Exploring the science of particle counting
“This article will highlight, analyze and explain the controversy that particle counting finds itself in when compared to other accepted scientific analytical measurements and diagnostic tools.”
By Holger T. Sommer
Particle counting has been applied as an analytical and process control tool for almost three decades and has found acceptance in many industries. Industries which depend in one way or another on particle counting are: semiconductor, data
storage, pharmaceutical, medical device, food process, aerospace and fluid power industries to name the major users of particle counters for contamination monitoring. In general, all applications for which control of cleanliness or contamination level of manufacturing environments, raw materials, process media or products are key to productive and profitable operations. New applications for particle counting are continuously discovered and tried with more or less success.
This article will highlight, analyze and explain the controversy that particle counting finds itself in when compared to other accepted scientific analytical measurements and diagnostic tools. The physics of particle counters, methods and procedures, limitations and application are partly responsible for the poor image this powerful analytical quality control technique could have if properly used.
This brief overview into the physics, performance and limitations of particle counters will refer to application examples where limitations of the technology might make or break the success of particle counting in an entire industry.
Particle counting technology
Counting objects is basic to human nature and one of the first intellectual challenges we face growing up. Because of the basic nature and simplicity of the counting process, everyone understands it and believes themselves to be counting experts. This is true and acceptable as long as we can use our senses to identify the objects and can eliminate most of the subjectiveness by avoiding counting results based on criteria other than the presence or absence of an object. Counting, for that matter, does not permit maybes and is strictly black and white — the true reason why, when objects are counted correctly, everybody will agree with the result.
As soon as different selection criteria are introduced, the counting results will differ and controversy will arise. As an example we refer to a test in which two individuals have the assignment to count the number of red balls in a box. The box contains four red balls, three blue balls and two yellow balls. Obviously the correct result is: four balls. However, only one of the two individuals comes up with the right answer. The other person`s counting result is seven. Why? This person is color blind. To him the red and the blue balls appear the same. Based on experience, he knows that red has a somewhat dark appearance, so he concludes the result must be seven (4(red)+3(blue) = 7).
In the world of automatic particle counting, minor differences in particle appearance to the instrument will result in a huge difference in counting results. Particle counter manufacturers intentionally or unintentionally overlook minor or even major limitations of their instruments. This leads to controversy between instrument results for certain applications. The simple basic counting process now becomes a complex and very subjective task depending on the perception and interpretation of observations by the individual instrument.
For the interested user of particle counters, the basic physics of optical single particle counting are summarized here to a level of knowledge that will enable him or her to assess the usefulness of particle counting for his/her specific application.
Most particle counters used in various industries utilize optical detection to count and size small individual particles suspended in air or any other fluid (gases or liquids).
Optical single particle counters have been developed around two detection methods: Light scattering and light extinction. These two methods are related to each other as shown below but have very specific advantages and limitations depending on the counting/sizing application. Below both methods are presented, and their relationship is explained.
Light scattering
Many text books and publications have been written around the physical processes surrounding the interaction of light and small particles. It is not the purpose of this article to indulge in the beauty of complex mathematics describing the theory of light scattered by particles larger than the wavelength l of an illuminating light beam (Lawrence-Mie Scattering: for most laser-based particle counters l > 0.63 µm ) or smaller than that wavelength (Rayleigh-Scattering: for most laser-based particle counters l < 0.63 µm). The two theories exist because quite different physical processes and parameters govern light scattering of large or small (relative to the light wavelength) particles.
As long as a particle does not emit light, the light scattered by a particle carries information about the particle and the light source. For our purposes here, light may be characterized by two quantities: wavelength (l) and intensity I. With the constant velocity of an electromagnetic wave (light ) in a specific media, a relationship between Frequency (n), Wavelength (l) and Velocity of light (c) is given :
c = l n
Each wavelength of light corresponds to a specific color and with a constant velocity (c) to a specific frequency.
Light intensity is defined as the radiative energy (E) per unit area (A) from rays of all wavelengths. The radiative energy (E) is the integral of all specific energies e(l) summed over all wavelengths.
with this equation we get for the intensity I:
When a particle is illuminated by a light beam, the energy of the radiation source is redirected or absorbed. Redirecting the energy is called scattering. Two physical mechanisms are responsible for redirecting light energy: reflection and refraction (see Figure 1).
Reflection redirects light through surface interaction. This component of the scattering process depends very much on the size, shape and surface conditions of the particle. Refraction redirects light energy through refractive interaction. Light which enters a particle will change its direction because of a change in the speed of light inside the particle media. This component of the scattering process strongly depends on the optical material properties of the particle or the index of refraction of the particle material.
