Abslute Precisin in Particle Size Analysis Authr: Dr Graham Rideal Funder and Senir Scientist Whitehuse Scientific Ltd Summary Abslute precisin in particle size analysis can nly be achieved when the tw principle parameters f length and cncentratin (e.g. weight) are directly traceable t Internatinal Standards. Furthermre, the particles t be analysed must be spherical s their size can be described by a single dimensin. Techniques that fulfil these criteria are called primary methds and include Micrscpy, Sieve Analysis and Sedimentatin Analysis. This paper evaluates the accuracy f a 100- stage spinning riffler using electrfrmed sieves and then examines Snic Sieving as a primary particle sizing methd in the certificatin f a 20-100 micrn spherical reference standard. Intrductin Primary techniques f particle size analysis are methds where the units f length and cncentratin (e.g. weight) are directly traceable t Internatinal Standards. Laser diffractin r sedimentatin using an ptical methd t assess cncentratin are bth secnd rder methds because, in the frmer, cmputer mdels are required in deriving particle size, while in the latter, there is nt always a direct relatinship between ptical density and cncentratin. Optical micrscpy is prbably ne f the best knwn primary techniques because a measuring graticule can be calibrated against an Internatinal Standard. Thus, when a particle is measured against the graticule, it s dimensins becme directly traceable t the Internatinal unit f length. Similarly, the penings f a highly accurate electrfrmed sieve can be calibrated using micrscpy. If the assciated electrnic balance is als calibrated with reference standard weights, then bth length and weight are directly traceable, making sieve analysis ne f the few primary sizing techniques. Until recently, brad distributin reference standards have nly been available in irregularly shaped quartz pwders s cmparisns between different sizing methds have never been gd. With the advent f new spherical reference standards, it is pssible fr the first time t measure a pwder by several primary techniques dwn t 0.1 f a micrn and thereby give abslute and precise definitins f size. The spherical reference standards can then be used t assess the perfrmance f ther high-speed methds f particle size analysis. In a recent prgram spnsred by the Bureau f Certified Reference (BCR) based in Brussels, the Eurpean Cmmunity is develping a new set f brad range spherical reference standards. Whitehuse Scientific was asked t prduce a duplicate set f standards, smetimes called Mirrr Standards t shrt-list nearly 40 applicant
Abslute Precisin in Particle Size Analysis Dr G Rideal, Whitehuse Scientific Ltd labratries frm a rund-rbin test int five Certifying Labratries fr each primary sizing technique. The five primary sizing methds were: Micrscpy and Image Analysis Sieve Analysis Electrical Sensing Zne methd Gravity Sedimentatin (Andreasen pipette) Centrifugal Sedimentatin Whitehuse Scientific was the nly Eurpean labratry t be selected n all five methds. The Gilsnic Autsiever was a recmmended primary methd because: it can measure dwn t 5 micrns the use f electrfrmed r micrprecisin sieves makes the methd very accurate the nbard cmputer remves peratr dependence giving excellent reprducibility This paper describes the perfrmance f the Gilsn Autsiever in the BCR rund-rbin exercise and cnfirms the technique as ne f the leading primary particle sizing methds available tday. Experimental (a) The Gilsnic Autsiever Unlike cnventinal sieving where the sieves are either rbitally shaken, r mechanically r electrmechanically vibrated vertically, the Gilsnic system hlds the sieves statinary while rapidly mving the pwder vertically thrugh a nest f sieves using acustic energy. This is achieved by hermetically sealing the sieve stack tp and bttm with highly flexible rubber latex diaphragms and using an audi ludspeaker t generate a rapidly scillating clumn f air thrugh the