| Search | Feedback | Contents | Deutsch |
 |
| Issue 36 |  |
|||
| Printable version | ||||
| Category | Titel | Author | ||
| Guest Article | In situ Particle Sizing of Concentrated Dispersions | Rolf Daniels |
1 Introduction
The knowledge of particle size is one of the key issues in characterizing disperse systems. A lot of analytical techniques are known for this purpose [1]. A major draw back of most of the common techniques, e.g. laser diffraction or electrical zone sensing, is that extensive dilution of the samples is required before measurement. This forces the particle size analyst to determine particle size off line. Moreover, it bears the risk that the physico-chemical equilibrium is disturbed dramatically. In turn, particle size can change significantly due to aggregation or coalescence. In this case a technique is required which allows size measurements on concentrated samples. Such a method is often referred to as in process, in line, or in situ method. Figure 1 shows the major steps which are required for an in situ method in comparison to a conventional off line technique. From this diagram it is evident that the most straightforward approach is using an in situ method to measure particle size in line.
Figure 1: Comparison of the required steps for particle size analysis.
(A) conventional method, off line
(B) in situ method, off line
(C) in situ method, in line
Our primary goal in determining particle sizes is to characterize emulsion systems which are stabilized with a suitable polymer. Although many of them were stable on storage, we cannot assume that extensive dilution would not have an effect on their particle size. Therefore, in 1989 we started this work with a rather simple in situ particle sizing technology, which was called „Scanning Laser Microscopy (SLM)“. Though the results looked promising, we encountered several restrictions concerning size range and resolution [2,3].
This encouraged us, in cooperation with Meßtechnik Schwartz, Düsseldorf, Germany, to further develop the in situ technology. Our combined efforts helped to improve the method substantially during the past years and resulted, in 2001, in a completely new instrument [4,5]. To distinguish this technology clearly from the previous one, we named it 3D ORM (Optical Reflectance Measurement) technology.
Although this technology seems to be rather helpful where conventional methods fail, it is so far scarcely described in literature. My intention in writing this article is therefore to provide a brief introduction to the principle of operation, the theory of data processing, and of course, the practical use of this innovative particle sizer.
2 Principle of operation
The complete set up of an 3D ORM instrument is illustrated in Figure 2. The main components are: sensor, signal processing unit, and a personal computer running the MTS/ LT software for data analysis.
Figure 2: A schematic diagram of the principle components required for an 3D ORM instrument [7].
The basic principle of the instrument is very simple and is illustrated in Figure 3. The light source is a laser diode which is built in the signal processing unit. The lasers used for 3D ORM instruments vary depending on the particular application. We use a 10 mW semiconductor laser (wavelength: 780 nm) for the analysis of emulsions with particles as small as 100 nm and dispersed phase volume fractions as large as 0.8.
The light is transmitted to the sensor using optical fibers. Within the sensor the laser beam passes consecutively a collimator lens, a wedge prism, a focusing lens and two sapphire windows. During operation the focusing lens rotates eccentrically. This makes the laser beam rotate and pass the wedge prism at sites with different path length. Consequently, the focal point of the laser beam moves in normal direction to the sensor window. Combined with the circular movement of the laser beam, the focal point scans across the sample on an elliptical orbit under a certain angle to the sensor window. This movement of the focal point in a three dimensional space in front of the sensor is the origin of the expression “3D ORM” for this technology.
Figure 3: A schematic representation of the basic principle of 3D ORM. The optical system makes the focal point scan across the sample on an elliptical orbit in front of the sensor.
If the vertical and radial moving focal point hits a particle, light is reflected. This light is then captured by the optical system inside of the instrument, and is passed to the detector. Its task is to transform the reflected light into an electrical signal which then can be further processed.
Figure 4 shows how the electrical pulses generated by the detector are related to the particles which are actually scanned by the laser beam.
Figure 4: Schematic diagram of the relation between scanned particles, corresponding raw signal, and filtered time-of-flight signal. The chord length of a scanned particle is proportional to the time of flight for a pulse.
In the first approximation, the chord length of a scanned particle lc can be accounted for by
where delta tS is the time of flight for a pulse. The proportionality constant in this equation is vS, the velocity of the scanning focal point.
