FAQ - General

Particle Counting Efficiency

Question
Why should counting efficiency be 50% at the most sensitive sizing threshold?

Answer

As real-world particle counters do not have perfect resolution, it is necessary to specify a point that corresponds to the nominal sensitivity. The point selected by Particle Measuring Systems and most other manufacturers is the 50% level of counting efficiency.

Before discussing the reasoning underlying the decision to set the most sensitive threshold at the 50% level of counting efficiency, it is important to understand what is referred to as "half count" calibration. Selection of the mean size of a narrow distribution of standard particles at or very near a sizing threshold is a standard procedure in the industry because it facilitates calibration of the OPC. If a sizing threshold is positioned at the mean size, and the calibration sample is otherwise free of background counts, the counts will be split evenly between the adjacent size classes sharing the threshold as a common boundary. This procedure is referred to as "half count" calibration. A relatively small shift in the sizing threshold under test produces an easily detectable difference in the counts in those adjacent size classes.

Selection of the 50% level of counting efficiency, to specify the most sensitive threshold of a particle counter, produces sizing results at that threshold consistent with the results obtained at the other thresholds. At the thresholds, 50% of the standard particles fall below the threshold and 50% fall above. At all thresholds above the most sensitive threshold, 100% of the particles should be sensed and counted, with half of each calibration population falling below its respective threshold and half falling above. The only difference at the most sensitive threshold is that the 50% that fall below are not counted.

Selection of the 50% level of counting minimizes dependence on sensor resolution when selecting the most sensitive threshold. With the use of the 50% level of counting efficiency, variations in resolution result in a different displacement from the nominal threshold for the 0% and 100% counting efficiency points, but do not result in moving the nominal threshold (Figure 1). At any selection other than 50% counting efficiency, a change in resolution results in a change in the nominal threshold. For instance, if the 100% level of counting efficiency were selected, the nominal most sensitive threshold would only be valid for the OPC on which the selection was made. Differences in optical and electronic components and design differences would result in movement of the actual most sensitive threshold because of a change in resolution. The same logic would apply to any other threshold setting where one is using a half count determination of that threshold.

In addition, selection of the size corresponding to 50% counting efficiency in an OPC is essential for meaningful comparison of OPC data with higher resolution instruments such as the LAS-X or the HSLAS. As you can see from curve B of Figure 1, an OPC with "good" resolution, which counts 100% at 0.14 microns, would count about 95% at 0.12, about 50% at 0.10, etc. The range from 0.10 to 0.14 microns, corresponding to the size difference between 50% and 100% counting efficiency for the "good" resolution OPC, equates to a sizing error spanning four HSLAS size classes! Thus, the selection of an OPC's most sensitive threshold at other than 50% counting efficiency will result in significant counting errors relative to data collected by high resolution instruments.

Contact us if you need more information or have questions.