- Email: [email protected]
Weighing imprecision and handleability of the sampling cassettes of the IOM sampler for inhalable dust
PII: S0003-4878(00)00052-1
Ann. occup. Hyg., Vol. 45, No. 3, pp. 241–252, 2001 2001 British Occupational Hygiene Society Published by Elsevier Science Ltd. All rights reserved Printed in Great Britain. 0003–4878/01/$20.00
Weighing Imprecision and Handleability of the Sampling Cassettes of the IOM Sampler for Inhalable Dust ¨ RAN LIDE´N* and GUNNEL BERGMAN GO National Institute for Working Life, S-112 79 Stockholm, Sweden
The weight stability of the sampling cassette of the IOM sampler for inhalable dust was tested in several weighing experiments. The results show that the reliability of repeated weighings was good, but the absorption of water vapour was slow and varied considerably among cassette specimen. The exponential time constant for water absorption was approximately 4 days, and 15–20 days were needed to obtain weight stability. With the help of cassette blanks the imprecision in dust weight could be held below 0.05 mg, if the cassettes were allowed one week’s storage in the weighing room before weighing, both before and after sampling. The IOM sampling cassettes seem to consist of a few subsets, each with identical relative weight increase in a weighing room. To keep the variability low it is important that both the blanks and the cassettes used for sampling come from the same subset. Experiments indicate that the conducting plastic of the IOM sampling cassette may be replaced with another kind of plastic with similar electrical conductivity, but whose humidity absorption is 30 times lower. A lid, which is weighed with the cassette, was designed so that the potential dust loss from the cassette proper to the commercial transport clip was eliminated. A flow adapter, which simplifies the measurement of the air flow during personal sampling, was designed. 2001 British Occupational Hygiene Society. Published by Elsevier Science Ltd. All rights reserved Keywords: dust sampling; inhalable dust; sampling cassette; weighing; reproducibility; IOM sampler; GSP sampler
INTRODUCTION
There are moves in many countries to specify inhalable dust sampling according to the internationallyagreed standards (International Organisation for Standardization, 1995; Comite´ Europe´en de Normalisation, 1993; American Conference of Governmental Industrial Hygienists, 1992). The Institute of Occupational Medicine (IOM) sampler is presently the most commonly used sampler for inhalable dust (Mark and Vincent, 1986). The outer part of the sampler consists of a cylinder with a diameter of 苲30 mm and a height of 苲15 mm. Inside the cylinder a sampling cassette is mounted which holds a 25 mm filter. The filter is placed at the bottom of the cassette, and upstream of the filter the cassette consists of a tube with a diameter of 15 mm and a length of 10 mm.
Received 16 February 2000; in final form 7 July 2000. *Author to whom correspondence should be addressed. Tel.: +46-8-730-9100; fax: +46-8-828-678; e-mail: [email protected]
The sampling cassette is thin-walled, 苲0.5 mm. When the cassette is mounted inside the sampler, its tube protrudes 1–2 mm outside the base of the cylindrical sampler. During sampling the sampler is mounted on the worker’s chest, and is intended to face horizontally outwards. The deposit consists of all dust which passes the inlet plane of the tube of the cassette, whether it is deposited onto the filter or the walls of the cassette. Therefore, the entire cassette must be weighed, and not only the filter. The IOM sampler has been produced commercially in several versions. Originally, Negretti Ltd produced a version in aluminium, which is however, no longer commercially available. The present manufacturer, SKC Inc. [Eighty-Four (PA), USA], has produced different versions, the first in a brown plastic which had a non-conducting surface, and the present version, in a black plastic which is electrically conductive. In addition, the sampling cassette itself is also available in stainless steel. The plastic sampling cassette of the IOM sampler weighs 苲0.8 g and the stainless steel version 3.8 g.
241
242
G. Lide´ n and G. Bergman
This is a considerable tare weight compared with typical filter weights, for example the Millipore [Bedford (MA), USA] 25 mm AAWP (0.8 µm pore size), which weighs 苲20 mg. IOM samplers ordered by occupational hygienists from our laboratory are not delivered with cassettes mounted in the samplers. During transport the cassettes are stored in a separate box, with SKC’s yellow transportation clip mounted in order to protect the inner surface of the cassette. After sampling, the clip prevents any sampled dust from escaping from the cassette. When sampling cassettes are sent from our laboratory to occupational hygienists for sampling at workplaces, previously two (presently three) blanks (cassettes through which no air shall be drawn) per 25 cassettes are included. The unexposed cassette blanks are used to correct for changes in temperature and relative humidity in the weighing room between the weighing before and after dust sampling. The dust load of each cassette is determined as the weight difference of the cassette after and before sampling, corrected for the average weight difference of the blanks at the two weighings. Prior to this investigation the cassettes were usually weighed after one or two days storage in the weighing room. After analysis at our laboratory, the sampling cassettes are washed in an ultrasonic bath, first with water mixed with a mild detergent. They are then rinsed in deionized water, and finally rinsed with a water–alcohol solution. The weight stability of the sampling cassettes with mounted filters, or rather the weighing imprecision, for aluminium sampling cassettes has previously been studied by Vaughan et al. (1989) and Mark (1990). Recently, Smith et al. (1998) and Li and Lundgren (1999) reported that a considerable mass of water absorbs to the plastic sampling cassette of the IOM sampler. Smith et al. (1998) found that 苲1.5 mg was absorbed by the plastic cassette when the relative humidity was increased from zero to 苲60%, and Li and Lundgren (1999) found that the cassette loses 苲2.5 mg when the relative humidity was decreased from 苲75% to zero. These results demonstrate the necessity of defining the requirements for obtaining a low imprecision in the weighing of IOM plastic sampling cassettes. A plastic with electrical conductivity is generally obtained by mixing the plastic proper with e.g. carbon black. It is the capacity of carbon black to absorb humidity which makes the plastic electrically conducting. Therefore, one cannot obtain an electrically conducting plastic which does not absorb water. However, the trade-off between electrical conductivity and absorption of humidity depends on the properties of the carbon black used. This report presents the results of weighing experiments on the two versions of the plastic sampling cassette for the IOM sampler and one experiment on the plastic sampling cassette for the Conical Inhalable Sampler (JS Holdings Ltd, Stevenage, UK), a British
plastic version of the German GSP sampler (Kenny et al., 1997). Additionally, practical experience gained and proposals to increase the handleability based on several years of workplace sampling with the IOM sampler are presented. METHODS
Several different experiments are reported in this paper. They were all performed with the same equipment, which will be described initially in this section. Cassettes The weighing experiments were mainly performed on the electrically conducting plastic (i.e. black) cassette manufactured by SKC Inc., but some experiments were also performed with the electrically insulating (i.e. brown) plastic cassette. Unless stated otherwise, empty cassettes (i.e. without any filter) were used in the experiments. The filters used were of cellulose acetate from Millipore, 25 mm AAWP (0.8 µm pore size). Several series of similar experiments are reported below, and in each series the cassettes used were chosen at random from the cassettes available. Unless stated otherwise, it was not our intention to use the same cassettes for repeated experiments. Weighing room Weighing was performed in a weighing room in which the temperature and humidity were controlled to 21.0±0.5°C and 55±2%, respectively. The humidity was designed to be as high as possible, in order to reduce the weighing bias caused by electrical charges on the weighed object, though lower than the deliquescence point for (metal) salts. (Outside the weighing room, unused cassettes were stored in ordinary laboratory rooms, for which only the temperature was kept at approximately 21°C by the general ventilation system. The humidity was neither modified nor registered.) All weighing was carried out with a Mettler– Toledo ME22 balance (Greifensee, Switzerland), with a resolution of 1 µg. At the beginning of each weighing session the balance was calibrated using a 100 mg weight supplied by the manufacturer. During the sessions, the balance was tared when necessary. In order to reduce the static electrical charge on the filter or cassette being weighed, a radioactive α-source [0.5 mCi 210Po, NRD Inc., Grand Island (NY), USA] was placed beneath the weighing pan. Unless stated otherwise, the cassettes were weighed after one or two days storage in the weighing room. Tolerance intervals For each series of filter or cassette weights presented below, the standard deviation refers to the
IOM sampler in precision and handleability
whole series and not its average. A tolerance interval for the series was also calculated. A tolerance interval is an estimate of the width of the interval, which for a normally distributed population, with confidence of 100*(1⫺2e)% contains 100*P% of the population (Hald, 1952). (This differentiates a tolerance interval from a confidence interval, which with 100*(1⫺2e)% confidence contains the estimation value of the population.) Symmetrical tolerance intervals are calculated as x¯±s, and is approximated with the following equation
冪c
⬇z1⫺e
n⫺1
冉 冊
2 1⫺P
1+
1 2n
243
taken as a measure of the effect of not using gloves when handling the IOM cassettes. Cassette weight increase The change in weight of sampling cassettes over time (⌬m = ⌬m(t)), when moved from a room with varying humidity to the weighing room with high and controlled humidity, was analysed with non-linear regression. The following model was used:
⌬m(t) = m1[1⫺exp(⫺t/t1)] + m2[1⫺exp(⫺t/t2)] (2)
(1)
