Contents 1.0 Introduction 1.1 Background 1.2 Objective 1.3 Approach 2.0 Equipment/ Methodology 2.1 facilities 2.1.1 DPG 2.1.2 ECBC 2.2 Simulant Materials 2.3 Disseminators 2.3.1 Dry Powder Disseminators 2.3.1.1 High volume Metronics Motor Blower Model 10 (Skilblower) 2.3.1.2 High shear but low volume air C0-axial nozzle (SRI) 2.3.1.3 Sympatec RODOS Disseminator 2.3.1.4 High Volume DPG CART 2.3.2 High volume Liquid Agricultural Sprayers 2.3.2.1 Gas powered Micronair (AU8000) 2.3.2.2 Electric powered micronairs (AU6349) 2.4 Sympatec Particle Measurement Detector 2.4.1 Principles, Capabilities, Advantages ‘Correct’ Background referencing 2.4.2 Sample rate, duration and timebase averaging Nature of Sprays (ie variability) – time base and duration averaging 2.4.2 Lens Dynamic Range (order of magnitude difference in size of dry vs wet requiring smaller focal length lenses for the drys) 2.4.2.1 Working distance 2.4.2.2 Optical Lens Protection 2.4.3 Laser Focusing 2.4.4 Diffraction pattern deconvolution 2.5 Dissemination Procedures 2.6 Methods of Data Analysis 2.6.1 MIE Interpretaion 2.6.2 Laser Diffraction (LD) 2.6.3 High Resolution Laser Diffraction (HRLD) 2.6.4 Forced Stability 3.0 RESULTS AND DISCUSSION 3.1 Material Physical and Chemical Properties 3.1.1Dry Powders 3.1.1.1 Milling 3.1.1.2 dry powder containers and sampling 3.1.1.3 Fluidizers – Prevention of Particle Agglomeration & Dispersion uniformity 3.1.1.4 Hygroscopicity (ie SAVS and proteins) 3.1.1.5 Electrostatics on agglomeration and dispersal during dissemination (ie. CART) 3.1.2 Liquids 3.1.2.1 Viscosity and Surface Tension 3.2 Disseminator Design and Operation 3.2.1 Shroud Effects 3.1.3.1 Dry powders 3.1.3.2 Liquid Sprays 3.2.2 Rotational Effects 3.2.1 Centrifugal force in Dry Powder blowers 3.2.2 Spin rate in liquid atomizers 3.2.3 Turbulence 3.2.3.1 Airflow Induced Mixing Vortices during large particle acceleration 3.2.3.2 Thermal Induced 3.2.3.3 Background Referencing 3.2.4 Dispersion Control 3.2.4.1 Feed rate Effects on concentration, obscuration and multiple scattering 3.2.5 Positioning of disseminator to detection laser beam 3.2.5.1 Orientation to Intersecting Laser Beam 3.2.5.2 Disseminator Setback or standoff from Laser Beam 3.2.5.2.1 Standoff at 6” vs 17.5” 3.2.5.2.2 Transport distance at 10meters 3.3 Lens Focal Length 3.3.1Working distance 3.3.2 Contamination 3.3.3 Vibration induced Beam instability Forced stability 3.4.4 Laser focusing 3.4 Data analysis and Interpretation 3.4.1 Number/ Mass/ Volume Distribution 3.4.2 Algorithm Comparison Dry (2010 R4 data) Liquid 3.4.3 Distribution smoothing [Spline fit] 3.4.4 Probability Distribution 3.4.5 Summary of September 2010 Sympatec Measurements 3.3.1.1 Dry Powders 3.3.1.2 Sprayer variability - Gas powered Micronair Wet Sprays Using R6 Lens 3.4.6 Summary of August 2011 Sympatec Measurements 3.4.7 Summary Comparison of 2010 and 2011 4.0 Summary and future efforts 5.0 Literature 6.0 Appendices Dry Powder and Liquid Droplet Size Characterization by Laser Diffraction at Dugway Proving Grounds and ECBC during 2010 and 2011 (Part 1) Michael Williamson and Paul J Deluca, US army ECBC Robert D. Doherty, SAIC Summary Particle size source distributions have been obtained for three high volume disseminators and compared to three lower volume yet higher shear disseminators using particulate and soluble biosimulant and inert geological materials in dry and liquid physical state. Although early 2010 distribution results in Dugway complemented those in 2011, operational problems associated with lens contamination were encountered. Temporary remedies included the brute force air expulsion and/or manual cleansing of the lens during measurement using a high volume M28 military HEPA filtered motor blower. Subsequently, a compressed air operated lens protection attachment has been designed, developed and used successfully during the trials in 2011. Particulate and droplet sourceterm size distribution is dependant on dispersion and deagglomeration efficiency, nature on material and of the disseminator itself particularly for sprays with respect to exact location of measurement and the geometry of the divergent spray cone. 1.0 INTRODUCTION 1.1 Background The work described herein was accomplished under the auspices of “Test Authorization for BioCharacterization, DTC Project No. 2005-DT-DPG-ARSPT-D0295” detailed according to the 18 March 2006 directive by the U.S. Army Developmental Test Command (DTC), Aberdeen Proving Ground (APG), Maryland. Discrepancies have been noted between results predicted by various models (VLSTRAK, SCIPUF, etc.) and the actual data collected by referee devices during the trials. In order to understand the problem, JP Bio has undertaken a large study to obtain accurate source term data on the dissemination systems used to aerosolize the test simulants. The necessary source term data includes the particle size distribution of various aerosol simulants measured directly at the dissemination point, and correlated with measurements obtained upto 85 meters downstream from the release. This data will be studied by modeling and simulation experts to determine where the discrepancies in the existing models exist, and why current models and referee equipment are not agreeing on actual field trial data. The Joint Program Office of Biodefense conducts frequent open air trials at DPG. Typical particle size analysis for biohazard analysis is frequently performed using various laser optical particle counters (OPC) and spectrometers such as the aerodynamic particle sizer (APS) focusing on persistent airborne particles from 0- ~5um and those of respirable importance up to and including ~10micrometers. These instruments have a theoretical dynamic range limitation of close to 20micron aerodynamic yet the primary size particles and droplets far exceed this at the point source of generation. Neglecting the origin or source generation of particles and the large particle fallout dismisses the potential for subsequent or future contribution to the persistent aerosol from ground contamination, resuspension and reaerosolization. Sourceterm particle distribution determination is coupled with the ability to comprehend the principles, mechanisms and differences of dissemination equipment with that of collection, sampling and detection for material behavior during delivery and presentation at the detector. Size determination of material dispersed from high volume disseminators is not a trivial process and is complicated by air flow, material characteristics such as surface tension, viscosity and density, high concentration and sampling. LD is a readily expanding modern day optical method of size analysis and offers advantages over other instrumentation methods particularly in the characterization of sprayers and blowers where non-invasive measurement is important. The expanded dynamic range available for larger primary droplets emanating from sprayer atomizers is an attraction other measurement techniques where particles of that size cannot be measured. Although, the technique has been known to provide rapid, real-time and precise detection, LD appears to excel at the larger particle sizes where scattering angles are small. All Instruments for sizing particles have their drawbacks, yet LD remains popular because of its relative simplicity and speed. Commercially available LD instruments are rapidly expanding including Malvern, Sympatec, Horriba, Beckman and Cilas. Numerous disseminators are commercially available that can be used intentionally or unintentionally to discharge materials to the environment. Disseminators differentiate from those primarily used for dry powders and those for liquids producing a wide range of particle and droplets sizes dependant on operational conditions and material properties and characterististics. The data provided does not represent an in depth evaluation of LD but the results over two very short 1.2 Objective The objective of this campaign was to obtain quality source term size distributions of aerosol particles generated from two different disseminators frequently used at the Dugway Proving Grounds to disperse high volumes of test simulants during open air trials. The disseminators of interest were the Micronairtm AU7000 gasoline powered liquid disseminator, a variant of the AU8000 agricultural disseminator and the Metronics Skilblower which is an old discontinued agricultural high volume powder disperser manufactured in the 1960’s by the Metronics Company. The Micronairtm AU6349 backpack electric liquid disseminator was also used producing a smaller volume than the au8000. 1.3 Approach Development methods for LD measurement of sprays from agricultural and powder dispersing sprayers were conducted in three phases: two at DPG in the summer and fall of 2010 and 2011 at both the ABT and JABT and one at the ECBC ambient breeze tunnel between June and August 2010. Disseminators included the H7500 (Figure 1. a modified version of the electric Micronairtm au6349), Au8000 gasoline powered micronairtm and the Metronics Skilblower (Figure 2) for the 2010 trials. In 2011 the AU800 gasoline micronairtm was omitted. In addition to water as a control material liquid, suspensions of Erwinia herbicola and MS2 in tryptic soy broth and water, respectively were used as liquid biosimulants. Dry powder simulants of kaolinite clay or Kaolin, Ovalbumin, bacillus thuringenesis var kurstaki (BTK), and various milled grades Arizona Road Dirt were disseminated with the powder disperser. The detection system was the f series HELOS laser diffraction instrument by Sympatec. 2.0 EQUIPMENT/ METHODOLOGY 2.1 Facilities 2.1.1 DPG Source term trials were conducted at the older ambient breeze tunnel (ABT) and at the rear of the newer joint ambient breeze tunnel (JABT) in 2010 without additional tunnel flow. The 2011 trials were conducted at the JABT concurrent with other instrumentation sampling and at an average flow of approximately 1.2m/sec. Angle of wet and dry dissemination trajectory is about 15 degrees to maximize distance transport as depicted in figure. The disseminator setback from the laser beam was about 6” in 2010 and moved to 17.5” in 2010. A metal scissors jack cart at between 200-300 pounds was used to support the sympatec in 2010 and 2011. Dugway skilblower model G12 s/n 02062805 with an on/off blower motor max speed switch used in 2011 was observed to be newer than previous skilblowers used at ECBC and Dugway in 2010. 2.1.2 ECBC The ambient breeze tunnel (ABT) at Bldg E5884 in the Edgewood area of Aberdeen Proving Grounds is smaller than the DPG tunnel at 14’ x 14’ x Size distribution measurements were conduct with and without flow at approximately 1.2m/sec with the sympatec placed on the concrete floor for better stability and reduced vibration. 2.2 Simulant materials Dry powder materials included several grades of Arizona road Dirt (ARD) at 2.65g/cc density: fine at 0-80, ultrafine at 0-5, and mid grade at 10-20 micrometers. The later mid grade was used at ECBC and not part of the official test materials. Kaolin containing the fluidizer cabosil, aka aerosil, and commercial grade Bacillus thuringenesis ssp. kurstaki (BTK) at proximated densities of 2.4 and 1.451 g/cc, respectively, were used at DPG and ECBC. These literature reported densities are for molecular species and does not address agglomerate aerodynamic behaviorial forms. Liquid materials included tap water, tryptic soy broth (TSB), male specific (MS2) coli bacteriophage in TNE2 buffer and several purity grades in growth media of Pantanoe agglomerans (aka Erwinia herbicola (EH)) in fermentation TSB broth. All materials were supplied by DPG. Aqueous materials were estimated at approximately 97% water. 1 2 Carrera M, Zandomeni RO, Sagripanti JL; J appl Microbiol 105(1); pp68-77 (2008) TNE Buffer recipe for the resuspension of pelleted virus containing Tris, sodium chloride (NaCL) and EDTA In 2011, a group of protein simulants were introduced as powders: Bovine Ovalbumin (BOVA), Porcine Ovalbumin (PORA), standard avian Ovalbumin (OVA) and rabbit albumin (RABA). 2.3 Disseminators 2.3.1 Dry Powder Disseminators 2.3.1.1 High volume Metronics Motor Blower Model 10 (Skilblower) An older discontinued Metronics powder dispersing skilblower was used in 2010 with a high volume military M28 motor blower unit with HEPA filtration. The skilblower uses a mechanical stirred feeder atop of a centrifugal blower unit with an outlet of approximately 2.5 inches. The high speed motor blower provided a FIGURE : Micromeritus Metronics powder means to assist in the removal of disperser ‘Skilblower’ configuration used in 2010 particles from the fourier lens and with the Sympatec HELOS vario and the M28 reduce measurement artifacts. military blower. Operated in a vacuum mode, aerosol was pulled and directed through the laser beam. This was not only a brute force lens cleaning method but removed aerosol from subsequent downwind samplers. To facilitate concurrent sourceterm size measurements with downwind transport sampling, the 2011 trials attempted to transition to an air curtain diversion system to provide better control and consistent measurements and allow the aerosol to pass thru the beam and progress down the tunnel. 2.3.1.2 High shear but low volume air C0-axial nozzle (SRI) Other dispersion nozzles and disseminators were used and compared to the high volume units for validation of dry powder measurements. These included the TOPAS (Dresden, Germany), Stanford Research International (SRI) designed venturi tube and the RODOS dispersion system designed and supplied by Sympatec. Measurements were conducted at ECBC in the ambient breeze tunnel and compared and overlayed to measurements obtained at Dugway using several new and used Metronics skillblowers. 