Evaluation of an optical particle sensor to determine the effect of nozzle shape on counting efficiency

Aerosol Science 40 (2009) 469 -- 476

Contents lists available at ScienceDirect

Aerosol Science

journal homepage: www.e lsev ier .com/ locate / jaerosc i

Evaluation of an optical particle sensor to determine the effect of nozzleshape on counting efficiency

Younggil Kima, Atul Kulkarnid, Kisoo Jeonb, Jinuk Yoonc, Sangwoo Kange, Juyoung Yune,Yonghyeon Shine, Taesung Kimd,∗

aGraduate School of Mechanical Engineering, Sungkyunkwan University, Suwon, Republic of KoreabGraduate School of Mechanical Engineering, Hanyang University, Seoul, Republic of KoreacHCT, Icheon, Republic of KoreadSchool of Mechanical Engineering, Sungkyunkwan University, Suwon, Republic of KoreaeKorea Research Institute of Standards and Science, Daejeon, Republic of Korea

A R T I C L E I N F O A B S T R A C T

Article history:Received 11 June 2008Received in revised form17 October 2008Accepted 17 December 2008

Keywords:Optical particle sensorNozzle shapeParticle sizeLight scatteringParticle counting

The optical particle sensor (OPS), using light scattering from particles, is widely used in clean-rooms and in atmospheric environmental monitoring. However, parameters that affect theparticle counting of the OPS have been less explored previously. The present study examinesthe effect of the nozzle shape on particle counting performance of the OPS. Experiments andsimulation studies were carried out for four different nozzle shapes in conjunction with threedifferent particle sizes. The effect of nozzle shape is evaluated not only by experiments ana-lyzing the intensity of scattered light and particle velocity but also by simulation. Fluent wasused to simulate the flow field and particle trajectories. We observed that particle velocity andtrajectory are strongly dependent on the input nozzle shape, which in turn has an effect onthe particle counting performance. Using these results, the design of OPS can be optimized toprovide higher counting efficiency.

© 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Optical particle counting is one of the main technologies employed in environmental monitoring and in the control of manyparticulate materials. Because of its ability to make in-situ measurements, the optical particle sensor (OPS) is widely used forcontamination analyses, e.g., water, clean rooms and hydraulic fluids, and for atmospheric analyses of liquid-borne or air-borneparticles (Tiwary & Colls, 2004). There are various optical methods for particle detection, such as measuring relaxation timeof the particle using optical aerodynamics (Mazumder & Kirsch, 1977), measuring the scattering intensity of particles (Jones,1999), imaging the particle size (including the holography; Thompson, 1974), and measuring the Doppler phase (Hirleman,1996). Among the above methods, light scattering intensity measurement has the lowest cost, is easy to operate and has theadvantage of in-situmeasurements without physically contacting the particles. It was therefore implemented to study compositecharacteristics of particles (Chang, Okuyama, & Szymanski, 2003), flame-formed particles (Flower & Hurd, 1987), liquid-borneparticles (Chernyshev, Prots, Doroshkin, & Maltsev, 1995) and microbes (Volten et al., 1998).

The OPS performance is dependent on various factors in spite of its simplicity. The minimum detectable particle sizeand the particle counting ability is strongly influenced by the laser source, photo detector, nozzle shape and particle size

∗ Corresponding author. Tel.: +82312907466; fax: +82312905889.E-mail addresses: [email protected] (Y. Kim), [email protected] (A. Kulkarni), [email protected] (K. Jeon), [email protected] (J. Yoon),

[email protected] (S. Kang), [email protected] (J. Yun), [email protected] (Y. Shin), [email protected] (T. Kim).

0021-8502/$ - see front matter © 2009 Elsevier Ltd. All rights reserved.doi:10.1016/j.jaerosci.2008.12.009

470 Y. Kim et al. / Aerosol Science 40 (2009) 469 -- 476

(Son, Kang, & Jeon, 1991). Apart from the proper selection of the laser source and photo detector, the nozzle shape is one ofthe critical factors, which affects the particle beam size and flow velocities in the detection area, and this may in turn affect thecounting efficiency of the particles. We therefore feel that there is a need to investigate the effect of the nozzle shape on thecounting performance of the OPS. To the best of our knowledge no such report is available and this study will be helpful to theaerosol researchers as well as the optical particle counter manufacturers.