However, a systematic error occurs when the length of a pulse is translated as the corresponding chord length by using this simple approach. This error depends on the relation of the diameter of the scanned circle and the chord length of the particle (Figure 5).
Figure 5: Schematic representation of the arc length and chord length of a particle. (The size of the particle has been emphasised for clarity.)
In general, the chord length lC and the arc length lA are related by:
The actually occurring error can be estimated by:
The raw signal from the detector is amplified and transmitted to an electronic filter system. This unit separates the pulses into two main fractions, namely the "good" pulses which are further considered and the "bad" ones which are discarded. Criteria for the selection include the symmetry of a pulse and the slope of the flanks of the signal.
Special electronic circuits convert the "good" pulses into a rectangular shaped signal which then contains the time of flight delta tS as a characteristic measure. These delta tS values are divided into appropriate size channels which are ordered in a geometric scale. This register typically consists of 250 channels. Since the scanning velocity of the laser beam is known, these channels correspond to distinct size bands [6,7].
At the end, the signal processing unit passes a chord length distribution to a computer, which performs the appropriate data analysis.
3 Data processing
If the laser beam randomly and vertically scans a single particle, a characteristic chord length distribution is produced. The origin of the chord length distribution can be understood by reference to Figure 6.
Figure 6: Derivation of the normalised chord length distribution of a singular cubic particle.
(A) Map of scanned chord lengths from the projected area of the particle
(B) Transformation into a two-dimensional chord length distribution
(C) Ordering the chord lengths due to their frequency
(D) Normalisation of the frequency distribution function
When the laser beam scans across a particle, it generates an ensemble of chord lengths which fit into the projected outline of the particle. The chord length lC of each scan is given by the difference of the corresponding upper (xu) and lower (xl) boundary where the beam enters or leaves the projected area. This map of chord lengths can be transformed in order to express the chord length as a function of the normal direction y, which is perpendicular to the scanning direction. Arranging these values in relation to the chord length gives the corresponding frequency plot. The final step is to normalize the x-values for the maximum chord length lc,max and the y-values for the maximum frequency delta ymax which then gives the normalized frequency distribution.
Using this principle, a characteristic chord length distribution can be generated for any particle shape [6]. This is the most simple example when we are dealing with spherical particles, because an analytical solution of the problem exists. The normalized chord length distribution function of a singular sphere f'(x) is given by
The resulting incremental and cumulative chord length distribution for a singular sphere is shown in Figure 7.
Figure 7: Plot of the incremental and cumulative chord length distribution of a singular sphere
For more complex shapes the chord length distribution has to be determined by Monte Carlo simulations. Figure 8 shows the resulting distribution function for a few relatively simple particle shapes. From this it is evident that the proper selection of the appropriate shape factor dramatically influences the result of the particle size analysis.
Figure 8: Plot of the incremental chord length distribution of singular particles of varying shape.
In a real experiment with perhaps as many as 10 to 12 particles/ml, the total chord length distribution will be the sum of the scans of all particles present. For a mixture of equally distributed spheres this is illustrated in Figure 9.
Figure 9: Plot of the incremental q(x) and cumulative Q(x) chord length distribution of a equally distributed sample of spheres. The green line shows the corresponding particle size distribution by number h(x).
Unfortunately, we also have to solve the opposite problem: A measured chord length distribution has to be transformed into an appropriate particle size distribution. Mathematically, this is called an inverse problem. This means that we need to find a particle size distribution which could have generated the data we measured.
The general correlation function is given by:
where:
h0 is the number distribution of chord lengths,
q0 is the number distribution of particles as a function of the particle size,
p describes the probability of measuring a chord length lC when scanning a particle of size lC,max,
lC is the chord length,
lc,max is the maximum chord length of an individual particle, and
lc,MAX is the maximum chord length of the largest particle.
This equation is called a FREDHOLM integral equation. It is valid for all convex shaped particles and uses the maximum chord length lc,max as a reference value.
There are several numerical approaches to solve the FREDHOLM integral equation. The most recent version of the MTS / LT© software uses the CHAHINE method for extracting the size distribution. Mathematically, it is rather complex and I will not discuss this in further detail. For further information see reference [9].