2.3.1.3 Sympatec RODOS Disseminator The RODOS, as a trademark of Sympatec Inc., is an engineered powder dosing delivery system and forms the basis for the Sympatec dry powder measurement system. The system integrates a vibratory feeder with controlled compressed air delivery to achieve a smooth, uniform and homogenous sample introduction through the sensing zone. A lamellar airflow under constant and uniform concentration through the narrow working distance eliminates abrasive contact with the optical lens. 2.3.1.4 High Volume DPG CART The CART high volume ventri sprayer was custom designed and fabricated at DPG. The name does not refer to an acronym and refers to the method of mounting for mobility considering its size and weight. The disseminator was only used in 2011 2.3.2 High volume Liquid Agricultural Sprayers Two Micronair (Bromyard Industrial Estate, Bromyard, Herefordshire, ENG) disseminators, electrical (AU6349) and a backpack type gas (AU8000) powered were used with liquids. The controlled droplet atomizer (cda) devices are characterized by screen covered variable rotating atomizers to 10,500 rpm controlling droplet size. A rotating atomizer creates droplets at sizes relative to the rotational speed selected by frequency and thrusts them outward and radially from the center. A rear airflow drives the droplets forward as the crossflow passes over the atomizer. The rotation rate and liquid feed rates are independently controlled in the electric while integrated in the gas powered Micronair. The electric version produces considerably greater output at twice the power requirements of 220volts. The electric version has a definitive cowling or shrouding just upstream of the atomizer affecting the generation of two visually distinct droplet populations. 2.3.2.1 Gas powered Micronair (AU8000) Dugway Proving Ground uses a modified version of the Micronair au8000 for liquid releases referred to as the H7500, in which the “backpack” of the Micronair (the 2 stroke engine and liquid reservoir) is replaced with a 4 stroke leaf blower engine to provide the forced air required to drive the spinning head atomizer. The spinning head woven wire gauze cylinder atomizer is taken from a Micronair au8000 and is fitted onto the leaf blower engine, and operates at a maximum of approximately 8400 RPM as determined by a stroboscope. A stream of droplets generated from the atomizer parallel to the detector laser beam is propagated thru the laser beam using a cross-flow of air generated by an onboard blower. The rotational speed of the head remained constant using approximately ¾ throttle corresponding to approximately 7000 RPM. Dugway supplied three different sprayers with the majority of tests conducted on a fairly new model with little or no field use, and 2 others which had seen many hours of field use. There were some notable differences in the data collected from the old and new versions. 2.3.2.2 Electric powered micronairs (AU6349) The Micronair au6349 sprayer uses a similar atomizing head to the au8000, but is powered by an electric motor and produces a much higher liquid output. Liquid feed to 2000 ml/min is provided by commercial peristaltic pumps for better control and longer release times. While not in current use at DPG, it is being studied as a possible replacement to the au8000 in field trials due to its higher output. Data is being collected on its performance to verify that the particle size distribution is similar to that of the au8000. Because it produces a much wider stream of particles than the au8000, the maximum distance was needed between the Sympatec detector and lens elements to keep the lens dry during measurements. To accomplish this, the R7 lens, with a focal length of 44.6”, was used. Even at this distance particles were still accumulating on the lens and other means were necessary to keep it clean. A plywood board placed 6” from the detector lens with a 3” hole to allow the beam to pass through blocked any stray particles from the lens without producing any detrimental effects on the data, and clean measurements were obtained. FIGURE 3: The Micronair au6349 high volume agricultural spray system. 2.4 Sympatec3 Particle Measurement Detector 2.41 Description On June 20 2011, the F-series HELOS configured system was calibrated and a new version of the software was purchased and delivered. The software designated as version 5 was a major upgrade over version 3 used in the 2010 trials allowing for the elimination of the full size EISA acquisition board in favor of USB connectivity and compatibility with the newer NT based operating systems. Version 5 provides better control and monitoring of laser health and longevity. The software, however, does not apparently allow for expanded analysis of the full dynamic range as does with the newer R-series instrument configuration where the software is designated as fraunhofer expanded (FREE) and mie expanded (MIEE) vs the older designated LD, HRLD FIGURE 4: Sympatec HELOS F-series Vario Laser Diffraction and mie. particle sizer The source laser and detector elements are housed separately and mounted on a 3’ aluminum rail to keep them properly aligned. The laser source and detector components are designed to be separated to accommodate large diffuse spraying systems and minimize lens contamination. All aerosol detection systems have an optimum resolution range among their dynamic range. The Sympatec HELOS Vario laser diffraction system is capable of particle measurement upto 6 mm using one of eight different fourier lenses, R1 to R8. As the lens number is increased for measuring progressively larger particles, so does the working distance at the expense of increasing laser beam instability due to the larger virtual focal lengths. 2.4.2 sample rate, duration and timebase averaging The highly variable nature of sprays dictates the necessity for significant averaging. Averaging is controlled through the use of time duration and timebase averaging. Data acquisition with sympatec LD is characterized by a high 2 khtz data throughput or temporal resolution. An ensemble particle diffraction spectral pattern is taken every 0.5milliseconds (msec) and packetized over a software selected time interval called a timebase. A timebase of 500msec was routinely used in these trials. A software 3 Sympatec Inc. Pennington NJ http://www.sympatec.com controllable two second measurement encompassing four timebase packets was routinely recorded over a time duration of 40 seconds. 2.42 Lens Dynamic Range The sympatec physical detector is of a fixed diameter for all lenses comprised of a fixed number (31) physical semi circular photosensitive concentric rings centrally radiating with continuous increasing widths to the outer permeter (Figure 2) to accommodate the lower light intensities of the diffracted light from smaller particles and droplets. The physical bin widths are inversely related to the particle sizes for which they represent. Data was collected with the Sympatec HELOS vario using four lens, the R3 and R4 for the smaller particle dry powders and the R6 and R7 for the larger liquid droplets. Long virtual focal lengths up to 2 meters with the R6 and R7 limit resolution above 10 and 20micrometers, respectively, yet increase the measurement capability Bin #1 is based on a sympatec reported lower detection of 0.5 micrometers for all lenses to 3.5mm Logarithm relationship between the dynamic ranges for each of the lenses At sufficiently high concentrations, typically 5-15% for dry powders not to exceed 35% for liquid droplets, materials transit the laser beam and a diffraction pattern is projected onto the light sensitive photodetector rings. The light intensities are computationally deconvoluted to virtual dynamic ranges dependant on the fourier lens used. Graphically visualized, the concentric ring physical position is proportional to the log of the upper bin diameters for each of the different lens. This allows for the seamless overlay of distributions for any material taken with any of the lenses. The corresponding ‘effective’ measurement ranges are 0.9 to 175 micrometers for the R3, 1.8 to 350 micrometers for the R4, 9 – 1750 micrometers for the R6 and 18 to 3500 micrometers for the R7 lens. 2.4.2.1 Working distance Since smaller particles diffract light at greater angles than do larger particles, particle measurement in restricted working distances unique to particular lenses is critical to achieving valid distributions. By vignetting, diffraction rays from smaller particles may not transit the fourier lens and strike the detector thereby not being detected. For diffraction rays of larger particles outside the working distance, such rays may transit the fourier lens but strike the detector surface at locations that translate to smaller particles thereby resulting in a subtle yet altered distribution. As the particle or droplet size increases, the effect diminishes from the very narrow diffraction angle. Schematic diagram of working distances for the Sympatec Vario and the concept of vignetting relative to the different lenses and focal lengths 2.4.2.2 Optical Lens Protection Any particles or droplets that attach to the lens during the measurement will produce anomalies. The high volume and large spray cones of sprayers and blowers necessitate some method and/or device for minimizing lens contamination. In 2010 at DPG, two approaches were used with the high volume M28 motor blower; 1) to pull the particles thru the laser using the vacuum, and 2) to push air across and adjacent to the lens. The longer focal lengths of the larger lenses (R6 and R7) coupled with the HELOS vario’s capability to spread the source and detector windows allows for the placement of the sprayer outlet further from the windows to minimize contamination. The placement of the Micronair head at approximately 12” from the detector lens benefited but did not completely eliminate droplet deposition due to the wide cones and high airflow. Sympatec corporation markets a clean air curtain specifically for the larger focal length lenses greater than R5, yet is reportedly not suitable for the reduced focal length lenses due to small working distance limitations. This air curtain is quite large and protrudes into the working distance restricting the small sensing volumes available to the smaller lenses. Alternative compressed air diffusing ‘air curtain’ designs were tested at the ECBC Breeze Tunnel during July and August of 2011 following the Dugway 2010 trials using the M28 military motor blower. The ultimate design used a convention small engine air cleaner filter (kohler WIX #42372) at 1.56 inches in height, outer diameter of 4.39 inches and inner diameter of 3.0 with a rating of 10 CFM. The lens cap was constructed by sandwiching the filter between two sheets of Plexiglas with an outer port for connection to a compressed air supply and an internal air stream diverter. Compressed air at 30psi was controlled using regulators supplying curtain airflows of 1.7m/sec at the detector and 2.5m/sec at the source lens. 2.4.3 Laser Focusing A single initial focus of the laser beam using software control at the start of each measurement series was routinely used to conserve material. At ECBC, the effect of laser intermittent focusing between measurements within a series with the high volume and variable sprays was examined and observed to be beneficial during the measurement of droplets using the R6 and R7 Lenses. 2.4.4 Diffraction pattern deconvolution Database stored diffraction patterns are subjected to deconvolution software algorithms to produce volume weighted distributions designated as ‘q3lg/’. Volume weighted distributions can be subsequently converted to number weighted distributions designated as ‘q0lg/’ equivalent to dN/dLogDp used in other particle measuring instruments. Alternatively they can be easily be converted by factoring the volume of a sphere. The F-series sympatec instrument allows for the selection of three algorithms, a mie based and two fraunhofer inversion algorithms. Although there is some confusion over the proper use of the fraunhofer algorithms, the LD for standard laser diffraction is thought to use primarily for dry powders and the HRLD designated as high resolution laser diffraction is better used for sprays. When performing standard LD on a diffraction pattern, HRLD is automatically invoked in an attempt to account for all the light at the detector centroid. In volume weighted distribution with large focal length lenses, this frequently results in a warning message that large particles may be present signifying a loss of accountable light not necessarily relating to large particles. These two algorithms are merged to one fraunhofer enhanced designated as FREE in the newer Rseries instrument configuration. 