34 エアロゾル研究（34）

Research paper Earozoru Kenkyu, 35 （1）, 34–42 （2020） doi: 10.11203/jar.35.34

Study on the Installed Filter Leakage Evaluation Criterion for Particle Counter and Aerosol Photometer

Muhammad Aiman MOHD NOR 1* and Yoshihide SUWA 2

Received 5 August 2019 Accepted 1 December 2019

Abstract　In order to examine the criterion of leak testing of installed filter systems shown in ISO 14644-3: 2005, evaluations of particle concentration at downstream of installed filter systems by particle counters and aerosol photometers were compared, assuming the same upstream concentra-tion distribution. As the result of numerical calculations, although particle counters and photometers detected different penetration rate values, a reasonably good correlation was observed between both instruments within the allowable requirements for the upstream particle concentration distribution stated in ISO 14644-3. A possibility to use the same evaluation criterion for both instruments was also studied. Based on the results obtained, it was found that the same standard leakage evaluation criterion could be applied to the filters with a performance of H14 (BS-EN 1822-1: 2009) or higher. However, for the filters with a performance of H13 and lower, it was considered unreliable to apply the same criterion. Currently, the ISO 14644-3: 2005 is too complicated, different criterion for particle counters and photometers are presented. Moreover, the criterion changes for particle counter at different filter grades used in the tested rooms. The findings obtained in this study will simplify the complexity exists in the current ISO standard.

Keywords : ISO Standards, Installed Filter Leakage Test, Discrete Particle Counter, Aerosol Photometer, Leakage Evaluation Criterion.

1 Department of Regional Environment System, Shibaura Institute of Technology

Toyosu 3-7-5, Koto-ku, Tokyo 135-8548, Japan2 Department of Mechanical Engineering, Shibaura Institute of

Technology Toyosu 3-7-5, Koto-ku, Tokyo 135-8548, Japan* Corresponding Author. E-mail: NA17101@shibaura-it.ac.jp (M.A. Mohd Nor)

determine whether there are leaks that present which can affect the cleanliness level of a cleanroom.

Currently, ISO14644 Part 3: 2005 3) “B.6 Installed filter system leakage test” and JIS B 9917: 2009 4) Cleanrooms and associated controlled environments-Part 3: Test methods “B.6 Installed filter system, leakage test” have been used as the standard leakage test method of a high-performance air filter installed in a cleanroom. These tests are not being done to determine the efficiency of the filter medium but to reliably detect defects such as pinholes that occurred due to the improper execution during the filter installation process or filter degradation over time. Since this test is needed to be carried out as an in situ operation inside a cleanroom, it must be practical and easy enough to be implemented by the tester.

In the current standard, two different leak detection methods of using two different instruments are presented in Annex B.6 of the standard: the method using aerosol photometer (B.6.1.2 Using an aerosol photometer: referred to as photometer in the present study) and the method using particle counter (denoted as DPC in ISO standard) (B.6.1.3

1.　Introduction

The High-efficiency Particulate Air (HEPA) and Ultra Low Penetration Air (ULPA) filter have been set at the supply vents as the final stage air filtration of a cleanroom to keep a clean environment. These filters were initially tested and graded by the filter manufacturer according to a particular filter test standard such as EN1822-1:2009 1) in Europe or IES-RP-CC001.5 2) in the USA based on the filter penetration performances. However, the filter performances can be affected when being set to a real cleanroom. The filters may develop leaks and defects throughout the transfer and installation processes. These filters need to be tested to

35Vol. 35　No. 1（2020） （35）

Using the discrete-particle counter (DPC): referred to as the DPC method in the present study). The acceptance criterion for the leakage test evaluation by scanning in the photometer method is when a particle downstream mass concentration shows a value higher than 0.01% of the particle upstream mass concentration, while in the DPC method, that is when a particle downstream number concentration shows a value higher than the standard penetration of the test filter multiplied with a K-factor (usually K＝10) of the particle upstream number concentration that is being detected.

