WO1994020833A1

WO1994020833A1 - Method and apparatus for positional manipulation and analysis of suspended particles

Info

Publication number: WO1994020833A1
Application number: PCT/GB1994/000431
Authority: WO
Inventors: Ian Leslie John Holwill; Gareth Bowen Davies; Nigel John Titchener-Hooker; Michael Hoare
Original assignee: University College London
Priority date: 1993-03-05
Filing date: 1994-03-07
Publication date: 1994-09-15
Also published as: GB9304545D0

Abstract

In a method of particle handling, a source of acoustic oscillations is applied to a conduit through which a particulate fluid suspension passes, creating acoustic standing waves in the suspension thereby forming local inhomogeneities in the positional distribution of particles in said suspension which may be used to improve the signal-to-noise characteristics of an interrogating optical beam.

Description

tfethod and apparatus for positional manipulation and analysis of suspended particles This invention relates to methods of manipulation and analysis of the positions of particles suspended in a fluid and finds particular application in the in situ optical monitoring of suspensions of particles of differing sizes.

In this specification the term "bands" or "banding" is used to describe the formation of striations of particles in suspension by ultrasound. The term internodal light scattering

(ILS) is used to describe the coupling of optical analyses with such ultrasonic pre-preparation of the sample.

Photon correlation spectroscopy is a technique for measuring particle diffusion coefficients in suspension. It has been applied to many biological systems in the past as it is a fast, accurate and noninvasive measure which can be used for inferring particle size or velocity distribution. The technique may be applied to the problem of analysis of complex process streams on-line in a biochemical pilot plant. This is a useful technique for monitoring and control applications using information such as size distribution and concentration to optimise yield in a multi-stage purification process. Here a wide distribution of particle size is encountered. One of the problems faced is the fast and non-invasive sample preparation needed to remove unwanted particulates from the suspension prior to measurement.

Ultrasonic separation of such particles is presented here as a promising solution to this problem.

According to the present invention there is provided a method of particle handling in which a source of acoustic oscillations is applied to a conduit through which a particulate fluid suspension may be caused to pass to create acoustic standing waves in said suspension, thereby to create local inhomogeneities in the positional distribution of particles in said suspension. There is also provided apparatus for conditioning samples of particulates suspended in a fluid comprising a flow cell having input and output ports for the ingress and egress of fluid and primary and secondary sources of acoustic radiation to direct radiation at said flow cell to create acoustic radiation nodes and anti-nodes therein.

These local inhomogeneities may be used for the purposes of enhanced optical measurement of the fractionated particle suspension.

These local inhomogeneities may be used for purposes of measurement or particle removal and separation.

The invention will be particularly described with reference to the accompanying drawings, in which: Figure 1 is a graph showing correlogram data for different particle sizes; Figure 2 is a schematic diagram illustrating the principles of the internodal light scattering technique Figures 3a to 3d show the migration of suspended particles under the influence of acoustic radiation; Figure 4 is a schematic diagram of one specific embodiment of the invention; Figure 5 shows a practical cell design for use with the apparatus of Figure 4;

Figures 6 to 9 are results of measurements made on suspensions of different sized particles; and

Figures 10a to 13 are diagrams illustrating specific embodiments of the invention. Photon correlation spectroscopy measures intensity fluctuations from a suspension of particles undergoing Brownian motion and from these fluctuations a diffusion coefficient is calculated which can be related to particle size. The relation for the relaxation rate of the fluctuations and diffusion coefficient are shown below. The problem of inverting such data into the required distribution is an ill-posed question, that is to say, for one set of data many solutions can be fitted to within the noise of the experiment. The correlation function can be represented as below. The experimental procedure of photon correlation results in a correlation function which in the case of a range of particle size measurement is a sum of many exponential decays.

Y in k_BT γ = Dk² and D = for spheres (2) βirεRμ where γ is the decay constant corresponding to a particular diffusion coefficient and t is the experimental sampling interval. G(γ) is the required distribution.

A priori information about the size distribution may also be included to aid the deconvolution programs. However, improvement of the data quality must be the first step to a reliable measurement of the system.