2.5 Dissemination Procedures Powders and liquids were disseminated under various scenarios to simulate practical applications. All disseminators were positioned as close as possible to the detector yet minimizing artifacts from lens contamination and air flow. This was about two inches for the dry powders and from six inches to 10meters for liquids. Orientation was routinely selected as perpendicular to the laser beam of the detector with a few trials at ECBC using a offset of about 60 degrees to the laser beam with the gas powered micronair. The later was to avoid the transmission of heated exhaust gases thru the sensing zone of the detector. All disseminators were operated in static air (2010) and in the presence of an external 1.25m/sec breeze tunnel air flow (2011). All disseminators were positioned with a slight inclination upward (about 10degrees) to promoted lofting and transport of materials to 85meters. Dry powder disseminators were operated with and without an automatic hopper feeder. The automatic feeder was selected at two auger speeds designated as high and low speeds. Feed rates were calculated by weighing the hopper before and after dissemination. Manual feeding by hand was accomplished using preweighed one gram samples in a 50cc conical tube and slowly rotating the tube while monitoring the time to proximate 30 to one minute releases. The integrated air flow and feed in the gas powered micronair was routine selected at about three quarters throttle measured at 7000rpm using a stroboscope. At ECBC, other throttle positions were selected and measured between 940 and 8400RPM using a stroboscope. Full throttle was measured at 8400RPM. Feed rates for electric powered micronair were routinely selected for official trials at 0.5, 1 and 2LPM using a peristaltic pump yet feed rates of 50, 100 and 250ml/min were used at ECBC. Atomizer rotations of 7000 and 10500rpm selected by altering the motor frequency (180htz=10,500rpm) were routinely used. Although the rear air flow and fan speed of the electric micronair was adjustable, the factory preselected speed was used. 3.0 RESULTS AND DISCUSSION 3.1 Material Physical and Chemical Properties 3.1.1Dry Powders 3.1.1.1 Milling ………. commercial vs ECBC inhouse prepared BTK Size distribution of dry powders is primarily determined by the extent of milling or grinding. For biological organisms, one particle is almost never correlated with one organism but is an aggregate of many organisms. The grade of BTK used in these trials was that of one of the commercial products4 and a very coarse grain material. Freshly cultured and finely milled BTK (BTKW or BTKB) from ECBC shows a distinctly smaller distribution than BTK commercial (BTK COM). Material of significant different sizes or bulk density can have pronounced effects on perceived size distribution if not mixed thoroughly before sampling and measurement. Finer particles will settle more rapidly in a container from vibration causing coarser particles to be over represented at the surface. MSDS5 Validation of ARD Fine and Ultrafine size distributions BTK commercial available from Able, Biobit, Cutlass, Dipel, Foray, Javelin, Thuricide and Vectobac; http://npic.orst.edu/factsheets/BTgen.pdf; National Pesticide Information Center, Oregon University 5 Arizona Test dust MSDS’s (Fine and Ultrafine); Powder technology Inc.; 9 Jan 2008 4 Comparison overlays of geometric mean diameter as a function volume % for ARD ultrafine material measured with the Sympatec and as supplied by Particle technology Inc., Burnsville, MN using the Coulter multisizer Acccomp 1.19 Comparison overlays of geometric mean diameter as a function volume % for ARD fine material measured with the Sympatec and as supplied by Particle technoly Inc., Burnsville, MN using the Coulter multisizer Acccomp 1.19 3.1.1.2 dry powder containers and sampling The trials conducted in this report occurred over two years from material acquired from large stocks at DPG. Although sampling from a container is an important variable in size distribution and referenced in the ISO documentation, no knowledge is available on the specific lots or containers and how the material is stored. Samples removed at different stages from a single large volume container without sufficient mixing may result in variable size distributions occurring from settling the formation of a container gradient over time. Precision variability within the same material measured over the course of the two year period at ECBC and DPG is expected. 3.1.1.3 Fluidizers – Prevention of Particle Agglomeration & Enhanced Dispersion uniformity Open sensing detection requires control over dispersion to insure the equivalent of good mixing and uniformity such as in closed cell arrangements of other instruments. Sufficient velocity thru the sensing areas, particularly for larger particles, is important to avoid repetitive measurement from recirculation. This becomes challenging for variable sprays with wide spray cones and disseminators using centrifugal blowers and peristaltic feeders. Material type and their associated chemical and physical properties, origin and retrieval from large containers and the degree of milling for a particular material all result in size distribution differences relating to dispersion ability. The use of hydrophobic fluidizers such as the aerosils and cabosils for dry powders promote agglomerate breakup and fluidity. Monitoring distribution statistics and obscuration for a particular measurement provides an indicator of aerosol uniformity thru the sensing area. During a kaolin series of three trials with 16 two second measurements per trial with the R4 lens, the X50 in micrometers and obscuration as COPT were plotted as a function of time as displayed by measurement series number. The change in measurement statistics are in turn visualized and reflected in the distribution series overlay as a shift in the mode quantities and ratios. A change or drift in COPT results in a corresponding change or drift in X50 resulting in poor precision graphically visualize by poor density distribution overlays. Ideally, a perfect dispersion thru the sensing area is expected to yield a uniform X50 and COPT over time approximating that observed in trial 3. Radical changes in COPT, as observed in trial 1, are usually attributed to starting and stopping dissemination, exhausting material while the measurement is being conducted or manual feeding. The first or last measurement spectra in a series often results in phantom large particle modes and are consistently removed during post processing. Note the apparent inverse relationship of COPT and X50 is a measurement series. While the sympatec does not allow for real time monitoring of these statistics, the new Spraytectm by Malvern Corp does. Kaolin series trials in sep 2010 3.1.1.4 Hygroscopicity (ie SAVS and proteins) Environmental effects on dispersion are likely to have an effect on size distribution particularly with proteinaceous and toxin materials where molecular composition interacts with moisture which is in turn affected by temperature. These trials were conducted at DPG, UT with a ‘relative’ humidity of approximately 17%. While such low moisture is expected to minimize agglomeration from the highly hygroscopic materials, dry conditions tend to promote high electrostatic conditions having an impact on dry powder or the electronics of the instrumentation. a data rate of 1 sec was used using an accelerated handfed feed rate over a 15sec duration. Measurements with a COPT below 0.8% were not incorporated into the statistics during post processing. These were typically the first couple of measurements due to an early activation of measurement or the later measurement when the material was exhausted. Manual hand feeding was difficult to maintain continuous uniform concentration and resulted in the COPT falling below 0.8% toward the end of the measurement period. 3.1.1.5 Electrostatics on agglomeration and dispersal during dissemination (ie. CART) Non covalent interactions, electrostatic during dispersion and vander waals during prolonged storage Material chemical structure and the presence of surface functional groups interacting with other particles, surrounding medium and dispersion forces has an effect on physical characteristics. Residual salts, sugars and proteins are hygroscopic resulting in swelling, adsorption to other particles and surfaces (ie.Clumping) and retardation of drying. The net effect can be an overestimation of size and an increase in difficult to clean lens contamination in the case of the sticky albumins. The very large X50’s observed with ovalbumin is expected to be variable in different environments of temperature and humidity. The CART is a unique disseminator. While conducting trials, the CART operator experienced a massive discharge of static electricity and might explain the unusual behavior experienced by the Sympatec. The sympatec was positioned close to the large metallic superstructure of the CART with its metallic nozzle a few inches from the laser beam and electronics housing. Shortly after the onset of the CART trials, the Sympatec began to experience random errors only remedied by a total reboot. Removal and isolation of the instrument from the CART vicinity and operated along the walls of the JABT could not reproduce any malfunction. The CART trials were abandon until the last night with the idea of attempting to thoroughly ground both the CART and Sympatec although this could not be pursued due electrical storms and power outages. The apparent electrical disburbances on the Sympatec might explain in part the increased variations (sd) in measurements observed as compared to the skilblower. The limited amount of data taken with the CART does suggest additional breakup of material as expected from greater air shear. This is the case with kaolin and the finer ARD dirt but not with the coarse ARD 10-20 exhibiting virtual no change. BTK may also exhibit similar behavior as the coarser ARD 10-20 yet only one trial was executed before malfunction of the Sympatec. 3.1.2 Liquids 3.1.2.1 Viscosity and Surface Tension Quantities of salts, proteins, detergents and other constituents used to reduce surface tension influence fluid behavior and droplet formation. Bacterial and viral preparations in various stages of purification with different diluents of differential composition are likely to affect droplet size. During droplet measurement, the diffraction pattern may be altered/interfered by material refractive index governed by light transmission through the droplet after refraction and absorption. LD spectrometry is generally believed to operate on the principle of fraunhofer diffraction yet only when the particles are large compared to the wavelength of light and when the ratios of refractive indices of particles and transport medium are clearly different from unity. Thus fraunhofer approximation is expected to be better with larger, more opaque material and iregular shape as these parameters appear less susceptible to refractive index6. Conversely, as spherocity and droplet transparency increase, the requirement for the mie model of analysis increases7. The sympatec F-series configuration does not allow for ‘expanded’ mie analysis of diffraction patterns acquired from lenses larger than R4. The newer sympatec Rseries configuration or the Malvern ‘spraytec’ is perhaps better suited for aqueous phase droplets. “The effects of refractive index (RI) selection on reported particle size results are most pronounced when particles are spherical, particles are transparent, when the RI of the particle is close to the RI of the fluid, or when the particle size is close to the wavelength of the incident light. If particle characteristics deviate from any or all of these conditions, the refractive index value selected will have a smaller effect on the calculated particle size results.”37 Horiba Tech Note; “Guide to selecting Refractive Index” TN118; http://www.horiba.com/fileadmin/uploads/Scientific/Documents/PSA/TN118.pdf 7 Paul Webb; “A primer on particle sizing by Static Laser light scattering” micromeritics technical workshop series; Jan 2000; http://www.particletesting.com/docs/primer_particle_sizing_laser.pdf 6 Two biological simulants, EH and MS2, designated as ‘SAVs’were observed to have different colors dependant on purification degrees associated with different compositional characteristics. Exact purification and composition is unknown at this time. EH SAVs were designated as 24Dirty (Dark) and RYP Clean (light brown). MS2 SAV preparations were designated as A for Clear and B for Brown both of which were still translucent. Diffraction patterns subjected to two algorithms and visually compared with the degree of purification suggest droplet size is both influenced by absorptivity and surface tension with perhaps the later having a greater role. The absence of the large particle mode (~100micrometers) in the darker and perhaps more concentrated diluents of EH is expected with a greater decrease in surface tension leading to a preponderance of smaller ligament/ fragmentary droplets. This is consistent with manufacture specifications at the respective atomizer rotation with the percentage of larger 100 micrometer droplets decreasing yield the smaller VMD size of about 40micrometers at 10,500 RPM. No information is available on the purification of EH or MS2. The inorganic salt composition of TNE buffer used for MS2 is believed to have lesser effect on surface tension than the residual components of the salts, sugars and proteins in TSB broth. Hoffman et al. (2011) have incorporated detergent at concentrations of 0.25% to reduce surface during droplet measurements of water from flat fan nozzles. Additional research is needed. 