Although both of these instruments using light scattering techniques as a measurement principle, both instruments yield two different types of sampling results due to the different way of detecting and sampling aerosol particles. The photometer receives light scattered from an ensemble of particles illuminated simultaneously and yields a mass concentration of all particles in the real existing particle size range. In contrast, the DPC counts particles by detecting light scattered by a single particle in succession and converted it into an electrical signal to give a measurement of particle number concentration of only at a set of specific particle size range out of the real existing particles.

Although many studies have been conducted on the differences in measurement sensitivity of photometer and DPC so far, relatively only a few studies have been conducted on the comparison between these two instruments in the context of a leakage test of an air filter and contradicting results were observed. Some researchers 5–7) observed a good correlation between photometer and DPC, while others 8,9) found a significant difference between these two instruments. We also did a theoretical comparison on the filter leak test result based on the direct comparison between the mass concentration and number concentration 10) which is the yield measurement of photometer and DPC, respectively. It was found that a significant difference from leak evaluation of these two concentration measurements. DPC produced more severe result compared to the photometer method and if the same leakage criterion is used for the filter grade lower than H14 in EN 1822-1 1), this can easily be misevaluated as a leak by the inspector even though there is no leak on the filter. On the other hand, since the filter H14 or higher grade has a relatively higher performance, it is possible to use the same criteria of 0.01% for photometer and DPC.

All of the studies above lead us to do an intensive theoretical study on the filter penetration evaluation by photometer and DPC since there are such disparities in the result of the previous studies. Furthermore, different criteria used depending on instruments create complexity in the installed filter leakage test, thus recently a request towards unifying the criteria was also suggested. Moreover, the

instrument responses were not considered in our previous study. Therefore, a further investigation is done to describe more accurately on the different leak test evaluation by photometer and DPC method in the present study.

2.　Comparison of penetration evaluated by the photometer and DPC method

for installed filter leakage test

2.1　Upstream concentration particle size distributionIn ISO 14644 Part 3: 2005 3), the requirement of the

upstream aerosol of photometer is presented as “the Mass Median Particle Diameter (MMD) for this production method will be between 0.5 μm to 0.7 μm with a Geometric Standard Deviation (GSD) of up to 1.7”, while for DPC, “the Count Median Diameter (CMD) for this production will typically be between 0.1 μm to 0.5 μm with a GSD of up to 1.7”. In this study, in order to satisfy requirements mentioned, the particle number distribution of poly-alpha olefin (PAO) particles generated from the Laskin nozzle type polydisperse particle generator plotted against the logarithmic particle diameter as shown in Fig. 1 11) is used. The particle number distribution in Fig. 1 shows a curve having the form of log-normal probability density distribution with a GSD of 1.7, MMD of 0.53 μm and CMD of 0.23 μm express as below

log logexp .

π log logp

pD CMD

f DGSD GSD

2

2

12 2

(1)

Eq. (1) is the normalized form of the particle number distribution in Fig. 1 by the total number of particles NT. The particle number distribution CN (Dp) in Fig. 1 can be expressed in term of probability density function as in Eq. (1) as follows

Fig. 1 Particle number distribution as a function of particle diameter in log plot of PAO (Poly-alpha olefin) particles generated from Laskin nozzle type polydisperse aerosol generator.

36 エアロゾル研究（36）

log logd exp .d log π log log

pTN p

p

D CMDN NC DD GSD GSD

2

22 2

(2)

The particle number distribution in Fig. 1 is used throughout this study as the primary upstream concentration particle size distribution. In addition to data in Fig. 1, a parametric study is also done by varying the value of GSD and MMD (by changing CMD) to produce different upstream particle distributions following the log-normal probability distribution function as in Eq. (1).2.2　Filter Penetration Rate

Filters considered were equivalent to H13, H14, U15 and U16 of the European standard EN 1822: 2009 1) (These correspond to ISO 35, ISO 45, ISO 55 and ISO 65 by the classification of ISO 29463-11 2), respectively.). The penetration as a function of particle diameter for various filter standards including the filter mentioned above is shown in ref 13). Fig. 2 shows the standard penetration for each particle diameter plotted for filter grade H13 and H14 in a linear and log-log plot. The penetration curve represented by a linear plot shows a steep curvature change and it is difficult to use a simple function for a data fitting. However, by converting it to a log-log plot, the data fitting process can be easily done using a quadratic function expressed in the summation notation as follows

lnexpn

PP n

n

DP D B bB

2

0 (3)

where, B is the constant and bn is the coefficient of the quadratic equation at nth degree.

