to the experimental results and used to calculate the adhesion force of spherical particles, observing that the particle size distribution effected the number of contacts between particles and thus the degree of cohesive behavior of the bulk powder [24][25][26][27] .Shear, turbulence, and impaction are dispersion mechanisms that are relatively easy to impart on a particle system and can be created simply by controlling device design.However, it is seen that these mechanisms alone are not entirely suf ficient and further research is required, especially for particles under 1 μm in size.It is therefore necessary to also study the powder properties to allow for processing steps during powder production to be implemented so that the cohesion of the powder is less than the forces provided by an appropriate dispersion device.To consider practical ways to improve aerosol dispersion it is necessary to establish the forces involved that work against particle-particle separation.These forces include van der Waals, capillar y and electrostatic forces.Controlling the influence of these forces on particle systems has been extensively researched in literature 6,14,18,19,21,[28][29][30][31][32][33][34][35][36][37][38][39][40][41][42][43][44][45] and will be briefly discussed here.With the exception of a few surfaces, most surfaces are not atomically smooth, and even at the atomic level there is some inherent roughness due to the size of atoms and their spacing within the crystal structure.If we consider two approaching surfaces, it is understood that the smoother the surface, the greater the contact area between the particles.If the surfaces are hence roughened, or the contact area between the particles is lessened through the addition of a mediating spacer particle, the overall contribution of van der Waals forces can be drastically reduced.The classical model for the effect of roughness on van der Waals forces was first postulated by Rumpf 46,47) and is shown in Equation 1.
Here A is the Hamaker constant, H is the interparticle 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.Recent work by Rabinovich and coworkers 53,54,90) has further refined Rumpf model to better predict experimental results, and in its final form as the PERC Model is seen in Equation 2. Their work indicates that surface roughness on the nanometer scale (<2 nm rms) greatly increases the force of adhesion, and as the scale of roughness increases there is a subsequent decrease in adhesive forces.
The additional terms in this equation include the maximum peak-to-trough height between the surface asperities, k1, the root mean squared roughness, rms, and the average peak-to-peak distance between the asperities, λ .
The presence of moisture between surfaces will also have a profound effect on dispersion.As the presence of water increases between two surfaces, liquid bridges can form.If the contact angle of the liquid on the solid is less than 90 the surfaces will be drawn closer due to capillary action.
Work has also been done on methods to control or change the effect of capillary forces acting between two surfaces by increasing the surface roughness to minimize the formation of capillary bonds.Rabinovich and coworkers have investigated the effect of nanoscale surface roughness on the critical humidity required for the onset of capillary forces in such systems 39,[48][49][50][51] and observed that the presence of roughness on sur faces on a nanoscale had a profound effect on the critical humidity required to induce capillar y adhesion. Teir study validated a simple analytical expression developed to predict the force of adhesion as a function of particle size, humidity, and roughness, which is most relevant for practical systems.A visualization of the effect of surface roughness is shown in Fig 1 .Electrostatic forces between particles are also necessar y to consider for aerosol dispersion and have been broadly investigated 11,31,35,44,[52][53][54][55][56][57][58][59] .Static electricity arises when charges on the surface of a material are not free to move to neutralize each other or do so slowly 52) .These charges remain on the surface where they can interact with their surroundings.Particles with a similar net surface charge, greater in magnitude than the attractive forces, will experience columbic repulsion.However the affect of electrostatic forces on dispersion fall outside the bounds of the research presented here and will be left for future investigations.
The following investigation is directed at improving the dispersion of high aspect ratio particles by considering factors that affect dispersion as discussed above.A model system of aluminum flakes was used to investigate the effect of changing the interparticle contact area by the addition of 5-10 wt% spacer particles, the effect of relative humidity on the state of dispersion as well as the effect dispersion medium.