The objective of the current paper is to evaluate the counting performance of the OPS, experimentally as well as by simulationstudies, for different types of nozzle shapes and with different particle sizes. During the evaluation we used four differentinput nozzles in conjunction with three different particle sizes. The OPS results were compared with a standard calibratedcondensation particle counter (CPC) during the experiments. A commercial computational fluid dynamics software (Fluent) wasused to simulate the flow field and particle trajectories in accordance with the nozzle shape.

2. Methodology

In the present study, we used OPS which was designed and manufactured by Hyundai Calibration and Certification Tech-nologies Co., Ltd., South Korea (HCT). This is a commercial OPC intended for environmental and industrial monitoring. Thespecifications of the OPC provided by the manufacturer are listed in Table 1. A schematic of the OPS coordinate system with theinput and output nozzle is shown in Fig. 1.

The direction of the particle beam carried by fluid flow and the laser beam focused by the lensmodule is in the direction of the(–) Y-axis and X-axis, respectively. Light will scatter when the particle passes through the laser beam, which is in turn detectedby the photo-detector along the Z-axis, providing information on the particle counting, size and velocity. The chamber's innersurface has an anti-scattering material that prevents unwanted light scattering.

In order to investigate the effect of the nozzle on particle counting performance, this study considered four different cases ofinput nozzle shapes,while the output nozzle shape is kept constant throughout the study. The nozzle shape and size is determinedby d1, d2 and d3, where d1, d2 are the inlet and outlet diameter of the input nozzle, respectively, and d3 is the outlet diameter

Table 1Preliminary Specifications of the OPS provided by the manufacturer.

OPS modelnumber

Light source/power (mW)

Spectrum(�, nm)

Sampleflow (lpm)

Scatteringangle(range)

Photo detectorresponsetime (�s)

Spec. Sizerange (�m)

Spec. maxconc. (cm−3)

Countingmethod

4323 Laser diode/35 658 2.83 18–26 0.033 0.3–5.0 1000 Single count mode

Fig. 1. Schematic layout of the OPS for the coordinate system used in the simulation studies.

Y. Kim et al. / Aerosol Science 40 (2009) 469 -- 476 471

Fig. 2. Schematic of input and output nozzle, where d1 and d2 are the inlet and outlet diameter of the input nozzle, respectively, and d3 is the outlet diameter ofthe output nozzle (for dimensional details please refer to Table 2).

Table 2Input nozzle diameter used during simulation and experiments, where the output nozzle diameter is kept fixed at 1.5mm.

Case number Input nozzle diameter Remarks

Inlet (d1, mm) Outlet (d2, mm)

1 2 2 Reference model2 1 1 Small nozzle model3 4 2 Cone nozzle model4 4 4 Big nozzle model

of the output nozzle, as is depicted in Fig. 2. During simulation and experimentation we used standard polystyren latex (PSL)particles procured from Microgenics Corporation, Fremont, CA. The particle sizes used were 300, 500 and 700nm.

Case 1 is considered as a reference case and all other cases are compared to it accordingly. The inlet and outlet diameter ofthe input nozzle in case 2 are kept smaller than that of case 1. We selected a cone shape nozzle in case 3 to evaluate the effect ofthe focused particle beam on particle counting, where the outlet diameter of the input nozzle is smaller than the inlet diameter,making a cone angle of approximately 2.12◦. The inlet diameter and outlet diameter of the input nozzle in case 4 are kept biggerthan case 1. This will allow us to evaluate the effect of the big nozzle size on particle counting. The input and output nozzledimensions in all four cases are listed in Table 2.