Data retrieved from ref. 13) are in the range of 0.05 μm to 0.4 μm in particle diameter. In order to compare the characteristics between the photometer method and the DPC method aimed in this study, data on the large particle diameter side is insufficient. Therefore, the standard penetration data is plotted in a log-log plot and then a quadratic function is used as a fitting line as described by Eq. (3) to approximately extending the particle size range from 0.01 μm to 10 μm considered in the numerical integration. The penetration curve from the original data is not a symmetrical curve. Two quadratic curves are used to extrapolate the original data at the small particle size (0.01 μm to 0.05 μm) and large particle side (0.4 μm to 10 μm). The coefficients used for the extrapolation lines are tabulated in Table 1.

As can be seen from Fig. 2, the standard penetration of particles with a particle diameter of 0.05 μm or less and particle sizes of 2 μm or more are significantly decreased. When comparing the area under the penetration curve between the extrapolated particle range and the whole integration range (0.01 μm to 10 μm), the effect of small particle size is less than 1% and large particle size is less than 2%. These values indicate that even in the most extreme

Fig. 2 Standard penetration as a function of particle size distribution of H13 and H14 filter in (a) linear plot and (b) in a log-log plot.

Table 1　Coefficient of the fitting line used in extrapolation of the filter penetration curve

CoefficientH13 H14

0.01–0.05 μm 0.4–10 μm 0.01–0.05 μm 0.4–10 μmB 2.509 2.632 2.775 2.502

b0 －9.540 －7.049 －9.456 －8.337

b1 －13.098 －11.408 －14.157 －11.408

b2 －7.805 －7.819 －8.769 －7.431

37Vol. 35　No. 1（2020） （37）

case of 100% margin of error when absolutely no data representing small particles and large particles considered in the penetration curve function relatively small margin of error can be achieved. However, since the original data is smoothly extrapolated at both ends, a much smaller margin of error can be expected.2.3 Downstream concentration particle size distribution

Using the data in Figs. 1 and 2, the downstream particle concentration for each particle size was calculated as follows

d p u p pC D C D P D (4)

where, Cu (Dp) is the concentration of particles at the filter upstream side, Cd (Dp) is the concentration of particles at the filter downstream and P (Dp) is the penetration of a filter. These concentrations may be in the form of number or mass concentration seen by a particular measuring instrument. For example, substituting particle number distribution defined in Eq. (2) into Eq. (4), concentration of particles at the downstream of a filter media is CN,d (Dp)＝CN,u (Dp)×P (Dp)＝NT f(Dp)×P (Dp). This expression is an example of a direct calculation of the downstream concentration without considering any instrument response. Details on the reproduction of measurements by a photometer and DPC method considering the instrument responses are further explained in the next section.2.4 Photometer measurement and penetration rate eval-

uation methodA photometer receives light scattered by a number of

particles, simultaneously illuminated in an optical sensing volume, and provides a value of mass concentration. This measurement depends on the incident light, the geometry of the photo-detecting optical system and the aerosol physical parameters. Given the Mie theory of particle light-scattering, when particles with refractive index of m are illuminated by an unpolarised incident beam with a wavelength λ and intensity of I0, the intensity of light scattered collected by the collecting lens with a collecting aperture of semi-angle β and an axis of the collecting aperture inclined at an angle ϕ to the illuminating beam is given by,

, , , , , .πPHO p p N p pI I i D m i D m C D d dD

2

0 1 22 08

(5)

Where, CN (Dp) is the particle number distribution as in Eq. (2), i1 (Dp, θ, m) and i2 (Dp, θ, m) is the vertical and horizontal intensity functions of scattering light and ω (θ, ϕ) is the fraction of scattered light at angle θ collected by the collecting lens. Expressing Eq. (5) in term of a probability density function,

, , , , , .πPHO PHO p p p pI N I i D m i D m f D d dD

2

0 1 22 08

(6)

where, NPHO is the total number of particles per unit volume inside the optical sensing volume Vs of photometer and f (Dp) is the probability density function as in Eq. (1).