An approximate expression for experimental signal to noise ratio for a onodisperse suspension is shown below.

signal = ^•/ γT (3) noise

This accounts only for photon counting noise and not noise due to dust or any systematic errors. Here T is the experiment duration. This shows that a determining factor for noise on the correlogram is the largest particle size present (all other things being equal). Thus, apart from increasing the experimental time the only other way to improve the ratio is to increase gamma. This can be done by using larger angles. If the maximum particle size is reduced however this will increase gamma as there is an inverse relation between the two. Multiple sample time correlators are used so that a wide range of particle size distribution can be covered. Better resolution is possible however with a narrower size range and a correspondingly narrower spacing of sample times. Keeping T as low as possible is important in on-line applications. Better accuracy should therefore be accomplished in this situation by reducing the range of particle size to be measured and keeping the maximum size of particles in the range as low as possible. The effect of particle size on the standard deviation of the baseline measurement over fifty runs is illustrated in Figure 1 which shows the average standard deviation over the correlogram data versus the diameter of polystyrene particle used in the experiment. A clear increase is shown in the deviation with particle size. An interesting feature is that a longer experimental time may lead to larger deviations in the data, particularly for larger particle sizes.

An additional problem is the number of particles in the sample volume. When this reaches the limit where N-^**^* >> N then number fluctuations within the volume become a problem as the underlying theory of equation 1 is then not valid. Typically this region begins at a particle size of approximately 0.5μm diameter.

If the whole of the size range is required noise levels cannot be optimised by removing part of the distribution. However, an improvement may be made in terms of dust removal. Dust particles are always a problem in light scattering experiments. In particular it makes the noise level of measurement difficult to assess and this information is necessary in some data processing techniques. In most photon correlation instruments dust detection systems operate to shut down the data collection when dust is in the laser beam. This operates using estimates of dust size and the dimensions of the scattering volume. In biochemical engineering applications, smaller particles in the suspension are often those of most interest i.e. they are the product and the larger particles unwanted (although they provide useful information for process optimisation considerations). Examples are virus-Uke particles or hepatitis-B virus surface antigen vaccine particles both of which can be expressed in carrier cells such as yeast. Yeast homogenate covers a wide range of particle size from whole cell ghosts of about 5μm down to the 200nm region. For a photon correlation measurement the upper size limit is about 3 microns and the presence of particles larger than this compromise the result substantially. Usually filtration or centrifugation is required to remove the larger contaminants although their size and quantity are still an important measure in the present situation. More recently a two aqueous phase operation has been used to separate the smaller contaminants.

A miniaturised centrifuge may be used to remove unwanted components rapidly and for small sample volumes. Inline filters are also used to remove dust particles from the samples. We have found that the presence of these large particles tends to distort the resulting size distribution or indeed hide the smaller particles altogether so that sample preparation is essential prior to measurement. The errors in the baseline can do more than simply shift the distribution. The larger particles can contribute an almost constant offset to the baseline when their size is such that it is out of range of the correlator.

Acoustic standing waves cause migration of fine particles in suspension to the nodes of the sound field which are spaced at half-wavelength distances. The main driving force for migration is the radiation pressure exerted on each particle in suspension. Equation 3 describes the radiation pressure exerted on a spherical particle in a plane stationary wave.

P_r = 4ιra³ιEFsin(2κx) (4) where K is the phase constant; E, the mean energy density of the standing wave; a, the particle radius, x the time averaged position of the particle from the node; and F a correction 5 factor that accounts for the compressibility of the particle and the different impedances of the particle and fluid:

Λ + 5i(Λ - 1) 1

(5)

10 1 + 2Λ 3Λσ²

Pi Λ = , σ = (6)

,5

Where p represents density and c sound velocity; subscripts 0 and 1 refer to the fluid and particle properties respectively.

From equation 3 it is clear that there is a very strong dependence of the radiation pressure on the particle radius a. 0 The radiation pressure P_r is proportional to the cube of the particle radius, a, which in practice means that the separation driving force falls off rapidly with decreasing particle diameter. This factor can be used to separate particles of different diameters. The radiation pressure displays only a 5 relatively weak dependence on the particle density p. and sound velocity through the particle medium c . and therefore particle size is the determining factor in the migration of particles in the sound field.

Figure 2 is a schematic of the internodal light scattering 0 technique (ILS). Particles 21 migrate towards nodes 23 in the standing wave field where the pressure is minimum. The force on the particles is size dependent hence larger particles may be fractionated from the smaller and with appropriate choice of ultrasonic frequency the distance between the nodes allows a 5 focused laser beam 25 to pass through for light scattering analysis. The distance between the nodes is of the order of hundreds of microns and the diameter of the laser beam is of the order of tens of microns. Figures 3a-3d show the migration of particles at four stages from initial application of a standing wave field.