3.2 Disseminator Design and Operation The design of the disseminator affects agglomerate breakup, droplet shear, dispersion and plume shape, volume output and size distribution especially for on-demand droplet generators. Even for preprocessed milled dry powders, the extent of disseminator shear can result in a size distribution beyond its native inherent state from deagglomeration to communition. Size distribution has been initially observed at ECBC and Dugway to be controlled by disseminator type, model and serial number using the Topas, SRI nozzle, Skilblower, Sympatec RODOS, and several different AU8000’s during trials in 2010 and 2011. While this is expected due to the design and extent of shear on the material, differences in sizes of dry powders throughout the two year campaign is believed to be due largely to the small sizes below 10 micrometers, operation within narrow working distances and a desire to maintain lens cleanliness through the use of lens attachment devices (air curtains). With an exception of BTK commercial which surprisingly shows near identical medians (X50) among the two year at ECBC and Dugway, the various other powders are observed with greater medians (X50) in 2011 compared with 2010. With the truncation of the smaller size particles due to vignetting from the use of air curtains in 2011, this is expected as discussed later. Aside from the vignetting issue with smaller dry powders and small focal length lenses, Kaolin, with its lack of fluidity and platy characteristic, can result in greater differences among trials and diseminators. To reduce the impact, aerosil or cabosil fluidizers at about 5% have been added. Differences among disseminators in the case of larger droplets from the atomizers may be a different situation where the vignetting is unlikely due to sizes in the realm of 100 micrometers and much greater working distances used with the larger focal length lenses. At least three gas micronairs with different serial numbers were used in 2010 (designated as Micronairs 1-3) with X50’s ranging from 60 to about 83 micrometers (see statistical summary for gas micronair). With the integration of feed rate with RPM in the throttle positioning and the known inverse relation of droplet size and RPM, differences in size distribution might be expected. LD analysis exists both in cuvette and open sensing configurations where the former is much easier to attain the requirements for a narrow working distance. However this narrow working becomes more important as the size of the particle to be measured decreases. This limitation makes this technique ideal for the much larger primary droplets and dry particles emanating fro sprayers and blowers. Anomalies typically occur on either side of the distribution dynamic range and can be observed as an absence of data or incorrect sizing due to vignetting, or to additional modes from beam instabilities due to vibration and or refractive index differences within the dispersing medium. Beam steering from refractive index differences in the transport gas (air) can presumably occur from the formation of eddies or vortices caused by particle/ droplet energetics emerging at short distances to the sprayer/ blower outlet8. Environment influences primarily with thermal and air current disturbances are common with open sensing measurements some of which can be minimized by appropriate referencing or isolation by enclosure. 3.2.1 Shroud Effects An alternative approach to extending small particle measurement in widely energetic spray/blower cones using the R3 lens involves tapering the 2.5inch blower outlet to less than 1.5 inch prior to transmitting the dispersed powder thru the sensing zone. The resulting dispersed particle stream proximates the stream dimensions generated by the sonic SRI nozzle providing comparable distributions regardless of lens used. 8 Ghosh S. and Hunt, J.C.R., “Spray Jets in a Cross-flow”, J. Fluid Mechanics 365: 109-136, 1998 Cambridge Press 3.1.3.1 Dry powders Particle size distribution (dV/dlogDp) of ARD fine material disseminated with the SRI nozzle (solid) and tapered skilblower (dotted) using the R3 (black) and R4 (orange) lenses Tappering of the Skilblower outlet from 3” to 1.5” allows for the disseminated powder to traverse thru the narrow working distances of the smaller lenses. The industry standard for powder dispersal is the RODOS disseminator insuring constant and uniform flow and concentration thru the sensing zone. Overlaying a distribution generated with the Rodos dispersion standard (Blue dashed) in Jun 2011 to the tapered skilblower(dotted) and SRI nozzle (lines) shows excellent agreement. 9 3.1.3.2 Liquid Sprays 9 Leschonski K, Rothele S, Menzel U; A special feeder for diffraction pattern analysis of dry powders; Particle Characterization vol 1; pp. 161-166; 1984 Visualizing via digital videography10 shows two distinct populations with clear tap water and MS2 TNE buffer solutions, one described and referred to as a ‘Halo’ circumferencing yet perhaps asymmetrically about the rim of the cowling approximately 4 inches from the head of the atomizer and the ‘major flow’ being propagated from near center of the cowling. The ‘Halo’ is visually observed with a very rapid settling rate suggesting very large droplets. The ‘Halo’ flow is considered secondary to the spray emanating from the sprayer as it does not persists long enough to contribute to the airborne aerosol. Ground contamination and sprayer efficiency is affected greatly by the ‘Halo’. The rapid setting at the top of the ‘halo’ implies some combination with the major flow to be transported further downwind. Operating in a wind or breeze tunnel enhances the mixing of the ‘Halo’ with the central flow. Direct interrogation of the ‘halo’ by the Sympatec was performed by positioning the atomizer head approximately 6 inches from the laser beam (~2 inches from ‘halo’). This proved difficult from extreme obscurations, lens contamination and presumed aerodynamic air disturbances generated at close proximities by large droplets. The Eh or tryptic soy broth suspension and solutions are repeated observed with minimal ‘Halos’ and is translated to the distributions as more monomodal. This may be attributed to a lower compositional surface tension. Bimodality of centrifugal spinning disk and hollow cone atomizers have been observed, reported and attributed to film droplet and ligmentary atomization droplet formation modes11. The control over these modes are dependent on air shear and atomizer rotation. Electric micronair (AU6349) spray and cowling with associated positioning and setup of the sympatec laser beam at about 6" in 2010 Gas Powered Micronair (AU8000) spray plume and skeletal shroud Gas powered micronair produces different distributions dependent on orientation. Presented perpendicularly the same as the electric, a small shoulder appears at the same mode (~40micrometers) as the primary flow. There are obvious visual differences between the dispersion plumes of the gas and electric Micronair atomizers resulting from differences in cowling or shrouding designs around the atomizers. The purpose of the cowling in the electric micronair is not clear but could be simply for mounting purposes and/or to control cross-sectional uniformity of the plume. Bimodal droplet populations are more clearly distinct in the electric than the gas powered yet both interfere with the measurement process and the resulting perceived distribution. Others have compared distributions across the plumes of cda atomizers 12 and smaller anvil spray 10 Videography performed and provided independently by DPG Personnel Frost A.R. “Rotary Atomization in the Ligament Formation Mode”, J Agricultural Engineering Research 26(1), 63-78, 1981 12 Enaburekhan J.S. and Kaul R.N.; “Effect of shroudingon the volume distribution pattern and droplet size of two commercial c.d.a herbicide applicators”; Tropical Pest Management 29(4) pp. 339-345, 1983 11 heads13 showing distinct differences among shrouds. We have attempted to assess cross sectional plume distributions but found extreme difficulty using the high volume controlled droplet atomizers (c.d.a.) atomizers and contaminating the lens or disrupting the airflow thru the sensing zone causing artifacts. 3.2.2 Rotational Effects 3.2.1 Centrifugal force in Dry Powder blowers The skilblower uses a mechanical stirring attachment setting atop of a high speed centrifugal blower to forcibly disseminate material to a cloud. The internal motor provides sufficient centrifugal force to promote impaction/ caking of material to the inside walls. As time progresses, some material is more than likely released by sloughing to form large flakes or chunks This has been most frequently observed with the ARD fine powder and typically and apparently manifests itself at the 260 micron mode. Interestingly, this 260 mode was not observed in the 2011 trials. Rational might either be due to the few short and quick trials executed or to the newer Skilblower used. The 260 mode occurs frequently and repeatably with other skilblowers yet intermittently over a 40 second trial where perhaps only one third of the 10-15 2-sec measurements may show this mode. The occurrence of this mode has also been observed with BTK at ECBC but only at very high feed rates with lower frequencies. Difficulties in powder dispersal in regard to agglomeration and increasing with smaller particle size have been discussed and are the primary purpose of the RODOS14. Uncertainty remains as to whether this 260 mode is real or an anomaly occurring from beam steering in the presence of eddies generated during measurement of sprays at close proximity to the disseminator outlet. Inside of the outlet chute of the Metronics Skilblower disperser (left) displaying an accumulation of Kaolin undoubtedly resulting in the shedding and impaction of large particulate flakes during dissemination with subsequent fallout onto a concrete floor (right) Volume Size distribution for ARD fine material measured with the R4 lens showing the 260micron large particle mode presumably caused by flaking or sloughing off of impacted material in the throat of the skilblower Kippax P, Parkin S and Tuck C.; “Particle size characterization of agricultural sprays using laser diffraction”; ILASS-Europe 9-11 Sep 2002 14 Leschonski K, Rothele S, Menzel U; A special feeder for diffraction pattern analysis of dry powders; Particle Characterization vol 1; pp. 161-166; 1984; http://www.sympatec.com/docs/LaserDiffraction/publications/LD_1984_RODOS.pdf 13 3.2.2 Spin rate in liquid atomizers Micronair atomizers are among a class of controlled droplet atomizers (c.d.a.) for the ability to control droplet size by atomizer rotation thereby controlling small droplet drift and larger droplet fallout and ground contamination. The early work by Bode et al (1972) has contributed to the development of modern day high volume atomizers. 15 As liquid is extruded from the woven wire gauze cylinder, ligaments are pinched off into smaller droplets as rotation increases. The pinch off process is influenced by surface tension, viscosity and density. Bode et al among others have observed the cross sectional non uniformity of the spray plume with a greater volume at the periphery than at the center. This is expected has larger droplets are thrown outward by centrifugal force. The largest droplets thrown beyond the shroud and into more static air are abruptedly removed by sedimentation causing the center flow to be perceived as the dominant flow increasing with feed rate. The volume mean diameter (VMD) of the droplets is inversely proportional to the rotation from approx 2000 to 10,500 rpm.16 Two rotational speeds were used in 2011, 7000 and 10,500 rpm, showing similarities with the manufacturer specifications yet at the first mode of the bimode and proportional to feed rate. VMD, however, representing a combination of modes was typically about between 60 and 80micrometers dependant on sprayer serial number. Referring to the visual videography plumes of the electric micronair with its clear dual population, mode one referred to as central and primary flow at 40 and 50 micrometers closely corresponds to the manufacturer VMD specifications for 10,500 and 7000 rpm, respectively. This suggest the importance of atomizer position or ‘set back’ from the laser beam providing the very large droplets an opportunity to be eliminated. The gasoline powered atomizer in contrast does not show a well defined ‘Halo’ as the larger droplets are able to experience the airflow longer from the skeletal type shroud. Manufacturer stated VMD for a given atomizer rotational speed This smaller fragmentary primary population mode of 40 micrometers at higher rotations of 10,500 is shifted larger to near 50micrometers at reduced 7000rpm as expected. The 97micron mode observed with water and MS2 appears absent with the EH at higher RPM’s. This might be due to reduced surface tension of the increased media concentration and composition (~3%) in combination with the high air entrainment resulting in cleaner ‘pinch offs‘ of the evolving droplets. Ligamentary atomization is likely more efficient at higher rpm in the presence of surface tension reducing media accounting for some bimode at 7000 yet none at 10,500 rpm in the EH material17. The apparent lack of the bimodes with the R7 lense may be more due to a lack of resolution at the low end of the dynamic range. Alternatively, Teipel (2002) observed difficulties in determining the size of water mist droplets with a median diameter of 20 micrometers if fraunhofer approximation is used without the full mie theory18. Translucent water and MS2 preparations at higher surface tensions are observed here to produce bimodes at both 10,500 and 7000rpm. Uncertainty remains whether surface 15 Bode LE, Zimmerman TL, Goering CE and Gebhardt MR, The effects of flow rate on the distribution pattern and drop size spectrum of a spinning atomizer” Transactions of the ASABE, 1972 16 Micronair Direct Drive Industrial atomizers Operator’s Handbook and Part Catalogue; Micronair Sprayers LLD, Bromyard Industrial estate, Herefordshire, England 17 Muhammed I, Muhammed A and Sayyed A.H.; Effect of Power on droplet Size by Hand Held Spinning Disc Sprayer; Pakistan Jounal of Biological Sciences 8(40: 567-570; 2005 18 Teipel U.; “Problems in characterizing transparent particles by laser light diffraction spectrometry”; Chemical Engineering & Technology, 25(1), pp. 13-212002 http://www.dbi-net.dk/media/Doctoral_thesis_Bjarne_Paulsen_Husted_Main_Part.pdf tension or refraction within translucent droplets is the cause. The gas-powered Micronair (AU8000) yields droplet sizes inversely proportional to atomizer rotation as expected with other spinning disks arrangements. Using the R6 lens, successively increasing rotation produces smaller droplets with transition modes. Atomizer droplet size in the absence of rear airflow Agricultural atomizer sprayers are essentially droplet generators in crossflows provided by a rear fan and airflow. As the droplets emanate in a stream from the atomizer, the rear air or cross-flow over the atomizer primarily functions to remove the droplets from the stream and propels them forward. The rate the droplets leave the stream is dependent on size and inertia. The smaller droplets in the stream, influenced more by the airflow by their reduced inertia, are removed first with the large particles centrifugally thrown further outward. 