When particle density ρp is known, NPHO can be expressed in term of mass concentration M of particles in the optical sensing volume Vs as follow,

.π

sPHO

p p p p

MVND f D dD

306

(7)

Substituting Eq. (7) into Eq. (6), for a monodisperse particle distribution,

, , , , , , .PHO mono p pp p

MI K i D n i D n dD

1 23

1

(8)

where, K (＝3I0Vs/4π3) is the constant on the photometer performance. From Eq. (8) the theoretical response of photometer can also be calculated for a polydisperse particle distribution

, , .PHO poly PHO mono p p pp

I I D f D dDD

3

03

1 (9)

For the present study, the response of HUND TM digital μP (H. Hund, GmbH, Wetzlar, F.R.G) is theoretically calculated for the present investigation. The detailed specification of this photometer is given in Table 2. Gorner et al. 14) studied HUND TM digital μP and presented the theoretical way of calculating the response of photometer by combining the theoretical equations explained previously with the single experimental measurement of an arbitrary particle. They introduced the calibration index k/CF which allows one to calculate the photometer response for any other particle with known physical parameters. The coefficient k is the ratio of transformation between scattered light intensity IPHO,poly obtained by the calculation and the same intensity expressed in the voltage output (mV) given by the photometer in the experiment. The conversion factor CF is a value given by the factory setting which is an analogic-numeric conversion of the electrical signal to the numerical display. The ratio of the calibration constant k and CF yields a calibration index k/CF of the photometer, relating the intensity of scattered light to the mass concentration mg/m3. Therefore, using the calibration index, the concentration measured by the photometer for a range of monodisperse particles is

, , .CFPHO mono p PHO monokC D I (10)

Table 2　Optical parameters of HUND TM digital μP

Wavelength of illumination light beam, λ 950 nmAngle of detection of scattered light, φ 70°Semi-angle of light collection, β 10°

38 エアロゾル研究（38）

Solving Eq. (10), the photometer response of PAO particles was calculated as shown in Fig. 3. The decline in the response with increasing particle diameter above about 1 μm is common to all photometers 15,16). Complex interactions between the incident light and particle results in a wavy response curve. Then, by the integration of the product of CPHO,mono (Dp) over the particle distribution as in Eq. (1), the photometric concentration corresponding to a polydisperse particle is simply obtained as follows

, , .CFPHO poly p PHO mono p p p

p

kC D I D f D dDD

3

03

1 (11)

The penetration measured by the photometer for a polydisperse particle distribution is calculated by dividing the total fluxes scattered by particles at the downstream and upstream of a filter using Eq. (11). This will automatically cancel out the shape factor and constant instrument factor (calibration index k/CF) of the photometer. The penetration evaluated by a photometer is

, , , ,/

, , , , ,.

, , , , ,

PHO PHO poly d PHO poly u

p p p p p

p p p p

P C C

i D m i D m f D P D d dD

i D m i D m f D d dD

1 20

1 20

(12)

2.5 DPC measurement and penetration rate evaluation method

As mentioned in the introduction, DPC samples particles differently from photometer by counting particles one by one through a relatively small optical sensing cell. The scattered light intensity is detected and converted into an electrical signal by a photodetector. The electrical signal is then amplified and recorded, allowing particle counting and size estimation yielding the particle number concentration per interval of particle diameter, which is called the channel. The response of a DPC can be evaluated by each channel counting efficiency. An ideal DPC would have a counting efficiency function same as a step-function which represents a 100% probability for larger particle size and 0% for smaller particle size than a specified particle size as illustrated as the dashed line in Fig. 4(a). This line will represent as boundaries for channels in a DPC. For this ideal DPC, a channel will produce a similar shape to a Dirac delta function represented in Fig. 4(b).