Figure 4 is a schematic diagram of the ILS apparatus. An RF signal generator and amplifier 41 feed a transducer 42 connected to the light scattering cell 46 by way of a waveguide 44 and reservoir 42. The laser 45 and detector 47, part of the Malvern 4700 photon correlation spectrometer, are set up to perform light scattering analysis on the internodal spaces. The oscilloscope 49 is used to monitor the voltage fed to the ultrasonic transducers.

By controlling the energy density of the standing wave field E produced any particles over a chosen size can be immobilised or "banded" at the nodal plane of the stationary wave leaving the smaller particles free (unhanded) in the inter-nodal spaces. Appropriate choice of ultrasonic frequency and laser beam dimensions allows the particles in the inter-nodal spaces to be analysed by photon correlation spectroscopy. It is possible to do this rapidly and in situ so that a sample can be put into the apparatus without filtration or centrifugation steps and their associated problems. In the present application this may in some cases enhance and in other cases make possible measurement of soluble or insoluble products further up the unit operations chain. The technique of coupling laser analysis with ultrasonic standing wave clarification has been termed internodal light scattering (ILS).

Ultrasonic frequencies in the region of 1MHz were chosen due to relatively large distance between the nodal planes of the generated sound field and because of the efficient separation that occurs with sound of this frequency. An Audley Scientific Stabilised Frequency Signal Generator and Dual Ultrasonic Amplifier were used to drive the PC4 ceramic transducer (Morgan Matroc Ltd) producing the sound field. The initial work was carried out with a Malvern 4700 photon correlation instrument with a 633nm HeNe 30mW laser and an IBM PS2 for data processing. This technique is capable of particle sizing in the range 3μm down to about 3nm. The ultrasonic sample cell was set up in the index matching bath of the Malvern 4700 which reduces unwanted light reflections at the sample cell wall and fine vertical adjustment was facilitated by means of a Melles Griot optical component fine positioner. Thus the light beam could be passed through any part of the sample cell required. Polystyrene latex particles were chosen for use in this investigation as the aim of the work was to remove large particles from mixtures containing sub icron particles. Initial experiments were conducted using mixed mono odal distributions of very small latex particles (39nm) and larger (4.000 to 15.000nm) particles. The difference in average size between the two component distributions in suspension was then reduced to assess the lower size limit for the effective removal of the larger particles from the inter-nodal spaces. This included the use of 85nm particles, 2,000 and 15,000nm particles. Finally yeast homogenate diluted 1: 10 with filtered phosphate buffer was investigated with ILS analysis. The transducer voltage used was in the region of 25 volts peak-to-peak for the experiments using polystyrene latex particles and considerably higher (50 volts p-p) for the yeast homogenate experiments. The higher power levels were required due to the lower values of F associated with such biological particles.

Further work required the design of a self-contained ILS banding device which allowed more experimental flexibility and also incorporated a flow through cell to allow on-line sampling and analysis. Cooling was included to avoid undesired heating of the sample. A range of resonant frequencies for the transducers were chosen (1.25MHz, 21MHz and 3MHz). Higher frequencies allow more efficient banding (see equation 3) but reduce the internodal spacing and hence make it more difficult to perform internodal light scattering. A schematic diagram of the cell is shown in Figure 5, which depicts a vertical cross-section through the device. Ultrasound is guided through aluminium blocks 51, cooled by coolant which flows in the direction A to A', to the sample which flows in the direction B to B' through a narrow (f = 3mm) glass tube 52 surrounded by water 53. The water reduces flare and reflections where the laser beam hits the outer walls of the tube. Glass windows 54,55 are present in both the straight through position for transmission measurements and at 90 degrees for light scattering measurements. 0-rings 56 prevent the sample and index-matching water from mixing. The approximate dimensions of the device are 130mm from top to bottom and 50mm across and deep. The results obtained for a mixture of polystyrene latex spheres are shown in Figure 6. Particle size analysis of a mixture of 4μm and 39nm latex spheres. The solid line shows the analysis without banding. The two dotted traces are the internodal analysis and an analysis of 39nm spheres alone for comparison. The internodal analysis gives rise to a slightly broadened distribution probably due to scattering at the nodal planes. The distribution means for the lower two traces are 42.2nm for the onodisperse and 53.5nm for the internodally analysed mixture. A bimodal distribution is not shown for the normal analysis of the mixture as the software method used here was the cumulants method which is strictly for unimodal distributions but indicates the presence of the larger particles in the suspension by returning a higher distribution mean of 496.6nm. Data for the results of a unimodal analysis on a mixture of 39nm and 4,450nm particles without the application of an ultrasonic standing wave clearly indicate that the presence of the 4,450nm particles has seriously distorted the properties of the measured particle size distribution. The result of the ILS analysis of the mixture of 39nm and 4,450nm particles is shown in the same figure. The mean of the distribution is given as 53.5nm and the overall distribution is slightly broader than the 39nm suspension distribution. However the result shows that the ultrasonic standing wave has effectively moved the 4,450nm particles from the analysis. The slightly broader distribution of the ILS analysis results may be due to stray light scattered from the concentrated bands of particles at the nodes either side of the inter-nodal space entering the detector. However the distortion is

