For a detailed discussion of the principles involved in measuring particles in liquids and gases refer to "Optical Particle Monitors, Counters, and Spectrometers: Performance Characterization, Comparison and Use" by Dr. Robert Knollenberg and Dr. Donald Veal. This summary will cover volumetric and in-situ measurement techniques as related to the measurement of particle sizes and concentrations in liquids. It will touch upon the basic physics of the optical design of the several PMS sensors. These principles are universal and can be used to compare measurement techniques of all manufacturers of optical particle counters.

Liquid opticalparticle counters are divided into two basic groups: Volumetric and in-situ. - Volumetric liquid sensors are unique in that they count and size all of the particles passing through the sensor, assuming that the particles are larger than the minimum size threshold. This is because all of the liquid passes through a capillary, a section of which is entirely illuminated by a laser. All of the particles in the liquid pass through this section of the capillary and as the do they are counted and sized. Refer to Figure 1.

• In-situ liquid sensors differ from volumetric sensors in that only a portion of the liquid passing through the sample cell is illuminated. Refer to Figure 2. The portion illuminated is called the viewing volume. Because not the entire sample is illuminated, it is essential that the portion of the sample that is measured is accurately defined. Both volumetric and in-situ sensors have two sub groups: spectrometers and monitors. The sizing resolution and sample volume of these sensors are basically what differentiate one from the other.
• Spectrometers size and count particles very accurately. This is simply because the sample of liquid being tested is illuminated with very uniform light intensity. Both volumetric and in-situ sensors can be spectrometers. In the volumetric case the entire capillary is uniformly illuminated and in the in-situ case a separate optical system is employed to insure that only particles in the uniformly illuminated portion of the sample are counted and sized.
• Monitors can be very sensitive and/or have relatively large sample volumes but do not size particles accurately. Because they illuminate the sample volume with the entire laser beam, some of the particles pass through the edges of the beam and, therefore, scatter less light than others that pass through the center (i.e.: the highest intensity portion of the beam). These sensors invariably exhibit a phenomenon called sample volume growth. Figure 5 graphically depicts this tendency. Sample volume growth will be discussed in the section covering monitors. Note that only the center of the laser beam is used to illuminate the capillary. This center portion of the beam has a field strength variation of less than 10%.

Figure 1: Graphical illustration of the illumination of a capillary in a volumetric sensor

Figure 2: Graphical illustration of the illumination of the sample cell in a non-volumetric sensor

Basic Principles

The fundamental consideration in measuring particles in liquid or, for that matter, also in gases or on surfaces, is signal-to-noise ratio. This is a measure of the signal generated by light scattered and collected from a particle relative to the noise inherent in the optical and electronic system employed. Sensitivity, sample volume, and cost are directly driven by signal-to-noise ratio.

Volumetric Sensors

Volumetric sensors were developed first. Figure 1 shows the basic optical system employed in our volumetric sensors. The liquid in these sensors is delivered to the illuminated sample volume through a small capillary. This technique allows measurement of particles down to about 0.2 microns in size. Light scattered from the internal walls of the capillary ultimately determines the background noise and thus the particle signal-to-noise limit. Figure 1 also shows that the capillary is illuminated by only the uniform intensity portion of the laser light which has a Gaussian light intensity profile. In Particle Measuring Systems' volumetric sensors, only the center of this light source (less than 10% intensity variation) is used to illuminate the sample. Because of this uniform illumination, particle can be accurately sized. That is to say, all particles of the same size pass through the same intensity light and, therefore, scatter the same amount of light. All particles pass through the uniform intensity section, therefore, all particles are counted and accurately sized (assuming that they are larger than the minimum detectable size). This is illustrated by Figure 3 which is a typical spectrometer characteristic curve. To produce this curve two things are required; first, the particles detected must be in a uniformly illuminated volume. Secondly, the sample volume must be independent of particle size.

The light scattering characteristics of particles changes from Rayleigh to Mie scattering as the size of the particles transit through the wavelength of the illuminating light source (refer to Figure 4). The non-monotonic potion of the scattered light characteristic is influenced by the angle over which the scattered light is collected by the light detector. A common technique for reducing the signal-to-noise ratio is to reduce the collecting optics' solid angle, thus reducing the stray light. This reduces the light scattered by the many surfaces next to the illuminated sample volume, thus reducing noise. Unfortunately, the optical resonance increases as the solid angle of collected light decreases. This has a direct influence on the sensor's ability to accurately size particles. Figure 4 shows the scattering characteristics of particles at two different solid angles for a single wavelength.

Figure 3: Characteristic response curve of a typical 0.2 micron spectrometer.

Figure 4: Calibration response function.

