CHAPTER 1
25 Years of Particle Size Conferences
B. Scarlett
DEPARTMENT OF CHEMICAL ENGINEERING, TECHNICAL UNIVERSITY, P.O. BOX 5045, DELFT, 2600 GA, THE NETHERLANDS
1 1966 AND ALL THAT
It is almost precisely twenty five years since the first of these conferences, organised then by the Particle Size Sub-Committee of the Society of Analytical Chemistry which has now grown into the Particle Size Characterisation Group of the Analytical Division of the Royal Society of Chemistry. The sub-committee had recently completed a review of 74 different methods of particle size analysis and these efforts were the inspiration both for the first conference and for the formation of the permanent subject group. This is the seventh conference in the series and it is entirely appropriate that it should be held in the same place, at Loughborough University. The proceedings of each of the conferences is permanently documented by a bound book of conference proceedings. Several authors feature in all the seven conferences.
The closing address at the first conference was given by Professor Harold Heywood, an address which was based on many years of experience and which could be given, almost without modification, at the end of this conference and still be entirely appropriate. Thus, some truths do not change and although they must be learnt anew by each succeeding generation, they remain essentially the same.
The developments which he foresaw have largely occurred over the intervening twenty five years but it may be that now those basic messages have even more practical importance and that the rate of development grows exponentially. In this paper, my intention is to re-state some of the message of Heywood and to illustrate it in the circumstances of today.
2 WHY OH WHY?
The first quotation I would like to take concerns why particle size measurement is important, why it commands the attention of professional groups, companies, journals and conferences specifically devoted to its development. Heywood wrote:-
"However, it must be realised that particle size analysis is not an objective in itself but is a means to an end, the end being the correlation of powder properties with some process of manufacture, usage or preparation."
A particle size measurement does not have a meaning unless the objective of the measurement is also specified. Thus, the techniques which should be used depend entirely on the accuracy which is required and the circumstances of place and time in which the measurements must be made. There is no such thing as the "best" particle size technique unless the circumstances are also specified. A particle analysis laboratory cannot operate on the same basis as some normal analytical laboratories, samples come in and numbers go out. In a particle analysis, the question must always first be addressed. "What do we need to know?" Another way of saying this is that, eventually, particle analysis is an engineering tool, not a basic science. Engineers must use all the known laws of science to solve the particular problem with which they are confronted in the most elegant manner.
In this respect, it may be that the circumstances are more difficult than they were twenty five years ago. At that time, the number of instruments was more restricted than now as was certainly the number of companies who supplied those instruments. Current instrument manufacturers must try to provide a package of hardware and software which will handle a large number of powders and applications. The commercially available instruments are thus, inevitably, a compromise between what is desirable and what is possible, the same instrument being sometimes too sophisticated and sometimes inadequate. To be a manufacturer of particle measuring instruments is not an easy life and it can only get worse because the tendency will be for the applications to become more diverse and sophisticated. Why should the applications become more diverse? That comes partly from the development of new technologies, of course, but it also comes from a gradually increasing sophistication in the use of the measurements. (Fig. 1).
At the simplest level we use particle size measurements to monitor their concentration or to control the reproducibility of a product. Thus, we compare what we find with what we expect and if the two do not coincide we reject the product. The science of powder technology, however, is concerned to use the microscopic properties of the system, for example the particle size distribution, to interpret the bulk behaviour of the powder. If it is to be used in dilute circumstances, then the bulk behaviour can be derived by integrating the behaviour of the individual particles but usually this is not so and the relationship between the microscopic and macroscopic properties must take account of the particle interactions. By observing the difference in particle size distribution of samples which exhibit a different bulk behaviour, we begin to make a "correlation" between the two which, whether empirical or theoretical, quantitative or qualitative, involves interpretation of the mechanisms involved. Somewhere between these two purposes usually lies the purpose of a particle size measurement. There is, however, a far more ambitious level at which powder technology must eventually operate and, as yet, rarely does. That is to design the particles and the particle mixture to produce required properties, to use the relationships between microscopic and macroscopic properties in a predictive manner. It is the more rigorous use of particle size measurements which introduces the real diversity and which requires the measurements to be carefully matched to the problem. The increased diversity does not alter the basic needs which Heywood described. He wrote:-
"Finally, we should examine the problems awaiting solution in the study of particle sizing and applications to industrial production".