19 This results in the formation of the primary and smaller droplet central flow of the plume with a larger droplet ‘halo’ centrifugally positioned around the outer perimeter of the plume. A direct measurement of the emitted atomizer droplet stream was performed by operating the unit without a rear air flow positioning the atomizer such that the droplet stream is perpendicular to the laser beam. Results provide further evidence the two modes are inherent products of the atomizer as expected. A shift from bimodal for the water and MS2 materials to a monomode with the TSB suggest an effect of surface tension resulting in a cleaner ‘snap off’ of the ligaments as surface tension decreases. The monomodal shift with the TSB is observed to proximate the manufacturer reported 40/50 micrometer mode and may be related to fraunhofer diffraction and the extent of material light absorbancy (more opaque) discussed later. Size distribution of rotating atomizer without rear air crossflow Size distribution of ejected droplets of water (left) and 3% tryptic soy broth (standard 30gm/L) (right) at 500ml/min using the R6 lense from the electric micronair without a rear airflow and positioned such that the direction of ejection was perpendicular to the laser beam. Atomizer rotation was set at 180htz or ~10500rpm. Spectra taken on 28 June 2011 at 15:22:40 and 15:31:28, respectively. 3.2.3 Turbulence The divergent nature of sprays as they spread conically away from the nozzles increases the risks of detector lens contamination. Compromises between optimal working distances for a specific lens and sufficient space to allow for the divergent spray are necessary in addressing large particle anomalies from disturbances in large sensing volumes. At the JABT in Dugway, anomalies were plentiful perhaps in part due to the larger tunnel entrance and tunnel volume potentially impacting flow fields. In particular at various times due to external wind gusts, we perceived oscillating wind speeds at 22meters from the entrance even with a constant control room set point of ~1.25m/sec. The control of flow fields in open sensing is of particular concern and addressed with the use of the RODOS in the ISO 13320 document20. Straddling a sympatec across a wind tunnel similar to and referenced by Hoffman et al, 2011 to measure sprays may reduce the non uniformity of air. 19 20 Ghosh S. and Hunt, J.C.R., “Spray Jets in a Cross-flow”, J. Fluid Mechanics 365: 109-136, 1998 Cambridge Press http://www.sympatec.com/docs/LaserDiffraction/publications/LD_1984_RODOS.pdf 3.2.3.1 Airflow Induced 3.2.3.1.1 Mixing Sprays are inherently variable attributed to the droplet formation and ligament fragmentation, shear breakup and dispersal process. Droplet sizes range from the smallest of satellites to very large coalescence modified. The high variability require significantly longer measuring times to average out the variability. Air entrainment in sprays, producing turbulence among particles/droplets results in complex air-droplet dynamics. While some turbulence is desirable for mixing and to insure homogeneity, excess turbulence leads to recirculation artifacts within the sensing zone. 3.2.3.1.2 Vortices from retardation of large particle/droplet acceleration A phenomenon called ‘beam steering’ caused by Schlieren effects due to density differences21 (ie. Thermal or pressure) in air is a consideration in open sensing measurements where environmental influences are present. Vortices depicted by Ghosh S. and Hunt, 1988 are suggested here to be areas of differential pressure causing beam steering if atomizer position to sensing area is too close Sprays contain eddies or small pressure cells of differential refractive. As the diameter of a moving particle or droplet increases, so does the amount of turbulence generated by its resistance in the airflow. Although eddies can be numerous close to the sprayer outlet they typically dissipate quickly22. Cognizance of the numerous eddies at the nozzle dissipating quickly to about 20cm from the nozzle is important for identifying an appropriate measurement location (ie. Setback) and reducing artifacts from pressure cell refractive index differences. Disseminator proximity positioning to Beam can possibly influence large particle refractive index associated anomalies. Vortices can also be introduced to the sensing area from anthropogenic activity or the placement of objects upwind of the sensor. 3.2.3.2 Thermal Induced Introduction of heated air to the sensing zone is particularly important in the measurement of droplets from internal combustion engines. 3.2.3.3 Background Referencing Background referencing or establishing appropriate controls to substract non-specific signals is an an important metric in any analytical procedure and is further discussed and highlighted in the ISO 13320 guidelines. Referencing the sensing zone in the presence of the sprayer airflow without feed provides notable differences in elimination of phantom counts caused by the introduction of thermal gradients pulled thru the sensing zone. The real time and instantaneous nature of open sensing detection coupled with the randomness of turbulence as the driving force of dispersion does not lend itself to a complete elimination of anomalies from background referencing. Anomalies caused by pressure eddies in close proximity to the sprayer outlet cannot be referenced out as they are a direct result of droplet production and transport thru air. Increasing the disseminator setback from the beam to 24 inches might eliminate the 1 mm mode but will undoubted create more lens contamination problems from the high volume sprayers.23 3.2.4 Dispersion Control According to the international standard Iso 1332024 standard for laser diffraction, flow rate and material concentration maintenance monitored as Optical Concentration (COPT) thru the beam is important for quality measurement. Particle Bjarne Paulsen Husted; Experimental measurements of water mist systems and implications for modeling in CFD”; Doctoral Thesis, Lund Univ Sweden 2007; http://www.dbi-net.dk/media/Doctoral_thesis_Bjarne_Paulsen_Husted_Main_Part.pdf 22 Ghosh S. and Hunt, J.C.R., “Spray Jets in a Cross-flow”, J. Fluid Mechanics 365: 109-136, 1998 Cambridge Press 23 Fritz B.K., Hoffman W.C. and Bagley W.E.; “Effects of Spray Mixtures on Droplet Size under Aerial Application Conditions and Implications on drift”; Appl Engineering in Agriculture 26(1), pp 21-29; 2010 24 Particle size analysis for laser diffraction methods (2009), International Organization for standardization (ISO) 21 recirculation in the sensing zone, particularly for small particles, and turbulence should be avoided.25 During the infancy in development of modern day LD, the Rodos powder disperser was developed to achieve this control in dissemination. However, such control is difficult to achieve due to the variable nature of dispersion for a particular disseminator or in the design of the feed mechanism such as with the use of peristaltic pumps. Dispersion in open sensing zones is based on real time or instantaneous ‘sample prep’ and mixing concurrent with measurement where the parameters of concentration, Turbulence and pressure should fall within an ideal range. TOO LOW Poor Signal/Noise <-Insufficient mixing <-Poor Agglomerate breakup <-- IDEAL TOO HIGH [Concentration] <-- multiple scattering/ difraction [Turbulence] <-- recirculation artifacts [Pressure for shear] <-- comminution or rare re-agglomeration Sufficient concentration is necessary to effect appropriate signal to noise but in excess results in multiple scattering. Turbulence is required for mixing but in excess results in recirculation and multiple measurement of the same particles. Sufficient pressure is needed for agglomerate breakup of dry powder from settling, yet in excess can result in communition and smaller than expected sizes. 3.2.4.1 Feed rate Effects on concentration and obscuration and multiple scattering Elevated quantities of material required for laser diffraction measurement might be considered a drawback in situations where materials are limited. As an ensemble measurement technique, achieving sufficient concentration within the sensing volume is important in achieving appropriate signal to noise. This is accomplished by monitoring the optical concentration (COPT) as reported by the instrument. Optical concentrations are dependent on droplet size and feed rate. As with other analytical instrumentation and when plotted as a function of concentration or feed rate, obscuration is linear within the designed detector response range up to the point of occlusion. Laser diffraction guidelines suggest a minimum of 5% COPT for dry powders, yet experience suggest COPTs as low as approximately ~ 0.8% are sufficient for valid measurements as in the case of the SAVS proteins where material was limited. Dry powder material distributions are typically an order of magnitude smaller at 10micron than liquid primary droplets at 100micron. This contributes to the optical obscurity since many more small particles/droplets can occupy the same equivalent volume for larger droplets. The highest recommended concentration of 35% for liquid droplets is greater than that of 15% for dry powders before multiple diffraction becomes an issue. Feed rate is not expected to affect distribution statistics as observed with the dry powders. Distribution mean (VMD) and median (X50) of the electric Micronair (AU6349) appear to decrease with increasing feed rate using translucent water and MS2 suspensions while increasing slightly with the same increasing feed rate for the slightly colored EH suspension (figure ). This is expected from the more monomodal characteristic of the EH presumably due to surface tension differences discussed later and bimodal characteristic of the translucent water and MS2. The apparent decrease in Effect of AU6349 Feed rate on COPT, X50 and VMD as measured using water and the R6 lens on 18 Aug 2011 in JABT distribution statistics in the electric micronair for water and MS2 is thought to be an effective disproportionate increase in the smaller mode from a rapid elimination of the larger mode due to settling. Larger droplets are centrifugally ejected into more 25 Heuer, M., Witt, W., Rothele, S., Extension of laser diffraction into the cm region, PARTEC 1995, Nuremberg; http://www.sympatec.com/docs/LaserDiffraction/publications/LD_1998_ExtensionofLD.pdf of a static air mass outside the shroud resulting in collisions, coalescence and fall out. Other than the expected general quantitative monomodal increase with the EH with increase in feed rate, flooding of the wire cage mesh atomizer at increased feed rate could result in an increase in VMD and X50 as more volume emerges from the screen to form droplets. No detectable change in VMD and X50 were observed in the monomodal distributions of the gas powered Micronair for the limited trials performed. However the gas powered micronair is difficult to study since both rpm and feed rate are unfortunately integrated to the throttle position and cannot be observed independantly. Figure with tap water shows the reported COPT is increasing linearly from 100ml/min to 1LPM as expected for a quantitative increase in feed rate. This corresponds to the linear yet normalized modal increase at the primary manufacturer reported flow of ~40 micrometers. The slight increase in VMD and X50 observed is again expected and presumably occurs from a flooding of the atomizer screen as discussed above. During the initial setup week and validation at DPG, an extra lower feed rate of 100 ml/min was used with the AU6349 and water that was not used in the official trials providing for an additional data point for graphical visualization and quantitative effects of feed rate (figure ). The data is supported by earlier data at ECBC showing the linear effect of feed rate from 50ml/min to 1LPM at the 40 micrometer mode when overlayed distributions are normalized to 97micrometers (Figure ). COPT disproportionately increases from a feed rate of 1 to 2 LPM with little or no change in distribution statistics of VMD and X50 suggesting a loss in detector response as COPT reaches 55% in excess of the recommended 35% (figure ). A slight observed decrease in statistics may be evident of multiple diffraction as expected. Newer LD instruments such as the Sympatec R-series configuration and Malvern spraytec apparently have algorithms to compensate for multiple diffraction. Tribalier et al.