Transferability of Particle Size Distribution Data Between Manufacturers and Models
The transferability of particle size distribution data between manufacturers and models becomes problematic when upgrading from an outmoded analyzer to a new one, or when the supplier and purchaser of raw materials (powder and granular materials) own different analyzers.
Even supposing that data transferability unavoidably cannot be obtained because of different principles of measurement, it is pretty hard to come to terms with the fact that data transferability is unobtainable due to differences in model even between instruments that apply the same principle, i.e., laser diffraction particle size analyzers (i.e. particle size analyzers that operates based on the laser diffraction/scattering method).
Basically, factors that influence the transferability of data between manufacturers and models are the same as the "factors that determine measurement results" mentioned in the last lecture. Small differences in each of the factors build up and eventually result in transferability being lost.
The performance of laser diffraction particle size analyzers, such as measurement range and resolution, has improved remarkably over the past ten years. Due to rapid evolution and advances such as this, it is difficult to sustain 100 % data transferability when developing new models. Also, the technical gap between manufacturers has expanded, too.
And, there are also pitfalls in laying the stress on the transferability and continuity of data.
When large particles could not be detected by an existing analyzer due to limitations in measurement range or sampler circulating mechanism, accurate measurement will once again not be attained if a new analyzer is selected with the priority given to transferability with those measurement results.
As a matter of fact, there is a high possibility that particle size distribution data can be transferred between manufacturers and models when the conditions shown in Table 2 below are satisfied.
Table 2 Conditions for Ensuring the High Possibility of Transferability of Particle Size Distribution Data Between Different Manufacturers and Models
|(a) The particle size distribution of the sample must be continuous (i.e. smooth) and the distribution width must not be too broad.|
|(b) Pretreatment conditions must be made as uniform as possible. Clear and quantitative criteria must be provided for not just the dispersant or dispersion medium but also for dispersing conditions and sampling (fractionation/loading) conditions.|
|(c) Do not simply set the same refractive index before performing measurement. The refractive index that is ideal both for the model and for the measurement sample must be selected.|
|(d) The particle size distribution of the sample must be located at the center of the measurement range of the model in question. It must be located away from the measurement lower and upper limits.|
It is a fact that differences in data between manufacturers and models exist, as shown above. Yet, there are still many cases where differences based on other factors are mistaken or misconceived as being caused by the analyzer.
First off, the instability of the measurement target (sample) itself is sometimes the cause of differences in data. Even if you intend to measure the same measurement sample (target) on a different model, variance between lots or changes/deterioration of the measurement target (sample) itself may sometimes result in a situation where different measurement targets (samples) are measured.
Furthermore, some measurement targets (samples) are extremely sensitive to pretreatment conditions. And, if a mistake is made in the fractionating/loading method, distribution deviated from the population will be measured.
When evaluating data transferability, quantitative criteria used for evaluation also must be sufficiently taken into consideration. If data having a wide distribution is evaluated at 50 % (median diameter) or at other arbitrary % particle sizes, differences between data will be overvalued. Inversely, differences will become smaller if data is evaluated by arbitrary particle size %. If the data of steep distributions is evaluated at an arbitrary % particle size, the difference between data will be smaller, and, inversely, if the data is evaluated by arbitrary particle size %, the difference will be greater.
In this way, how criteria are used sometimes leads to different conclusions even when measuring the same data.