Materials and Methods
An aluminum flake material was used as a model particle throughout this investigation (Sigma 1215, Sigma Technologies, Inc).These flakes are produced by physical vapor deposition and were delivered as a 70 wt% slurry in isopropyl alcohol.The flakes were rinsed 2-3 times by ultrasonication for 30 minutes (Misonix Sonicator 3000) in 50 mL of isopropanol (Fisher Scientific, USA) and separated by centrifugation (Beckman BH-2).The rinsing process removed any remaining product used in the PVD process.
For surface modification, 2% (vol.)solutions of n-octadecyltrimethoxysilane (ODTMS) were made as follows. 2 mL of silane and 0.5 mL of water was placed into 100 mL of ethanol (Fisher Scientific, 200 Proof), adjusted to a pH of 5 using glacial acetic acid (Aldrich, ACS Grade).10 g of powder was slowly added to the solution while stirring and the solution was allowed to continue mixing for one hour.After the allowed time, the mixture was solvent exchanged to remove any excess silane via dilution and centrifugation (Beckman JA-21) and dried using a supercritical fluid drying process described elsewhere 18) .The spacer material used was a 10 nm mean diameter hydrophobic fumed silica, (Aerosil® LE2, Degussa).The spacer was added at 5 and 10 wt% to slurries of aluminum particles suspended in isopropanol.These slurries were then supercritically dried in a custom designed super critical drier (University of Florida, Paar Instruments).
The degree of dispersion of the aerosol particles was characterized using a lab-scale air dispersion chamber, fabricated in-house to integrate into a Thermo Electron Magna 760 Fourier Transform Infrared (FTIR) Spectrometer.The cylindrical chamber had an internal volume of five liters and positioned in side the FTIR such that the IR laser passed through sapphire crystal windows of the chamber to the detector.
Change in contact area with increasing roughness.The illustration shows the decrease in contact area that occurs with increasing surface roughness and is attributed to the difference in interparticle forces between smooth and rough surfaces.

Fig. 1
The air dispersion chamber for the FTIR.A) The air dispersion chamber when positioned in the FTIR.B) A schematic diagram of the chamber showing the laser path through an aerosol cloud produced in the chamber.

Fig. 2
The FTIR Air Chamber is shown in Fig. 2.
Prior to experimentation, the chamber was purged with nitrogen for 15 minutes.Powders were then dispersed into the chamber by one of two methods.The first method inducted the powder into the FTIR chamber by evacuating the chamber to -12.5 psig.
The vacuum was quickly released dispersing the particles into the camber by the venturi effect.Particles were also dispersed into the chamber by high pressure injection of either nitrogen gas (5-90 psi) or carbon dioxide (800 psi, gas and liquid).
The characterization of the size and dispersion of mixtures posed a unique challenge.Each material has a unique index of refraction, thus laser diffraction could not be used to characterize dispersion.A new method was developed using the attenuation of laser light in a settling chamber.Once the powder was disseminated, measurements of the laser transmission through the particle cloud were made over time and graphed as the negative log of one divided by transmission.The slope of this line is indicative of the settling rate since the transmission increases as the powder settling out of suspension in the air 60) .A slower settling rate would result from a smaller particle size by increased dispersion.If a constant mass of material is used, and a constant concentration is assumed immediately after dissemination, the mass extinction coefficient of the Beer-Lambert Law, Equation 3, can be calculated and used to determine the airborne concentration over time.From this the degree of dispersion was determined and with it the efficacy of the processing and dissemination methods for that sample.
Here T is transmission of incident light passing through the dispersed sample, Me is the mass extinction coefficient, C is concentration, and PL is the path length of the laser through the aerosol cloud and assumed to be the diameter of the chamber.
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 applications; and one hundred seconds, a length of time considered to be in excess of most applications but interesting from the point of view of total aerosol lifetime.