In contrast, a real DPC will have a particular size resolution due to effects such as non-uniform light intensity in the sensing area where the sample air passes through the illuminating flux, variations in the sensitivity of the surface of a photodetector and electrical noise. These factors will underestimate and overestimate particle size, producing not a step function counting efficiency but rather a curve with a slope. The solid line in Fig. 4(a) and 4(b) shows the counting efficiency function and resulted in a channel response of the PMS LAS-X particle counter by the Particle Measuring

Fig. 3 HUND TM digital μP photometer response CPHO,mono as a function of the particle diameter calculated for monodisperse particles of PAO.

Fig. 4 (a) Counting efficiency curves and (b) channel responses of DPCs resulted from two counting efficiency curves as the lower and upper size limit boundaries. The dashed and solid lines represent an ideal particle counter and PMS LAS-X particle counter, respectively.

39Vol. 35　No. 1（2020） （39）

Systems used in this study. This model has been discussed and studied extensively 17,18). PMS LAS-X is capable of particle sizing in the diameter range from 0.1 μm to 7.5 μm which the size range is divided into 16 channels. A sigmoidal function is found to be the best fit for the counting efficiency function,

,p bineff p

D DD A erf

W

1

2 (13)

where the function erf is the error function, A is the half of maximum and minimum value and W is the measure of the distribution width producing different size resolution in the counting efficiency curve at a particular channel boundary size Dbin. Assuming that a DPC with q number of channels counts a discrete set of measurement in each channel, the total number particle detected by DPC in a given sample is

, , , , ,...,q

upper lowerDPC N p i eff p i eff p

iC C D D D i n

01 2

(14)

with , ,( ) ( )upper loweri eff p i eff pD D is the difference between the up-

per and lower counting efficiency boundary corresponding to the counting efficiency of an ith channel. Therefore, the pene-tration of the filter as evaluated by the DPC is then expressed as

, ,

, ,

, ,

/DPC DPC d DPC u

qupper lower

p p i eff p i eff pi

qupper lower

p i eff p i eff pi

P C C

f D P D D D

f D D D

0

0

(15)

2.6　Numerical experiments calculationEqs. (12) and (15) can be solved theoretically, given

empirical expressions for the filter penetration (Fig. 2), DPC

response (Fig. 4), photometer response (Fig. 3) and aerosol distribution functions (Fig. 1). It is more expedient, however, to use numerical integration on a computer, and for this, we calculated the integrals over particle size distributions. As for DPC, we did an integration for particle size between 0.3 μm to 10 μm. The minimum integration limit of 0.3 μm was selected because Suzuki et al. 19) found that 0.3 μm was the most suitable particle size to determine pinhole leaks on test filter and 0.3 μm is also widely available in the DPC channel specification. As for photometer, we did the integration between 0.01 μm to 10 μm. In a preliminary study, we found that the ratio between DPC and photometer method converged at a particle size equal to 5.8 μm. Integration at where the maximum limit more than of this particle size will give the same value, regardless of any particle size. For comparison, the number and mass concentration evaluation method without considering instrument responses conducted in the previous study 10) are presented together with the with results of the present study.

3.　Result and discussion

Fig. 5 shows the upstream and downstream particle concentration distribution of the H13 filter (standard penetration at MPPS: Most penetrating particle size is 0.05%) evaluated by the number concentration and DPC (Eq. (13)). Similarly, Fig. 6 shows the upstream and downstream particle concentration distribution evaluated by mass concentration and photometer (Eq. (11)) methods. The penetration for the number concentration and DPC methods in Fig. 5 are calculated by comparing cumulative for particles larger than 0.3 μm shown by the area of the shaded regions. However, for the mass concentration and photometer methods, the integration across the whole particle range are considered in

Fig. 5 Particle concentration distribution at the upstream and downstream side of an H13 filter (Standard penetration at MPPS: 0.005%) as a function of particle diameter evaluated by (a) direct Number concentration method and (b) number concen-tration measured by PMS LAS-X particle counter. The penetration for both methods is calculated by comparing the area of the shaded regions for particle size ≥ 0.3 μm at the upstream and downstream of a filter.

