Evaluation of the Classification Performance of the New Centrifugal Particle Mass Analyzer Jonathan P.R. Symonds Cambustion Ltd, J6 The Paddocks, 347 Cherry Hinton Road, Cambridge CB1 8DH, UK [email protected] PSL spheres have in the past been used to characterise the prototype CPMA (Olfert et al., 2006), and both the commercial Kanomax APM 3600 (Tajima et al., 2011) and APM 3601 (Tajima et al., 2009) instruments. This is an initially appealing method; PSL is a certified standard for size (and hence if the density is known, mass can be estimated). The monodispersity of the PSL peak minimizes the population of doubly charged particles (of twice the setpoint mass) which may pass through the DMA. Note that the DMA is still required to remove the “surfactant mode” from the PSL and to ensure all particles are charged. Using a mass equivalent size metric, the resolution of the CPMA may be narrower than that of the DMA, and almost certainly will be narrower than the PSL size distribution: Penetration radial location (m) 90% + 6.00E-01 5.00E-01 1+ 4.00E-01 3.00E-01 0 .03 4400n n mm 0.04 0.05 0.0 6 41 41nm nm Desired Mass 0 .07 0.0 8 0 .09 0.1 length along classifier (m) Too Heavy unstable transfer fn Too Light The CPMA, by contrast to the APM, sets up a system of forces which balance across the entire classification region. It achieves this by spinning the inner cylinder slightly faster than the outer cylinder: Static Penetration (losses by diffusion) 50% 40% Dynamic Penetration 30% 2+ 0.00E+00 60 65 70 75 80 85 0 90 50 95 Dp* (nm) The other drawback is that the NaCl particles are cubic, and this affects their classification in the DMA. However, their dynamic shape factor can be used to correct for this: 100 150 DMA Size (nm) 200 250 The next plot shows the losses only attributable to dynamic operation of the CPMA, calculated by correcting the total losses by the static losses. Dynamic Particle Loss (diffusion corrected) 100% 90% χ f , NaCl = 1.08 d meCc (d me ) χf = d veCc (d ve ) Response (a.u.) 0.02 60% 0% 1.00E-01 0% 3399n n mm 70% 10% 0% 0 .01 80% 20% 2.00E-01 0.052 5 0 Penetration 100% 7.00E-01 100% 100% 0.0 5 In the plot below, the blue data points show the penetration with the CPMA not rotating and the voltage switched off. Losses in this case represent those due to diffusion in the 1 mm gap, and other transport losses such as impaction. This then defines the d50, static ≈ 20 nm. The pink points show the penetration with the CPMA operational (at Rm=3), corrected for multiply charged particles and the difference in the widths of the transfer functions of the DMA and CPMA. 80nm NaCl Use of Polystyrene Latex Spheres 0 .05 5 ω inner = ω outer Penetration & Particle Loss Penetration (%) However, as described, the system of forces is unstable. Only particles entering exactly in the correct location between the cylinders will be correctly classified. As the centrifugal force increases with increasing diameter, and the electric force increases with decreasing diameter, then particles entering closer to the outer cylinder will be subjected to a larger than desired centrifugal force, and those entering closer to the inner cylinder will be subjected to a larger than desired electric force; in either case particles of the desired mass:charge ratio will be lost to the walls, and the transmission efficiency of the device is reduced. two drawbacks to using NaCl. The aerosol is not monodisperse, so there may be multiply charged particles exiting the DMA. These will be counted by the first CPC, but not the second when the CPMA is set to the peak mass, so these must be accounted for when calculating the transmission efficiency. This can be done by extending the CPMA scan to include peaks from the multiply charged particles, and adding their post-CPMA population onto the singly charged particle population for the purposes of efficiency calculation. Particle Loss (%) The Centrifugal Particle Mass Analyzer (CPMA, Olfert & Collings, 2005) classifies aerosol particles by their mass:charge ratio. Like the