sieves. A high frequency f 3600 scillatins per minute ensures rapid presentatin f the pwder t the sieve penings while amplitude is cntrlled t give excellent fluidizatin, even fr the highest density metal pwders. The nbard cmputer cntrls amplitude ramp up, duratin f test and amplitude ramp dwn, all f which can be prgrammed and recrded fr future use In additin, fr mre chesive r very fine pwders, vertical and hrizntal tapping can be implemented t keep the sieves clear. The sieving actin is schematically shwn in Figure 1 while the cmplete instrument and the cntrl panel are shwn in Figures 2 and 3, respectively. figure 1: Snic sieving actin. The cntained air clumn pulses the particles abve and thrugh the sieve meshes at 3600 times per min figure 2: GilSnic AutSiever figure 3: Cntrl Panel
Abslute Precisin in Particle Size Analysis Dr G Rideal, Whitehuse Scientific Ltd Fr the high precisin wrk required in certifying the reference standards, the three electrfrmed sieves per stack were sealed with PVC tape t ensure minimum pwder lss, especially fr the sub-20 micrn fractin. A 1g sample frm a 100-stage spinning riffler was first sieved n 90, 53 and 45 micrn sieves. The sub-45 micrn fractin was then sieved thrugh 32, 25 and 20 micrn sieves. Having ptimised the sieving prgram, the cnfiguratin was entered int the cmputer and repeated five times t check the reprducibility f the methd. (b) Sieve Calibratin Fr sieve sizes belw 100 micrns weaving technlgy is pushed t the limit, and variatins in aperture size becme mre significant. It is fr this reasn that the BCR insisted n the use f electrfrmed sieves fr the Certificatin prgram. figure 4: Micrscpe cmparisn f 20μm wire (left) and electrfrmed sieves (right) membrane in Gilsn sieves supplied by Buckbee Mears C., St. Paul, MN) The difference between the tw types f 20 micrn sieve is clearly shwn in Figure 4. Here a digitised micrscpe image f each sieve is reversed s that the sieve apertures can be cunted as particles. The Whitehuse Image Analysis prgram (ShapeSizer 4) measures the length and breadth at the centre f each hle t an accuracy f 0.1 f a micrn, calibrated against the Natinal Physical Labratry (UK) particle sizing graticule. The minimum dimensin is chsen as the aperture size as it is this dimensin that will hld back spherical particles. The accuracy f the electrfrmed sieve is clearly seen in Figure 5 where hld size variatin is 18.8 t 19.9 micrns cmpared with 18.4 t 24.1 micrns fr the wire sieve. figure 5: Size cmparisn f 20μm wire (left) and electrfrmed sieves (right)
Abslute Precisin in Particle Size Analysis Dr G Rideal, Whitehuse Scientific Ltd The unifrmity f shape f each aperture is als much better. Ninety-nine per cent f the electrfrmed sieve apertures have a maximum t minimum ratin belw 1.05 cmpared t 1.27 fr the wire sieve. The cmparisn is qualitatively and quantitatively illustrated in Figure 6. figure 6: Shape cmparisn f 20μm wire (left) and electrfrmed sieves (right) Results (a) Spin Riffling In rder t check the perfrmance f any particle sizing equipment, it is f paramunt imprtance that the pwders being presented fr analysis d nt vary frm sample t sample. Whitehuse Scientific was appinted as the main cntractr fr the subdivisin f the 10kg master batches because its purpse-built 100 stage rifflers culd reduce the 10kg t 1g in just tw stages. The reprducibility f the electrfrm sieves and the perfrmance f the 100 stage spinning riffler was determined as fllws: As the accuracy f a spinning riffler is much mre critical with carse pwders than with fine, the reprducibility f the electrfrm sieving system was first checked by subdividing a 10g sample f a 150-650 micrn reference standard eight ways n an 8- stage riffler and analysing each sample. The average weight lss, bth by sieve weighing and fractin recvery, was remarkably lw at 0.3% while the mean standard deviatin f the eight data sets was utstanding at 0.43% - see Table 1 and Figure 7. Reprducibility f Gilsnic Electrfrm Sieving 8 Stage Riffler Analysis 200-700 micrn Micrsphere Nminal Size Measured Size Average Fractin (%) Average Cumulative %> 710 712.6 2.880 97.120 600 602.3 6.672 93.330 500 498.8 6.244 87.085 425 424.2 12.679 74.407 355 354.0 29.414 44.992 300 300.7 23.627 21.365 250 250.7 10.082 11.279 212 214.3 6.784 4.495 180 180.2 4.166 0.329 Mean Standard Deviatin f 8 Data Sets = +/-0.43% Average SampleWeight=1.577gm Lss (sieve weighing) = 0.32% Lss (sample recvery) = 0.29% Table 1: Sieve Calibratin & Results