where z1⫺e is the 100*(1⫺e) percentile of the normal distribution, and c21⫺P is the 100*P% percentile of the chi-squared distribution (with n⫺1 degrees of freedom). Tolerance intervals, with 95% confidence (i.e. e=0.025) for 95% of the population (i.e. P=0.95), are listed below with their half-widths, i.e. s. Reliability The reliability of weighing was investigated by weighing one empty cassette 30 times within half an hour. During this time the air temperature rose by 0.2°C and the relative humidity by 0.1%. As a comparison, a 100 mg calibration weight was weighed 30 times within three quarters of an hour. During this time the air temperature rose by 0.3°C and the relative humidity decreased by 0.9%. Handling cassettes with and without gloves Two experiments were carried out, each involving 60 cassettes being weighed on two consecutive days. During weighing the cassettes were handled with a pair of tweezers. In between weighing, 30 of the cassettes were handled with hands and fingers as would be the case in a laboratory and at a workplace. This means that the yellow transportation clip was mounted and dismounted twice. However the cassette was not mounted inside a sampler. In the first experiment gloves of the label ‘Golden Hand latex examination gloves’ (Kebo Lab AB, Stockholm, Sweden) were used, and in the second the cassettes were handled with bare hands. The same person carried out the weighing. No attempt was made to obtain a representative selection of people with varyingly greasy or dirty fingers. During weighing the relative humidity was 53.4±0.3% and 54.9±0.3% for the experiments with and without gloves, respectively. The temperature was 21.5°C and 21.4°C in the two experiments. For each experiment (gloves and bare hands, respectively) the results are presented as the weight difference among the cassettes between the two weighing days. The difference in weight change overnight between cassettes handled with bare hands and those only handled with a pair of tweezers was
where t is the time since the change in relative humidity, ⌬m(t) is either the absolute weight difference between the weight at time t and the first weight at t>0, or ⌬m(t) is the ratio of the absolute weight difference at time t and the first weight at tⱖ0, m1 and m2 are the weight changes resulting from two exponential processes with time constants t1 and t2, respectively. The calculations were carried out with the non-linear curve-fitting module of the computer program SigmaPlot [Jandel Scientific, San Rafael (CA), USA]. Several different series of experiments were performed to analyse the temporal behaviour of the plastic sampling cassettes. In Series 1, 30 IOM sampling cassettes with filters and 30 without filters, and 10 filters in glass Petri dishes, which had all been stored in another room at the laboratory for several months, were weighed in that room which happened to have a humidity and temperature of 29.6% and 22.1°C during the weighing. The cassettes and the Petri dishes with filters were then taken to the weighing room. During the following three weeks the cassettes and the filters were weighed every working day. During this period the humidity and the temperature of the weighing room were 54.6±0.5% and 21.6±0.1°C (average±standard deviation) during the time of weighing. The results were calculated as the difference in weight for each cassette or filter relative to its weight on the first day. In Series 2, 40 sampling cassettes were weighed over a period of 18 days. After one month in the dust generation laboratory 10 of these were weighed again over a period of 18 days (Series 3). During the weighings the relative humidity and temperature in Series 2 were 57.1±1.3% and 20.9±0.8°C, respectively, while in Series 3 the relative humidity and temperature were 51.9±0.8% and 20.8±0.03°C, respectively. Thus, for Series 2 the relative humidity was above the required level, and for Series 3 the relative humidity was slightly below the required level of 55±2%. In Series 4, ten new (unused and unwashed) IOM sampling cassettes and five used (and washed) IOM sampling cassettes were used and in Series 5, 15 new filter holders for the Conical Inhalable Sampler, CIS, were used. Initially, all 30 cassettes were stored out-
G. Lide´ n and G. Bergman
244
side the weighing room for two weeks. They were then moved into the weighing room, and were weighed ten times over a period of 49 days. During the weighings, the relative humidity and temperature were 54.6±0.4% and 20.8±0.1°C, respectively. All cassettes/filter holders were electrically conducting and without filters. The CIS filter holder weighs considerably more than the IOM sampling cassette, 苲3.4 g compared to 苲0.8 g. The CIS filter holder is essentially a ring with a rectangular cross section, 6.7 mm with a 42 mm outer diameter. According to the data sheets for the plastics of the conducting IOM sampling cassette and the CIS filter holder (supplied by the manufacturers of the samplers), the resistivity of the two plastics were similar, 苲1 k⍀ cm.
Weighing imprecision as a function of storage time in the weighing room In order to determine the error in weighing dust samples, 30 sampling cassettes were exposed to quartz dust in a previously described dust chamber system (Lide´ n, 1993). All sampling cassettes had different dust deposits in the range 0.18–3.17 mg. Ten additional cassettes (incl. filters) were used as blanks. Both before and after the exposure to dust, during 18+18 days, all 40 cassettes were weighed every day for the first working week, and thereafter every other working day. (The unexposed cassettes constitute Series 2 and 3 presented above.) For each day of weighing, before and after the exposure to dust, one dust weight, mdust, was calculated for each cassette, k, according to the following equation
再
mdust(k,tB,tA) = mk(tA)⫺mk(tB)⫺⌬mREF(tB,tA) ⌬mREF(tB,tA) = mREF(tA)⫺mREF(tB) (3)
where tB and tA are the time since the cassettes were stored in the weighing room before and after the exposure to dust, respectively, and mREF is the average weight of the blank sampling cassettes at the time in question (tB or tA). The dust weight obtained at times tB and tA was compared with that obtained when the cassette weights had stabilised after 18 days in the weighing room, mdust(k,⬁,⬁). For all combinations of weighing days, before and after exposure of the cassettes to dust (tB and tA), the root mean square (RMS) average cassette weight difference ⌬mdust(tB,tA) was calculated from the equation
冘
WEIGHT CHANGES—RESULTS AND DISCUSSION
Reliability The results are presented in Table 1. They show that the reliability is sufficiently good to obtain a limit of quantification equal to a dust deposit of 0.02 mg, and thence the cassette only needs to be weighed once before and once after sampling. Handling cassettes with and without gloves The results are shown in Table 2. The effect of handling the cassettes with bare hands (苲0.02 mg), is statistically significant at the 5% level, but at this level the effect of using gloves is not significant. A weight increase of 0.02 mg might be considered insignificant, but a requirement that the cassettes be handled only with gloves informs all who are handling the cassettes of the necessary precautions. A question is how the cassettes would be handled at dusty workplaces if gloves were not required. It is possible but troublesome to use gloves when weighing in the laboratory. However, the laboratory staff will not know whether the occupational hygienist handled the cassettes with gloves or bare hands at the workplace. Our laboratory always supplies gloves in the transport case with the cassettes, together with written instructions to use them. However, on more than one occasion the cassettes have been handled with bare hands during sampling. The fact that the cassettes may have been touched by bare (unclean) hands implies the possibility of a larger, but unknown, error. Variability of blanks used in conjunction with cassettes exposed to dust The weight difference of the pair of blanks belonging to 37 sets of insulating cassettes and 34 sets of conducting cassettes ordered by occupational hygienists for their dust sampling purposes over a period of six months was determined from recorded weighing data. In some cases the weight of the filter of the blank sampling cassette was also determined. The results are presented in Table 3 as the difference in weight change of the two blanks between the two weighing occasions for each pair of blanks. Both
Table 1. Results from repeated weighing (30 times), of one IOM sampling cassette within 45 min and one 100 mg calibration weight within 30 min
n
1 ⌬mdust(tB,tA)2 = [m (k,t ,t ) nk = 1 dust B A ⫺mdust(k,⬁,⬁)]2
Average (mg)
SD (µg)
Tolerance interval (µg)
806.009
2
5
100.013
1
2
(4)
where n is the number of exposed cassettes (=30). The limit of quantification (LoQ) is taken as ten times the population standard deviation.
IOM sampling cassette 100 mg calibration weight
IOM sampler in precision and handleability
245
Table 2. The weight difference of 30 IOM sampling cassettes weighed on two consecutive days, when handled with gloves and bare hands
Gloves Untouched cassettes Effect of gloves Fingers Untouched cassettes Effect of fingers
Average (µg)
SD (µg)
Tolerance interval (µg)
16 15 1 33 14 19
4 2 1a 9 2 2a
9 5 22 4
a
The standard deviation for the effects of gloves and fingers.
averages and standard deviations for the two plastic versions of the sampling cassette of the IOM sampler are 15–20 times higher than the corresponding values for the filter of the sampling cassette. The imprecision (population standard deviation) for the blank filters is better than 0.01 mg, whereas the imprecision for the sampling cassettes of the IOM sampler is considerably larger, with a pooled value for both cassette versions equal to 苲0.08 mg. Accommodation to humidity by blanks (Series 1) The data are summarised in Table 4 and for the cassettes with filters (Series 1b) are shown graphically in Fig. 1. Our results confirm the findings of Smith et al. (1998) and Li and Lundgren (1999) concerning the amount of moisture absorbed by the cassettes, and the time scale involved. This experiment shows that if an IOM sampling cassette is weighed after a different number of days of acclimatisation in the weighing room before and after sampling, a systematic error occurs. For dust weights of less than 2 mg the uncorrected weight difference of the IOM sampling cassette may even appear to be negative. After 16 days in the weighing room the day-to-day variation was reduced to less than 10 µg. Not until after 18 days had the weight stabilised and only random variation persisted. In order to obtain correct cassette weights, 18 days of acclimatisation would hence be needed, both before and after use. In contrast, membrane filters in Petri dishes acclimatise in less than one day, and the pooled standard deviation among the filters over the 21 days was 5.2 µg.