Light when scattered by a single particle into a specific direction in relation to its original direction has therefore a unique signature which relates to the size, shape, and material of the particle. Because the effect of size is dominant, the light intensity collected through a fixed aperture of the receiving optics of an instrument can be calibrated to the diameter of a sphere of reference material (polystyrene latex; see below). Although there is a dependency of the scattering signal on particle shape and material, it is commonly accepted to report particle size information in equivalent polystyrene latex sphere diameters.
Mathematically, the interaction of a particle is expressed through complex equations that exceed the scope of this article. However, to illustrate some dependencies, a few equations must be listed.
The signal Sscat collected through an aperture produced by a single particle interacting with a light beam of wavelength l is
The Scattered light intensity (Iscat) is a complex function of direction (f), collection angle (q), geometry (L), wavelength (l), particle size (d), optical properties (nFluid), (nParticle) and the incident light intensity (Energy/Area) I0
I0,f,q, L, l are reasonable constants for a particular sensor; d, nParticle, are variables with d having strongest impact on Iscat. dA is a surface element of the aperture area A across which the integration must be executed. The signal Sscat only depends on properties of a single particle and parameters of the apparatus.
Figure 2 shows the basic configuration of a light scattering sensor. The light intensity of scattered light is a function of direction and aperture size. The larger the aperture (collection angle) through which light is collected and concentrated on a detector, the stronger the signal for all particle sizes will be. In addition, the direction under which light is collected effects the signal. The near forward scattering direction of the light beam produces the largest signals. Especially for particles larger than 1 µm, this direction produces orders of magnitudes larger signals than in all the other directions. However, for a particular particle counting instrument, scattering direction and collection aperture are fixed. Calibration to a reference material is used to eliminate the unknown effects from particle shape and particle material properties.
Light extinction
Extinct light is defined as the amount of light energy removed from the original light beam through redirection (scattering) and absorption by a single particle. The light extinction signal is created by the light energy scattered in all directions (except in the direction of the detector) and the amount of light removed from the light beam through absorption inside the particle. Because the light extinction signal is integrated over a large scattering angle, it is a weaker function of the particle shape and material property than light scattering. This weaker dependency on particle material and a larger size range are the main reasons why light extinction sensors are more suitable for certain applications. A radiation energy balance around a single particle in a light beam reveals how the two single particle sizing and counting methods relate to each other.
Ein=Eout+Escatter+Eabsorb+Eother
Definition:Eextinct=Ein-Eout
–>Eextinct=Escatter+Eabsorb+Eother
Extinction signal
Figure 3 illustrates this formula. The rays of the collimated beam change direction through surface reflection; refraction at the interface between the fluid and the particle are due to material differences inside the particle. Absorbing material will retain some of the energy which penetrates into the particle (see also Figure 1).
Applications
CleanRooms magazine readers are presumably most familiar with particle counters monitoring the cleanliness of air in cleanrooms or other controlled manufacturing environments. The light scattering technique is used to detect, size and count individual particles in a sample air stream. Most complaints I have recorded from users of particle counters relate to instrument reliability and lack of trust in numbers these instruments report. Counting results from counters of the same manufacturer and model tested, in the same cleanroom in close proximity of each other may show large discrepancies.
Particle counter manufacturers are constantly defending the performance and function of their instruments but are hesitant to admit their basic limitations. I believe that in some cases, manufacturers have intentionally mystified particle counting for lack of rational explanations or results, calling it art rather than an accurate science. For most applications, particle counters are very powerful tools for providing valuable information that is essential for process control and economic manufacturing.
Particle counters as scientific instruments lack standardization. For a long time, particle counter manufacturers have failed to agree on basic performance requirements for their products. Instruments used in different applications face different requirements. An instrument optimized in performance for one application might fail in another because it lacks certain requirements which were sacrificed for the first application. This conflict can be illustrated by comparing the requirements for a particle counter used in the semiconductor industry and in the pharmaceutical industry.
The semiconductor industry is responsible for advancing particle counting to levels of sensitivity never anticipated in the past. A 0.05 µm sensitivity at 1 CFM (28.3L/min) sample flow rate is today`s state of the art. This advanced level of sensitivity borders on the limits of light scattering physics. Instruments with hardly any safety margin at their performance limit are subject to failure and malfunction. In addition, objective instrument evaluation becomes impossible for the average user completely relying on performance guaranties from the vendor.