Aerosol Particle Mass Analyzer (APM, Ehara et al., 1996), it uses opposing centrifugal and electrical fields. Pre-charged particles enter a classification region between two concentric cylinders which are rotating. The rotation causes a centrifugal force to be applied to the particles deflecting them outwards, whilst a potential difference applied between the cylinders deflects them inwards. Only particles of the correct mass:charge ratio such that the two forces balance will emerge from the classification section, otherwise they will be lost to either of the cylinder walls. using the drift limited model proposed by Reavell et al. (2011). The resolution parameter, Rm is defined as below: Output/Input Ratio [s] The CPMA mDMA = π 6 ρd ve 3 80% 70% 60% 50% 40% 30% 20% 10% 0% 0 Mass Accuracy 50 And at Rm=5: The data below were taken using NaCl, firstly at Rm=3, with 1.5 lpm flow through the CPMA: 100 150 DMA Size (nm) 200 250 Penetration 100% 90% Normalised Dp 0.01 Therefore, the CPMA at its peak will only ever allow a certain fraction of the total aerosol to pass; this has to be taken into account when calculating the penetration efficiency. 0.09 0.1 length along classifier (m) neutrally stable For an otherwise identical analyzer, Olfert et al. (2006) experimentally proved that this method does increase the penetration efficiency of the analyzer at the desired mass setpoint. In 2011, Cambustion launched a commercial version of the CPMA, with a 200 mm long classifier, 120 mm in diameter and a 1 mm gap between the electrodes. This paper characterizes the classification performance of this version of the CPMA. 100% 90% 80% 40% 30% 20% 10% 1.41 1.61 1.81 2.01 2.21 2.41 2.61 Mp* (fg) The test aerosols used here are Polystyrene Latex Spheres and Sodium Chloride. The CPMA is scanned for mass by counter-varying the speed and voltage; this ensures a constant resolution across the entire scan, unlike scanning the voltage alone. The appropriate speed and voltage for the desired mass and resolution is automatically calculated by the CPMA interface Here one considers the use of a Sodium Chloride aerosol instead. The advantages of this aerosol are that only one species is present, it can be used at smaller sizes than PSL, and the technique is reliant on the accuracy of the DMA alone. There are, however, Acknowledgements: The author would like to thank Dr Jason Olfert of the University of Alberta for useful discussions and the use of some of the line drawings above, and Kingsley Reavell of Cambustion. Presented at the American Association for Aerosol Research (AAAR) Conference, Minneapolis 2012 200 160 140 120 80 90 100 150 200 250 Dynamic Particle Loss (diffusion corrected) 70% 60% 50% 40% 30% 20% 50 Conclusions 0% -10% -20% -30% -40% 100 1000 DMA Size (nm) And then at Rm=5: Mass calculated from DMA size (fg) 0.01 0.1 1 10 10 Mc = 1.036 Md + 0.005 R2 = 0.998 1 60 50 40 30 0.1 200 250 The CPMA method agrees with the calculated DMA mass to within 5% across the range of sizes tested (20 nm to 200 nm). No appreciable loss in accuracy is observed for the smaller particles, unlike results reported by Tajima et al, 2009 (for the APM3601) which show increasing disagreement with the DMA mass as the particle size is reduced, reaching a 50% error at 20 nm. Results for the APM3600 (Tajima et al 2011) show good mass accuracy at smaller sizes, so it is not known what causes this issue in the 3601. The CPMA does use two autoswitching voltage ranges as it was found that a single voltage supply was not accurate across the entire range from 0.1 to 1000 V. If the voltage and speed and classifier dimensions are well controlled, then the accuracy of the CPMA should exceed that of the DMA method with assumed particle shape and density. Some loss of particles is observed at the highest speeds (= smallest particles & narrowest resolution), probably due to dynamic forces in the inlet and outlet regions of the classifier. Bibliography • J.S. Olfert & N. Collings. New method for particle mass classification—the Couette centrifugal