Department of Anaesthetics and Intensive Care Medicine, University of Wales College of Medicine, Heath Park, Cardiff CF14 4XN, UK
Declaration of interest. The author is funded by the Medical Devices Agency. Manufacturers of some of the devices tested in this article have provided travel and subsistence expenses to visit their factories. The Department of Anaesthetics and Intensive Care Medicine, University of Wales College of Medicine, undertakes commercial testing of some of these products.
Accepted for publication; May 20, 2002
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Method. The penetration of sodium chloride particles through 12 breathing system filters was measured by two different techniques (using either forward light-scattering laser photometers or a neutral hydrogen flame photometer).
Results. The geometric means of the penetration values for the 12 filters varied from 0.0039% to 22.6% and from 0.0004% to 20.6% for the two techniques, respectively. For 10 of the 12 filters, with penetration values greater than 0.03%, the penetration values measured by, and the repeatability of, the two techniques were similar. The ratio of the penetration values measured by the two techniques (calculated from the mean difference in log10(penetration) between the two techniques for these 10 filters) was 0.93 (95% confidence interval 0.38 to 2.30). There is therefore only a small difference (at most a factor of about two either way) between the two techniques compared with the thousand-fold range in penetration values of the breathing system filters. For the remaining two filters, penetration values obtained using the flame photometer were less, and were close to or below the detection threshold of the laser photometer.
Conclusion. The neutral hydrogen flame photometer provides similar results to the forward light-scattering laser photometer technique.
Br J Anaesth 2002; 89: 5415
Keywords: equipment, heat and moisture exchange filters; measurement techniques, filtration
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
An example of test equipment suitable for measuring filtration performance is given in the standard (Model AFT 8130, TSI Inc., St Paul, USA). The filtration performance of 33 breathing system filters has been measured using this equipment.2 That study showed that penetration differed between the two types of filter (pleated hydrophobic and electrostatic), and demonstrated that penetration was less for filters with larger surface areas.2 However, it was not clear whether the penetration values obtained applied only to the particular test equipment used in the study.
Other test equipment is available for measuring the penetration of sodium chloride particles through filter media. In particular, a test rig is available (Moores Test Rig (CEN Bench Rig), SFP Services, Christchurch, UK) which was designed to measure the filtration performance of respiratory protective devices to the British and European standard, BS EN 143. This test rig may also be suitable for determining the filtration performance of breathing system filters, although the aerosol generators and the detection systems used to determine the penetration of sodium chloride particles are different in the two test rigs. New measurements on 12 filters by this alternative test rig were compared with previous measurements on the same 12 filters obtained using the TSI test equipment. This was carried out to determine whether the penetration of sodium chloride particles through filters measured by the two techniques is similar, provided the flow of air used is the same for both.
![]() |
Methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
In the present study, the penetration of sodium chloride particles through five unused samples each of 12 of the different filters used in the previous study (Table 1) was measured using a second test rig (Moores Test Rig (CEN Bench Rig), SFP Services, Christchurch, UK). The filters consisted of six pleated hydrophobic and six electrostatic filters, and covered the range of penetration values obtained during the previous study.2 This test rig was designed to measure the performance of respiratory protective devices to the British and European standard, BS EN 143.3 A Collison nebulizer generated sodium chloride particles from a 1% w/w sodium chloride solution. According to the manufacturers specification, the diameter of the particles measured using the longest diagonal varied from 0.04 to 1.2 µm, with a mass median diameter of 0.6 µm. The mass median aerodynamic diameter is 0.4 µm with a mass concentration of particles of 13 mg m3. The penetration of the particles through the filter media was measured using a neutral hydrogen flame photometer. The intensity of the flame is proportional to the mass of sodium in the sample. The flame was viewed through a narrow-band interference filter with a half-peak bandwidth of 3 nm at 589.3 nm. Neutral density filters were used to prevent the light transmission flooding the photomultiplier. The minimum detectable penetration was 0.000005%. Penetration was again measured with a flow of 30 litre min1 of air passing through the filter under test. The output from the photomultiplier was displayed on an analog meter, and the reading was recorded within 30 s of initiating flow through the filter. The penetration (%) was calculated from tables provided by the manufacturer (SFP Services), taking into account the neutral density filter that was used.