The pharmaceutical industry`s objective in using particle counters is the satisfaction of FDA Good Manufacturing Practice. Here accurate numbers of particles are more important than accurate size determination. Because of the physics applied to detect, size and count small particles, the requirements for sensitivity, count accuracy and size resolution are inseparable. Maximizing for one desired performance property might reduce another.
There have been several documents developed which address calibration and use of particle counters. In RP-14 of the Institute of Environmental Sciences, ASTM F 649-80 and ASTM F 328-80 (1989) procedures and methods are described which allow for the calibration of particle counters under ideal laboratory conditions. Particle counters calibrated and used within the parameters set by these standards are performing and will produce reasonably accurate results. However, as soon as one or another parameter changes, major differences in counting results can occur because each particle counter (even of the same model) has slightly different adjustments, manufacturing tolerances and performance of components. Cumulatively, they are partly responsible for the differences in signals from particles and their interpretation in conjunction with particle counter electronics. Together with procedural variations and counting statistics, total particle counting results between instruments can vary drastically. More than 50 percent up to 200 percent have been observed and are the source of frustration of particle counter users. Manufacturers are at a loss to explain and are sometimes embarrassed to admit the perceived poor performance of their products. The reason for their embarrassment might be explained by the limited knowledge some manufacturers have about their own products and the applications these products are used for.
Lack of standardization
For many years particle counters were delivered to the pharmaceutical industry with counting efficiency performance leading to inaccurate particle count results. Instruments with counting efficiencies of less than 20 percent for 0.3 µm are currently used all over the world, even though the industry`s accepted value for counting efficiency at the sensitivity limit of any particle counter should be 50 percent. Because of the lack of common performance standardization, manufacturers are free to state their own specification. users depend on the accuracy of the vendor`s information because it is very difficult and expensive to verify and validate these specifications.
The Japanese Industry Standards JIS-9921-1989 (instruments counting airborne particles) and JIS-9925-1991 (instruments counting liquid-borne particles) are two instrument performance standards that focus on eliminating the instrument`s performance as a source of counting and sizing discrepancies. These two standards provide minimum requirements for particle counters combining the three most important instrument parameters for consistent performance, regardless of manufacturer and model (in referring to our example in the introduction, colorblindness would be eliminated).
The three instrument specifications are sensitivity, size resolution and counting efficiency. Because these three specifications are related to each other, performance boundaries must be met simultaneously for each specification. In JIS 9921-1989 and JIS 9925-1991 means are also described on how to best measure these parameters for instrument performance evaluation.
Sensitivity is defined as the smallest particle of size dsens the instrument will detect and count at nominal 50 percent (range 30 percent to 70 percent) of all the particles when challenged with particles of size dsens. The instrument will count only half of the particles because the sensitivity size threshold is set in the middle of the distribution produced by the instrument when challenged with particles of size dsens. (See Figure 4.)
1) For particles of sensitivity size (dsens) the counting efficiency is by definition 50 percent. Two other conditions must be fulfilled simultaneously:
2) The separation from the noise must be acceptable. This means the ratio Nn/Np of smallest counts between the noise peak (Nn) and the particle peak (Np) and the counts at the particle peak (Np) must be equal to or smaller than 12.
3) The counting efficiency of particles 1.5 times larger than the sensitivity particle size dsens has to be nominal 100 percent (range 80 percent to 120 percent).
The third condition implies a certain size resolution for the instrument since the slope of the increase of counting efficiency with particle size is directly related to size resolution.
The relationship of the size calibration curve (“Signal” axis; left) and counting efficiency (right vertical axis) is illustrated in Figures 5 and 6 for low and high resolution sensors. For the same calibration, a low resolution sensor will produce a counting efficiency curve, which is shallower (smaller rising slope) than a high resolution sensor. However, high size resolution, which is commonly perceived as a high quality instrument, will make the instrument more sensitive to counting errors! A small shift in size thresholds level (voltage) will lead to large count errors for narrow particle size distributions, especially when this type of distribution is produced by HEPA filters of a cleanroom installation near the sensitivity limit of the particle counter. The question is: what is important for the application, accurate particle size or accurate and reliable particle numbers?
Misperceptions
A newcomer to the family of industries currently using particle counters as a process control tool goes through frustration and very painful and expensive experiences because of misperceptions and misinformation. For example, the drinking water treatment industry is using particle counters to monitor the treatment process (sand filtration) with regard to two ranges of particle sizes: 2 µm to -4 µm and 8 µm to 16 µm. These are the ranges in which disease infecting parasites appear. The purpose of monitoring the filtration process with particle counters is to remove particles in the parasite size range at a required efficiency, thus also preventing their distribution to consumers.