The resulting particle size distribution can be presented in many different forms, e.g. volume distribution, number distribution, RRSB distribution, or a trend signal for in process applications. Figure 10 shows an example of a screen shot of the Win ORM software.
Figure 10: Screen shot of the Win ORM© software
4 Range of particle sizes and resolution
3D ORM instruments from MTS measure particles in a nominal range from 100 nm to a few millimeters. In order to increase resolution, a less wide spread range should be preferred. We use an instrument qualified for pharmaceutical emulsions (ECA 817 and LabScan 2007). It analyses droplets from 0.1 to 125 µm divided into 1024 bins. Some of the results which are presented in the applications chapter are from an earlier model (ECA 01015), which measures in the range from 0.4 to 125 µm.
5 Experimental aspects
The experimental set-up of a particle size analysis may have a considerable impact on the overall performance of a measurement. Although the following issues were selected more arbitrarily than systematically, they represent most of the aspects which we believe have to be worth mentioning from our experience with the 3D ORM technology.
5.1 Sampling considerations
The 3D ORM technology enables the determination of the particle size in the disperse systems in situ. This, however, does not mean that we encounter no problems with proper sample selection. When we use an in situ method, the sampling procedure is where we install our sensor (Figure 11, Figure 12), e.g. adjacent to a homogenizer or somewhere in a storing tank.
Near the homogenizer the particles move around the sensor very rapidly in a turbulent stream, whereas in a storing tank they stay almost stagnant. For laboratory uses, we place a sample of at least 200 ml in a glass beaker, where baffles are included, and mix it with a stirrer which produces a stream directed opposite the sensor head (Figure 13). This allows us to present a representative sample, when using the 3D ORM scanning laser beam. In principal, it is also possible to measure smaller samples, i.e. in a test tube, when the smallest MTS sensor with 7.85 mm diameter is used (Figure 14).
Figure 11: Sensor installation directly in the mixing area of a pilot scale homogenizing unit.
Figure 12: In line installation of the MTS 3D ORM Fingerprint Sensor Technology.
Figure 13: LabScan sensor immersed in a glass beaker for off line measurement in the laboratory. The stirrer produces a stream directed opposite the sensor head.
Figure 14: 3D ORM sensors designed for varied applications.
Another issue which should be addressed in this context is where the focal zone is located in relation to the traveling distance to the vertical focus movement. In general, it should be preferred to have the focus deep (= traveling distance) within the sample in order to allow particles of all sizes to approach the focus at random. However, this might be limited to a too low intensity of reflected light, due to scattering effects. Moreover, the traveling distance to the 3D ORM focal point from the sapphire lens should not be less than half the diameter of the biggest particles in order to make sure that – at least theoretically - all possible chord lengths of these particles can be scanned.
In principal, there is maximal optimal distance to the focus for every sample. However, for practical reasons it might be straightforward to always use a 3D ORM instrument with a long traveling distance to the focal point to ensure uniform value. From our experience, it turns out that we obtain reliable results on most pharmaceutical emulsions when the traveling distance to the focus was 500 micron for the ECA 817 and 250 micron for the LabScan 2007 moving from the sensor window to the sensor head.
5.2 Calibration
Instruments using the 3D ORM technology do not need a calibration on an operation.
If calibration is required this can be done using standard particles, e.g. BCR standard. However, the user must be aware that there might be differences in the optical behavior (reflectivity) of this standard and the sample of interest. If necessary, it is possible to make a software calibration to such standards. In the same manner it is possible to calibrate the instrument to already existing results from other particle size analysis. This enables the user to switch the instrument in an ongoing project. Such a special calibration can help to decrease the error for a specific sample but might also bring about a severe malfunction of the instrument with all other samples. For this reason, it seems to be more appropriate to use 3D ORM data in a comparative sense, and bear in mind that there is likely to be a significant discrepancy between the measured size and the so called „true particle size“. However, this is a rather common issue because each method for particle size analysis can only provide its particular equivalent diameter which is normally not identical to the so called „true size“ (or whatever is meant by this expression).