Accommodation of humidity by blanks (Series 2 and 3) After having been stored twice in the weighing room for 18 days, the average and standard deviation of the 10 blank cassettes’ weight difference (final weight in Series 3 minus final weight in Series 2) were ⫺0.19±0.03 mg, and was mainly due to the 5% difference in relative humidity. Figure 1 shows the average weight increase from the initial weight for Series 2 and 3. The numerical values obtained when the data in Fig. 1 are modelled with Eq. (2) are presented in Table 5. The considerably larger uncertainty in the estimated numerical values for Series 2 may have been caused by the climate in the weighing room being more unstable during that series (see above). In summary, the data indicate a short time constant of approximately 5 h, and a longer time constant of approximately 4 days. The shorter time constant may be associated with the adsorption of water vapour onto the surface of the plastic cassette, and the longer time constant with the absorption/diffusion of water vapour into the interior of the plastic cassette. After what appeared to be complete moisture accommodation in Series 3 (Fig. 1) the cassettes were kept in the weighing room for a further 56 days, and were weighed three times in this period. During this period the weight of the IOM sampling cassettes increased, though at a much lower rate than previously. Over the period of 56 days the weight of the cassettes increased by 30 µg/week, although this might have been caused by a slight increase in relative humidity over the period, from 50.5 to 53.2%.
Table 3. The difference in weight difference between pairs of blanks weighed before and after the use of the accompanying non-blanks
Insulated IOM sampling cassette Corresponding filter Conducting IOM sampling cassette Corresponding filter
Number
Average (µg)
SD (µg)
Tolerance interval (µg)
37 24 34 9
⫺31.1 1.6 71.8 5.1
65.7 4.4 105 6.0
163 10.9 261 14.9
G. Lide´ n and G. Bergman
246
Table 4. Weight increase of IOM sampling cassette with and without mixed cellulose acetate filter, Series 1b and 1a, and filters only, (average and SD) as a function of time in the weighing room Filter in glass Petri dishb
IOM sampling cassette With filtera Time (day) 0.8 1.7 2.9 3.8 6.9 7.9 8.9 9.9 10.9 13.9 14.8 15.9 16.9 17.9 20.9
Without filtera
Average (mg)
SD (mg)
Average (mg)
SD (mg)
Average (mg)
SD (mg)
0.709 0.954 1.129 1.230 1.534 1.568 1.607 1.651 1.656 1.694 1.704 1.718 1.738 1.722 1.724
0.050 0.061 0.073 0.080 0.102 0.105 0.108 0.111 0.113 0.114 0.114 0.114 0.114 0.113 0.112
0.589 0.852 0.996 1.109 1.421 1.460 1.500 1.538 1.552 1.589 1.606 1.616 1.638 1.628 1.629
0.041 0.056 0.065 0.072 0.092 0.096 0.099 0.101 0.102 0.104 0.104 0.104 0.104 0.103 0.103
0.207 0.200 0.196 0.195 0.199 0.194 0.196 0.206 0.200 0.203 0.199 0.198 0.200 0.191 0.191
0.015 0.016 0.012 0.010 0.014 0.011 0.010 0.012 0.015 0.013 0.010 0.014 0.013 0.011 0.013
a
Thirty cassettes. Ten filters.
b
Fig. 1. Absolute weight increase of IOM sampling cassettes (with filters) as a function of storage time in the weighing room. Curves are based on non-linear regression, Eq. (2). Error bars indicate 95% confidence limits.
Fig. 2. Standard deviation of absolute weight increase versus average absolute weight increase of IOM sampling cassette for Series 1b, 2 and 3. Lines shown have slopes equal to relative standard deviation listed in Table 6.
Weighing imprecision as a function of storage time in the weighing room After approximately two days in the weighing room, the standard deviation of the blank cassettes was found to be proportional to the average weight
increase (see Fig. 2). The values are listed in Table 6. The relative standard deviation varied between 7 and 10%, but appeared to be independent of the weight increase after two to three weeks. The absolute standard deviation increased with the time spent in the weighing room.
Table 5. Parameters (point estimate and SD) from modelling the weight increase of the IOM sampling cassette with the model in Eq. (2) Series 1b 2 3
Comment
m1 (mg)
t1 (h)
m2 (mg)
t2 (day)
r2
Cassette with filter Cassettes unexposed to dust Blanks for cassettes with dust
0.51 (0.03) 0.19 (0.06) 0.19 (0.03)
5.5 (0.7) 5.0 (2.5) 4.9 (1.4)
1.24 (0.03) 0.58 (0.06) 1.06 (0.02)
4.04 (0.02) 2.65 (0.42) 4.75 (0.36)
0.999 0.993 0.999
IOM sampler in precision and handleability
247
Table 6. Results from three series of weighing of the IOM cassette in a weighing room for approx. three weeks Series
1b 2 3
Comment
Cassettes incl. filter Cassettes unexposed to dust Blanks for cassettes with dust
Number of cassettes
Final average weight increase (mg)
Average variability over test period
SD (%)
Tolerance interval (%)
30 39
1.7 0.75
6.6 7.9
16.8 19.4
10
1.2
10.1
34.2
Regarding the weight increase of the IOM cassette, the tolerance intervals were calculated to be approximately 17–35%. These values may seem to be too high to allow the correct estimation of the weight corrections from blanks, even if the cassettes were allowed to accommodate in the weighing room for two to three weeks, both before and after dust sampling. The dust deposits (Series 2 and 3) were therefore analysed with Eqs (3) and (4) in order to determine whether the use of blanks would reduce the weighing error caused by the slow accommodation to humidity of the cassette to within acceptable limits. Figure 3 shows how ⌬mdust(tB,tA), from Eq. (4), varies with tB and tA. ⌬mdust will, by definition, be ⬅0 for tB=tA=18 days. The maximum value was 0.15 mg, and the average value was 0.08 mg. The highest values were obtained for short times in the weighing room after exposure to dust, and long times before exposure to dust. When weighing both before and after exposure to dust was performed after the same number of days in the weighing room, ⌬mdust⬇0.07 mg for the first three days. ⌬mdust had decreased to below 0.04 mg after one week. Our results indicate that the standard deviation in the weight of cassette blanks was s⬇100 µg when the
Fig. 3. Root mean square of the difference between weighed dust deposits in IOM sampling cassettes [mg] after complete humidity accommodation (18 days), and when the cassettes were weighted on any day less than 18 before and after deposition of dust. (In the light grey area in the upper right corner, the RMS value is less than 0.025 mg.)
blanks are weighed after usually one day of accommodation in the weighing room (see Table 3). The standard deviations among the blanks from Series 2 and 3 after one day of accommodation in the weighing room were similar, s⬇90 µg. However, the RMS deviation between the fully accommodated dust weight and the weight after one week in the weighing room was ⬇40 µg. If three blanks are used instead of 10, the correction to the calculated standard deviation becomes √[1 + 1/3]/[1 + 1/10]⬇1.1 (International Organization for Standardization, 1999), which can be neglected. By combining the value of 40 µg with the standard deviation among the ten blanks after 19 days in the weighing room, s⬇30 µg, the standard deviation of the determination of a dust weight (in a cassette) after one week in the weighing room was obtained as 50 µg. Based on this standard deviation, the limit of quantification is calculated as 0.5 mg.
Humidity absorption of used and unused IOM sampling cassette (Series 4) and another plastic sampling cassette with a similar conductivity (Series 5) The results show that after 49 days in the weighing room, the 10 unused IOM sampling cassettes (Series 4a) could be subdivided into three subsets with identical relative weight increases (ratio of cassette weight increase to initial cassette weight), 0.25, 0.31 and 0.36%, respectively, based on initial and final weighing. (Within each subset the average was at least 250 times larger than the standard deviation.) Subsets within a series will be labelled with an upper-case letter appended to the series number. There was no correlation between relative weight increase and initial cassette weight. For the five used cassettes (Series 4b) no subdivision was discernible. The absolute weight increase for the unused and used IOM cassettes were similar, 2.37 and 2.09 mg. Due to the existence of a subdivision of the unused IOM cassettes, the standard deviation among the unused (unwashed) cassettes was much higher than for the used (washed) cassettes, 13.9 and 5.9%, respectively. Reanalysis of the used cassettes of Series 1, 2 and 3 with the statistical computer program JMP 3.1 [SAS
G. Lide´ n and G. Bergman
248
Institute Inc., Cary (NC), USA] revealed the existence of several subsets after a period of approximately three weeks in the weighing room, see Table 7. The results for Series 3 were almost independent of whether the analysis was based on the blanks only, or if the cassettes with dust deposits were also included. (All individual cassettes were partitioned identically between high and low relative weight increase in Series 2 and 3.) This means that each cassette remains within its subset also after being loaded with dust. However, the actual amount of relative weight increase for each subset depends mainly on the subset’s history of relative humidity outside the weighing room. Subsets A–M have been labelled differently in Table 7, since they are from different experiments. However, for all IOM sampling cassettes, the number of subsets is presumably much lower, probably 3–5. For example, as noted above, it was found that except for one cassette, subsets G=I and H=J. Figure 4 depicts how a combination of a histogram with a normal quantile plot reveals the existence of several subsets. After discussions with the two companies selling the two samplers it was evident that for both samplers, the number of manufactured batches was much lower than the found number of subsets. It is thus unknown whether the basis for the differentiation into subsets is anything other than the total (i.e. including pre-experimental) histor-
ies of accommodation to different humidities of different groups of cassettes. The relative weight increases for the subsets of Series 1b, 2, 3, 4a and 4b are shown graphically in Fig. 5. (Series 1a is not shown for better clarity.) It can be seen from the figure that the new (unused and unwashed) IOM cassettes (Series 4b) had not reached equilibrium, even after 7 weeks of accommodation to water humidity. The relative increases in weight of the unused (unwashed) CIS filter holders (Series 5) are much lower than for the IOM sampling cassette after 49 days in the weighing room, and were for the three subsets 0.0080, 0.090 and 0.17%, respectively. See Table 7 for the final data pertaining to three weeks storage. The results are presented graphically in Fig. 6. (Within each subset the ratio of the average to the standard deviation was in the range 12–170.) For the CIS filter holder there is a correlation between the relative weight increase and the initial filter holder weight. For the two subsets of the CIS filter holders with the highest water absorption (5-Y and 5-Z), accommodation was not completed even after 49 days in the weighing room. The numerical values obtained when the data in Fig. 5 are modelled with Eq. (2) are presented in Table 8. 75% of the weight increase has a time constant of 苲26 days, and 25% has a time constant of 苲30 h. The electrical surface conductivity of the three sub-
Table 7. Average and standard deviation of the cassette weight increase after approximately three weeks in the weighing room. Data for all experimental series divided into subsets Series
Comment
Number of cassettes
Average weight increase (‰)
SD (‰)
Cassettes without filtera Subset A Subset B Subset C Cassettes with filterb Subset D Subset E Subset F Cassettes unexposed to dust Subset G Subset H Blanks for cassettes with dustc Subset I Subset J Used cassettes New/unused cassettes Subset K Subset L Subset M
30 4 18 7 30 5 4 20 39 17 22 10 5 5 5 10 5 2 3
2.0 1.7 2.0 2.1 2.1 1.9 2.0 2.1 0.90 0.84 0.95 1.5 1.4 1.6 2.4 2.6 2.2 2.5 2.8
0.11 0.14 0.01 0.01 0.12 0.01 0.03 0.01 0.07 0.07 0.02 0.13 0.10 0.03 0.11 0.38 0.01 0.01 0.01
New/unused filter holder Subset X Subset Y Subset Z
15 6 6 3
0.55 0.081 0.67 1.3
0.46 0.003 0.026 0.011
IOM sampling cassette 1a
1b
2 3 4a 4b
CIS filter holder 5
a
One outlier not included in subsets A–C. One outlier not included in subsets D–F. c One outlier not included in subsets I–J. b
IOM sampler in precision and handleability
249
Fig. 4. Histogram and normal quantile plot of final (after 21 days in weighing room) relative weight increases for Series 1b (30 IOM sampling cassettes with filter). The Y scale of the histogram and the normal quantile plot show relative weight increase. Note that the data seem to consist of three subsets and one outlier.