slight. This result indicates the utility of using ILS to pre-condition a sample for analysis of the smaller particles in suspension.

The results of ILS analysis on a mixture of 39nm and 2,000nm particles are shown in Figure 7. Between the nodes the suspension is evidently clear of 2μm particles and a distribution mean of 49.3nm is measured. The deviation from the expected size of 39nm is most likely due to scattering from the larger particles at the nodal planes. A measurement with the laser beam adjusted to intentionally scatter from the nodal region is also shown. This gives a diameter of 18.5μm for the distribution mean. This is indicative of the high concentration present in the bands as multiple scattering in a dynamic light scattering analysis usually gives rise to an underestimate of particle size. Diffusion is being hindered either due to ultrasonic forces or to interparticle interaction such as charge repulsion. Such hindrance will give rise to a larger effective diameter. Data from the analysis performed in the internodal space shows a clear peak close to 39nm indicating complete removal of the 2,000nm particles from the analysis volume.

Results from an analysis performed on the concentrated bands of particles accumulated at the nodes of the standing wave field show a mean of 18,500nm, about nine times the actual size of the 2,000nm particles. This result may be attributed to a number of phenomena, such as may be caused by the high concentration of the particles in this region resulting in particle interaction effects such as particles aggregating to form floes of particles; or an effect of the ultrasonic field hindering the normal diffusion of the particles. In cases of high particle concentration multiple scattering of the light from the sample may occur which invalidates the analysis technique used by the instrument as it is based on singly scattered light. However multiple scattered light, when interpolated in a dynamic light scattering experiment, will give rise to a lower than expected particle size rather than a very much greater one, hence one of the aforementioned causes is likely to be the reason for the large particle size returned by the analyser.

The results of the normal and ILS analyses of a mixture of 85nm and 15,000nm particles are displayed in Figure 8. (Internodal analysis (solid line) and normal analysis of a mixture of 85nm and a small quantity of 15μm latex spheres.) The inter nodal result shows a greatly increased contribution in the 80nm region demonstrating removal of a substantial fraction of the larger particles. However a broad peak at around lμm is apparent. This is due to other contaminants in the sample.

The dotted line represents a normal analysis (without ultrasound) of the mixture and shows a bimodal distribution indicating the presence of 85nm particles. The very broad peak from 5,000nm upwards reveals the presence of contaminants and as well as the 15,000 nm particles in the mixture.

The ILS analysis (solid line) shows a dramatic increase in the 85nm peak indicating the removal of the 15,000nm particles from the inter-nodal space. The remaining signal at 200-1,200 nm represents the contaminating particles in the suspension whose distribution has now been resolved by the removal of the 15,000 nm particles by the standing wave field. The results of an ILS analysis on a mixture of 85nm and l,400nm particles are given in Figure 9, a chart which shows internodal analysis of a suspension of 1.4μm and 85nm latex spheres. The original distribution (not shown) is very similar and shows that, for this experimental set-up, the banding limit for latex spheres is about 1.4μm. It will be possible to extend banding to lower particle sizes by appropriate modification to the design of the ultrasonic cell.

