Overview Article
Particle Instruments & Diagnostics Research
For the past twenty-five years, researchers at the University of Hertfordshire, Hatfield, U.K., have been engaged in the development of instruments and systems for the characterisation of airborne particles. This work, now carried out within the University’s Centre for Atmospheric and Instrumentation research (CAIR), has been based largely on the use of spatial light scattering analysis as a means of achieving real-time characterization and classification of individual particles at high particle throughput rates. This article summarises the past results of this work and some of its areas of application. Please refer to Current and Recent Projects for more detailed information on our research work.
Optical Particle Counters (OPC’s) are tools widely used by the aerosol science community to provide particle concentration data and, in some cases, particle size spectra. Almost invariably, these instruments function by delivering a sample airflow containing the particles through an illuminating beam and recording the pulse of light scattered by each particle to a suitably placed detector. Correct design of the airflow system can ensure that particles pass through the beam in single file and that particle coincidences in the beam are minimal. In many cases, the magnitude of the scattered light signal is used to derive a first order assessment of particle size (usually with the assumption that the scattering particles are spherical).
However, there is known to be far more information about the particle embedded within the spatial distribution of the scattered light, and, over the years, advances in optoelectronic device performance, coupled with the continuing increase in low-cost computing capacity, have presented opportunities for this information to be extracted and used in the real-time characterization of complex aerosol systems on a particle-by-particle basis. We have been amongst those attempting to exploit these opportunities.
Spatial Light Scattering Profiles
The spatial distribution of light scattered by a particle, often called the scattering profile, is a complex function of the size, shape, structure, and orientation of the particle, as well as of the properties of the illuminating radiation (wave-length, polarization state). The examples given in figure 1 illustrate the wide variations these patterns can assume. If then, we are seeking to characterize or identify a particle on the basis of its size, shape, and structure, it is necessary to control or specify as many of the other variables as possible. For example, monochromatic, linear polarized radiation from a laser is frequently employed, and some attempt is usually made to control of limit the orientation of the particle with respect to the direction of illumination. The importance of particle orientation control can be appreciated from the examples given in Figure 2 which show the scattering from physically identical fibre particles, the only difference being the angle of each particle to the illuminating beam. Because of this sensitivity to particle orientation, several methods have been developed to improve orientation control, mostly based upon aerodynamic focussing of the sample airflow. In such cases, elongated particles can be made to align axially with the airflow.
One of the main challenges of using spatial light scattering to characterize aerosol particles comes from the need to interrogate the scattering profiles from individual particles at high rates, possibly many thousand particles per second. The earliest instruments developed at the University therefore opted for the simplest configuration of detectors, namely three detectors in a ring around the illuminating beam axis, as indicated in Figure 3. Despite their simplicity, these instruments proved to be very capable of characterizing aerosols into particle sets corresponding to broad scattering asymmetry: spherical particles, which scatter equally to all three detectors are ascribed an Asymmetry Factor of 0, whilst at the other extreme, long fibres which scatter predominantly to one detector only, are ascribed an asymmetry factor of 100. Since an approximate particle size can be also derived from the sum of the three detector outputs, it is possible to plot a graph of particle size versus particle asymmetry factor for the whole particle population, as illustrated schematically in Figure 3. ‘Low’ spatial resolution systems of this type form the basis of high-throughput real-time particle analysers now manufactured by Bristol Industrial and Research Associates Ltd., in the U.K.
The Particle Instruments and Diagnostics research is also involved in the development of instruments for monitoring airborne particles from aircraft platforms. ‘SID’, the Small Ice Detector, (SID1) developed for the Meteorological Research Flight at Farnborough, for use on the UK FAAM (Facility for Airborne Atmospheric Measurement) aircraft. SID1 employs a set of six high sensitivity detectors arranged about an illumination beam (from an Nd:YAG laser) to assess particle shape, and a further two detectors to provide trigger and particle size signals. The instrument is designed to classify cloud particles in the range from ~1 m to ~80 m diameter, and is intended to provide, for the first time, an ability to discriminate between micrometre sized ice crystals and super-cooled water droplets in, especially, cirrus cloud. (It is known that such particles play a vital role in the radiative properties of these clouds and therefore will affect climate behaviour as a whole, but since the scattering properties of each type are very different, the ability to model and predict such behaviour requires a knowledge of the actual particle types in the cloud). Figure 4 shows a schematic diagram of the mechanics of SID1, with an indication of where the instrument fits on the aircraft.