(2003) in working with Malvern’s spraytec observed limitations to these compensating algorithms.26 In 2011droplet size distribution measurements were made for three liquids using all possible combinations of two lenses, two atomizer rotations and three feed rates. Lens resolution, determined by dynamic range and the effective range and location of the individual concentric rings of the detector and the influence in resolving the 40/97micron mode in addition to the appearance of the larger 1 mm mode, has been discussed previously. Feed rate, as has also previously been discussed, affects both the optical concentration in the sensing volume proportionately as expected and the distribution statistics, X50 and VMD. The translucent feed materials of water and MS2/ TNE buffer typically better resolves the 40 micron mode and increases with increasing feed rate thereby resulting in a decreasing overall mean and median. More opaque liquids as with EH appear to typically loose the resolution of the 40 micron mode resulting in an opposing observed increase in mean and median. Some evidence of the 40-50 micron mode is present at 7000 rpm with the R6 lens yet absent at the higher 10500 RPM. Atomizer rotation consistently supports the evolution of larger droplets with decreasing RPM as expected from ~80 to ~110 micrometers for 10500 and 7000 rpm, respectively, for the secondary mode. A size of 97micrometers was observed for this mode at 10600 rpm at ECBC yet again might be explained with different disseminators. Similarly, the first mode of 40micrometers with the R6 at 10500 rpm was observed to increase to ~50micrometers at 7000 RPM. The apparent absence of the 50 micron mode using the R7 lens for both water and TNE buffer while apparently present with both lens at 10500 is not clear. 3.2.5 Positioning of disseminator to detection laser beam 3.2.5.1 Orientation to Intersecting Laser Beam Orientation of the disseminator relative to the laser beam Triballier, K.C, Dumouchel and Cousin, J.; “A technical study on the spraytec performance: influence of multiple light scattering and multimodal drop-size distribution measurements”; Experiments in Fluids, 35(4), pp. 347-356, 2003 26 impacting particle presentation to the optics can have effect on apparent size distribution. Some disseminators because of their size cannot be oriented in any other way then directly perpendicular without contamination of the optics. The gas micronair as a consequence of its heated internal combustion gases and associated adverse anomalies from refractive index beam steering was preferably oriented approximately 60 degrees to the laser beam. Differences in the distribution portrayed in figure between 60 and 90 degrees to the beam suggest a role of droplet shape factor as the material is being presented to the optics. Although LD guidelines reported no significant measurement differences with shape owing to the symetricality of the diffraction patterns, measurements and associated phenomenon are based on a perpendicular dispersement thru the laser beam. Not shown in the figure for the orientation perpendicular to beam axis (diamonds), distributions tend to portray a slight shoulder (more than shown in this distribution) at 40/50 microns similar to the electric micronair. This might be explained by examining the images of the plumes for both micronairs with their distinction spatial arrangement. In the gas micronair, the large and small droplets are more intermixed emphasizing the larger mode. Distributions of the electric micronair are more influenced by standoff positioning from the laser beam where the extent of the 97 micron mode is determined by the extent of large droplet rapid settling or fallout. 3.2.5.2 Disseminator setback or standoff from Laser Beam Disseminator proximity placement to the laser beam, referred to as setback or standoff, is important to reduce the effect of air velocity on distribution. Air velocity thru the sensing zone reduces with increasing setback. Hoffman (2010) observed an increases in droplet size with velocity increase from 0.5 to 6.7m/sec and attributed the increase to overestimating larger droplets as the smaller droplets have reduced resident times I the sensing zone27. In any case, a distribution must be annotated with the distance it was measured. 3.2.5.2.1 Standoff at 6” vs 17.5” Comparing distributions obtained in 2010 and 2011 at an atomizer head setback distance from the laser beam of ~6 and 17.5”, respectively, the 1mm mode is present at both distances but the magnitude and frequency of the event is reduced as expected with increasing setback distance as eddies dissipate quickly. Although differentiation between real or phantom remains problematic, the cause may perhaps be due to refractive index differences within eddies caused by turbulences at close proximity. If these modes occur from large droplet eddies close to the outlet as a consequence of their slower acceleration, one might expect the incidence to be reduced with distance as observed and previously discussed. The 1 mm mode is present in distributions using the R6 and R7 lenses and appears to increase as feed rate increases to 2 LPM compared with that at lower feed rates of 0.5LPM as expected. However suggestive the evidence is for beam steering, uncertainty remains as to real or artifact and provides rational for further investigation. Effect of AU6349 distance positioning from the sympatec laser beam designated as ‘setback’ on the appearance and magnitude of the 1mm mode as performed at Dugway in 2010 and 2011 does not necessarily exclude the phenomenon of artifacts from high visual obscurity or beam steering from refractive index differences in pressure cell eddies produced by droplets emanating from the nozzle . Setback 6 inches atomizer to laser (2010) Setback 17.5 inches atomizer to laser (2011) Hoffman, W.C.; Fritz, B.K.; Bagley, W.E. and Lan, Y.; “Effects of Air Speed and Liquid temperature on Droplet Size”; J. ASTM International; 8(4) 2010 27 3.2.5.2.2 Transport distance at 10meters (30ft) from dissemination The electric micronairTM was placed approximately 10meters upwind of the sympatec in the ECBC breeze tunnel and operated at three feed rates between 0.5 and 2LPM at 10,500rpm using the R6 lens, tap water and TSB broth. Measurements consistently yielded obscurations less than optimal with obscuration (COPT) drift during the sequence when focusing between measurement was turned off. Obscurations increased from about 0.5-0.8% at 0.5LPM to about 1% at 1 LPM and 1-3% at 2LPM for tap water. TSB broth was slightly higher at 2-4% and 2LPM presumably resulting from the increase in yellow color. Three out of ten spectra for the single trial at 0.5LPM producing acceptable obscurations >1% using tap water and the R6 lens (Figure … ECBC tunnel 29Jun2011 15:48) showed distributions similar to that at closer proximities to the detector with COPTs greater than 10%. The two bimodal distributions had an average X50 of 52 micrometers with the monomodal distribution at 48. COPTs were 2.1 and 1.4, respectively. All six sequential trials on the same date measured at 1-2 LPM resulted in near log normal monomodal distributions as indicated by ratios of X50 to VMD being close to unity. Sympatec reported X50’s between 58 an65 micrometers, close to those reported in Dugway 2011at a standoff distance of 17.5 inches. A total of 18 spectra yielded an average X50 of 64 micrometers and a standard deviation of 6.4. Obscuration drift at 10 meter from the micronair disseminator is unknown yet suggest beam instability at the larger focal lengths. This did provide encouragement that sufficient material was being transported thru the sensing zone for downwind measurements providing the feed rate was maintained greater than 1 LPM. This lends support for further research. 3.3 Lens Focal Length Expected particle and droplet size determines the selection of the lens to be used. Smaller lenses with very small focal lengths are require to capture the wide angle diffraction rays onto the detector. This represents the Achilles heel of laser diffraction where manufactures appear to over exaggerate the limits. As with other spectrometers, working at the extremities of the dynamic range can be misleading and unreliable. Selection of lens so that the distribution lies central to the dynamic is important. In the sympatec f-series instrument, two lenses were only available for liquid droplets, R6 and R7. While R7 was used primarily to identify very large millimeter sized droplets, the R6 was slight larger than preferred. An R5 lens, similar to that used by Texas A&M between 2008 and 2010 for c.d.a measurements is preferable shifting the distribution more central to the dynamic range leaving less question to the first mode at 40micrometers. 4 Water No Tunnel Flow 250ml/min 3.5 R7 Jun22 15:05 X50=72 3 2.5 2 1.5 1 0.5 0 1 10 100 1000 10000 3.3.1Working distance (Vignetting) All dry powders were measured with an R4 lens in 2010 at the expense of a lower effective sizing limit of 1.8 micrometers . With an increase in working distance from 3.8inches of the R3 to 5.1 inches of the R4, minimal vignetting is believed to have occurred regardless of the wide and energetic spray cones and with the assistance the high volume blower to divert particles away from the lens. In an effort to lower the sizing capability to less than one micron, the R3 lens was attempted in combination with a lens protection cap protruding about 2 inches into the working distance. Without sufficient space for the 2.5 inch skilblower outlet, much of the dispersed powder was outside the working distance resulting in vignetting and a gross ‘roll off’ of particle sizes less than approximately 4 micrometers.28 R3 size distribution data provided in this report should be viewed with caution. The clean air curtain designed to reduce material impacting the lens in 2011 was not used in 2010. The width of the curtain being approx 2 inches reduces the working distance allowance for wide mouth disseminators. This is more evident when the R3 with a 3.8inch compared with the R4 with a 5.1 inch working distance was used in 2011 (figure ). ‘Roll off’of the small particles is observed in both situations, albeit more so with the R3 lens. To alleviate the issue of working distance in LD and extend the sizing capability below one micron, wet dispersion is accomplished using a cuvette where the working distances can be reduced to less than an inch. This is only useful for material without disseminator characterization. The distributions reported by Ling et al (2012) using wet dispersion on a coarse grade of kaolin agrees well with our data above two micrometers with the R4. The provides additional sizing to 0.5 micron using the CILAS 1180 LD system, fraunhofer analysis, and a small focal length lense.29 This approach, however, does not lend itself to disseminator (sprayer or blower) characterization and can result in large particle agglomeration if extended sonication is used in wet dispersion measurements. 3.3.2 Contamination Some form of lens protection is required for droplet size measurements from wide angle sprayers when measuring at close proximity. While any attachment to the lens impacts the narrow working distance for small particles causing vignetting, material impaction and lens damage from abrasion becomes more serious with large outlet energetic sprayers. Dry powders because of their small size and lack of moisture are less prone to accumulation on surfaces as compared with moist droplets yet the potential for lens abrasion remains high. The occurance of lens contamination is visually in the spectra by the appearance and accumulation of millimeter sized particles. The laser source and detector compartments of the Sympatec varios F-series model used in these trials can be separated at a distance exceeding 50 inches promoting the measurement of large divergent sprays. The choice of small focal length lenses (R3 and R4) have the advantage of measuring distributions from ~1-10micrometers but require working separation distances of less than ~5inches that are non-conducive to overspray. With an increase in separation distance the added sensing volume is sensitive to larger particles and/or disturbances in airflow. 28 Personal communication with Jim Garten, APL Ling F.N., Kassim K.A., Karim A.T.A.; Size Distribution Analysis of Kaolin using Laser Diffraction Technique; Advanced Materials Researc, Vols 341-342; pp. 108-112, 2012 29 For dry powders using the Skilblower and droplets using the AU6349, a distance not to exceed ~26inches appeared to work well at ECBC for achieving distributions with minimal anomalies. At times the separation distance was increased by as much as 32 inches for the R6 and 44inches for the R7. The increase in separation was selected in an attempt to capture the entire cone spray within the sensing volume rather than cutoff a portion as was done earlier at ECBC. Further research is needed on the impact of vignetting with large focal length lenses. IMAGE OF CONTAMINATION SPECTRA 3.3.3 Vibration induced Beam instability Phantom large particle counts occur as a result of light rays striking the inner detector rings producing a progressively larger size number with increasing lens focal length and dynamic size range. With the largest focal length lens (R7), there are two distinct phantom mode regions of the LD spectra appearing to correlate with air related refractive index beam steering: 1) the 1 mm mode and 2) modes from the last two detector elements corresponding to >2mm. As the dynamic range is decreased with the R6 lens, these modes are unresolved with the upper size at 1750 micrometers. . As the beam travels further with the large focal length lenses, they are more susceptible to vibrations causing the beam to ‘waffle’ or oscillate resulting in beam steering or ‘wings’ as light rays strike the adjacent two rings of the photodetector closest to the centroid. These ‘wings’ or phantom large particles interfere with diffraction rays of real large particles striking this area. Two software techniques are available to apparently identify and remove these phantom modes; 1) software algorithm and 2) ‘forced stability’. The later manual forced removal effectively reduces the desired dynamic range redistributing the volume contribution over the remaining dynamic range. This technique was used for all of the droplet distributions obtained with the R6 and R7 lens resulting in good coefficient of variance of results when multiple trials were performed on the same material and disseminator. A similar mode, albeit unlikely, at 260micrometers is observed with only the ARD fine dry powder and the R4 lens with an upper size of 320micrometers. High resolution Fraunhofer LD (HRLD) is an algorithm apparently designed for special emphasis to the large particle modes and for that reason is preferable in droplet spray measurement. Large particle phantom counts observed in the standard Franhofer LD algorithm are frequently removed by HRLD. The HRLD algorithm was used for real time processing and visualizations during all the trials, with subsequent repost processing using standard LD for further analysis and statistical purposes. Comparing the number with the volume distributions, large phantom modes are the result of exceedingly few particles with very large volumes. They are seldom, if ever, displayed in the instrument converted number distribution. Phantom large particle counts are frequently not completely removed by HRLD, masking the volume distribution of the smaller droplets. 