TOP 10 ERRORS IN PARTICLE ANALYSIS AND HOW TO AVOID THEM?

Particle analysis is an integral part of the quality control of bulk materials and is routinely performed in numerous laboratories. The methods used have often been established for years and are hardly ever questioned. Nevertheless, the procedure should be critically reviewed from time to time because a whole series of sources of error can negatively influence the results of a particle analysis.

1. SAMPLING : When sampling inhomogeneous bulk materials, it must be ensured that the properties of the laboratory sample taken correspond to those of the total quantity. During transport, for example, small particles move down the interstitial spaces due to vibration and collect at the bottom of the container. In bulk cones, one usually observes a concentration of the small particles inside the cone. Subsamples are frequently taken from several locations and mixed together to counteract the effect of segregation. Suitable aids such as sampling lances can further improve the situation.

2. SAMPLE DIVISION : The sample amount available for particle analysis is usually too large for the measuring instruments used. In many cases, the quantity must be further reduced in the laboratory. Poor or unperformed sample division is one of the main sources of error in particle analysis, especially for materials with wide size distributions. The use of sample dividers can remedy this situation. Even a simple sample splitter leads to significantly improved reproducibility when several subsamples are analyzed.

3. DISPERSION : Dispersion is the separation of particles to make them accessible to measurement. Particles that stick together due to different attracting forces are called agglomerates. It is usually desirable to break up these agglomerates before measurement. Agglomerates can also occur in suspensions. This can often be prevented by selecting a suitable dispersing medium (carrier fluid). Agglomerates that are still present in the suspension can be broken up by using ultrasound. Most modern particle sizers have powerful ultrasonic probes built in, so that sample preparation is done entirely inside the instrument. In general, the larger the particles, the higher the probability of error in sampling and sample splitting. With finer particles, the susceptibility to error is more likely to occur during dispersion.

4. SIZE DEFINITION : Strictly speaking, particle size is only unambiguously defined for spherical structures, namely as the diameter of this very sphere. For non-spherical particles, different measured values can be obtained, depending on the orientation and the measuring technique used. Even more advanced particle measurement methods use different "size models". In sieve analysis, particles ideally orient themselves so that their smallest projected area fits through the smallest possible mesh. Sieve analysis thus tends to determine particle width. In laser diffraction, all diffraction signals are evaluated as if they were generated by ideally spherical model particles. In contrast to image analysis, the particle shape cannot be determined. Furthermore, laser diffraction evaluates a signal generated by a particle collective with particles of different sizes. Calculation of the size distribution is therefore indirect. Different methods for particle measurement inevitably produce different results. While laser diffraction and sieve analysis are difficult to correlate, the results of sieve analysis and image analysis are often very close, since imaging techniques can determine particle width and sieve analysis tends to be a width measurement.

5. INCORRECT SAMPLE AMOUNT : Using too much or too little material can negatively influence the measurement result. In laser diffraction, too high a particle concentration can lead to multiple scattering, and if too little sample is used, the signal-to-noise ratio is poor. However, modern laser analyzers indicate the ideal concentration for measurement and warn users as soon as the amount is too high or too low. In image analysis, you can't actually use too much sample. If too little sample is analyzed, the result will be unreliable and poorly repeatable due to the small number of detections. Since the required amount of particle detections depends on the size of the particles and even more on the distribution width, it is difficult to make general recommendations here. Repeatability tests are helpful, especially looking at the "rough end" of the distribution.