The original distribution (not shown for reasons of clarity) was very similar to the ILS result. This observation would suggest that there exists a minimum size limit for the removal of particles from the internodal space for this particular experimental set up. This result does not define the absolute limit for the technique, only of this particular piece of apparatus. Redesign of the ILS cell, incorporating the use of two transducers and more effective ultrasonic frequencies, would be expected to reduce the minimum size limit to the order of hundreds of nm by inputting a greater ultrasonic energy density and hence exposing the particles in suspension to very much greater radiation pressure forces. Experiments on the ILS analysis of unclarified yeast homogenate were performed to assess the possible improvements to be gained from the use of ILS in yeast based product particle six distribution analysis. Results of numerous experiments showed that at the high ultrasonic energies required to effect banding of the homogenate particles it was difficult to maintain the stability of the ultrasonic standing wave for sufficient time to record consistent and repeatable results. This observation may be attributed to the relatively high ultrasonic intensities causing the formation of gas bubbles in the homogenate suspension which in turn move spasmodically and chaotically within the starting-wave field disrupting the regular banding patterns of the particles.

Results of the ILS measurements returned a mean particle size of 636 ± 215nm with a polydispersity of 0.438 ± 0.051 for the yeast homogenate. Normal analysis (without ultrasound) of the homogenate gave a mean size of 1003 ± 328.2nm and a polydispersity of 0.535 ± 0.186 (The polydispersity is a measure of the deviation of the particle distribution from a perfect onodisperse distribution). The 90° scattered light from the normal and inter-nodal samples was 129.9xl0³ ± 14.4xl0³ and 79.8xl0³ ± 14.5xl0³ (counts per second) respectively indicating that particles contributing to nearly half of the scatter intensity had been shifted to the nodes of the standing wave field. Applications of this device for on-line sampling in biochemical pilot plant operations are wide.

1. VLP monitoring after homogenisation requires removal of a substantial fraction of the large yeast debris particles and it has been shown that yeast homogenate bands readily in the standing wave field.

2. Whole yeast cells can be banded leaving areas totally free of the cells in which proteins secreted out of the cell walls may be monitored. 3. Column eluants often contain column fines which interfere with a light scattering experiment and these may be banded away from the beam.

4. Dust particles especially for forward light scattering applications may be held away from the beam. This will be particularly useful in quality assurance and control experiments.

5. Monitoring of biomass in fermentation systems by optical means is often hindered by the presence of solid substrate in the media. A two step measurement of the system with substrate plus biomass and then, after banding the cells away, the substrate alone would allow subtraction of that part of the signal due to the substrate.

6. Monitoring low levels of particulates which may be below detection levels could be aided by concentrating particles for measurement at the nodes.

To set up the standing wave in the sample one ultrasonic transducer and a reflector or two transducers acting in opposition are operated across a sample chamber as shown in Figure 5. A standing wave field is created resulting in a periodic variation of pressure through the sample. The particles in suspension migrate to where the pressure gradient is minimum which is at the vibration nodes of the standing wave. These are spaced at half-wavelength intervals through the sample. The speed of sound in water is approximately 1500ms^"1 and the frequency of sound used is typically from 500Hz up to 10MHz. A 1MHz standing wave in water therefore has an inter-nodal spacing of 750mm.

The use of light scattering to measure the properties of particles or other components in fluids offers a rapid and convenient method for making non-invasive measurements. Often such experiments are hindered by the presence of large particles which may either be part of the sample or contaminating "dust" particles which are difficult to eradicate even after careful sample preparation. Ultrasonic standing waves maybe used to segregate particle suspensions in terms of size forming striations of particles in the sample of alternately large and small particle sizes as described above. The width of these striations is of the order of the acoustic wavelength, typically several hundred microns. The acoustic forces acting on the particles are strongly dependent on size and the energy density of the acoustic field may be varied to choose a particular size cut-off above which particles will move into striations leaving the smaller species to diffuse freely between these "bands". The smaller particles or other components may then be examined by light scattered from a focused laser beam which may be aimed into the suspension between the striations of large particles. The diameter of a focused beam at visible wavelengths may be easily designed to be of the order of lOOnm. The distance between the centres of the striations is λ/2 where λ is the acoustic wavelength in the medium. For a 2MHz frequency wave in water this is 375mm. The scattered light may be detected with a complementary optical set-up and hence a light scattering experiment carried out.

Since the length of the sample tube needs only to be a few acoustic wavelengths very small sample volumes only are needed. In many applications small sample volumes are a necessity.