We used Fluent software (version 6.2.16) to evaluate the particle trajectories for various nozzle shapes, with assumption thatthe flow is turbulent and incompressible. The volumetric flow rate at the inlet is considered to be 0.3 lpm and the gradientsare assumed to be zero for the outlet of all the cases. No slip boundary condition is used for the chamber's inner surfaces. It isassumed that particles colliding with the surface are trapped. The finite volume method (FVM) and SIMPLE algorithm were usedto simulate the flow. A standard k − � model was used to analyze a turbulent flow field. Individual particle trajectories weretracked using discrete phase model (DPM) in Fluent, which employs a Lagrangian approach that considers gravity and drag forceusing the Stokes–Cunningham equation.

The schematic of the experiment set-up used during this study is described in Fig. 3. The experimental set-up consisted of aDMA (Differential Mobility Analyzer, TSI DMA 3081), a CPC (Condensation Particle Counter, TSI CPC 3025A), an atomizer (HCTmodel 4810), a diffusion dryer, and the OPS under test. The BNC2110 is used to provide the connectivity between the OPS and aPC using a National Instruments data acquisition system. The digital and analog signals captured from the OPS under test weresampled at approximately 250,000 samples per second to evaluate the experimental results for counting. The particles generatedfrom the atomizer are classified as a specific particle size after passing through the DMA. The OPS under test is located betweentheDMAand CPC. The particles therefore travel through theDMA, OPS, and CPC sequentially. Themeasurement datawas acquiredfrom the OPS and CPC in real time. The particle's flow rate is kept at 0.3 lpm and is within the flow limit of both the particlecounters. It is well known that light scattered from a particle follows theMie theory for particles sizes used in this experiment andis a function of the particle's size and shape, as well as, the ratio of the particle's refractive index to the transport media refractiveindex. Hence, the experimentally observed results were finally verified by applying the Mie theory. Bohren and Huffman (1983)have developed a program based on theMie theory and this programwas used to calculate the theoretical response of the OPS forthe three particle sizes. The refractive index of the particle and aggregation of particles has an effect on the scattering intensity

472 Y. Kim et al. / Aerosol Science 40 (2009) 469 -- 476

Fig. 3. Schematic diagram of the experiment set-up. The OPS under test is connected before CPC.

(Liu & Daum, 2000; Quinten, Friehmelt, & Ebert, 2001). Our study, however, considered that the scattering is due to a single PSLparticle only and that there is no aggregation of these particles.

3. Results and discussions

We systematically evaluated the effect of selected nozzles on the performance of the OPS during simulation. The simulatedfluid flow velocity distribution inside the OPS is depicted in Fig. 4 for all the four cases and for 700nm PSL particles. We foundthat, in all the four cases, the flow velocity reduces after the outlet of the input nozzle due to a wide chamber area. It is clearlyseen that the flow velocities experience no change in the X-axis and Z-axis directions.

For simulation of particle trajectory the physical properties of the PSL (700nm) particle were used and 10 particles whichwere equally spaced along the X-axis at the inlet were tracked. Fig. 5 shows the simulated particle trajectory for all four nozzlecases, it reveals that there are no major differences in the particle trajectories for the various nozzle shapes. It is obvious that theparticle trajectories are spreading in the chamber area; however, in case 2, we observed that this phenomenon occurred after thedetection area in the chamber and this is mainly due to the inertial force on the particle.

The 700nm particle velocities along the Y-axis are depicted in Fig. 6 as an example. The chamber shape is shown in the darkgrey color background for ease of understanding. The flow field and the trajectories are in the (–) Y-axis direction for all thefour cases. We observed that the particle velocities are dependent on nozzle dimension. Since the outlet diameter of the outputnozzle is the same in all the cases, we observed that the particle velocity distribution is also the same at the output nozzle area.In case 1, the particle velocity is uniform in the inlet nozzle area with little variation in the detection area, where we estimatedthemean velocity to be about 1.86m/s. The particle velocity is higher in case 2 than in case 1. This observation is obvious becausecase 2 has a small inlet and outlet diameter of the input nozzle in comparison to case 1. The estimated mean velocity in thedetection area is about 5.91m/s, which is much higher than in case 1, because the particle beam size is smaller than in case 1.Case 3 illustrates the effect of cone shape nozzle on the particle velocity. Since the outlet diameter of the input nozzle in case3 is equalized with that of case 1, the particle velocity gets accelerated and then subsequently moves similar to case 1 after thedetection area. Particle trajectories are focused within the input nozzle area, but the particle beam size in the detection areais similar to case 1; therefore, particle counting is anticipated to be similar to case 1. The particle velocity and trajectory arethe same in the detection area for cases 1 and 3, even though they differ in their inlet nozzle area. Therefore, we found thatthe outlet diameter of the input nozzle has a strong influence on the velocity as well as on the trajectory in the detection area.The estimated mean velocity in the detection area for this case is about 1.88m/s. The large inlet and outlet diameter of case 4, ascompared to other cases, causes a drop in particle velocity as low as approximately 0.56m/s. We observe from the figure that theparticle velocity remains unchanged in the detection area. Particle trajectory is found to follow a straight line in the detection