Results and Discussion
High aspect ratio particles, such as flakes, pose unique challenges for dispersion because their inherent high surface area to mass ratio.More specifically, the thickness of the flake is at least an order of magnitude smaller than the diameter the interparticle contact area between the flakes can be much higher while the mass which dispersive forces can be applied is very low.Associated with the high surface area are also high attractive surface forces, in particular van der Waals forces.For these reasons, once the surfaces come into contact, dispersion can become impossible.If the contact area between flakes is decreased, adhesion between the particles will be decreased and dispersion can be improved.van der Waals forces are generally accepted to be significant only at short ranges, typically at separation distances less than 50 nm.Therefore, by maintaining some separation between the particles the van der Waals attraction will decrease.One way to achieve such an effect is through enhancement of surface roughness, shown in Fig. 1.First, the asperities on the surface lead to a less intimate contact and a lower total contact area between the par ticles.Second, the distance between the average surface planes is increased.The effect of roughness on van der Waals force can be seen by comparing the force distance curves calculated for smooth and rough surfaces, as shown in Fig. 3.
Here the theoretically calculated van der Waals force of attraction is plotted versus separation distance.Plot 1 shows the attractive forces for two smooth sur faces and Plot 2 shows the attractive Calculated van der Waals forces of a smooth and rough surface.Surface roughness has a large effect on interparticle forces.Plot 1 (solid) shows the calculation of van der Waals attraction for two smooth surfaces.Plot 2 (dotted) shows the effect of surface roughness incorporated by the PERC model 53,54) which not only decreases the contact area, but also separates the average surface plane of each body.A significant decrease in the calculated interaction force is seen.
Fig. 3 forces for the same size flakes with a 4 nm RMS surface roughness as calculated by the PERC Model, Equation 2 21) .This increase in surface roughness is shown to have dramatically reduced interparticle attraction with ultimately leads to less agglomeration.By applying this concept to the high aspect ratio flakes under investigation, decreasing the attractive forces will lead to an improvement in dispersion.
The sur face roughness of the aluminum flakes cannot be altered easily, however, by adding spacer particles between the flakes, the effect of roughness can be simulated, as illustrated in Fig. 4. The fumed silica employed for this investigation was chosen because it had a diameter approximately the same size as the flake thickness (10 nm) and has a lower Hamaker constant than aluminum: 6.6×10 -20 J as opposed to 1.5×10 -19 J respectively 23) .It was expected that the combination of these two effects would reduce the interparticle forces.A significant decrease in the magnitude of the theoretically calculated forces was found, as shown in Fig. 5.This graph shows a comparison of the calculated force distance curve for smooth aluminum flakes 12 μm in diameter and 12 nm thick along with the addition of 5-10 wt% fumed silica.The calculated forces go through an energy minimum at 5 wt% silica after which there is a slight increase in interaction forces at 10 wt% silica.At higher loadings of silica, there is a higher number of interparticle contacts and therefore an increase in the attractive forces.Based on these calculations, experiments were conducted to investigate the effect of adding fumed silica to the flakes.
For the first set of experiments, samples of bare aluminum flakes and those containing 5 wt% percent silica were disseminated using a par tial vacuum pressure.As expected, the addition of silica resulted in a reduction of interparticle forces.Fig. 6 shows a comparison of the airborne concentration for the vacuum disseminated aluminum and aluminum con-taining the silica.The addition of the silica lead to a sixteen percent increase in airborne concentration of particles.It should be noted, however that complete dispersion was not achieved and other factors must be considered.
The effect of interparticle agglomeration due to capillar y adhesion was subsequently investigated.Capillar y forces between par ticles 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 [24][25][26][27] . Tis was readily observed by the change in particle size distribution of aluminum flakes in Fig. 7 after preconditioning at 20, 50, and 80% relative humidity for 48 hours.The particle size is seen to increase 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.

Fig. 4
Calculated van der Waals forces with the addition of silica between two flakes.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 silica spacer material.The Plot 1 (solid) shows the calculated van der Waals force for two smooth disks and Plot 2 (dotted) shows the calculation for two disks separated by 5 wt% of 10 nm fumed silica particles.Plot 3 (dashed) shows the calculation when 10 wt% silica is added.The spacer has the same effect as increasing surface roughness, but done without having to alter the flakes themselves.The Hamaker constant of silica is also lower than that of aluminum which also contributes to a decrease in van der Waals forces.

Fig. 5
Airborne concentration of aluminum particles by vacuum dissemination.The addition of 5 wt% silica to the aluminum demonstrated a 10-20% increase in airborne concentration.
Fig. 6 with relative humidity as capillary forces increase the adhesion between the particles.The primary particle size is shown for comparison.The equation for the attractive forces due to the presence of liquid bridges between two spheres is given below 50) .
Here R is the radius of the capillary, γlv is the surface tension of the liquid phase, θ is the contact angle of the liquid at the particle surface, and the sign of the equation represents the force direction, which is cohesive in nature.By this reasoning, dispersion can be improved by controlling the capillary adhesion, i.e. increasing the contact angle through surface modification or by increasing the dispersion force through the use of a volatile medium.Research on the use of surface modification to control the contact angle is reported in detail elsewhere 19) however only the relevant results will be discussed below.For this investigation, aluminum flakes were coated with a hydrophobic silane and the dispersion properties were examined using the FTIR Air Chamber.The use of a volatile medium between the particles presents the opportunity to develop a radiating dispersion force from inside the agglomerate in addition to the traditional forces, such as turbulence and drag, an agglomerate would experience during dispersion.Liquid carbon dioxide was chosen as the volatile medium due to its environmental inertness, relatively low boiling point, high molar expansion ratio and low toxicity.In itʼs liquid form, CO2 is stored at approximately 800 psi at room temperature and during release to the atmosphere The liquid undergoes a rapid molar volume increase of the order of 500%.
The application of liquid CO2 as a dispersive medium is shown illustrated in Fig. 8.
The airborne concentration of the bare and OD-TMS surface modified flakes at 50% relative humidity disseminated using liquid CO2 can be seen in Fig. 9.
No significant change in airborne concentration was observed.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 Fig. 10, no significant change in airborne concentration or settling rate can be seen with increasing humidity.These results demonstrate that the use of liquid carbon dioxide to disperse and disseminate powder overwhelmed any effect of prior powder conditioning such as agglomeration due to capillary forces on dispersion.
The airborne concentrations, although independent of preconditioning humidity, was comparable to the dr y aluminum alone and thus has room for Particle size of aluminum flakes at 20, 50, and 80% relative humidity.The graph represents change in particle size of the aluminum flakes after supercritical drying and conditioning for 48 hours at 20% relative humidity (dotted), 50% relative humidity (dashed), and 80% relative humidity (dotted and dashed).Plot 1 (solid) represents the primary particle size measured in isopropyl alcohol and provided as a comparison.The standard deviation for the measurement is ±3μm.The change in particle size and distribution due to changes in the relative humidity can readily be seen.