He then went on to classify the problems into three groups. The first was:-
Standards through Standards
"First there is the development of some standard method of analysis which gives an absolute measure of particle size distribution. This may already exist but, if so, the validity needs establishing beyond question. Time of operation is not a vital factor for such a procedure, as it would mainly be used for fundamental research".
Of course, the fundamental method has always existed and is indeed terribly tedious. In a review of the 1981 conference, Leschonski wrote:-
"It would, therefore, be most desirable to describe size more often, based on the volume of the particle, e. by its volume diameter".
Thus, if it were possible to measure the volume of each individual particle, for example by weighing and if the density is known then the result can be presented either as a distribution by number, or a distribution by mass, of the equivalent volume diameter. Thus, in order to carry out an "absolute" method of particle size measurement, all that is required is an accurate balance. Of course, the method is impractical and, as the particles become smaller, impossible but it is the ideal against which other methods can be assessed and can be calibrated. Over the past twenty five years, the view has increasingly prevailed that we should regard the equivalent volume diameter of the particle as the basic size and that other equivalents are dependent upon both the size and shape of the particle. To accept this view requires only two simple steps. First to accept that a measurement of the size of a particle assigns only one scalar number to a particle and is, therefore, bound to contain a limited amount of information about the particle. The 'size' of the particle means, basically, how much material is contained in the particle, that is its volume. The second step is to accept that other equivalent spheres appertain to particular conditions. For example, the equivalent settling diameter depends on the Reynolds number and thus on the fluid in which it is settling. The equivalent light scatter diameter depends on the scattering parameters and thus the wavelength of the incident light and the refractive index of the surrounding medium. Any equivalent diameter relates exactly only to the precise circumstances in which it is determined and the translation of it to different circumstances requires some assumptions to be made.
In our modern circumstances, the need for an absolute method of measurement can be interpreted as the need for calibration and reference materials which are certified on the basis of traceable measurements, measurements which can eventually be related back to the standard kilogram and standard metre. In creating the coarser BCR fractions Leschonski's suggestion, based in turn upon the suggestion of Andreason, has been adopted and the samples are certified in terms of the equivalent volume diameter, traceable to the standard kilogram. Some of the samples of standard spherical particles which are available can also claim to be certified on the basis of measurements directly traceable, in that case to the standard metre. The current efforts of organisations such as BCR and IFPRI to create more reference and calibration materials must be encouraged and intensified. At the same time, standards must be written, and adopted by ISO, which describe clearly how those materials should be used to calibrate and control the commonly used methods of particle size measurement. In this way, the standard method of analysis will not only exist, it will also become available. When materials which have been certified by a traceable method are presented to another instrument, we are in fact comparing that instrument with Heywood's absolute method. If we simply compare, the materials are being used as a reference material whereas if the results are used to adjust the output of the instrument then we are calibrating. In general, distributions of particles are more useful as reference materials and monosized fractions to calibrate but I believe that the procedures which we must develop will require both and will present them, mixed in different proportions, in order to assess three parameters:-
reproducibility accuracy sensitivity.
The existing BCR materials are already being used to assess reproducibility and accuracy, for example as reported in this conference by Allen and Davis. Before we can adequately assess all three factors and, if necessary calibrate, we need also the spherical reference materials which are currently being produced as well as much narrower fractions of spherical particles. It is also necessary to elaborate the use of those materials and the development and description of these various uses of the materials is just as important as the certification of the materials. The ISO standards are the obvious medium for this second task. Because the range of materials which is available at present is quite restricted, the way in which we use them is similarly restricted.