|
Log10(penetration) values were used for analysis because, with the large range in penetration values obtained, this led to more consistent variances over the range of penetration. The variances of the log10(penetration) values of each filter measured by each of the two techniques were calculated. The variances of the log10(penetration) values from the two techniques of measurement were compared using the paired t-test (StatView 5, SAS Institute Inc., SAS Campus Drive, Cary, USA).6 The null hypothesis was that there is no difference between the variances for the two techniques. If the hypothesis was correct, the comparison of the two techniques could then be made.
The mean log10(penetration) for each filter measured by each technique was calculated. The difference between the two means for each filter was plotted against the average of the two (Fig. 1). The mean difference (d) and the standard deviation (SD) of the differences between the two techniques were calculated. When repeated measurements are made, the limits of agreement are wider than those given by d±1.96SD, because the averaging of the repeated measurements has removed some of the random error associated with each individual measurement.4 The corrected standard deviation of the differences, SC, is (SD2+
SA2+
SB2), where SA and SB are the standard deviations for the two different methods A and B.4 SA and SB were taken to be the square root of the mean of the variances of log10(penetration) for each method. The limits of agreement are then d±1.96SC. This gives the limits of agreement for a pair of single measurements on a given filter by the two techniques.
|
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
The plot of difference against mean of log10(penetration) for the two techniques showed that there were two obvious outliers, with the penetration values measured by the SFP test rig less than those measured by the TSI equipment (Fig. 1). All 20 measurements of penetration for these two filters by the two techniques were less than 0.03%, giving a filtration efficiency of greater than 99.97% for both filters. A filter with a filtration efficiency greater than 99.97% is the highest category according to the National Institute for Occupational Safety and Health (NIOSH) standard for respiratory protective devices.7 Both techniques placed the two filters into this category, but the difference between the techniques for these two filters skewed the analysis of the comparison. The penetration values were also close to the limit of sensitivity of the TSI test equipment (0.001%). These filters were therefore removed from the following analysis, and the calculations on the difference between the two techniques were made on the results from the remaining 10 filters.
The two methods of measurement were compared using the method reported by Bland and Altman,4 described briefly above. The mean difference in log10(penetration) between the two techniques was 0.0294 (Table 3). The standard deviation of the differences between the means of log10(penetration) values measured by the two techniques (SD) was 0.162. The standard deviation from the SFP rig, SA, was 0.117 and from the TSI rig, SB, was 0.201, giving the corrected standard deviation from which to calculate the limits of agreement, SC, as 0.20. The limits of agreement were therefore 0.42 and 0.36. The antilog of the mean difference was 0.93, and antilogs of the limits of agreement gave 0.38 and 2.30. Therefore, on average, the SFP technique gave penetration values 0.93 times that of the TSI technique, although the ratio between a pair of single measurements by the two methods on any given filter may be as little as 0.4 or as much as 2.3.
The pooled SD was 0.165, giving an SE of 0.074 from which to calculate the 95% CI associated with the difference between and the mean of the log10(penetration) values from the two techniques.
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The TSI equipment measures the light scattered by any particle in the air. Particles released by the breathing system filter itself could be counted as penetration particles. Hygroscopic salts added to the heat and moisture exchanger layer in the filter may be released during the test, causing an apparent, but incorrect, increase in the penetration value. However, the flame photometer only measures particles containing sodium, and, as there was good agreement between the two techniques, the error associated with the release of particles from the filters must be small.
The penetration of particles though a filter depends on, amongst other factors, the size of the particles in the challenge. The size distribution of particles used to challenge the filter is specified in the standard to have a CMD of 0.075 (SD 0.020) µm and a geometric SD not exceeding 1.86. Particles with different diameters penetrate through filter media with varying efficiencies. Particles that are the most penetrating have a diameter in the range 0.050.5 µm.8 The most penetrating particle size depends on the face velocity (volume flow rate of air per unit area of filter media), the diameter of the fibre and the packing density of the fibres.8 Therefore, the size of the most penetrating particle will vary for different filters.