Unfortunately, the water treatment industry adapted a calibration and procedure for monitoring treatment efficiency which had its origin in the controlled laboratory environment and was used by trained lab technicians and scientists. When this method was transferred to the field for large scale process monitoring and control, the original objective for using particle counters (counting particles in two relatively large size categories) got lost and more attention was focused on accurate sizing rather than counting. This resulted in a very confused and upset industry which was told by vendors that particle counting is the solution to their problems. Vendors did not voice warnings early enough and neglected to educate their new application customers in a timely manner for fear of losing sales.
A similar example can be cited from the semiconductor and data storage industry. A change of the use of particle counters took place during the late 1980s and early 1990s. During the 1980s mostly single particle counters on carts were used to monitor selected locations in cleanroom installation. The results from these measurements were mostly evaluated relative to each other and in view of a baseline performance value of the cleanroom for which product yield and process control had been accepted. Only after relative changes of particle counts from the established process background level were observed by the same instrument that had established the background level, was action initiated to bring the process back under control. Relative particle numbers — relative to accepted values established using experience with the process — were important for contamination control and product yield.
Pushing toward smaller dimensions, higher product yield and larger wafers semiconductor manufacturers included contamination control and monitoring as design parameters into the specifications of process lines. Process engineers, designing for productivity and minimizing loss of product due to particles, set cleanliness specifications in the form of absolute particle numbers. These specifications asked, in the opinion of some particle counter manufacturers, the impossible from an instrument that was always used to determine relative particle counts but that had never been intended to determine absolute correct number of particles that could be compared with counts from other counters counting particles accurately. Some manufacturers made this transition and invested in equipment and research and development to improve their instruments, while others did not. The users of particle counters in the these industries simply educated themselves and in many cases now know more about particle counters than the manufacturers.
User mistakes
Particle counter manufacturers are not always the bad guys. About 1992, the need for cleaner and more economically run manufacturing environments led some clever cleanroom operators to the conclusion that by reducing makeup air, energy can be saved and less dirty fresh air must be cleaned. Their logic was correct with only one little flaw: by recycling more air in the cleanroom system, chemical vapors and gases are enriching raising the level of molecular contamination. Particle counters, which are optical instruments with sensitive optical surfaces exposed to the cleanroom air suddenly face a problem not experienced before — vapors from process chemicals condensed on lenses and mirrors leading to rapid instrument failure due to contamination in environments which had no particles. The users blamed the manufacturers for instrument failure but would not admit that the vapors in their cleanroom air were responsible, because cleanroom personnel were exposed to the same vapors.
Conclusion
Particle counters are useful process and contamination monitoring tools if the veil of magic is removed and this technology is based on solid scientific grounds. As long as vendors interpret the counting process as art, the door for subjective interpretation is open and the facts of science are distorted by alternative motives.
Three changes must be implemented to re-gain and re-establish the scientific respect particle counting deserves as a measurement technique:
1) The need for a strict particle counter performance standard similar to JIS 9921-1989 will streamline the comparability of data produced by particle counters designed, built and tested to this standard. Counting results from such an instrument can be trusted to represent counts as close to the true counts of particles as technically feasible.
2) Eliminating counting errors which stem from flawed procedure, sample and data handling, and using particle counters in the wrong applications can be achieved by open and honest advice, admittance of instrument limitations, in depth education and solid training.
3) Better understanding of particle count statistics which fundamentally limit count accuracy of instruments, and the interpretation of data in view of each individual application will prevent unrealistic expectations users might have and will give particle counter manufacturers reasons to accept and explain instrument limitations.
Editor`s Note:
The options expressed in this article are those of the author and do not necessarily reflect the opions of the publisher.
Dr. Sommer is a contamination consultant and president of TEAM Engineering, Inc. in Merlin, OR. He is an international authority in the field of particle counting instrumentation and has developed 19 patents in this area. Dr. Sommer has published extensively on particle counting and used his knowledge of instrument design and development in specific applications such as fluid power and lubricant contamination analysis. His background in mechanical engineering provides the basis for a clear understanding of machine condition monitoring applications. Dr. Sommer taught fluid mechanics, heat transfer and turbulent combustion for many years in Germany, at Carnegie Mellon university and at the university of Maryland. As engineering manager, B.P. Engineering., and vice president of technology and business development at Met One, Inc. and Hiac/Royco, Dr. Sommer pioneered the field use of optical single particle counters for machine condition monitoring.
 |
 |
 |
 |
 |
 |
 |
 |
 |