6. UNDERESTIMATING TOLERANCES : Every measuring instrument shows certain systematic uncertainties and tolerances which must be taken into account when interpreting the results. Test sieves made of wire cloth are manufactured according to the standards DIN ISO 3310-1 or ASTM E11. These standards specify how the real mesh size of each sieve is to be tested. Each test sieve is inspected by an optical method before delivery and a specified number of meshes are measured. The mean value of the measured opening width must be within prescribed tolerances around the nominal mesh size. Calibration certificates are available for each sieve which contain information on the real mesh sizes and their statistical distribution

7. OVERESTIMATING SENSITIVITY : A frequent issue in particle analysis is the detection of oversize particles, i. e. a small amount of particles that are larger than major part of the distribution. Here, the sensitivity of the measurement method plays a decisive role. Imaging methods offer the advantage that each particle detected represents a "measurement incident” and is thus also shown in the result. The contributions of the individual particle sizes are superimposed and an iterative procedure is used to calculate the size distribution. If the amount of oversize particles is small, the contribution of these particles is not sufficient (signal/noise ratio) to show up in the result. For a reliable detection of oversize particles with laser diffraction, the contribution should be >2 %.

8. WRONG DENSITY DISTRIBUTION : Particle size distributions can be represented graphically in several ways, with the particle size always on the x-axis. Intuitively easy to access is the histogram representation, where the bar width corresponds to the lower and upper limit of the measurement class and the height corresponds to the amount of particles in the respective size interval. These size intervals are often determined by the performance and resolution of the measurement system used. Popular with many users is the representation as distribution density, often succinctly and incorrectly referred to as a "Gaussian curve". The distribution density is the first derivative of the cumulative curve. Where the cumulative curve rises steeply, the density distribution has a maximum; where the cumulative curve is flat, the density distribution has a minimum. It is important here that a true density distribution shows the slope of the cumulative curve. Thus, the quantity in the measurement class must be divided by the class width. The accuracy of the density distribution increases with the number of measurement classes.

9. TYPES OF DISTRIBUTION (NUMBER, VOLUME, INTENSITY) : Particle analysis results are usually given as a percentage, either as a percentage per measurement class or as a proportion greater or smaller than a certain size x. However, these percentages can have very different meanings. It makes a huge difference whether these values refer to mass, volume or number. Which type of distribution is present depends strongly on the measuring system used. In sieve analysis, the weights of the sample in each fraction are determined by backweighing and are then converted into mass percentages. These are identical to a volume-based distribution, provided there are no density differences between particles of different sizes. Other methods, such as hand measurement with a caliper, provide number-based distributions based on the number of particles in each measurement class. The difference between number-based and mass/volume-based distributions lies in the fact that for volume distributions, large particles have a stronger weighting, while for number distributions, small particles are weighted stronger. Laser diffraction relates all signals to a sphere of equal effect and thus provides volume-based distributions. Since a collective signal and not individual incidents are evaluated here, laser diffraction cannot determine number distributions. The situation is different for single particle measurement methods, such as image analysis. Since image analysis covers different size definitions, it is reliably possible to carry out this conversion with a suitable volume model (usually a prolate rotational ellipsoid). This makes image analysis data comparable to sieve data or laser diffraction.

10. WORKING WITHOUT SOPS : As with all other analytical methods, a uniform, standardized procedure is also a prerequisite for consistent and meaningful measurement results in particle measurement. Such Standard Operating Procedures (SOPs) always guarantee the same, defined measurement processes and work steps. A prerequisite is that all instrument settings are stored by the software and can be retrieved. However, an SOP comprises more than just instrument settings. Specifications for sampling, sample division, sample preparation and evaluation should also be precisely specified here. It is advisable to create work instructions that are as precise as possible to guarantee consistent quality of the measurement results.

Various methods are used for particle analysis, the most common being laser diffraction, dynamic image analysis and sieve analysis. Successful analysis and meaningful results can only be achieved if preparatory steps such as sampling, sample division and sample preparation are carried out correctly. The selection of the appropriate method for the sample material and a meaningful evaluation of the measurement data finally lead to a successful particle analysis. Microtrac MRB is one of the leading suppliers of particle measurement technology from the fields of laser diffraction and dynamic light scattering as well as static and dynamic image analysis and offers the complete portfolio for particle characterization from a single source.