Additionally for automated systems sample delivery and extraction are very simply effected with a stopped flow system. No membrane changes are required and no moving parts for sample delivery or extraction and the set-up can be simply operated in a contained fashion which may be important when dealing with volatile, corrosive or toxic samples. The sample to sample separation characteristics are also fixed by the energy density of the acoustic field. Acoustic streaming forces which are generated due to non-uniform radiation pressure across the sample due to a non-uniform radiation from the driving transducer may cause flow effects throughout the sample similar to thermal convection currents and such flows may affect the particle size measurement. In practice we have found that reducing the area over which banding occurs will aid reduction of such disturbances however a larger area allows a greater energy density in the acoustic field and hence better separation.

Cavitation, the formation of microbubbles due to acoustic energy may also disrupt the scattered light signal. Such cavitation can be avoided at higher energy densities by using higher frequency ultrasound which is less prone to such instabilities and by degassing the dilution water used in the experiments. At higher frequencies the effects of streaming are of more concern.

The desire for higher ultrasonic frequencies to avoid cavitation is at odds with the need for a sufficient space between nodes in which to focus a laser beam. The higher the frequency, the less space available and hence the more likely is spurious scattering from such bands entering the detector. The frequency range ultimately chosen is a compromise between the two. We have found that at 3MHz with an internodal distance of approximately 180mm it is difficult, although not impossible to perform light scattering experiments. A 2MHz frequency offers a satisfactory compromise between separation efficiency and internodal space and at 1MHz the alignment is made very much easier but cavitation events were far more frequent at the higher transducer drive voltages. The choice of frequency therefore becomes dependent on the sample to be examined such that if only very large particles need to be removed from the measurement volume and low drive voltages sufficient, the lower frequencies will suffice. Conversely as one moves to separate smaller particles down to the sub icron region, higher frequencies are essential and the alignment more crucial. Different optical measurement techniques operate at different wavelengths therefore the geometry of the set-up will vary.

A priori information about the size distribution may also be included to aid the deconvolution programs, however improvement of the data quality must be the first step towards the realisation of a reliable measurement of the system.

In particular the measurement can often provide more information if the range of particle size to be measured is reduced. Also the presence of particles above the upper size threshold at which the sizing technique operates decreases the signal to noise ratio at the detector by scattering spurious light. The removal of such particles can therefore provide better data from which to calculate the size distribution of the sample. The free diffusion of the smallest particles is unaffected by the ultrasonic forces. The measurement of optical density is prevalent in many areas in order to determine concentrations of solutions or suspensions. The Beer-Lambert law is usually used to describe the reduction in transmitted light intensity due to scattering and absorption in dilute systems. When the sample is more concentrated multiple scattering occurs and the Beer-Lambert law breaks down. In such cases alternative theories must be used or a calibration made.

There are two situations where the ILS method may be useful here. The first concerns the measurement of solutions or small particulates in the presence of large and the second the measurement of low concentrations of particles which may be enhanced by concentrating them with the ultrasound into the measurement volume. The former reflects a similar situation to that found in DLS in that the presence of larger particles in the suspension cause the detected signal to be dominated by light scattered from such particles and further information may be available if these particles are removed. The latter situation may be useful where a low concentration of particles open to manipulation by ultrasound may only be detected by increasing their local concentration thus rending a higher OD or scattered signal .

The geometry for such measurements is varied. For looking at concentrated bands a focused beam on to the banded sample will be necessary but for conventional O.D. measurements either several beams focused between the striations of large particles or a mask created for the purpose of blanking the banded areas out of the transmitted signal, thus allowing an equivalent reference to be measured (see figure 10a to 10c). This shows an incident beam I.- passing through a mask M. The transmitted beam I.*- will exhibit shadow portions S. Under normal flow through conduit C, suspended particles are uniformly distributed, (Figure 10b) but when the acoustic radiation is applied, striations form at the nodes N.

One application for the measurement of refractive index is the estimation of component concentrations of small species in solution. A simple expression for the concentration dependence of refractive index is shown in equation 1. Alternatively, taking into account the polarisability of the molecules, the Clausius-Mossoti equation may be used (for spherical particles). If large particles are present they are likely to disrupt the signal and concentration estimates compromised, but displacement by ultrasound will alleviate this.