Fig. 5. Relative weight increase of IOM sampling cassettes as a function of time in weighing room. Series labelling as in Table 7. Error bars indicate 95% confidence limits.
G. Lide´ n and G. Bergman
250
CASSETTE HANDLEABILITY—EXPERIENCE AND DISCUSSION
Fig. 6. Relative weight increase of CIS filter holders as a function of time in weighing room. Series labelling as in Table 7. Curves are based on non-linear regression, Eq. (2). Error bars indicate 95% confidence limits.
sets (5-X, 5-Y and 5-Z) of the CIS filter holder was measured with a resistance meter, and all were found to be conducting. The relative weight increase of the best subset of the CIS filter holder was 30 times lower than the weight increase of the best subset of the IOM cassette. It would therefore be possible to manufacture the IOM sampling cassette in a plastic which absorbs considerably less humidity, but would still be electrically conducting. The standard deviations of both types of new/unused cassettes were high. For the IOM sampling cassette the standard deviation was four times greater for unused cassettes than for used cassettes. It is unknown whether this high value is random, or whether using the cassettes reduces the variation between the subsets. If so, repeated washing of new cassettes before they are taken into use may be important. In order to reduce the high variability among the IOM sampling cassettes two options are available. (1) Terminating the use of plastic cassettes in favour of those in stainless steel. This would require a substantial capital investment as these cassettes are approximately six times more expensive than the plastic cassettes. (2) Ensure that all plastic cassettes used on one occasion (both cassettes through which dust is sampled and corresponding blanks) belong to the same subset. This would require a substantial labour effort as it would take several weeks to determine which cassette specimen belong to each subset.
Transport clip The yellow transport clip manufactured by SKC is used during transport in order to prevent the interior of the cassette from becoming contaminated and to prevent sampled dust from escaping from the cassette. See Fig. 7, left side. It is impossible to remove the clip from the cassette, or to replace it without touching the cassette with one’s fingers. Therefore, gloves should be used when sampling with the IOM sampler. We have found that during transport dust may migrate from the sampling cassette to the interior of the clip. This dust is not weighed, since it may be difficult to detect if it is light in colour, and it may be difficult to get it back into the cassette. This would entail additional work which would probably lead to unknown errors. In order to ensure the integrity of the sampled dust a cover (lid) which fits tightly over the cassette has been manufactured in Deldrin. See Fig. 7, right side. It is possible to write the number of the sample on the cover for simple identification of the cassette. The cover is then weighed together with the cassette and, in this way, all sampled dust is weighed. The cover weighs 苲0.6 g, and typical values of its weight increase after two days in the weighing room is 苲0.15 mg. After two days of accommodation the weight of the cover is constant. The cover has been tested at several workplaces. Before the cover is replaced on the cassette after sampling, loose dust on the outer surface of the cassette (that part which extends outwards from the front of the sampler) must be removed in order not to be covered by the cover. During sampling, dust may be deposited on the outer surface of the inlet tube of the cassette. (This has only been observed for the plastic version.) Deposition occurs mainly on the part of the tube which extends outwards from the front of the sampler, but also occurs closer to the filter end of the tube, since the diameter of the tube is slightly smaller than the opening in the front. Any dust deposited here should be removed before weighing the cassette. This problem has been reduced by the introduction of the conducting version as the difference in diameter between the tube of the cassette and the opening in the front is smaller.
Table 8. Parameters (point estimate and SD) from modelling the relative weight increase of the CIS filter holder with the model in Eq. (2) Series-subset 5-X 5-Y 5-Z
m1 (‰)
t1 (days)
m2 (‰)
t2 (days)
r2
0.076 (0.002) 0.25 (0.021) 0.46 (0.034)
1.2 (0.25) 1.2 (0.20) 1.3 (0.19)
0.76 (0.025) 1.49 (0.041)
26.2 (3.0) 26.1 (2.5)
0.961 0.999 0.999
IOM sampler in precision and handleability
251
Fig. 7. Photograph showing IOM sampling cassette with SKC transport clip mounted (left), IOM sampling cassette with our cassette cover unmounted (middle), and IOM sampling cassette with our cassette cover mounted (right).
Measurement of flow rate Originally there was no adapter for measuring the flow rate through the IOM sampler. After some time however, SKC Inc. introduced a flow adapter. The adapter is pressed tightly against the IOM sampler by a spring, and a gasket makes the seal tight. The spring pulls from the back of the sampler. However, this adapter can only be used when the sampler is not mounted on the worker. With the spring dismounted from the adapter, it must be held in place hard against the front of the IOM sampler with one or two hands, while simultaneously another hand holds a flow meter, and yet another hand, if necessary, corrects the pump flow. Another flow adapter has therefore been constructed in deldrin. This adapter fits tightly onto the inlet tube of the sampling cassette, in a similar fashion to Hetland’s flow adapter for the Personal
Inhalable Dust Spectrometer (Hetland, 1996). At the other end of the flow adapter, a plastic or rubber tube is connected to the flow meter. See Fig. 8. This flow adapter is secured to the IOM sampler without the need for a supporting hand, and the occupational hygienist has both hands free to measure the flow rate and correct it if necessary. This flow adapter has been used at several workplaces. The leak flow between the adapter and the cassette is generally less than 2%, and can be disregarded. CONCLUSIONS
The repeatability of weighing the IOM sampling cassette is very good. However, the cassette absorbs a considerable amount of water vapour, and hence takes a long time to accommodate to the humidity of a weighing room after being used at a workplace. The
Fig. 8. Photograph showing IOM sampling cassette with flow adapter mounted (right) and the unmounted flow adapter only (left).
252
G. Lide´ n and G. Bergman
exponential time constant for humidity accommodation is four days, and in order to obtain equilibrium between air humidity and the cassette, 15–20 days are needed before the cassette may be weighed correctly. The standard deviation in weight increase between the cassettes is considerable, 5–10% of the weight increase caused by an increase in humidity. Despite this large variability, it seems to be possible to obtain a low weighing imprecision by the use of blanks. The cassettes should be stored in the weighing room for one week before and after sampling until they are weighed. In this way, an imprecision of 0.05 mg can be obtained, and hence a limit of quantification of 0.5 mg. Experiments indicate that the conducting plastic of the IOM sampling cassette may be replaced with another type of plastic with similar electrical conductivity, but whose humidity absorption is 30 times lower. The IOM sampling cassettes tested seem to be derived from several subsets with clearly distinguishable relative weight increases as a function of time. However, the cause for the differentiation into subsets was not found. To reduce the imprecision substantially below 0.05 mg, either metal sampling cassettes shall be used, or for plastic sampling cassette, specimen shall not be mixed between the subsets. A cover which is weighed together with the cassette has been designed and tested to ensure the integrity of the sample during transportation. A new flow adapter has been designed and tested for which no hands are needed to keep the flow meter attached to the IOM sampler. Acknowledgements—This project was supported by the Swedish Fund for Working Life Research, project number 95-0186. We wish to thank Anders Lindquist (National Institute for Working Life) for the parts he has manufactured which have allowed us to test whether the IOM sampler may be improved for workplace sampling.