Y. Kim et al. / Aerosol Science 40 (2009) 469 -- 476 473

Fig. 4. Fluid flow velocity distribution using Fluent for all four cases. The velocity bars (meter/sec) shown correspond to each case for 700nm PSL particle.

474 Y. Kim et al. / Aerosol Science 40 (2009) 469 -- 476

Fig. 5. Simulation results of the particle trajectory for all four cases with equally spaced 700nm PSL particles.

Fig. 6. Particle velocity distribution for all four cases with particle position on the Y-axis using 700nm PSL particle where one particle is tracked from the centerof the inlet.

area. The outlet diameter of the input nozzle is bigger than both the other cases and the laser beam diameter; therefore, particlecounting is expected to be low due to the loss of particles which do not pass through the laser beam.

The experimentally evaluated results for particle counting efficiencies for all the cases and particle sizes are depicted inFig. 7. The number concentration of PSL particles were measured by CPC during experiment is in the range of 400–500particles/cm3 and is well below the specified concentration limit of the OPS, above which coincidence may occur. The plotshows a dependence on particle size and nozzle shape for particle counting efficiency. It is well known that there is lower scat-tered light as the particle size decreases, and may be beyond the detection limit of the detector. The observed particle countingis, therefore, found to be very low for 300nm particles as compared to the other two particle sizes, and in such cases it is verydifficult to analyze the effect of the nozzle. Alternatively, there is enough scattering light for the 500 and 700nm particles to beable to evaluate the nozzle effect on the counting efficiency. The particle counting efficiency is high in case 2. Laser light intensityhas a Gaussian distribution and therefore, the scattered intensity is different according to the particle path through the laserbeam even though the particle sizes are equalized (Jones, 1999; Holve & Self, 1979).We know that the particle beam size of case 2is smaller than that of case 1 from the particle trajectory analysis using Fluent. Thus, the probability of a particle passing throughthe middle of laser beam for case 2 is much higher than for case 1. The particle counting efficiency of case 4 is lower than that ofcase 1 for the same reason. We also found that there is no effect of using cone shape to focus the particle beam since the particlecounting efficiency was the same for cases 1 and 3. Like the preceding simulation results, we can also experimentally concludethat the outlet diameter of the input nozzle is an important factor that affects the counting and trajectory in the detection area. Ingeneral, we observed that the particle counting efficiency increases with particle size, irrespective of the nozzle size and shape.

Upon investigating the voltage peaks from the photo detector output during particle counting we observed that there isno coincidence effect. This is due to the low concentration of particles 400–500/cm3, which is well below than the specified

Y. Kim et al. / Aerosol Science 40 (2009) 469 -- 476 475

Fig. 7. Experimental results for particle counting efficiency shows distinct variation for all four cases and particle sizes.

Fig. 8. Particle residence time in the detection area.

maximum concentration value of the OPS. Also the sample flow during experiment is maintained at 0.3 lpmwhich is much lowerthan the specifications of OPS, which resulted in an isolated voltage peak for single particle counting.

The particle residence time in the detection area will also affect the particle counting performance of the OPS, so we evaluatedthe particle residence time experimentally and the results are depicted in Fig. 8. The residence time is obtained from the meanpulse width during experimental measurements. We observed that there is a clear correlation between cases 1 and 3 and whichindicates that the cone shape of case 3 does not affect on the residence time. It is impossible to analyze the residence time forcase 4, since many particles pass through the edge of the laser beam and limit good experimental data needed for comparison.Note that the larger particles move slower than the smaller particles, as is expected (Wang & Hencken, 1986).