40 エアロゾル研究（40）

the penetration calculation as illustrated by the shaded regions in Fig. 6. These figures depict the case for a filter barely satisfies a standard performance with exactly no leakage such as pinhole or tear. If there are any leakages, the downstream concentration curve shown in the graph will indicate a larger value than presented in the figures.

As can be seen from the figures, all evaluation methods show a decreasing trend for larger particle size after peaking at certain particle size. In the case of filter upstream and downstream distribution, mass concentration and photometer evaluation method peaks at a larger particle size compared to the number concentration and DPC evaluation method. Furthermore, mass concentration and photometer method have a significantly lower gradient for small and large particles with photometer having a higher value at the peak around particle size 0.5 μm and a steeper gradient for particle larger and smaller than peak particle size compared to the mass concentration method. Photometer response decreases for particles smaller and larger than particle size around 0.5 μm which gives a less weighted value compared to mass concentration evaluation method. Since the penetration is the ratio between the area under the curves shown as two different hatched regions in each figure, the evaluation of the number concentration and DPC methods are expected to be more severe than the mass and photometer methods.

Table 3 shows the penetration for each filter from H13 to U16. As expected Number concentration and DPC method produced larger values but within the same order compared to mass concentration and photometer method. DPC method has relatively the same value as the number concentration method. The difference between these two measurements is caused by the resolution of the counting efficiency curve of a particle counter as shown in Fig. 4. The photometer method has a penetration value of approximately half of the mass concentration method. This is due to the different weighted value applied to particle correspond to the photometer response curve (Fig. 3).

In the current ISO 14644-3 3), an evaluation based on the detection of downstream particle concentration exceeding 10 times the standard penetration of the test filter is currently used. This is based on the experimental result of Suzuki et al. 19) which became the basis of this standard. Suzuki et al. 19) found that pinhole leakage could not be determined clearly unless a leakage evaluation criterion about ten times of the standard penetration is being used. Therefore, based on the calculated penetration, the leakage criterion of 0.01% for H13 or higher performance filter perfectly matched to this condition. However, in order to clearly assign the unification of leakage evaluation criterion further investigation across allowable GSD, MMD and CMD range presented in ISO

Fig. 6 Particle concentration distribution at the upstream and downstream side of an H13 filter (Standard penetration at MPPS: 0.005%) as a function of particle diameter evaluated by (a) mass concentration and (b) photometer (HUND TM digital μP). The penetration for both methods is calculated by comparing the area of the shaded regions for whole existing parti-cle size range considered (0.01 μm to 10 μm) at the upstream and downstream of a filter.

Table 3 Penetration rate of each filter evaluated by the number concentration, DPC, mass concentration and photometer method

Standard penetration of filtersPenetration rate

Number Conc. DPC Mass Conc. PhotometerH13 (0.05% at MPPS) 2.86×10－5 2.90×10－5 2.03×10－5 1.07×10－5

H14 (0.005%) 2.86×10－6 2.90×10－6 2.03×10－6 1.07×10－6

U15 (0.0005%) 2.86×10－7 2.90×10－7 2.03×10－7 1.07×10－7

U16 (0.00005%) 2.86×10－8 2.90×10－8 2.03×10－8 1.07×10－8

41Vol. 35　No. 1（2020） （41）

14644-3 3) were done.Fig. 7 illustrates the penetration seen by the instruments

for different polydisperse aerosols, all with GSD＝1.7 for H13 filter evaluated by Eqs. (12) and (15) for the photometer and DPC method, respectively. Both DPC and number concentration method measurements produced relatively flatter curve compared to mass concentration and photometer methods. DPC and number concentration methods agree well with each other which is expected. Photometer penetration curve resembles that of a mass concentration measurement method but detects maximum penetration at a much smaller MMD. As explained before, both methods give more weight to the larger particles and see the aerosols as having a larger size than number concentration and DPC methods. However, since photometer utilizes light scattering to observe particles, the particle size dependence weight is different from a normal mass concentration method due to the response function as illustrated in Fig. 3.