3.3.4 Laser focusing Sensing volume increases with lens focal length and beam diameter increases. Monitoring COPT through a measurement series provides information on the stability and uniformity of dispersion required for good precision. Large focal lengths apparently reduce the instruments ability to maintain beam focusing at the centroid of the photodetector at least in the case of larger liquid droplets. Prior to measurement, the beam is focused to the centroid of the detector and is assumed to be maintained throughout the measurement period. Each measurement trial is composed of 16 2-second spectral series views. Focusing the beam only once prior to start of the series results in a gradual increase in obscuration over the coarse of the series. To somewhat mediate this problem, refocusing the beam between each 2-second spectral measurement has been observed to stabilize the obscuration while reducing beam steering and subsequent phantom large particles. The incidence of this phenomenon while using the R6 and R7 lenses and not with the shorter focal lengths support beam stability. Better referencing, however, with the sprayer fan on without the atomizer may improve obscuration drift. Additional focusing requires additional time and can extend the total time for each trial by a factor of four. Intermediate focusing is not always possible with high output sprayers and limited quantities of material because of the length of time required to obtain a focus. This approach was not used at Dugway in 2011 due to restricted material quantities. 3.4 Data analysis and Interpretation 3.4.1 Number/ Mass/ Volume Distribution The method used to describe an aerosol is dependant on the purpose. Small particles dominate the number distribution while large particles dominate the mass or volume distributions. While the terms can be difficult to comprehend, aerodynamic mass and volume are emphasized during transport, dispersion and aerosol sampling because of inertia. Number distribution is important in clean rooms and is predominantly used in generic particle counters due to their simpler design and cost. Number distribution is limited to information about the particles such as behavior. Since volume is a cubic function of number where 1000 one micrometer particles would occupy the same volume one ten micrometer particle, the large particle mode represents a small percentage of the total number yet a high percentage of total volume. This exceedingly small number of particles represented by the large volume mode is emphasized in one ARD fine trial over extended measurement duration where the mode occurs only about one-third the time. Low occurrence has apparently been observed to adversely effect lidar systems.30 Comparing the various depictions of a distribution for any particle population is important before making conclusions. ARD fine (2010) using the HRLD algorithm number (q0) vs volume (q3) probability density functions (pdf) showing the large particle volume modes as a result of just a few particles in the number distribution. 3.4.2 Algorithm Comparison 3.4.2.1 Dry Powder ARD Fine Sympatec supplies three algorithms for the deconvolution of diffraction patterns to produce cumulative percent size distributions over the dynamic range for a specific lens. Each algorithm focuses on a specific size range to assist in differentiation. Standard LD and High resolution LD (HRLD) are based on fraunhofer theory of laser diffraction. Note the LD and HRLD versions in the F-series configuration used here and dropped in the newer R-series configuration to MIEE and FREE designations referring to 30 Communications with Evan Thrush, JHUAPL enhanced versions. Standard fraunhofer (LD) is recommended in most situations, particularly the smaller particle dry powders, supplementing with HRLD to better define the large particle fraction and identify phantom modes. 31 However, HRLD has been generally reported to be better for sprays 32 where larger particle modes predominate. With HRLD, large particle phantom modes are generally removed or minimized that would normally be observed with LD. Our results indicate subtle differences when the LD and HRLD are compared, with the HRLD tending to slightly shift the distribution to the larger sizes. The mie algorithm is most suitable for very fine spherical particles where density and the real and imaginary refractive index components are known. Mie use is nearly exclusive for spherical translucent water droplets less than 10 micrometers22. Using a tapered output to the skilblower to insure the ARD fine aerosol was being confined to the narrow 3.8inch working distance, the acquired diffraction pattern was subjected to three algorithms. Standard LD and HRLD fraunhofer analysis show nearly identical distributions with the HRLD removing intermittent large modes observed more frequently LD. The reduction in fine particles less than 2micrometers by mie analysis may not be neccessarily accurate due to crude literature approximation of density at 2.65, and real (n) at 1.5 and imaginary (k) at .0233 34 which may be at the high end of the range. Notice a similar reduction of fines less than 2 micrometers is observed when fraunhofer distributions are overlayed with distributions obtained with the R4 using the Rodos and tapered skillblower (Figure ..). Unfortunately data is not available at this time for the Rodos using the R3 lens. These analysis were performed with the older F-series sympatec configuration with software designated as mie, LD and HRLD where the analysis under mie is limited to smaller particle sizes acquired by lenses of R4 or lower. This might suggest incomplete or poor analysis in the upper size ranges of even the lower lenses. The newer R-series configuration software now combines and replaces the fraunhofer LD and HRLD for an apparent extended dynamic range analysis using software designated as FREE. Likewise mie in the F-series has been replaced with the extended version ‘MIEE’. Fraunhofer, however, has been observed to over- or under- estimate ultrafine particles dependent on particle size particularly less than two micrometers. While mie maybe better suited for liquid analysis with smaller and more spherical particles, fraunhofer appears preferable for sprays providing they are sufficiently opaque35. The choice of standard LD vs HRLD to use for deconvolution of diffraction patterns has been a subject of confusion. Sympatec has acknowledged the apparent preference for initially using LD presumably in the smaller sized dry powders and supplementing with HRLD but the differences in the two algorithms have yet to be fully realized. HRLD is apparently automatically executed with the selection of LD by its annotation in the warning message of coarse particles when the instrument is not able to account for all the light. 31 Personal Communication Stephan Roethele, Sympatec, Aug 2011 Particle Size Analysis (PSA) Seminar, Pennington, NJ, October 19-20, 2011 33 WG Egan, Appl Optics 1982 34 Tuminello P.S, Arakawa E.T., Khare B.N., Wrobel j.M. and Querry M.R.; “Optical Properties of Bacillus subtilus spores from 0.2 to 2.5 micrometers”; Applied Optics, 36(13), may 1997 35 Gerben B. J. de Boer, Cornelis de Weerd, Dirk Thoenes, Hendrik W. J. Goossens, “Laser Diffraction Spectrometry: Fraunhofer Diffraction Versus Mie Scattering”, Part. Charact. 4 (1987) 14-15 [http://alexandria.tue.nl/repository/freearticles/620871.pdf] 32 3.4.2.1 Liquid 3.4.3 distribution smoothing [Spline fit] 3.4.4 Cumulative Probability Distribution Cumulative probability distributions (CDF) to obtain statistical population percentage contributions within particular size ranges can be displayed either as Cum% as a function of particle size or as the particle size as a function of cumulative probability less than the stated size. The later is preferred where the different depictions of the population as number, volume and mass can easily be visualized on the same graph. Since displaying non lognormal or bimodal distributions as a probability distribution using this approach provides a non-linear solution, the method is only applicable to log normal distributions producing a straight line. However, separate treatment of the individual modes of a non-log normal distribution is an acceptable practice.36 Since MS excel does not offer a routine, a clever spreadsheet was developed by Patrick O Shaunessey and made public by BGI. Figure .. provides a typical size distribution of MS2 in TNE buffer (Sep 2010 22:25) using an R7 lens illustrating two size modes at the 1 mm and 18-86 micrometer mode where the few large droplets in number mode overwhelms the volume distribution. The graphical display of the entire distribution as a log cumulative probability distribution clearly shows non linearity. Individual treatment of the modes of this non lognormal distribution, yields linear log cumulative probability distributions where the cumulative percentage fractions of the distributions can easily be obtained. 36 US Enviromental Protection Agency; http://www.epa.gov/apti/bces/module3/distribu/distribu.htm 3.4.5 Summary of September 2010 Sympatec Measurements 3.3.1.1 Dry Powders Characteristic volume size distributions of the dry powder materials measured with the R4 lens in 2010 using the HRLD inversion algorithm are provided (above). These represent one each of a successive series of the dry powders provided in the statistical summary using the LD inversion algorithm. Each dry powder was measured at least six times for a duration of about 40 seconds. The coefficient of variation of the X50 in the table for Kaolin, ARD Fine, ARD Ultrafine and BTK is 7, 3, 2 and 3 %, respectively, and is of reasonable agreement with the 5% variation of LD provided in the ISO guidelines. Note a forced stability of three was used in the analyses of ARD Fine where the 260micron mode is removed and might be considered a phantom coarse mode. The higher variation with the kaolin might be due to the platy character of the material. 37 Only three of the six total trials are used in the statistics for BTK because of the increased number of coarse particles observed in the vicinity of 260micrometers where the COPT was reduced below 10%. Using these trials with a reduced COPT, the coefficient of variation for X50 becomes 41%. When the LD algorithm is used, the HRLD algorithm is simultaneously accessed in an attempt to account for all the light spread over the detector. Failure to account for all the light implies absence or blockage resulting in a warning that coarse particles might exist. In the ARD fine and similar to BTK, a warning of coarse particles is obtained in the later 3 trials when the COPT is reduced to less than 7%. Ling FNL, Kassim KA, Karim AT, “Size Distribution of Kaolin using Laser Diffraction Technique”, Advanced Materials Research, Vols 341-342, pp. 108-112, 2012 37 3.3.1.2 Gas powered Micronair Wet Sprays Using R6 Lens Sprayer differences for same model and manufacturer Statistical Summary for the Coefficient of Variance of trials for the Gas Powered Micronair (AU8000) during the 2010 DPG Trials at the ABT i. ii. iii. iv. v. vi. vii. R6 lens, three materials, Three successive days in the DPG ambient breeze tunnel (ABT); ~7000 RPM ~3/4 throttle standard Fraunhofer LD deconvolution Omitting light contribution in the three closest ring elements (ie forced stability) of the detector centroid from beam 'waffling' 20% of 604 spectra were unusable due to triggering, lens contamination, referencing, insufficient material, etc %CV of the X50 is between 3-5% for water and MS2; EH significantlty deviated at 18% CV with a generally increased distribution broadening as reflected by the sg clues as to an effect from surface tension and absorbtivity Obscurations were <35% 98% of the time 3.4.6 Summary of August 2011 Sympatec Measurements Statistical summary for standard dry powder size distributions obtained in 2011by dissemination using the Skilblower and DPG’s engineered Cart. Data may not accurately reflect the fine particle fraction due to lens protection cap protruding into the 3.8 and 5.1 inch working distances of the R3 and R4 lens, respectively. Red Font style highlights discrepancies in Dugway's Log files for feedrates recorded in an isolated control room against those recorded at the instrument and supported by sympatec optical concentration (COPT) values VMD is the volume ‘mean’ diameter derived as a quotient of the fourth divided by the third moment about ‘0’ (Hinds) X50 is the volume ‘median’ diameter derived as the upper bin diameter less than the 50% cumulative The deviation between X50 and VMD reflects the departure from log normality The sqrt of the quotient of the 84th divided by the 16th cumulative percentages are provided only as a ‘What if’ log normal distribution, yet is better described using the ½ the single sd statistics (X50/X16 and X84/X50) about the mean portraying the magnitude of ‘skew’ and the ‘non log normal’ behavior. 