REFERENCES American Conference of Governmental Industrial Hygienists. 1992–1993 Threshold Limit Values for Chemical Substances and Physical Agents and Biological Exposure Indices. Cincinnati (OH): American Conference of Governmental Industrial Hygienists; 1992. Comite´ Europe´ en de Normalisation. EN481 Workplace Atmospheres—Size Fraction Definitions for Measurement of Airborne Particles. Brussels: Comite´ Europe´ en de Normalisation; 1993. Hald A. Statistical Theory with Engineering Applications. New York: John Wiley and Sons, 1952. Hetland S. Flow Adapter in Deldrin for the PIDS. Norway: Norwegian National Institute of Occupational Health; 1996 (personal communication). International Organisation for Standardization. ISO7708 Air Quality—Particle Size Fraction Definitions for HealthRelated Sampling. Geneva: ISO; 1995. International Organisation for Standardization. ISO/CD15767 Workplace Atmospheres—Gravimetric Analysis of Aerosol Collection Media. Geneva: ISO; 1999. Kenny LC, Aitken R, Chalmers C, Fabrie`s JF, Gonzalez-Fernandez E, Kromhout H, Lide´ n G, Mark D, Riediger G, Prodi V. A collaborative European study of personal inhalable aerosol sampler performance. Annals of Occupational Hygiene 1997;41:135–53. Li S-N, Lundgren DA. Weighing accuracy of samples collected by IOM and CIS inhalable samplers. American Industrial Hygiene Association Journal 1999;60:235–6. Lide´ n G. Evaluation of the SKC personal respirable dust sampling cyclone. Applied Occupational and Environmental Hygiene 1993;8:178–90. Mark D. The use of dust-collecting cassettes in dust sampling. Annals of Occupational Hygiene 1990;34:281–91. Mark D, Vincent JH. A new personal sampler for airborne total dust in workplaces. Annals of Occupational Hygiene 1986;30:89–102. Smith JP, Bartley DL, Kennedy ER. Laboratory investigation of the mass stability of sampling cassettes from inhalable aerosol samplers. American Industrial Hygiene Association Journal 1998;59:582–5. Vaughan NP, Milligan BD, Ogden TL. Filter weighing reproducibility and the gravimetric detection limit. Annals of Occupational Hygiene 1989;33:331–7.
Ann. occup. Hyg., Vol. 45, No. 3, pp. 241–252, 2001 2001 British Occupational Hygiene Society Published by Elsevier Science Ltd. All rights reserved Printed in Great Britain. 0003–4878/01/$20.00
Weighing Imprecision and Handleability of the Sampling Cassettes of the IOM Sampler for Inhalable Dust ¨ RAN LIDE´N* and GUNNEL BERGMAN GO National Institute for Working Life, S-112 79 Stockholm, Sweden
The weight stability of the sampling cassette of the IOM sampler for inhalable dust was tested in several weighing experiments. The results show that the reliability of repeated weighings was good, but the absorption of water vapour was slow and varied considerably among cassette specimen. The exponential time constant for water absorption was approximately 4 days, and 15–20 days were needed to obtain weight stability. With the help of cassette blanks the imprecision in dust weight could be held below 0.05 mg, if the cassettes were allowed one week’s storage in the weighing room before weighing, both before and after sampling. The IOM sampling cassettes seem to consist of a few subsets, each with identical relative weight increase in a weighing room. To keep the variability low it is important that both the blanks and the cassettes used for sampling come from the same subset. Experiments indicate that the conducting plastic of the IOM sampling cassette may be replaced with another kind of plastic with similar electrical conductivity, but whose humidity absorption is 30 times lower. A lid, which is weighed with the cassette, was designed so that the potential dust loss from the cassette proper to the commercial transport clip was eliminated. A flow adapter, which simplifies the measurement of the air flow during personal sampling, was designed. 2001 British Occupational Hygiene Society. Published by Elsevier Science Ltd. All rights reserved Keywords: dust sampling; inhalable dust; sampling cassette; weighing; reproducibility; IOM sampler; GSP sampler
INTRODUCTION
There are moves in many countries to specify inhalable dust sampling according to the internationallyagreed standards (International Organisation for Standardization, 1995; Comite´ Europe´en de Normalisation, 1993; American Conference of Governmental Industrial Hygienists, 1992). The Institute of Occupational Medicine (IOM) sampler is presently the most commonly used sampler for inhalable dust (Mark and Vincent, 1986). The outer part of the sampler consists of a cylinder with a diameter of 苲30 mm and a height of 苲15 mm. Inside the cylinder a sampling cassette is mounted which holds a 25 mm filter. The filter is placed at the bottom of the cassette, and upstream of the filter the cassette consists of a tube with a diameter of 15 mm and a length of 10 mm.
Received 16 February 2000; in final form 7 July 2000. *Author to whom correspondence should be addressed. Tel.: +46-8-730-9100; fax: +46-8-828-678; e-mail: [email protected]
The sampling cassette is thin-walled, 苲0.5 mm. When the cassette is mounted inside the sampler, its tube protrudes 1–2 mm outside the base of the cylindrical sampler. During sampling the sampler is mounted on the worker’s chest, and is intended to face horizontally outwards. The deposit consists of all dust which passes the inlet plane of the tube of the cassette, whether it is deposited onto the filter or the walls of the cassette. Therefore, the entire cassette must be weighed, and not only the filter. The IOM sampler has been produced commercially in several versions. Originally, Negretti Ltd produced a version in aluminium, which is however, no longer commercially available. The present manufacturer, SKC Inc. [Eighty-Four (PA), USA], has produced different versions, the first in a brown plastic which had a non-conducting surface, and the present version, in a black plastic which is electrically conductive. In addition, the sampling cassette itself is also available in stainless steel. The plastic sampling cassette of the IOM sampler weighs 苲0.8 g and the stainless steel version 3.8 g.
241
242
G. Lide´ n and G. Bergman
This is a considerable tare weight compared with typical filter weights, for example the Millipore [Bedford (MA), USA] 25 mm AAWP (0.8 µm pore size), which weighs 苲20 mg. IOM samplers ordered by occupational hygienists from our laboratory are not delivered with cassettes mounted in the samplers. During transport the cassettes are stored in a separate box, with SKC’s yellow transportation clip mounted in order to protect the inner surface of the cassette. After sampling, the clip prevents any sampled dust from escaping from the cassette. When sampling cassettes are sent from our laboratory to occupational hygienists for sampling at workplaces, previously two (presently three) blanks (cassettes through which no air shall be drawn) per 25 cassettes are included. The unexposed cassette blanks are used to correct for changes in temperature and relative humidity in the weighing room between the weighing before and after dust sampling. The dust load of each cassette is determined as the weight difference of the cassette after and before sampling, corrected for the average weight difference of the blanks at the two weighings. Prior to this investigation the cassettes were usually weighed after one or two days storage in the weighing room. After analysis at our laboratory, the sampling cassettes are washed in an ultrasonic bath, first with water mixed with a mild detergent. They are then rinsed in deionized water, and finally rinsed with a water–alcohol solution. The weight stability of the sampling cassettes with mounted filters, or rather the weighing imprecision, for aluminium sampling cassettes has previously been studied by Vaughan et al. (1989) and Mark (1990). Recently, Smith et al. (1998) and Li and Lundgren (1999) reported that a considerable mass of water absorbs to the plastic sampling cassette of the IOM sampler. Smith et al. (1998) found that 苲1.5 mg was absorbed by the plastic cassette when the relative humidity was increased from zero to 苲60%, and Li and Lundgren (1999) found that the cassette loses 苲2.5 mg when the relative humidity was decreased from 苲75% to zero. These results demonstrate the necessity of defining the requirements for obtaining a low imprecision in the weighing of IOM plastic sampling cassettes. A plastic with electrical conductivity is generally obtained by mixing the plastic proper with e.g. carbon black. It is the capacity of carbon black to absorb humidity which makes the plastic electrically conducting. Therefore, one cannot obtain an electrically conducting plastic which does not absorb water. However, the trade-off between electrical conductivity and absorption of humidity depends on the properties of the carbon black used. This report presents the results of weighing experiments on the two versions of the plastic sampling cassette for the IOM sampler and one experiment on the plastic sampling cassette for the Conical Inhalable Sampler (JS Holdings Ltd, Stevenage, UK), a British
plastic version of the German GSP sampler (Kenny et al., 1997). Additionally, practical experience gained and proposals to increase the handleability based on several years of workplace sampling with the IOM sampler are presented. METHODS
Several different experiments are reported in this paper. They were all performed with the same equipment, which will be described initially in this section. Cassettes The weighing experiments were mainly performed on the electrically conducting plastic (i.e. black) cassette manufactured by SKC Inc., but some experiments were also performed with the electrically insulating (i.e. brown) plastic cassette. Unless stated otherwise, empty cassettes (i.e. without any filter) were used in the experiments. The filters used were of cellulose acetate from Millipore, 25 mm AAWP (0.8 µm pore size). Several series of similar experiments are reported below, and in each series the cassettes used were chosen at random from the cassettes available. Unless stated otherwise, it was not our intention to use the same cassettes for repeated experiments. Weighing room Weighing was performed in a weighing room in which the temperature and humidity were controlled to 21.0±0.5°C and 55±2%, respectively. The humidity was designed to be as high as possible, in order to reduce the weighing bias caused by electrical charges on the weighed object, though lower than the deliquescence point for (metal) salts. (Outside the weighing room, unused cassettes were stored in ordinary laboratory rooms, for which only the temperature was kept at approximately 21°C by the general ventilation system. The humidity was neither modified nor registered.) All weighing was carried out with a Mettler– Toledo ME22 balance (Greifensee, Switzerland), with a resolution of 1 µg. At the beginning of each weighing session the balance was calibrated using a 100 mg weight supplied by the manufacturer. During the sessions, the balance was tared when necessary. In order to reduce the static electrical charge on the filter or cassette being weighed, a radioactive α-source [0.5 mCi 210Po, NRD Inc., Grand Island (NY), USA] was placed beneath the weighing pan. Unless stated otherwise, the cassettes were weighed after one or two days storage in the weighing room. Tolerance intervals For each series of filter or cassette weights presented below, the standard deviation refers to the
IOM sampler in precision and handleability