Fig. 7
Illustration of the dispersion effect of liquid carbon dioxide.Upon release, the liquid carbon dioxide will undergo a phase transformation from a liquid to a gas resulting in an increase in volume of approximately 500 times.The rapid vaporization of liquid carbon dioxide aids in the dispersion of the powder.

Fig. 8
Airborne concentration of surface modified aluminum flakes at 50% humidity.The plot shows the airborne concentration of bare aluminum flakes and ODTMS surface modified aluminum flakes.Surface modification does not improve dispersion over liquid CO 2 dissemination alone.
Fig. 9 further improvement.To do so the combination of both methods was used.The aluminum flakes and flake/silica mixtures were disseminated using the liquid CO2 and the results can be seen by the increase in airborne concentration in Fig. 11.Although the dispersion of the bare aluminum did not improve, the addition of the spacer shows a 20% 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 improved dispersion.
Further analysis of the results was performed with the aid of a statistical design software package.A 3D representation of the results within the design space can be seen in Fig. 12.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 the three factors is designated by the number at each corner inside the cube.Since humidity did not show a significant effect on the airborne concentration experiments, it was not included in the analysis.From these results, the best conditions are found using liquid CO2, 5 wt% silica, and using sur face modification.Based on the data, an empirical model describing the effect of each factor on dispersion was developed by the software and given in Equation 5.
Cair= 73.55+0.70A(Medium)-1.80B(SurfaceModifi-(5) cation)+3.95C(Spacer)+6.80AC+1.80BC In Equation 5, the factors that have a significant effect on airborne concentration are shown along with the magnitude of their effect according to the factor coef ficient.The standard deviation for the coefficients was ± 0.6.Half the coefficient value is the factor effect.It is seen that the spacer and the interaction between the spacer and the medium have the largest effect on dispersion.Using this model, a maximum airborne concentration can be calculated based on different levels of these parameters.Within this design space, a maximum airborne concentration of 85% is predicted and experimentally obtained.

Conclusions
Through this research, a systematic investigation of the effect of significant factors on powder dispersion in air has been conducted.An understanding of the ef fect of these factors on dispersion, and the magnitude of their effect, has been developed.Through the use of spacer particles, the interparticle contact area between the flake surfaces was reduced and in turn the adhesion forces were reduced.It was Airborne concentration of bare aluminum at 20, 50, and 80% humidity.Using liquid CO 2 dissemination, the usual effect of increasing relative humidity on increasing agglomeration is no longer a factor due to the improved dispersion method.

Fig. 10
Airborne concentration using liquid carbon dioxide and silica.The addition of 5 wt% silica has previously been shown to reduce the van der Waals forces between flakes by increasing the interparticle spacing.This also increases the volume of liquid carbon dioxide between the flakes and in turn the amount of pressure that can be generated upon dissemination.

Fig. 11
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 conditions for optimum dispersion can be seen to result from 5 wt% spacer and liquid CO 2 .
Fig. 12 demonstrated that the addition of 5 wt% of fumed silica as spacer particles resulted in a sixteen percent increase in airborne par ticle concentration using partial vacuum induction.The condensation of liquid after conditioning at elevated relative humidity levels lead to significantly agglomeration due to capillary adhesion.To overcome this challenge, a novel dispersion method was developed which employed the use of liquid carbon dioxide as a volatile dissemination medium.Through this method, the particles were dispersed in a liquid that under went a high molar volume expansion during dissemination because of its phase change to a gas, the thermodynamically favored state at standard temperature and pressure, once released from the devise.The expansion of gas added a radial force in addition to the shear normally produced through dispersion and dominated any effect of powder preconditioning, however it was not large enough to significantly improve dispersion.
Once the spacer was added and disseminated using liquid CO2 dissemination resulted in a significant increase in airborne concentration of 20% and a maximum airborne concentration of 85% was observed.

Future Work
Up to this point experiments have been performed to investigate the effect of important parameters on the interparticle size and airborne concentration.The effect of chosen parameters on interparticle forces has been theoretically calculated and compared to the experimental results.In order to obtain a better understanding of the effect of these factors on the interparticle forces, measurements of the forces on the macro and micro scale must be performed.A comparison of the change in magnitude of the interparticle forces as they relate to these factors will result in a correlation between the significant interparticle forces and their relationship to dispersion.