Abslute Precisin in Particle Size Analysis Dr G Rideal, Whitehuse Scientific Ltd figure 7: Reprducibility f electrfrmed sieving Table 2 and Figure 8 cnfirmed that bth the electrfrmed sieve system and the spinning riffler were f the highest calibre with recvery and reprducibility at 0.2% and 0.85% respectively. Nminal Size Gilsnic Electrfrm Sieve Analysis 5 Samples 200-700μm Micrspheres Measured Size Average Fractin Average Cumulative (%) %< 710 712.6 2.370 97.630 600 602.3 6.301 93.700 500 498.8 6.092 87.607 425 424.2 12.803 74.804 355 354.0 29.588 45.216 300 300.7 23.667 21.549 250 250.7 10.098 11.450 212 214.3 6.819 4.631 180 180.2 4.320 0.311 Mean Standard Deviatin f 5 Data Sets = +/-0.85% Average Sample Weight = 1.513gm Lss (sieve weighing) = 0.19% Lss (sample recvery) = 0.24% Table 2: Sieving Results fr the 100-Stage Spinning Riffler figure 8: Sample-t-sample variatin frm the 100-stage spinning riffler
Abslute Precisin in Particle Size Analysis Dr G Rideal, Whitehuse Scientific Ltd Samples f any size distributin culd nw be taken frm the 100-stage spinning riffler with abslute cnfidence knwing that sample-t-sample variatin was insignificant. It was then pssible t extend the analysis range f the Gilsnic Autsiever dwn t 20 micrns. Any sample variatin wuld then be due t the perfrmance f the sieving prcess rather than the samples presented t it. In the final sectin f this wrk, the Gilsnic Autsiever was used t certify the 20-100 micrn Whitehuse Reference Standard. The sieve calibratin and methdlgy have already been briefly described. Given that the pwder was s fine, the average sample lss after sieving five samples frm the 100-stage riffler was exceptinally lw at 0.4% (Table 3), while the standard deviatin f the data set was still very gd at 1.46%, Figure 9. Gilsnic Certificatin Sieve Analysis 5 Sample 20-100μm Micrspheres Nminal Size Measured Size Average Fractin (%) Average Cumulative %< 90 90.0 0.618 98.961 63 63.2 7.997 90.965 45 45.1 31.306 59.659 32 32.0 38.149 21.510 25 25.0 11.225 10.286 20 19.4 9.749 0.563 Mean Standard Deviatin f 5 Data Sets = +/- 1.46% Average Sample Weight = 1.155gm Lss (sieve weighing) = 0.396% Table 3: Gilsnic Certificatin Results fr a 20-100μm Standard Finally, tw additinal samples were taken frm the 100-stage carusel and analysed using tw ther primary particle sizing techniques specified by the BCR: Micrscpy and Image Analysis (Whitehuse Shapesizer4) and the Electrical Sensing Zne methd (Elzne). In the frmer methd, 6,000 particles were cunted, and the number based data was cnverted t vlume fr cmparisn, while in the latter, 50,000 particles were cunted and again cnverted t vlume distributins. In bth cases there was excellent agreement with the sieving methd, and an average standard deviatin f nly 2.2% was seen between the three methds, Figure 10. Acknwledgements The authr wishes t thank the fllwing individuals fr their assistance in this research prgram: 1. Dr. H. Mercus, Delf University, Hlland fr develping the electrfrm sieve analysis guidelines. 2. Prf. J. Ddds, CNRS, Nancy, France fr develping the Micrscpy and Image Analysis methd. 3. Dr. J. Llyd, Lughbrugh University, UK fr develping the Electrical Sensing Zne methd. 4. Mr C. Ward, Gilsn Cmpany, Inc., Ohi, USA fr his invaluable advice in ptimising the perfrmance f the Gilsnic Autsiever. figure 9: Reprducibility f the certificatin analysis figure 10: Cmparisn f three primary particle sizing techniques