In-Situ Sensors

As stated before, there is a practical limit to the signal-to-noise inherent in volumetric measurement techniques. Because it is easier to polish a flat surface than the curved surface of a capillary and thus reduce unwanted stray light, sample cell design has gone to flat optical surfaces for illumination and scattered light collection. Also, 90º collecting optics are used to further reduce unwanted scattered light. Figure 2 shows a typical in-situ type sample cell. With this type of sample cell it is undesirable to illuminate the cell walls because of the stray light that would be scattered from these side walls in the sample cell. This partial illumination technique introduces the in-situ type of measurement where only a portion of the liquid is used for sizing and counting, the sensor is a spectrometer. As stated before, in a spectrometer the sample volume does not change with the size of particles being measured. The method required to sort out only uniformly illuminated particles requires multiple optical paths as show in Figure 2. One path is used to size the particle while the other is used to detect only particles passing through the proper portion of the illumination beam, i.e., the uniformly illuminated portion. If all particles passing through the laser illumination are sized, the sensors is considered a monitor and exhibit a comprise in sizing accuracy. Figure 5 illustrates the difference in scattered light relative to where the particle passes through the beam in a monitor. If the sensitivity threshold is set as shows in Figure 5, the volume of liquid is defined, and changes, for each size particle. One can easily see that depending on the size of the particle, the number of particles sized increases as the size of the particles increases. This is to say that larger particles passing through the edge of the beam will scatter enough light to exceed the threshold and, therefore, will be counted and improperly sized. Figure 6 illustrates this effect. As you can see, the actual volume sampled grows as the size of the particles increases. This is what is referred to as sample volume growth.

The ability is properly size particles is also adversely affected by non-uniform illumination of the sample volume. Figure 7 illustrates the difference in sizing response of a monitor and a spectrometer with mono-dispersed sample of polystyrene latex spheres (PSLs). The spectrometer accurately sizes the particles with very good resolution. The monitor tends to spread out the distribution by under-sizing the particles which are not illuminated at the maximum intensity. This is an illustration of the different amount of light scattered from the same size particles depending upon the distance from the center (maximum illumination) portion of the laser beam.

Figure 5: Illustration of the relative amount of light scattered versus the distance from the center of the laser illumination

[Note that the sample volume changes with particle size (i.e.: more of the larger particles get counted than the smaller ones). This is an illustration of sample volume growth.]

Figure 6: Characteristic curves of a typical size of 0.1 µm monitor

Figure 7: Comparison of sizing ability of monitors and spectrometers [The width of the sized spectra is a measure of the resolution of the sensor.]

Other Measurement Considerations

One of three basic types of information is desired in measuring particle contamination in liquids: the distribution of the sizes of the particulate contamination, long term variation of particle concentration above a given size and the occurrence of episodic events for alarming purposes.

To get size distribution information, concentrations are generally sufficiently large that sampling statistics are not consideration. Here sizing and counting accuracy and resolution are critical, requiring the selection of a spectrometer. An example of the use of a sensor in such an application would be to characterize filters.

For long term monitoring of a process, the time response of the sensor will depend upon the concentration of particles in the size range of the sensor at a given flow rate. The statistical significance of the sample time interval is determined by sample volume and flow rate.

In the case of episodic event detection the same statistical consideration (as is the case for long term monitoring) are necessary. This leads to consideration of statistical efficacy of particle contamination measurements. To get a statistically significant sample (95% confidence level) one must have at least 20 events in a sample. This is according to small sample statistics using the Student’s T distribution. Where this comes into play is in determining how much liquid must be sampled to be confident in the measurement of particle contamination. This in turn indicates what flow rate is required to get a statistically significant measurement in a reasonable time given the sensor’s sample volume. For example, to be confident (at the 95% level that a liquid has fewer than 0.5 particles per mL at some size, 40 mL must be sampled to get 20 particles detected. It will take 40 hours to get a valid sample if the actual quantity of liquid sampled is one mL per hour.

In the case of episodic event detection, sample statistics play a critical role in determining the sample interval necessary to establish statistical significance. If a short sample interval is chosen to facilitate episodic event detection, few, if any, particles may be detected during that interval. Thus, the information reported relative to particle concentrate is statistically insignificant. The sum of several insignificant measurements is in itself insignificant. The only way around this dilemma is to increase the volume of the liquid sampled in a given time period or to increase the sensitivity of the sensor which will result in more particles being counted because of the exponential increase in the number of particles as the size of the particles decreases. Particle Measuring Systems' sensors sample the largest volume of liquid per time interval for any given size of particle. The bottom line is the number of particles detected per time period per dollar spent.

To reemphasize, there are basic physical considerations common to all optical particle sizing and counting instruments. In evaluating these instruments one must first clearly define what measurements are desired and then evaluate each instrument against these requirements.