whole series and not its average. A tolerance interval for the series was also calculated. A tolerance interval is an estimate of the width of the interval, which for a normally distributed population, with confidence of 100*(1⫺2e)% contains 100*P% of the population (Hald, 1952). (This differentiates a tolerance interval from a confidence interval, which with 100*(1⫺2e)% confidence contains the estimation value of the population.) Symmetrical tolerance intervals are calculated as x¯±s, and is approximated with the following equation
冪c
⬇z1⫺e
n⫺1
冉 冊
2 1⫺P
1+
1 2n
243
taken as a measure of the effect of not using gloves when handling the IOM cassettes. Cassette weight increase The change in weight of sampling cassettes over time (⌬m = ⌬m(t)), when moved from a room with varying humidity to the weighing room with high and controlled humidity, was analysed with non-linear regression. The following model was used:
⌬m(t) = m1[1⫺exp(⫺t/t1)] + m2[1⫺exp(⫺t/t2)] (2)
(1)
where z1⫺e is the 100*(1⫺e) percentile of the normal distribution, and c21⫺P is the 100*P% percentile of the chi-squared distribution (with n⫺1 degrees of freedom). Tolerance intervals, with 95% confidence (i.e. e=0.025) for 95% of the population (i.e. P=0.95), are listed below with their half-widths, i.e. s. Reliability The reliability of weighing was investigated by weighing one empty cassette 30 times within half an hour. During this time the air temperature rose by 0.2°C and the relative humidity by 0.1%. As a comparison, a 100 mg calibration weight was weighed 30 times within three quarters of an hour. During this time the air temperature rose by 0.3°C and the relative humidity decreased by 0.9%. Handling cassettes with and without gloves Two experiments were carried out, each involving 60 cassettes being weighed on two consecutive days. During weighing the cassettes were handled with a pair of tweezers. In between weighing, 30 of the cassettes were handled with hands and fingers as would be the case in a laboratory and at a workplace. This means that the yellow transportation clip was mounted and dismounted twice. However the cassette was not mounted inside a sampler. In the first experiment gloves of the label ‘Golden Hand latex examination gloves’ (Kebo Lab AB, Stockholm, Sweden) were used, and in the second the cassettes were handled with bare hands. The same person carried out the weighing. No attempt was made to obtain a representative selection of people with varyingly greasy or dirty fingers. During weighing the relative humidity was 53.4±0.3% and 54.9±0.3% for the experiments with and without gloves, respectively. The temperature was 21.5°C and 21.4°C in the two experiments. For each experiment (gloves and bare hands, respectively) the results are presented as the weight difference among the cassettes between the two weighing days. The difference in weight change overnight between cassettes handled with bare hands and those only handled with a pair of tweezers was
where t is the time since the change in relative humidity, ⌬m(t) is either the absolute weight difference between the weight at time t and the first weight at t>0, or ⌬m(t) is the ratio of the absolute weight difference at time t and the first weight at tⱖ0, m1 and m2 are the weight changes resulting from two exponential processes with time constants t1 and t2, respectively. The calculations were carried out with the non-linear curve-fitting module of the computer program SigmaPlot [Jandel Scientific, San Rafael (CA), USA]. Several different series of experiments were performed to analyse the temporal behaviour of the plastic sampling cassettes. In Series 1, 30 IOM sampling cassettes with filters and 30 without filters, and 10 filters in glass Petri dishes, which had all been stored in another room at the laboratory for several months, were weighed in that room which happened to have a humidity and temperature of 29.6% and 22.1°C during the weighing. The cassettes and the Petri dishes with filters were then taken to the weighing room. During the following three weeks the cassettes and the filters were weighed every working day. During this period the humidity and the temperature of the weighing room were 54.6±0.5% and 21.6±0.1°C (average±standard deviation) during the time of weighing. The results were calculated as the difference in weight for each cassette or filter relative to its weight on the first day. In Series 2, 40 sampling cassettes were weighed over a period of 18 days. After one month in the dust generation laboratory 10 of these were weighed again over a period of 18 days (Series 3). During the weighings the relative humidity and temperature in Series 2 were 57.1±1.3% and 20.9±0.8°C, respectively, while in Series 3 the relative humidity and temperature were 51.9±0.8% and 20.8±0.03°C, respectively. Thus, for Series 2 the relative humidity was above the required level, and for Series 3 the relative humidity was slightly below the required level of 55±2%. In Series 4, ten new (unused and unwashed) IOM sampling cassettes and five used (and washed) IOM sampling cassettes were used and in Series 5, 15 new filter holders for the Conical Inhalable Sampler, CIS, were used. Initially, all 30 cassettes were stored out-
G. Lide´ n and G. Bergman
244
side the weighing room for two weeks. They were then moved into the weighing room, and were weighed ten times over a period of 49 days. During the weighings, the relative humidity and temperature were 54.6±0.4% and 20.8±0.1°C, respectively. All cassettes/filter holders were electrically conducting and without filters. The CIS filter holder weighs considerably more than the IOM sampling cassette, 苲3.4 g compared to 苲0.8 g. The CIS filter holder is essentially a ring with a rectangular cross section, 6.7 mm with a 42 mm outer diameter. According to the data sheets for the plastics of the conducting IOM sampling cassette and the CIS filter holder (supplied by the manufacturers of the samplers), the resistivity of the two plastics were similar, 苲1 k⍀ cm.
Weighing imprecision as a function of storage time in the weighing room In order to determine the error in weighing dust samples, 30 sampling cassettes were exposed to quartz dust in a previously described dust chamber system (Lide´ n, 1993). All sampling cassettes had different dust deposits in the range 0.18–3.17 mg. Ten additional cassettes (incl. filters) were used as blanks. Both before and after the exposure to dust, during 18+18 days, all 40 cassettes were weighed every day for the first working week, and thereafter every other working day. (The unexposed cassettes constitute Series 2 and 3 presented above.) For each day of weighing, before and after the exposure to dust, one dust weight, mdust, was calculated for each cassette, k, according to the following equation
再
mdust(k,tB,tA) = mk(tA)⫺mk(tB)⫺⌬mREF(tB,tA) ⌬mREF(tB,tA) = mREF(tA)⫺mREF(tB) (3)
where tB and tA are the time since the cassettes were stored in the weighing room before and after the exposure to dust, respectively, and mREF is the average weight of the blank sampling cassettes at the time in question (tB or tA). The dust weight obtained at times tB and tA was compared with that obtained when the cassette weights had stabilised after 18 days in the weighing room, mdust(k,⬁,⬁). For all combinations of weighing days, before and after exposure of the cassettes to dust (tB and tA), the root mean square (RMS) average cassette weight difference ⌬mdust(tB,tA) was calculated from the equation
冘
WEIGHT CHANGES—RESULTS AND DISCUSSION
Reliability The results are presented in Table 1. They show that the reliability is sufficiently good to obtain a limit of quantification equal to a dust deposit of 0.02 mg, and thence the cassette only needs to be weighed once before and once after sampling. Handling cassettes with and without gloves The results are shown in Table 2. The effect of handling the cassettes with bare hands (苲0.02 mg), is statistically significant at the 5% level, but at this level the effect of using gloves is not significant. A weight increase of 0.02 mg might be considered insignificant, but a requirement that the cassettes be handled only with gloves informs all who are handling the cassettes of the necessary precautions. A question is how the cassettes would be handled at dusty workplaces if gloves were not required. It is possible but troublesome to use gloves when weighing in the laboratory. However, the laboratory staff will not know whether the occupational hygienist handled the cassettes with gloves or bare hands at the workplace. Our laboratory always supplies gloves in the transport case with the cassettes, together with written instructions to use them. However, on more than one occasion the cassettes have been handled with bare hands during sampling. The fact that the cassettes may have been touched by bare (unclean) hands implies the possibility of a larger, but unknown, error. Variability of blanks used in conjunction with cassettes exposed to dust The weight difference of the pair of blanks belonging to 37 sets of insulating cassettes and 34 sets of conducting cassettes ordered by occupational hygienists for their dust sampling purposes over a period of six months was determined from recorded weighing data. In some cases the weight of the filter of the blank sampling cassette was also determined. The results are presented in Table 3 as the difference in weight change of the two blanks between the two weighing occasions for each pair of blanks. Both
Table 1. Results from repeated weighing (30 times), of one IOM sampling cassette within 45 min and one 100 mg calibration weight within 30 min
n
1 ⌬mdust(tB,tA)2 = [m (k,t ,t ) nk = 1 dust B A ⫺mdust(k,⬁,⬁)]2
Average (mg)
SD (µg)
Tolerance interval (µg)
806.009
2
5
100.013
1
2
(4)
where n is the number of exposed cassettes (=30). The limit of quantification (LoQ) is taken as ten times the population standard deviation.
IOM sampling cassette 100 mg calibration weight
IOM sampler in precision and handleability
245
Table 2. The weight difference of 30 IOM sampling cassettes weighed on two consecutive days, when handled with gloves and bare hands
Gloves Untouched cassettes Effect of gloves Fingers Untouched cassettes Effect of fingers
Average (µg)
SD (µg)
Tolerance interval (µg)
16 15 1 33 14 19
4 2 1a 9 2 2a
9 5 22 4
a
The standard deviation for the effects of gloves and fingers.
averages and standard deviations for the two plastic versions of the sampling cassette of the IOM sampler are 15–20 times higher than the corresponding values for the filter of the sampling cassette. The imprecision (population standard deviation) for the blank filters is better than 0.01 mg, whereas the imprecision for the sampling cassettes of the IOM sampler is considerably larger, with a pooled value for both cassette versions equal to 苲0.08 mg. Accommodation to humidity by blanks (Series 1) The data are summarised in Table 4 and for the cassettes with filters (Series 1b) are shown graphically in Fig. 1. Our results confirm the findings of Smith et al. (1998) and Li and Lundgren (1999) concerning the amount of moisture absorbed by the cassettes, and the time scale involved. This experiment shows that if an IOM sampling cassette is weighed after a different number of days of acclimatisation in the weighing room before and after sampling, a systematic error occurs. For dust weights of less than 2 mg the uncorrected weight difference of the IOM sampling cassette may even appear to be negative. After 16 days in the weighing room the day-to-day variation was reduced to less than 10 µg. Not until after 18 days had the weight stabilised and only random variation persisted. In order to obtain correct cassette weights, 18 days of acclimatisation would hence be needed, both before and after use. In contrast, membrane filters in Petri dishes acclimatise in less than one day, and the pooled standard deviation among the filters over the 21 days was 5.2 µg.