Medium to High Resolution Spatial Detection
It is evident from the scattering profiles shown in the above figures that there is more information available in the patterns than could be extracted using just three or even six detectors. Interestingly, past research carried out at UH has indicated that there is also an upper limit to the number of detectors which may usefully be employed. Above ~ 40-60 spatial detectors (the number depends in part on their spatial arrangement), the ability to classify particles based on their scattering profiles is not improved significantly by addition of more detectors unless thousands of discrete detectors are used (as, for example, in a digital camera). Furthermore, even at the level of a few tens of detectors, the data processing demands become very significant and the achievable particle analysis rates begin to fall unless expensive hardware processing is acquired. Nevertheless, by adoption of advanced detector technology and custom electronic data acquisition chips, we are exploiting the capabilities of these medium to high resolution systems in the field of aerosol monitoring.
Figure 5 shows a simplified scattering geometry which we use to develop prototype medium to high resolution instruments. The scattering profile detectors have taken a number of forms but are typically multi-element custom designed photodiode arrays exhibiting both wedge elements and segmented annular rings. A laser is used to illuminate the sample particle flow which is again arranged to ensure that particle coincidence in the beam is rare. The light scattered in the forward direction from the particle is directed onto the detector array. Particle trigger electronics and parallel data read capabilities ensure that the transient spatial scattering pattern is captured simultaneously for all detector elements. Figure 6 illustrates, using false colour rendition, the data which is captured by a typical detector array for various particle types.
The magnitudes of the signals received by each detector element constitute the data which are sent to the processor for analysis. Given that for each particle there will be several tens of bytes of data, and that the particle throughput may be several thousand per second, the real-time processing of data to provide the required monitoring and characterisation of the sampled aerosol can present a challenge.
SID2 Probe
As an example of a medium resolution scattering detection instrument, a second generation probe SID2, has been constructed under funding from NERC and the UK Met Office, and it to is available on the FAAM aircraft. SID2 employs a custom 31-pixel spatial light scattering detector which allows the morphology of particles down to micrometre sizes to be classified at rates up to 8,000 particles per second. For atmospheric ice crystals, for example, the classes would typically be columns, plates, rosettes, etc.
The SID2 (Small Ice detector Probe)
Processing Spatial Scattering Data
The choice of processing to be employed will depend to some extent on the hardware processing power available. The results of one of the simplest approaches and one which requires only modest (by today’s standards) processing power, is shown in Figure 7. Here, starting with data of the form given in Figure 6, a value R is evaluated (being the distance from the pattern centre to the centroid of intensity of the pattern), together with an ‘Asymmetry Factor’ (being a root-mean-square measure of the variation of signal values across all detector elements). These require modest processing but are in many cases capable of yielding useful classification of the data, as shown.
Figure 7: Classification of aerosol particles from spatial scattering data. (Purple=droplets; Blue=irregular cubic particles; Orange=fibrous particles; Green=flake-like particles.
In many other cases, a more effective (but more ‘processor hungry’) approach is to analyse the scattering data using an artificial neural network. In these cases it is necessary to ‘train’ the network using selected training-set data which represent the particle classes of interest. For example, if a particular type of particle is being sought in a sample atmosphere, the network would first be trained by passing through the instrument a sample aerosol containing just that particle type. The network would thus ‘learn’ to recognise the scattering characteristics of those particles and the likely variation in scattering brought about by both the inevitable physical variations between the particles and the orientational effects on scattering. We have used this approach successfully for asbestos fibre detection as well as in the assessment of certain environmental aerosols.
We are currently developing further the capabilities of the spatial scattering techniques by the additional simultaneous measurement of other parameters such as particle fluorescence (using UV laser illumination), Raman spectroscopy, degree of deformation of droplets in sample airflow, and measured aerodynamic size. Data such as these would be fed to the artificial neural network along with the spatial scattering data to enhance real-time particle identification and classification.
Current instrumentation developments include: the SID2H atmospheric probe, an enhanced version of the original SID2 probe designed for higher flight speeds and greater sub-micron particle sensitivity (SID2H is being constructed for the US HIAPER research aircraft); and the SID3 atmospheric probe, being built for the Met Office and UK academic research communities under NERC funding to allow high-resolution in-situ characterisation of cloud aerosol.
Contact
Prof. Paul Kaye
T: +44 (0) 1707 284173
F: +44 (0) 1707 284185
E: p.h.kaye@herts.ac.uk
PI Group home | Overview | Projects | Publications | Sponsors