A striking feature is that starting from particular particle size, the DPC method seems to agree with photometer method for larger MMD and CMD. Photometer produced penetration of the same order to DPC starting around MMD＝0.2 μm (CMD＝0.1 μm) and DPC method intersects with photometer method at CMD approximate to MPPS. The shaded region represents the allowable MMD range (0.5 μm to 0.7 μm) for photometer method as presented in ISO 14622-3 3). In this region, DPC and photometer methods have the same order in penetration value but the difference increases as the MMD increases. For the allowable CMD range (0.1 μm to 0.5 μm) of a DPC method, the values of penetration are very close to the leakage evaluation rate criterion (0.01%) represented by the dashed line which can be ambiguous if the same criterion

shall be used for both photometer and DPC methods.Fig. 8 summarizes the results for log-normal aerosols

concentration distribution having standard deviations in the range of 1.05–1.7 for an H13 filter. Again, the ratios of the DPC to photometer method penetration values are indicated as a function of MMD. Predicted photometer method values ranging from about 0.3 times larger than DPC method for smaller MMD and as much as 3.3 times smaller for larger MMD. As the aerosol comes closer to being monodisperse, the DPC and photometer method penetration ratio approaches unity regardless of the aerosol size. This convergence corresponds to the first limiting case for current calculation; when regardless of any measurement methods only concentration for corresponding particle size is calculated. For any given GSD, there is also an aerosol of a certain size that produces a ratio of unity. Such a family of curves exists having different penetration function. Although DPC and photometer method differed in values but correlated reasonably well within the allowable MMD range, which is unexpected since both yield a different unit of measurement. However, since we made a comparison based on the cumulative calculation across particle size range, the difference that was thought significant when the previously compared result between DPC penetration value at a particular size to a single photometer penetration value for the whole existing particle size range was significantly diminished.

Given all the results discussed above, although the H13 filter penetration result obtained by DPC and photometer method are smaller than 0.01%, however, within the allowable CMD range, the DPC method produced penetration that is too close to the standard leakage evaluation criterion of 0.01%. There is a high possibility of misjudgment of leakage during an inspection, even when there are no leaks on the

Fig. 8 Comparison of filter penetrations detected by the DPC and photometer for eight lognormal challenge aerosols (standard deviations are indicated for each curve) for H13 filter. The dashed line indicates the unity correlation between DPC and photometer.

Fig. 7 Theoretical penetration calculation by three mea-surement methods for poly-disperse aerosol for H13 filter.

42 エアロゾル研究（42）

test filter. For this reason, it is considered unreliable to apply for the same standard leakage evaluation criteria of 0.01% as the photometer method to the filter with a performance of H13 or less for the DPC method. On the other hand, since the installed filter leakage test is not done to evaluate the performance of the filter itself, leakage evaluation criterion of 0.01% is considered applicable to the filters with H14 or higher performance (U15 and U16).

4.　Conclusion

The responses of the aerosol photometer and DPC were considered to investigate theoretically the leakage evaluation of air filters. The theoretical approach involves calculating the filter penetration in which an instrument measures by numerical integrations of aerosol size distributions, filter penetration, photometer and DPC response functions. Based on the calculation, the DPC produces a penetration value that gives equal importance on each set of particle range, regardless of its size. On the other hand, the photometer, which emphasizes more on the larger particles in a distribution, yields a value more closely associated with the particle mass concentration penetrating the filter. The differences between these two instruments are further dependent on the MMD or CMD of the challenge aerosol distribution. When test aerosols have a CMD greater than MPPS for the filter, the DPC can be expected to have higher penetration values than photometer. In contrast, when aerosols have a CMD less than the MPPS, the photometer shows higher penetration values than the DPC.