The geometry of this set-up depends upon the method of refractive index measurement but one possible method is shown in Figure 11. A sample cell 111 is mounted on a glass block 113 of refractive index N. Acoustic waves are generated by a transducer 115. Incident optical I-_j radiation is focused by a lens 117. The emergent radiation I_t is measured by a diode array 119. Another possible application is infra-red spectroscopy. The detection method for this is the absorption of infra-red radiation due to the optical resonances of chemical bonds. With knowledge of components present and precalibration, multi-component mixtures may be quantified. Again for complex suspensions where soluble components are of interest, the larger particulates often hinder the accuracy of the detected signal. Removal of large particulates will improve the accuracy. Of importance here is the geometry of the system as regards wavelength, which runs from approximately 0.7mm to 25mm for near IR, 25-50mm for id-IR and >50mm for far-IR. The size of a spot able to be focused depends upon the optical set-up and can be expressed by equation x. Hence as wavelength increases the spot size will grow past the internodal distance. Thus either operation with a mask as in the O.D. measurement is necessary or a back reflection measurement is required as shown in Figure 12. In this configuration due care needs to be taken with the reflection received from the thick band of particles behind the sample region. The incident beam I.^* is back-scatter by the sample I_s and by the particulate wall I_w.

Low angle and multiple angle laser light scattering may be used to determine molecular weight and elucidate particle shape, as shown in Figure 13, which shows a sample 131 held in a cylindrical cell 133. Light from a laser beam is scattered through an angle θ by the particles. A direct analogy of the case of dynamic light scattering may be drawn here when attempting to measure small components in a complex mixture. In particular for low angle measurements where scattering from large particles is greatest it will be most useful in removing unwanted scatterers.

The technique is not restricted to clearing suspensions for dynamic light scattering experiments alone. Static light scattering would also be possible in the simplest case for optical density measurements. The technique should find applications in other areas as well.

Claims

Claims : - 19 -

1. A method of particle handling in which a source of acoustic oscillations is applied to a conduit through which a particulate fluid suspension may be caused to pass characterised in that acoustic standing waves are created In said suspension thereby forming local inhomogeneities in the positional distribution of particles in said suspension.

2. A method of particle handling according to claim 1 characterised in that a sample-to-sample separation characteristic is controlled by changing the energy density of the field of said acoustic standing waves.

3. A method of particle handling according to claim 1 characterised in that the frequency of the radiation of said acoustic standing waves is greater than 1MHz.

4. A method of particle handling according to claim 1 characterised in that the frequency of the radiation of said acoustic standing waves is in the range l-3MHz.

5. A method of measuring a characteristic of particles in a fluid according to any one of the preceding claims characterised in that optical radiation is directed at said fluid at a region of non-uniformity of distribution of said particles and that measurement 1s made of the optical radiation emerging from said fluid.

6. A method of measuring a characteristic of particles in a fluid according to claim 5 characterised in that said region is a region of minimum density of the larger particles in said suspension.

7. A method of measuring a characteristic of particles in a fluid according to claim 6 characterised in that said particles are incorporated in a mixture having a component of larger size than said particles, whereby said particles are at least partially separated from the particles of said component.

8. A method of measuring a characteristic of particles in a fluid according to claim 7 characterised in that said component is yeast.

9. A method of measuring a characteristic of particles in a fluid according to claim 7 characterised in that said component is dust.

10. A method of measuring a characteristic of particles in a 5 fluid according to claim 5 characterised in that fluid is a column eluent.

11. A method of monitoring a fermentation biomass characterised in that it comprises measuring an optical characteristic of said biomass followed by making a further measurement by a method

10 according to any one of the preceding claims 1 to 5.

12. Apparatus for conditioning samples of particulates suspended in a fluid comprising a flow cell having input and output ports for the ingress and egress of fluid characterised in that it includes primary and secondary sources of acoustic radiation to

15 direct radiation at said flow cell to create acoustic radiation nodes and anti-nodes therein.

13. Internodal light scattering measuring apparatus characterised in that it includes conditioning apparatus according to claim 12.

20 14. Internodal light scattering measuring apparatus according to claim 13 characterised in that it includes a mask (M) to restrict an incident light beam to pre-determined regions of a sample cell (C).

15. Internodal light scattering measuring apparatus according to 25 claim 13 characterised in that it includes a sample cell (111) mounted on a glass block (113) and that emergent radiation is measured by a diode array (119).

16. Internodal light scattering measuring apparatus according to claim 13 characterised in that it includes a source of infra-red

30 radiation.