3.4.7 Summary Comparison of 2010 and 2011(explanation of differences) Attempting to maintain lens cleanliness using various methods between 2010 and 2011in an effort to obtain consistent and precise measurements resulted in considerable difference in the dry powder material statistics. These differences were minimal for material having particle sizes greater than about 7 micrometers and included BTK, ARD 10-20 and to a lesser effect on ARD fine. This is predominantly from sufficiently narrow diffraction angles as particle size increases to maintain light ray impaction on the photodetector whether inside or outside the working distance. Table… for 2010 and table … for 2011 dry powders show considerable deviation for particles <5micrometers as the diffraction rays miss the photodetector by vignetting causing removal of fine particles from the distributions. The 2011 X50 data for especially Kaolin and ARD ultrafine are overestimated by about 4 micrometers. Values represent averages of multiple 2 second measurements with the R4 lens. Distributions are narrower in the 2011 data and are shifted larger attributed to vignetting in 2011 with the use of the wide lens protection system. ARD fine is processed using forced stability of three to remove the 260micron mode. Percent Volume (q3/lg) density Distribution Comparisons of Dry Powders using the R4 Lens with Different Skilblowers at Dugway during the 2010 and 2011 trials Kaolin ARD Fine BTK ARD Ultrafine Appreciably more particle size ‘roll off’ occurred in 2011 compared to 2010 with the R4 lens and a 5.1 inch working distance in combination with an approximate 2 inch lens protection cap with the 2.5 inch skilblower outlet. The ‘roll off’ in 2011further results in an increase in X50 for all four dry powders as particles less than about 5micrometers are eliminated or reduced from the distribution. The requirement for some lens protection system to reduce lens contamination artifacts from wide spray and energetic blowers dictates the need for further development and reduction in size of the lens cap. The lens protection systems needs to be sufficiently small to allow for the complete aerosol to be within the narrow working distance for the selected lens focal length while minimizing recirculation and turbulence effects from the outflow. Sympatec Inc does market a lens protection system but due to its larger size greater than our first attempt, is not suitable for lenses less than about the R6 38. Total Particle counts as a function of particle size for BTK measured at DPG in 2010 and 2011 using the R4 lens showing less “roll off” at the smaller sizes in 2010. The motor blower in 2010 was manually tapered to about 1.5” to allow for the 2.5” outlet. This is just greater than the 3.8 “ working distance. Some ‘roll off’ remains in 2010 with the use of the motor blower for lens protection but is not as severe in 2011 with a cap of 2 inches. This is further confirmed when the sympatec RODOS powder disperser is used to discharge a narrow plume thru the narrow working distance. 38 Communications with Sympatec during Oct 2011workshop Comparison of droplet distributions relative to feed rate, atomizer rotation and lens selection for Erwinia herbicola (EH) disseminated using the AU6349. Absence of the 1mm mode in the R6 lens is from removal of the later two bin volumes by ‘forced stability’. Note the more monomodal distribution at 10,500 RPM compared to the bimodal of water and MS2. Comparison of droplet distributions relative to feed rate, atomizer rotation and lens resolution for translucent materials water (Top) and MS2 (bottom) disseminated using the AU6349. Absence of the 1mm mode in the R6 lens is from removal of the later two bin volumes by ‘forced stability’. Size distribution comparison for material disseminated using Four Disseminators [TOPAS sag410 (30% feedrate; 25psi; two Metronics Skilblower (Dugway sep 2010 and ECBC Jul 2011); and Rodos ]as measured by Sympatec Laser Diffraction with an R4 Lens 4.0 Summary and future efforts Laser diffraction is a promising method for non-intrusive particle and droplet sizing high volume sprayers using guidelines and procedures described in ISO13320. However, obtaining accurate and consistent size distributions of material disseminated from high volume blowers and sprayers is not trivial contributing largely to the wide spray cones, spray variability, and narrow working distances for particles less than 10 micrometers. Distributions are highly dependent on material properties, disseminator, conditions of dissemination and instrument theory and therefore must be explicitly defined. Coefficient of variations for the X50’s of dry inert material simulants of kaolin, arizona dirt and a commercial coarse grade of Bacillus thuringenesis show good agreement with the ISO standard when the material is presented to the system within the appropriate working distance. Similar good agreement is also established when compared with disseminators having narrow spray cones that adapt easier to the narrow working distances. Low wt/vol (<3%) liquid suspensions of biomaterials using high volume cage atomizers show an order of magnitude difference in size distributions than the dry materials disseminated with the blower necessitating larger focal length lenses. Susceptibility to beam oscillations and phantom counts might be reduced with a preferred R5 lens with a cutoff of ~1mm providing for better positioning of the distribution within the mid spectral region similar to the design of the newer Malvern Spraytec laser diffraction system. Emerging algorithms to assist in fine particle fraction definition under ~5 um including deconvolution, vignetting and multiple diffraction are on the horizon. Future efforts will involve the exploration of these algorithms and use the source term size data to correlate with downwind transport behavior. Enhancement of the clean air curtain technology to reduce the size is in progress to enable improved non-intrusive fine particle sizing from blowers and short focal length lenses and working distances. Better LD methods are being engineered to cross sectionally interrogate high volume, divergent spray plumes to reduce obscuration and large particle phantom anomalies. LD is an established measurement technique for larger particles greater than about 30-50 micrometers but is continually being challenged for its ability to size smaller particles. The method using fraunhofer approximation excels when the particles are large and opaque where the refractive index ratio of particle and transport gas are significantly different from unity. There have been disputes and clarifications on the assumptions being made for LD39,40, the small focal lengths lenses required for small particles necessitating very small working distances and algorithms required for small particle measurement. The large lenses with much large focal lengths and larger cross sectional beams have drawbacks limiting the smallest particle to be measured and susceptibility to vibration and subsequent artifacts. Regardless of the extraordinary care needed in the use and interpretation of LD and its apparent results, it remains a powerful, rapid and precise method of particle/ droplet sizing from high volume sprayers and blowers. Future efforts should focus on particulars of the method including the many measurement factors such as control over sampling, material properties, environment and dispersion. Nearly all optical methods and instruments for size characterization are based on equivalent size and single particle measurement. However, this can be misleading based on the material chemical and physical properties. 41 Particle shape and agglomeration have an important role on characterization. Accurate and representative sizing requires the material to be presented to the optics without loss due to inertial and sedimentation impaction and isokinetic and isoaxial sampling. Open sensing LD remedies loss particularly for large particles by allowing for direct presentation without aspiration or transmission loss. Laser Diffraction instrumentation use both mie and fraunhofer theories to deconvolute the diffraction patterns with fraunhofer apparently providing better performance with large particles sizes generally ten times that of the wavelength of the laser source. This equates to about 6 micrometers in this case of the sympatec with its He-NE laser at 632nm42. Although Mie is reportedly more exact and covers a broader dynamic range excelling as particles decrease in size, it requires prerequisites that are not realistically or easily attainable18. Using LD with its large particle capability in conjunction with other sizing methods for small particles is an option to obtain a more complete and perhaps accurate particle size distribution. Johns Hopkins APL has attempted this and is further described in another report43. Identical principles of droplet formation are easily seen when comparing the two different micronairs with identical rotating screen atomizers. Modal ratios in distributions and methods for measuring are quite different dependent primarily on positioning to the sensing system as a consequence of cowling design and the exhaust gases produced by the gas powered unit. Both large and small droplets are observed in both units more distinctly defined in the electric powered version and intermixed in the gas powered version. When large droplets are removed by rapid settling with transport distance, the resulting small modes are Kelley R.N. and Etzler F.M.; “What is wrong with Laser Diffraction? A critical review of current laser diffraction methods for particle size analysis”; http://www.donner-tech.com/whats_wrong_with_ld.pdf 40 Paul Kippax; “Appraisal of the Laser Diffraction Particle-Sizing Technique”; Pharmaceutical Technology; March 2005; http://www.pharmtech.com/pharmtech/data/articlestandard/pharmtech/112005/150841/article.pdf 41 http://www.chemeurope.com/en/whitepapers/61205/measuring-particle-size-using-modern-laser-diffraction-techniques.html 42 http://www.linkedin.com/groups/Mie-Theory-Fraunhofer-diffraction-in-147418.S.134764466 43 R. E. Eager, J. F. Garten, Z. Chaudhry, T. K. Cossio, K. H. Keller, N. Salciccioli, E. P. Thrush; Source-Term Characterization Based on Aerosol Modeling Test 1 Measurements, FPD-R-12-0017; April 2012 (in Press) 39 predominant and appear nearly identical for a given atomizer rotation. Regardless of the absence of trials to segregate the larger droplets by transport, clues exist in the data. Further work is needed on engineering controls to measure these high volume divergent sprayers perhaps similar to the recent work published by Hoffman et al (2011) using a wind tunnel. 5.0 LITERATURE http://www.sandrelli.net/transport.htm Philips, J.C., “Agricultural Spray Droplet Dispersion in Turbulent Windflow”, University of Birmingham ENG, September 1997 Bauer, T.J., “Biological Slurry Spray Characterization”, 6th Singapore International Symposium on Protection Against Toxic Substances, 8-11 December 2009 Cristini V., Guido S., Alfani A., Bławzdziewicz J., and Loewenberg M., “Drop breakup and fragment size distribution in shear flow”, J. Rheol. 47(5), 1283-1298 September/October (2003) “Results Obtained with a New Instrument for the Measurement of Particle Size Distributions from Diffraction Patterns”, Dipl.Ing. Michael Heuer, Prof. Dr.-Ing. Kurt Leschonski, Particle Characterization 2, 1985, p. 7 (available at http://www.sympatec.com/EN/LaserDiffraction/Publications.html) Pöschl U. and Mikhailov E., “Water interactions of aerosol particles composed of protein macromolecules and salts: hygroscopic growth, microstructural rearrangement, electric and kinetic effects”, Geophysical Research Abstracts, Vol. 7, 05095, 2005 Mikhailov E., Vlasenko S., R. Niessner R., and Poschl U., “Interaction of aerosol particles composed of protein and salts with water vapor: hygroscopic growth and microstructural rearrangement”, Atmos. Chem. Phys., 4, 323–350, 2004 Volume (q3/lg) Density Comparisons for Dry Powders using the R4 Lens with the Skilblower and CART at the Dugway 2011 trials perhaps support the idea of poor dispersability of small particle powders due to agglomeration and adhesion forces and the effect of higher shear forces on further breakup Kaolin BTK ARD Fine ARD Ultrafine ARD 10-20 6.0 APPENDICES Statistical summary for the droplet size distributions obtained by dissemination using the AU6349 and post processed using ‘forced stability’ to remove volume in the last two bins for redistribution in remaining bins * Brown Shading: There were three Repeats for EH at 7000 rpm and 0.5 LPM as the prior two had spurious what we had believed were artifacts from presumably some lens contamination. A closer look at the data reveals the second of the three trials is more of the anomaly with the first, although variable, shows the 1 mm mode. The High (VMD) and sd’s are attributed to coarse particle anomalies at 1mm or residual large modes from photodetector elements not removed by the initial forced stability of two