The TSI equipment complies with the requirement of the standard, generating an aerosol of sodium chloride particles with a CMD of 0.07 µm and a geometric SD not exceeding 1.83. SFP claims that its test equipment generates an aerosol of sodium chloride particles with a range of particle size of 0.041.2 µm when measured using the longest diagonal, and a mass median aerodynamic diameter of 0.4 µm. However, an aerosol of particles with a CMD of 0.07 µm and a geometric SD of 1.83 has a mass median aerodynamic diameter of 0.34 µm (see Appendix). Therefore, from theory, the two techniques should provide nearly the same results if the total challenge is similar (product of mass concentration and time). In the present study this was so, over the range of penetration values 0.0325%.
![]() |
Acknowledgements |
---|
![]() |
Appendix |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
lng = {
ni (ln di ln dg)2/(N 1)}0.5
where di is the diameter of the ith particle, dg is the geometric mean diameter (which equals CMD for a lognormal distribution), ni is the number of particles in the ith group, and N is the total number of particles. For the lognormal distribution, 95% of the particles lie between CMD÷g2 and CMDx
g2.
The mass median diameter (MMD)
The MMD is the diameter for which 50% of the total mass of the particles is contributed by particles with a diameter larger and 50% by particles with a diameter smaller than this value. The MMD can be obtained from the CMD from the following HatchChoate equation.8
MMD = CMD exp 3 (lng)2
The aerodynamic diameter
An alternative approach is to quote the diameter of the particle in aerodynamic terms, that is, an equivalent diameter so that, regardless of the density or shape of the particle, all particles with the same aerodynamic diameter will behave in the same way, for example, by having the same terminal velocity under gravity in air.
The aerodynamic diameter, da, of a single spherical sodium chloride particle with a physical diameter, dp, is given by:8
da = dp{Cc (dp) / Cc (da)}0.5 {p /
0}0.5(1)
where 0 is the standard density (1000 kg m3),
p is the density of the particle (2165 kg m3 for sodium chloride) and Cc is the Cunningham (slip) correction factor, a factor that has to be included when the size of the particle is close to the mean free path of the air molecules. Cc is given by:8
Cc = 1 + ( /d) {2.34 + 1.05 exp(0.39 d/
)}
where d is the diameter of the particle (µm), and is the mean free path of the air molecules, which is 0.0665 µm at 20°C and 101.3 kPa.
Since the aerodynamic diameter, da, occurs on both sides of equation (1), it needs to be calculated iteratively. A shape factor has to be included for particles that are not spherical.8
Example
If CMD=0.07 µm and g=1.83, then 95% of the particles lie between 0.021 and 0.23 µm. The equivalent MMD=0.21 µm and the mass median aerodynamic diameter is 0.34 µm.
|
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
2 Wilkes AR. Measuring the filtration performance of breathing system filters using sodium chloride particles. Anaesthesia 2002; 57: 1628[ISI][Medline]
3 British Standards Institution. Specification for particle filters used in respiratory protective devices (BS EN 143:1991). Milton Keynes, UK: British Standards Institution, 1991
4 Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; i: 30710
5 Wilkes AR. Factors affecting the filtration performance of breathing system filters. Br J Anaesth 2000; 84: 280P
6 Bland JM. Comparing within-subject variances in a study to compare two methods of measurement. http://www.sghms. ac.uk/phs/staff/jmb/jmb.htm, 22 April, 2001
7 National Institute for Occupational Safety and Health (NIOSH). Respiratory Protective Devices. Code of Federal Regulations, Title 42, Part 84. Morgantown, West Virginia, USA: National Institute for Occupational Safety and Health, 1995
8 Hinds WC. Aerosol Technology. Properties, Behavior, and Measurement of Airborne Particles, 2nd Edn. New York: John Wiley and Sons, 1999