The possibility of a unified criterion for a leakage test of an installed filter system by considering the characteristics of DPC and photometer was studied through the trends and values of the calculated penetration. As a result, it was found that for a filter having H14 or higher performance in European standard EN 1822-1 1), the same leakage evaluation standard 0.01% as the photometer method can be applied in the DPC method. On the other hand, it was found that it was not possible to apply the uniform criterion for filters with H13 or lower performance, therefore, it was necessary to establish criteria for each filter grade instead.

References

1) European Committee for Standardization. BS-EN 1822-1: 2009: High Efficiency Air Filters (EPA, HEPA and ULPA). Part 1; Classification, Performance Testing, Marking, Brussels, Belgium: CEN

2) Institute of Environmental Sciences and Technology. IEST-RP-CC001.5: HEPA and ULPA Filters. Arlington Heights, IL: IEST; October 2009

3) International Standards Organization. ISO 14644-3: Clean-

rooms and Associated Controlled Environments – Part 3: Test Methods. Geneva, Switzerland: ISO; 15 December 2005

4) Japanese Industrial Standard. JIS B 9917-3: Cleanrooms and Associated Controlled Environments – Part 3: Test Methods. Tokyo, Japan: JIS; 20 March 2009

5) Gail, L. and Ripplinger, F.: Correlation of Alternative Aerosols and Test Methods for HEPA Filter Leak Testing, Proc. of Annu-al Technical Meeting of the Institute of Environmental Sciences and Technology, 369–376 (1998)

6) Ward, S.: Photometer vs LSAPC APRIL 2018 SCW Rev1 Final, ISO/TC 209/WG3 N374, 2018

7) Hargreaves, J. and and Thaveau, B.: French Annex on Compar-ative Leak Testing Results, ISO/TC 209/WG3 N370, 2018

8) Kase, T., Suzuki, Y., Hitoshi, E., Nobuyoshi, O., Takeshi, S. and Satoki, S.: A Study on Testing of HEPA Filters, Kuki Seijo to Kontamineshyon Kontororu Kenkyu Taikai Yokoshu, 9, 221–224 (1990) (in Japanese)

9) Greiner, J.: HEPA Filter Leak Testing Using the Particle Counter Scan Method, CleanRooms, 4, 36–39 (1990)

10) Mohd Nor, M. A. B. and Suwa, Y.: Study on the Possibility of a Unified Criterion for a Leakage Test of an Installed Filter System, J. of Air Cleaning Assoc., 56, 29–34 (2019)

11) Reference material of the 5th. Kanomax International Aerosol Workshop

12) International Standards Organization. ISO 29463-1: High-efficiency Filters and Filter Media for Removing Particles in Air – Part 1: Classification, Performance Testing and Mark-ing. Geneva, Switzerland: ISO; 7 October 2011

13) Vokes Air Hepatex CR: Setting The Quality Standards, http://www.filterpak.fi/images/tuotetiedot/hepa&ulpasuodattimet/Hepatex_CR_EN.pdf

14) Görner, P., Bemer, D. and Fabries, J. F.: Photometer Measure-ment of Polydisperse Aerosols, J. Aerosol Sci., 26, 1281–1302 (1995)

15) Armbruster, L.: A New Generation of Light-Scattering Instru-ments for Respirable Dust Measurement, Ann. Occ Hyg., 31, 181–193 (1987)

16) Armbruster, L., Breuer, H., Gebhart, J. and Neulinger, G.: Pho-tometric Determination of Respirable Dust Concentration with-out Elutriation of Coarse Particles, Particle & Particle Systems Characterization, 1, 96–101 (1984)

17) Szymanski, W. W. and Liu, B. Y.: On the Sizing Accuracy of Laser Optical Particle Counters, Particle & Particle Systems Characterization, 3, 1–7 (1986)

18) Hinds, W. C. and Kraske, G.: Performance of PMS Model LAS-X Optical Particle Counter, J. Aerosol Sci., 17, 67–72 (1986)

19) Suzuki, K., Suwa, K., Okui, K. and Yanagimoto, A.: A Proposal of Simplified Procedure for In-Situ Leakage Test of Installed Filter System, J. Japan Air Cleaning Assoc., 49, 35–44 (2012) (in Japanese)