particle mass analyzer, Journal of Aerosol Science 36 (11) pp1338–1352 (2005) Comparison with “DMA Mass” 50% 0.01 100 150 DMA Size (nm) • K. Ehara, C. Hagwood & K.J. Coakley. Novel method to classify aerosol particles according to their mass-to-charge ratio — aerosol particle mass analyzer, Journal of Aerosol Science 27 pp217–234 (1996) DMA Size (nm) • J.S. Olfert, K.St.J. Reavell, M.G. Rushton & N. Collings. The Experimental Transfer Function of the Couette Centrifugal Particle Mass Analyzer, Journal of Aerosol Science, 37 pp1840–1852 (2006) 40% % difference in mass The experiments in this paper use a DMA to classify (and pre-charge, by virtue of the bipolar neutralizer) aerosol by electrical mobility diameter, and then the CPMA is step scanned, producing a mass versus concentration spectrum for particles of that diameter (e.g. Olfert et al., 2007). 70 60 50 40 10% CPMA Mass (fg) Tandem DMA-CPMA Experiments There are a number of issues with using PSL for this kind of validation work. Firstly, though a certified standard for size, the tolerance in size quoted, when converted to mass can become unacceptable. For example, Thermo Scientific PSL at 50 nm has a standard deviation of 7 nm (or 15%), which corresponds to 45% in the mass domain. Secondly, the aerosol produced by nebulizing PSL also contains a large concentration of particles consisting of impurities in the water and surfactant chemicals used in the preparation of the PSL suspension. These particles will be of a different density to the PSL particles of interest, and therefore can cause artefacts when selected by the CPMA; this is especially a problem for smaller PSL particles. Thirdly, the success of this method depends on the accurate overlap of the DMA transfer function with the PSL distribution — the accuracy is as much a function of the accuracy of the DMA as the accuracy of the PSL. 100 DMA Size (nm) 0 20% -50% Use of Sodium Chloride 50 0% 30% 10 As well as showing agreement in terms of mass and transmission efficiency, this also validates the drift based model of Reavell et al (2011) used to calculate the speed and voltage setpoints of the CPMA from the desired mass and resolution. 0 10% 40% 50% 0% 80% Comparison with “DMA mass” 50% 60% 0% 1.21 DMA Size (nm) Peak (DMA at PSL peak) = 1.86 fg CPMA peak = 1.89 fg Error = 1.8% Upscan Downscan Model Dynamic Penetration 30% 90% CPMA Mass Scan, 150nm PSL, Rm= 5.13 70% 40% 10% Particle Loss (%) 0 .08 Static Penetration (losses by diffusion) 50% 20% 0.001 200 0.07 180 0.06 160 41 n m 0.05 140 40nm 0.04 120 0.03 100 39nm 0.02 90 0.01 60% 100% 80 0 70% 0.01 70 0 .05 0% 0% 1 0.1 30 0.0525 R = 0.999 20 (CPMA) 10 10 2 % difference in mass 1/radius radial location (m) ωα 100% 100% 1 MC = 1.047 Md + 0.0006 The figure below shows an upscan and a downscan of 150 nm Thermo Scientific PSL, classified with a TSI 3081 DMA, compared with the theoretical “transmission” which takes into account the difference in the widths of the source aerosol, and the transfer function of the DMA and CPMA. Output/Input Ratio [m] 0.055 0.1 CPMA Mass (fg) The forces are therefore neutrally stable: Mass calculated from DMA size (fg) Penetration (%) 80% 30% 20% • J.S. Olfert, J.P.R. Symonds & N. Collings. The effective density and fractal dimension of particles emitted from a light-duty diesel vehicle with a diesel oxidation catalyst, Journal of Aerosol Science 38 (1) pp69–82 (2007) 10% 0% • K.St.J. Reavell, J.P.R. Symonds & M.G. Rushton,. Simplified approximations to Centrifugal Particle Mass Analyser performance, European Aerosol Conference, Manchester (2011) -10% -20% • N. Tajima, N. Fukushima, K. Ehara, H. Sakurai. Mass Range and Optimized Operation of the Aerosol Particle Mass Analyzer, Aerosol Science and Technology 45 pp196–214 (2011) -30% -40% -50% 10 100 DMA Size (nm) 1000 • N. Tajima, H. Sakurai, N. Fukushima and K. Ehara. Determination of the Standard Operating Conditions of a miniaturized Aerosol Particle Mass Analyzer, American Association for Aerosol Research conference (2009)
© Copyright 2026 Paperzz