When used to assess reproducibility, there is not a great difference between spherical and irregular reference materials. However, in assessing accuracy there is some difference. In using spherical particles we are comparing what the instrument produces with what it is expected to produce. This is because the concept of the instrument is almost certainly based on the idea that the answer will be in terms of some equivalent spherical diameter. On the other hand, if a sample of irregular particles is used as a reference material and the recorded results diverge from the certified results, the explanation can often be advanced that a different equivalent sphere has been measured. We should regard a big divergence as an indication that the instrument must be calibrated to measure particle size, that is equivalent volume diameter. In one way then, spherical materials present a fairer reference material for an instrument in that it can be expected to produce the correct result, irregular particles probe its intrinsic accuracy. When it comes to using the materials to calibrate, on the other hand, the relative role of spherical and irregular particles is reversed. The spherical particles calibrate the size axis accurately only to measure spherical particles. If accuracy is required for other materials, then the calibration must be achieved with that same material. In order to calibrate a method for particle size, monosized fractions must be created. For spherical particles this is relatively easy but to create samples of irregular particles, all of which have the same equivalent volume diameter, is difficult and requires further development of the methodology. Furthermore, a calibration must concern not only the size axis but also the distribution axis. This can also be achieved by the use of monosized fractions, mixed in different proportions.
Thus, we see that not only has the absolute method always existed but also that, with imagination, we can compare and calibrate our more practical methods to it. However, we must also now extend our first tenet, that the particle size measurement is not an end in itself but must be related to the usage. Reproducibility, accuracy and sensitivity are also not absolute objectives and must also be related to the usage.
Beware the Software
The next need which Heywood listed was:-
"Secondly, there is the development of methods of analysis for routine procedures, for which repetitive accuracy is most important, though absolute accuracy should not be ignored".
The past twenty five years have seen an explosion in the diversity of the equipment available and a complete step change .in the design and convenience of those instruments. I think that it is true to say that the instruments of today do not really use any physical principles or laws which were not used twenty five years ago. Rather the development has come about by the availability of modern components, lasers, optic fibres, photodiodes, and piezo crystals for example. Above all, however, engineering has been revolutionised by the development of the computer and this generality applies also to particle size measurement. The ability to carry out complex calculations and to record and manipulate large amounts of data in a short time has completely changed the concept of what is possible in a particle size measurement and this is a process which is still in its infancy. It is this development which has also compounded the problem. Previously the basic problem was that techniques which are based on different physical principles give different answers with the same sample. Now, instruments which are ostensibly the same give different results, both in substance and form, due to the disparate software packages which process and manipulate the basic data and eventually produce it in an attractive and convincing form.
If we have now accepted the basic tenet that both the measurement and its quality must match the problem, and we have, and if also the problems are becoming more numerous and sophisticated it follows that the diversity and sophistication of the measurement techniques will continue to increase. Preversely, this will not be in the hardware of the instruments which may well become, increasingly, standard arrays of photodiodes, optic fibre bundles, piezo transducers and conductance electrodes. Rather the difference will be in the software. The basic configuration of the instruments cannot change much.
On the one hand we can try to present the particles to the detector individually and count and characterise them or, on the other hand, we can maintain the particles as a complete mixture and then must address the problem of deconvoluting the signal. In general the first method produces more accuracy if a number distribution is required, the second can probably handle a larger sample. A combination of the two can always produce a better result than either individually. The intermediate state involves making some sort of partial separation, either within the measuring region or a separation in time, as in the classic sedimentation techniques. These basic possibilities will not change but what has and will continue to change is the number of signals which can be presented to the sample and the number of detectors with which it can be surrounded. If all these signals are used to estimate, check, interogate in a rigorous manner then the quality of our results can only improve. If they are so manipulated, smoothed, adjusted that the final result resembles what we wish we had rather than the reality, then we would be better to stick to the sieve and the microscope. Which of these situations appertains depends upon our software package. If I must identify today one problem which is significantly different than in Heywood's day, it is that the software packages we use are still rather closed and it is not obvious precisely what they do to the data. The development which must occur is that an instrument must be able to utilise a whole library of software which is added to and borrowed from by a host of people as in any library. The present situation is that the library often has one book, written by the librarian.