Accommodation of humidity by blanks (Series 2 and 3) After having been stored twice in the weighing room for 18 days, the average and standard deviation of the 10 blank cassettes’ weight difference (final weight in Series 3 minus final weight in Series 2) were ⫺0.19±0.03 mg, and was mainly due to the 5% difference in relative humidity. Figure 1 shows the average weight increase from the initial weight for Series 2 and 3. The numerical values obtained when the data in Fig. 1 are modelled with Eq. (2) are presented in Table 5. The considerably larger uncertainty in the estimated numerical values for Series 2 may have been caused by the climate in the weighing room being more unstable during that series (see above). In summary, the data indicate a short time constant of approximately 5 h, and a longer time constant of approximately 4 days. The shorter time constant may be associated with the adsorption of water vapour onto the surface of the plastic cassette, and the longer time constant with the absorption/diffusion of water vapour into the interior of the plastic cassette. After what appeared to be complete moisture accommodation in Series 3 (Fig. 1) the cassettes were kept in the weighing room for a further 56 days, and were weighed three times in this period. During this period the weight of the IOM sampling cassettes increased, though at a much lower rate than previously. Over the period of 56 days the weight of the cassettes increased by 30 µg/week, although this might have been caused by a slight increase in relative humidity over the period, from 50.5 to 53.2%.
Table 3. The difference in weight difference between pairs of blanks weighed before and after the use of the accompanying non-blanks
Insulated IOM sampling cassette Corresponding filter Conducting IOM sampling cassette Corresponding filter
Number
Average (µg)
SD (µg)
Tolerance interval (µg)
37 24 34 9
⫺31.1 1.6 71.8 5.1
65.7 4.4 105 6.0
163 10.9 261 14.9
G. Lide´ n and G. Bergman
246
Table 4. Weight increase of IOM sampling cassette with and without mixed cellulose acetate filter, Series 1b and 1a, and filters only, (average and SD) as a function of time in the weighing room Filter in glass Petri dishb
IOM sampling cassette With filtera Time (day) 0.8 1.7 2.9 3.8 6.9 7.9 8.9 9.9 10.9 13.9 14.8 15.9 16.9 17.9 20.9
Without filtera
Average (mg)
SD (mg)
Average (mg)
SD (mg)
Average (mg)
SD (mg)
0.709 0.954 1.129 1.230 1.534 1.568 1.607 1.651 1.656 1.694 1.704 1.718 1.738 1.722 1.724
0.050 0.061 0.073 0.080 0.102 0.105 0.108 0.111 0.113 0.114 0.114 0.114 0.114 0.113 0.112
0.589 0.852 0.996 1.109 1.421 1.460 1.500 1.538 1.552 1.589 1.606 1.616 1.638 1.628 1.629
0.041 0.056 0.065 0.072 0.092 0.096 0.099 0.101 0.102 0.104 0.104 0.104 0.104 0.103 0.103
0.207 0.200 0.196 0.195 0.199 0.194 0.196 0.206 0.200 0.203 0.199 0.198 0.200 0.191 0.191
0.015 0.016 0.012 0.010 0.014 0.011 0.010 0.012 0.015 0.013 0.010 0.014 0.013 0.011 0.013
a
Thirty cassettes. Ten filters.
b
Fig. 1. Absolute weight increase of IOM sampling cassettes (with filters) as a function of storage time in the weighing room. Curves are based on non-linear regression, Eq. (2). Error bars indicate 95% confidence limits.
Fig. 2. Standard deviation of absolute weight increase versus average absolute weight increase of IOM sampling cassette for Series 1b, 2 and 3. Lines shown have slopes equal to relative standard deviation listed in Table 6.
Weighing imprecision as a function of storage time in the weighing room After approximately two days in the weighing room, the standard deviation of the blank cassettes was found to be proportional to the average weight
increase (see Fig. 2). The values are listed in Table 6. The relative standard deviation varied between 7 and 10%, but appeared to be independent of the weight increase after two to three weeks. The absolute standard deviation increased with the time spent in the weighing room.
Table 5. Parameters (point estimate and SD) from modelling the weight increase of the IOM sampling cassette with the model in Eq. (2) Series 1b 2 3
Comment
m1 (mg)
t1 (h)
m2 (mg)
t2 (day)
r2
Cassette with filter Cassettes unexposed to dust Blanks for cassettes with dust
0.51 (0.03) 0.19 (0.06) 0.19 (0.03)
5.5 (0.7) 5.0 (2.5) 4.9 (1.4)
1.24 (0.03) 0.58 (0.06) 1.06 (0.02)
4.04 (0.02) 2.65 (0.42) 4.75 (0.36)
0.999 0.993 0.999
IOM sampler in precision and handleability
247
Table 6. Results from three series of weighing of the IOM cassette in a weighing room for approx. three weeks Series
1b 2 3
Comment
Cassettes incl. filter Cassettes unexposed to dust Blanks for cassettes with dust
Number of cassettes
Final average weight increase (mg)
Average variability over test period
SD (%)
Tolerance interval (%)
30 39
1.7 0.75
6.6 7.9
16.8 19.4
10
1.2
10.1
34.2
Regarding the weight increase of the IOM cassette, the tolerance intervals were calculated to be approximately 17–35%. These values may seem to be too high to allow the correct estimation of the weight corrections from blanks, even if the cassettes were allowed to accommodate in the weighing room for two to three weeks, both before and after dust sampling. The dust deposits (Series 2 and 3) were therefore analysed with Eqs (3) and (4) in order to determine whether the use of blanks would reduce the weighing error caused by the slow accommodation to humidity of the cassette to within acceptable limits. Figure 3 shows how ⌬mdust(tB,tA), from Eq. (4), varies with tB and tA. ⌬mdust will, by definition, be ⬅0 for tB=tA=18 days. The maximum value was 0.15 mg, and the average value was 0.08 mg. The highest values were obtained for short times in the weighing room after exposure to dust, and long times before exposure to dust. When weighing both before and after exposure to dust was performed after the same number of days in the weighing room, ⌬mdust⬇0.07 mg for the first three days. ⌬mdust had decreased to below 0.04 mg after one week. Our results indicate that the standard deviation in the weight of cassette blanks was s⬇100 µg when the
Fig. 3. Root mean square of the difference between weighed dust deposits in IOM sampling cassettes [mg] after complete humidity accommodation (18 days), and when the cassettes were weighted on any day less than 18 before and after deposition of dust. (In the light grey area in the upper right corner, the RMS value is less than 0.025 mg.)
blanks are weighed after usually one day of accommodation in the weighing room (see Table 3). The standard deviations among the blanks from Series 2 and 3 after one day of accommodation in the weighing room were similar, s⬇90 µg. However, the RMS deviation between the fully accommodated dust weight and the weight after one week in the weighing room was ⬇40 µg. If three blanks are used instead of 10, the correction to the calculated standard deviation becomes √[1 + 1/3]/[1 + 1/10]⬇1.1 (International Organization for Standardization, 1999), which can be neglected. By combining the value of 40 µg with the standard deviation among the ten blanks after 19 days in the weighing room, s⬇30 µg, the standard deviation of the determination of a dust weight (in a cassette) after one week in the weighing room was obtained as 50 µg. Based on this standard deviation, the limit of quantification is calculated as 0.5 mg.