5.3 Reproducibility
To check the reproducibility of the measurement, we divided an emulsion sample consisting of 2.5 kg into ten equal portions. The measurement was then replicated ten times using a standard operation procedure consisting of the following steps:
• cleaning sensor head
• filling sample into the beaker
• starting stirrer at 400 rpm
• starting measuring cycle
• storing data
• stopping stirrer and removing sample
• cleaning sensor
Figure 15 shows the results of these experiments. Obviously, the reproducibility of the ECA 817 was fairly good and we calculated a standard deviation for the mean volume diameter of 0.64 % [10].
Figure 15: ECA 817 results of replicated measurements on a 20 % o/w emulsion demonstrate that in situ particle size analysis highly reproducible.
6 Applications
We use 3D Optical Reflectance Measurements mainly to characterize oil-in-water emulsions which are stabilized with polymeric emulsifiers, e.g. hydroxypropyl methylcellulose (HPMC). A lot of experiences have been accomplished during this work. In order to illustrate the possibilities as well as the limitations of this technology, I will stress two examples where we have also compared the data to other methods like Transmission Electron Microscopy (TEM) or laser diffraction.
6.1 Particle size measurement on highly concentrated o/w emulsions [11]
For this study we used o/w emulsions with varying oil volume fractions, i.e. 20, 40, 60 and 80 %. The aqueous phase consisted of a 2.5 % solution of hydroxypropyl methylcellulose (HPMC). Thus a decreasing amount of polymer had to stabilize an increasing amount of oil.
After the preparation, all emulsions containing up to 60 % oil appeared macroscopically homogenous. The microscopic image showed the typical features of a partly flocculated system. Emulsions containing 80 % oil were still of the oil-in-water type and behaved like a mayonnaise. However, these samples could not be sufficiently homogenized due to their high viscosity. The consistency of the other emulsions varied from liquid like with 20 % oil to soft cream like with 60 % oil.
Particle size was measured the with an ECA 01015. In parallel, we prepared freeze fractured replicas of the samples and examined the specimens by TEM. The results are depicted in Figure 16.
Figure 16: Volume distribution and corresponding TEM micrographs of o/w emulsions with 20, 40, 60, and 80 % oil content; bar: 5µm.
Emulsions with 20 % oil phase revealed the most shallow size distribution. By increasing the oil content the spread broadens and the mode of the distribution moves towards smaller values. 80 % emulsions gave almost the same particle size distribution as the 60 % emulsions. Only a small but reproducible peak in the size range from 50 to 80 µm indicated the inhomogeneity of this preparation. However, this is not sufficient for a quantitative characterization of the system.
The TEM micrographs confirm the results of the particle size analysis qualitatively. Emulsions up to 60 % oil contain almost spherical droplets in the size range from 1 to 10 µm. The internal phase is formed of mainly singular droplets and some few aggregates of two or more spherical particles. 80 % emulsions consist of drops which differ greatly in size and are in close contact. As is typical for a mayonnaise, the drops are polyhedral.
In addition, emulsions with up to 60 % oil content were monitored for more than 2 years while stored at room temperature (Figure 17). After 6 months storage, the macroscopic and microscopic appearance, as well as their consistency, did not change. Accordingly, particle size analysis revealed only minor changes. In detail, 20 % emulsions show a ripening process during the first month of storage. This is a typical feature for this type of emulsion. Beyond that time, particle size did not change significantly. Emulsions with 40 % oil behaved in a similar way. Owing to their higher viscosity, the effect is less pronounced and takes place over a longer period of time. In contrast, the 60 % emulsions were not sufficiently stable. This is indicated by a second peak in the particle size distribution which developed during storage.
Figure 17: Particle size of the model emulsions after preparation,
6 months, and 27 months storage.
(A) emulsions containing 20 % oil, (B) emulsions containing 60 % oil.
The measurement enables to monitor coalescence in 60 % emulsions as well as Ostwaldt ripening which occurs in 20 % emulsions.
From these results, we conclude that 3D ORM technology is able to in situ characterize emulsions with up to 60 % oil content. The method is sufficiently sensitive to monitor severe changes due to coalescence as well as to trace small changes produced by Ostwaldt ripening. Emulsions with 80 % oil could not quantitatively be characterized.
However, we must bear in mind that these emulsions are susceptible to dilution. Thus, reliable results can only be expected when the samples can be analyzed without extensive dilution. Thus, this challenging problem cannot be solved by currently existing methods. The results of the ORM-Measurement, however, can be taken as a fingerprint which is characteristic for this sample and allows to identify changes which occur during manufacturing or storing.