As the nozzle shape in case 2 showed highest counting efficiency, the average pulse heights produced by three size particlesfrom case 2 are superimposed on the scattering cross section calculated from the Mie theory for sizing performance evaluation(Fig. 9). A general agreement between the experimental and theoretical result was observed. The pulse height of 300nm particlesis slightlymore deviated from theoretical calculation compared to other particle sizes probably due to the relatively small amountof light scattering, which also resulted in low counting efficiency.

4. Conclusions

This study numerically evaluates the particle velocities and trajectory in the OPS and compares them with the experimentalresult. We also evaluated the effect of nozzle size and shape on the particle counting efficiency. The results indicate that particlecounting performance is affected by particle velocities, their trajectories in the detection zone, and the inlet nozzle size andshape. Out of the four cases studied, particle counting efficiency is highest for case 2 with the smallest particle beam, which is

476 Y. Kim et al. / Aerosol Science 40 (2009) 469 -- 476

Fig. 9. Validation of theMie theorywith experimental results for the case 2. Black dots represent the pulse height with respect to particles used in the experiment.

related to the Gaussian distribution of the intensity in the laser beam. The particle residence time measurement in the detectionarea showed that the response time of OPS is fast enough to detect particles for case 2 with highest velocity. The findings fromthis study can greatly aid in the development of an OPS to improve its performance.

Acknowledgments

This work was partly supported by the GRRC program of Gyeonggi province and HCT through research program and instru-mental support

References

Bohren, C. F., & Huffman, D. R. (1983)). Absorption and scattering of light by small particles. New York: Wiley.Chang, H., Okuyama, & Szymanski, (2003). Experimental evaluation of the optical properties of porous silica/carbon composite particles. Aerosol Science and

Technology, 37, 735–751.Chernyshev, Prots, Doroshkin, & Maltsev, (1995). Measurement of scattering properties of individual particles with a scanning flow cytometer. Applied Optics.

Lasers, Photonics, and Environmental Optics, 34, 6301–6305.Flower, W. L., & Hurd, A. J. (1987). In situ measurement of flame-formed silica particles using dynamic light scattering. Applied Optics, 26, 2236–2239.Hirleman, E. D. (1996). History of development of the phase-doppler particle-sizing velocimeter. Particle & Particle Systems Characterization: Measurement and

Description of Particle Properties and Behavior in Powders and Other Disperse Systems, 13, 59–67.Holve, D., & Self, S. A. (1979). Optical particle sizing for in situ measurements part 1. Applied Optics, 18, 1632–1645.Jones, A. R. (1999). Light scattering for particle characterization. Progress in Energy and Combustion Science, 25, 1–53.Liu, Y., & Daum, P. H. (2000). The effect of refractive index on size distribution and light scattering coefficients derived from optical particle counters. Journal of

Aerosol Science, 31, 945–957.Mazumder, M. K., & Kirsch, K. J. (1977). Single particle aerodynamic relaxation time analyzer. Review of Scientific Instruments, 48, 622–624.Quinten, M., Friehmelt, & Ebert, (2001). Sizing of aggregates of spheres by a white-light optical particle counter with 903 scattering angle. Journal of Aerosol

Science, 32, 63–72.Son, J. Y., Kang, & Jeon, (1991). Characteristics of optical single particle sizing techniques. Journal of the Korean Physical Society, 24, 137–144.Thompson, B. J. (1974). Holographic particle sizing techniques. Journal of physics E: Scientific Instruments, 7, 781–788.Tiwary, A., & Colls, J. J. (2004). Measurements of atmospheric aerosol size distributions by co-located optical particle counters. Journal of Environmental Monitoring,

6, 734–739.Volten, H. et al. (1998). Laboratory measurements of angular distributions of light scattered by phytoplankton and silt. Limnology and Oceanography, 43,

1180–1197.Wang, J. C. F., & Hencken, K. R. (1986). In situ particle size measurments using a two-color laser scattering technique. Applied Optics, 25, 653–657.