6.2 Influence of formulations variables on the size distribution of o/w emulsions [12]
Laser diffraction (LD) has become one of the most popular particle sizing techniques and whenever any method is discussed, corresponding data from LD measurements are requested. For this reason, the following example compares 3D ORM data to LD measurements. The results originate from a study where the influence of ethanol on the properties of o/w emulsions was evaluated (Table 1). Such an addition of ethanol to o/w emulsions is frequently used in order avoid other preservatives.
Table 1: Composition of model o/w emulsions containing different amounts of ethanol.
Apart from improving the microbiological quality, adding ethanol changes the physico-chemical properties and might have a substantial impact on the appearance of the emulsions. Accordingly, particle size analysis can help to identify the changes that are likely to occur.
The 3D ORM measurements were performed on a Meßtechik Schwartz ECA model 817 (0.1 - 125 µm; 18 bins). For LD a Malvern Mastersizer was used and data were analyzed according to either Fraunhofer and Mie theory. The presented results are based on Fraunhofer analysis, however, Mie analysis produced distribution curves which were rather similar, except there was a slight shift towards smaller particle sizes.
The results of particle size measurement on the model o/w emulsions are shown in Figure 18. Both methods reveal a decreasing mean volume diameter with increasing amount of ethanol. This is consistent with images from light microscopy. The absolute values for D10%, D50%, and D90% are close together with always slightly increased values with the 3D ORM measurement. Both methods deliver highly reproducible results. However, on closer inspection there are remarkable deviations between both methods. 3D ORM indicates that D10% decreases when the amount of ethanol added increases, whereas this value remains unchanged with LD. Consequently, the span, which is defined as D10% and D90% range divided by the mean, decreases with increasing ethanol content, when calculated from LD. In contrast, the span from 3D ORM measurement appears to be almost invariable with the ethanol content. In addition, LD suggests bimodal distributions for emulsions without and with 5 % ethanol, while the corresponding 3D ORM data show monomodality. These differences can be explained by the underestimated smaller particles which are a typical feature of LD instruments and will appear predominantly when the particle sizes vary greatly [1]. Hence, in this example 3D ORM data seem to be somewhat closer to reality than the results from LD are.
Figure 18: Comparison of volume distribution and span of model emulsions containing different amounts of ethanol obtained from ORM measurement or laser diffraction.
6.3 Ethylcellulose-stabilized O/W- and W/O- emulsions [13]
Ethylcellulose is used for its ability as an oil soluble polymer to stabilize surfactant-free o/w emulsions. However, the consistency of these emulsions varied enormously depending on the processing temperature. Emulsification at 30° C produced liquid-like systems. Preparations at 15° C, however, delivered a semisolid cream. Further investigations revealed that the fluid lotions are of the w/o type whereas the semisolid systems represent o/w emulsions. Photomicrographs illustrate the structural differences. The fluid lotion is characterized by numerous small individual droplets, each surrounded by the coherent phase. In contrast, the semisolid cream exhibits an extremely wide particle size distribution. The corresponding results from particle size analysis are shown in Figure 19. The droplets of the lotion are measured to be smaller than 4 µm whereas the cream consists of droplets as large as 15 µm. Furthermore the particle size of the cream is extremely wide spread. Thus the measurement with the ORM-System as able to quantify what can be seen from photomicrographs only in a qualitative manner.
Figure 19: Volume distribution of a w/o type lotion and a o/w cream which have identical composition and use both ethylcellulose as polymeric emulsifier but they were prepared at different temperatures (15 or 30 °C).
6.4 Suspensions and Nanoemulsions
Finally, it should be demonstrated that the 3D ORM-Technology is not only able to measure particle size in polymer-stabilized emulsions. The method can also be used to characterize other dispersed systems, e.g. tooth pastes, as illustrated in Figure 20.
Figure 20: Volume distribution of two commercial samples of tooth pastes.
Figure 21 gives a look into the future. The measurement of a silicon-in-water nano-emulsion shows that this technology has also a potential in characterizing sub-micron sized dispersions. Further conclusions need, however, additional systematic studies.
Figure 21: Measurement on a sub-micron sized silicon-in-water emulsion demonstrates the possible applications of 3D ORM technology in the future.
7 Conclusions
3D ORM technology is primarily intended for in situ measurements. Since in real systems almost none of the other particle sizing methods can be used in a similar way, this feature can hardly be compared. However, 3D ORM technology also enables off line measurement. When such results are compared to TEM and LD measurements it turns out that 3D ORM is an equipotent method. It produces results which are reliable and highly reproducible when emulsions are analyzed together with droplets in the range from 0.4 to 40 µm and up to 60 % internal phase. For more concentrated Systems it is possible to create a fingerprint which represents the sample. In all cases, 3D ORM has the unique advantage that no sample preparation is required for an in real time and in situ measurement. This reduces not only the work load, but also helps to avoid artifacts when particulate systems are analyzed which are critical to dilution. This includes, of course, a lot of emulsion systems, and is most vital for suspensions in order to avoid uncontrolled dissolution of particles. The MTS 3D ORM sensors can also be sterilized under in situ conditions.
First results on sub-micrometer particulate systems with the nanoScan 2009 look also promising. However, the final decision over future applications depends on additional data.
8 References
[1] Washington, C., Particle size analysis in pharmaceutics and other industries, Ellis Horwood London, 1992
[2] Daniels, R., Barta, A., Erfahrungen beim Einsatz der in situ-Partikelmessung sowie der Messung der elektrischen Leitfähigkeit bei der Rezepturoptimierung von Emulsionen, Chemie-Anlagen + Verfahren 8/93, 62 - 66, 1993
[3] Daniels, R., Barta, A., Herstellung, Charakterisierung und Stabilitätsprüfung von O/W-Emulsionen mit Methylhydroxypropylcellulose als Emulgator, Pharm. Ztg. Wiss. 4, 177 - 183 1991
[4] Daniels R., Gerber M., Karwoth R., Empt U., Vorrichtung zur Messung von Partikelabmessungen in Fluiden. Deutsches/Europäisches Patent DE 196 23 095 A1/EP 0 813 052 A3
[5] Schwartz; F.H., Braun; M., Apparatus for measuring particle dimensions in fluids. United States Patent 5,900,933
[6] Karwoth, R., Handbook of ORM Technology, Meßtechnik Schwartz, D-Düsseldorf, 1996
[7] Lenzing, M; Handbook of 3D ORM Technology, Meßtechnik Schwartz, D-Düsseldorf, 2002
[8] Karwoth, R., Internal documentation, Meßtechnik Schwartz, D-Düsseldorf, 1996
[9] Twomey, S., Introduction to the mathematics of inversion in remote sensing and indirect measurement, Elsevier Scientific Publishing Amsterdam, 1977
[10] Rimpler, S., Pharmazeutisch-technologische Charakterisierung von O/W-Emulsionen mit Methylhydroxypropylcellulose als Polymeremulgator, Thesis University of Regensburg, 1996
[11] Rimpler, S., Daniels, R., In situ particle sizing in highly concentrated oil-in-water emulsions, Pharmaceutical Technology Europe .8(9), 72 - 80, 1996
[12] Wollenweber C., Einfluss von Ethanol auf Methylhydroxypropylcellulose stabilisierte Öl-in-Wasser Emulsionen. Dissertation Technische Universität Braunschweig 2000. http://opus.tu-bs.de/opus/volltexte/2000/91
[13] Melzer E, Herstellung und physikochemische Charakterisierung von W/O-Emulsionen unter Verwendung von Ethylcellulose als nichtionischem Polymeremulgator. Dissertation Technische Universität Braunschweig 2000. http://opus.tu-bs.de/opus/volltexte/2000/146
Author
Prof. Dr. Rolf Daniels
Prof. Dr. Rolf Daniels has a Ph.D. degree in Pharmaceutics. Before continuing his academic career, he worked for Pfizer in the department of pharmaceutical development for 2 years. In 1995 he became Professor of Pharmaceutics in the Institute of Pharmaceutical Technology at the Technical University of Braunschweig. His main interests are in the field of surfactant-free emulsions, stability assessment of semi-solids, and controlled delivery of insect repellents. Since 1997 he has been head of the department Dermocosmetics of the Society of Dermopharmacy (GD).
top