Humidity absorption of used and unused IOM sampling cassette (Series 4) and another plastic sampling cassette with a similar conductivity (Series 5) The results show that after 49 days in the weighing room, the 10 unused IOM sampling cassettes (Series 4a) could be subdivided into three subsets with identical relative weight increases (ratio of cassette weight increase to initial cassette weight), 0.25, 0.31 and 0.36%, respectively, based on initial and final weighing. (Within each subset the average was at least 250 times larger than the standard deviation.) Subsets within a series will be labelled with an upper-case letter appended to the series number. There was no correlation between relative weight increase and initial cassette weight. For the five used cassettes (Series 4b) no subdivision was discernible. The absolute weight increase for the unused and used IOM cassettes were similar, 2.37 and 2.09 mg. Due to the existence of a subdivision of the unused IOM cassettes, the standard deviation among the unused (unwashed) cassettes was much higher than for the used (washed) cassettes, 13.9 and 5.9%, respectively. Reanalysis of the used cassettes of Series 1, 2 and 3 with the statistical computer program JMP 3.1 [SAS
G. Lide´ n and G. Bergman
248
Institute Inc., Cary (NC), USA] revealed the existence of several subsets after a period of approximately three weeks in the weighing room, see Table 7. The results for Series 3 were almost independent of whether the analysis was based on the blanks only, or if the cassettes with dust deposits were also included. (All individual cassettes were partitioned identically between high and low relative weight increase in Series 2 and 3.) This means that each cassette remains within its subset also after being loaded with dust. However, the actual amount of relative weight increase for each subset depends mainly on the subset’s history of relative humidity outside the weighing room. Subsets A–M have been labelled differently in Table 7, since they are from different experiments. However, for all IOM sampling cassettes, the number of subsets is presumably much lower, probably 3–5. For example, as noted above, it was found that except for one cassette, subsets G=I and H=J. Figure 4 depicts how a combination of a histogram with a normal quantile plot reveals the existence of several subsets. After discussions with the two companies selling the two samplers it was evident that for both samplers, the number of manufactured batches was much lower than the found number of subsets. It is thus unknown whether the basis for the differentiation into subsets is anything other than the total (i.e. including pre-experimental) histor-
ies of accommodation to different humidities of different groups of cassettes. The relative weight increases for the subsets of Series 1b, 2, 3, 4a and 4b are shown graphically in Fig. 5. (Series 1a is not shown for better clarity.) It can be seen from the figure that the new (unused and unwashed) IOM cassettes (Series 4b) had not reached equilibrium, even after 7 weeks of accommodation to water humidity. The relative increases in weight of the unused (unwashed) CIS filter holders (Series 5) are much lower than for the IOM sampling cassette after 49 days in the weighing room, and were for the three subsets 0.0080, 0.090 and 0.17%, respectively. See Table 7 for the final data pertaining to three weeks storage. The results are presented graphically in Fig. 6. (Within each subset the ratio of the average to the standard deviation was in the range 12–170.) For the CIS filter holder there is a correlation between the relative weight increase and the initial filter holder weight. For the two subsets of the CIS filter holders with the highest water absorption (5-Y and 5-Z), accommodation was not completed even after 49 days in the weighing room. The numerical values obtained when the data in Fig. 5 are modelled with Eq. (2) are presented in Table 8. 75% of the weight increase has a time constant of 苲26 days, and 25% has a time constant of 苲30 h. The electrical surface conductivity of the three sub-
Table 7. Average and standard deviation of the cassette weight increase after approximately three weeks in the weighing room. Data for all experimental series divided into subsets Series
Comment
Number of cassettes
Average weight increase (‰)
SD (‰)
Cassettes without filtera Subset A Subset B Subset C Cassettes with filterb Subset D Subset E Subset F Cassettes unexposed to dust Subset G Subset H Blanks for cassettes with dustc Subset I Subset J Used cassettes New/unused cassettes Subset K Subset L Subset M
30 4 18 7 30 5 4 20 39 17 22 10 5 5 5 10 5 2 3
2.0 1.7 2.0 2.1 2.1 1.9 2.0 2.1 0.90 0.84 0.95 1.5 1.4 1.6 2.4 2.6 2.2 2.5 2.8
0.11 0.14 0.01 0.01 0.12 0.01 0.03 0.01 0.07 0.07 0.02 0.13 0.10 0.03 0.11 0.38 0.01 0.01 0.01
New/unused filter holder Subset X Subset Y Subset Z
15 6 6 3
0.55 0.081 0.67 1.3
0.46 0.003 0.026 0.011
IOM sampling cassette 1a
1b
2 3 4a 4b
CIS filter holder 5
a
One outlier not included in subsets A–C. One outlier not included in subsets D–F. c One outlier not included in subsets I–J. b
IOM sampler in precision and handleability
249
Fig. 4. Histogram and normal quantile plot of final (after 21 days in weighing room) relative weight increases for Series 1b (30 IOM sampling cassettes with filter). The Y scale of the histogram and the normal quantile plot show relative weight increase. Note that the data seem to consist of three subsets and one outlier.
Fig. 5. Relative weight increase of IOM sampling cassettes as a function of time in weighing room. Series labelling as in Table 7. Error bars indicate 95% confidence limits.
G. Lide´ n and G. Bergman
250
CASSETTE HANDLEABILITY—EXPERIENCE AND DISCUSSION
Fig. 6. Relative weight increase of CIS filter holders as a function of time in weighing room. Series labelling as in Table 7. Curves are based on non-linear regression, Eq. (2). Error bars indicate 95% confidence limits.
sets (5-X, 5-Y and 5-Z) of the CIS filter holder was measured with a resistance meter, and all were found to be conducting. The relative weight increase of the best subset of the CIS filter holder was 30 times lower than the weight increase of the best subset of the IOM cassette. It would therefore be possible to manufacture the IOM sampling cassette in a plastic which absorbs considerably less humidity, but would still be electrically conducting. The standard deviations of both types of new/unused cassettes were high. For the IOM sampling cassette the standard deviation was four times greater for unused cassettes than for used cassettes. It is unknown whether this high value is random, or whether using the cassettes reduces the variation between the subsets. If so, repeated washing of new cassettes before they are taken into use may be important. In order to reduce the high variability among the IOM sampling cassettes two options are available. (1) Terminating the use of plastic cassettes in favour of those in stainless steel. This would require a substantial capital investment as these cassettes are approximately six times more expensive than the plastic cassettes. (2) Ensure that all plastic cassettes used on one occasion (both cassettes through which dust is sampled and corresponding blanks) belong to the same subset. This would require a substantial labour effort as it would take several weeks to determine which cassette specimen belong to each subset.
Transport clip The yellow transport clip manufactured by SKC is used during transport in order to prevent the interior of the cassette from becoming contaminated and to prevent sampled dust from escaping from the cassette. See Fig. 7, left side. It is impossible to remove the clip from the cassette, or to replace it without touching the cassette with one’s fingers. Therefore, gloves should be used when sampling with the IOM sampler. We have found that during transport dust may migrate from the sampling cassette to the interior of the clip. This dust is not weighed, since it may be difficult to detect if it is light in colour, and it may be difficult to get it back into the cassette. This would entail additional work which would probably lead to unknown errors. In order to ensure the integrity of the sampled dust a cover (lid) which fits tightly over the cassette has been manufactured in Deldrin. See Fig. 7, right side. It is possible to write the number of the sample on the cover for simple identification of the cassette. The cover is then weighed together with the cassette and, in this way, all sampled dust is weighed. The cover weighs 苲0.6 g, and typical values of its weight increase after two days in the weighing room is 苲0.15 mg. After two days of accommodation the weight of the cover is constant. The cover has been tested at several workplaces. Before the cover is replaced on the cassette after sampling, loose dust on the outer surface of the cassette (that part which extends outwards from the front of the sampler) must be removed in order not to be covered by the cover. During sampling, dust may be deposited on the outer surface of the inlet tube of the cassette. (This has only been observed for the plastic version.) Deposition occurs mainly on the part of the tube which extends outwards from the front of the sampler, but also occurs closer to the filter end of the tube, since the diameter of the tube is slightly smaller than the opening in the front. Any dust deposited here should be removed before weighing the cassette. This problem has been reduced by the introduction of the conducting version as the difference in diameter between the tube of the cassette and the opening in the front is smaller.
Table 8. Parameters (point estimate and SD) from modelling the relative weight increase of the CIS filter holder with the model in Eq. (2) Series-subset 5-X 5-Y 5-Z
m1 (‰)
t1 (days)
m2 (‰)
t2 (days)
r2
0.076 (0.002) 0.25 (0.021) 0.46 (0.034)
1.2 (0.25) 1.2 (0.20) 1.3 (0.19)
0.76 (0.025) 1.49 (0.041)
26.2 (3.0) 26.1 (2.5)
0.961 0.999 0.999
IOM sampler in precision and handleability
251
Fig. 7. Photograph showing IOM sampling cassette with SKC transport clip mounted (left), IOM sampling cassette with our cassette cover unmounted (middle), and IOM sampling cassette with our cassette cover mounted (right).
Measurement of flow rate Originally there was no adapter for measuring the flow rate through the IOM sampler. After some time however, SKC Inc. introduced a flow adapter. The adapter is pressed tightly against the IOM sampler by a spring, and a gasket makes the seal tight. The spring pulls from the back of the sampler. However, this adapter can only be used when the sampler is not mounted on the worker. With the spring dismounted from the adapter, it must be held in place hard against the front of the IOM sampler with one or two hands, while simultaneously another hand holds a flow meter, and yet another hand, if necessary, corrects the pump flow. Another flow adapter has therefore been constructed in deldrin. This adapter fits tightly onto the inlet tube of the sampling cassette, in a similar fashion to Hetland’s flow adapter for the Personal
Inhalable Dust Spectrometer (Hetland, 1996). At the other end of the flow adapter, a plastic or rubber tube is connected to the flow meter. See Fig. 8. This flow adapter is secured to the IOM sampler without the need for a supporting hand, and the occupational hygienist has both hands free to measure the flow rate and correct it if necessary. This flow adapter has been used at several workplaces. The leak flow between the adapter and the cassette is generally less than 2%, and can be disregarded. CONCLUSIONS
The repeatability of weighing the IOM sampling cassette is very good. However, the cassette absorbs a considerable amount of water vapour, and hence takes a long time to accommodate to the humidity of a weighing room after being used at a workplace. The
Fig. 8. Photograph showing IOM sampling cassette with flow adapter mounted (right) and the unmounted flow adapter only (left).
252
G. Lide´ n and G. Bergman
exponential time constant for humidity accommodation is four days, and in order to obtain equilibrium between air humidity and the cassette, 15–20 days are needed before the cassette may be weighed correctly. The standard deviation in weight increase between the cassettes is considerable, 5–10% of the weight increase caused by an increase in humidity. Despite this large variability, it seems to be possible to obtain a low weighing imprecision by the use of blanks. The cassettes should be stored in the weighing room for one week before and after sampling until they are weighed. In this way, an imprecision of 0.05 mg can be obtained, and hence a limit of quantification of 0.5 mg. Experiments indicate that the conducting plastic of the IOM sampling cassette may be replaced with another type of plastic with similar electrical conductivity, but whose humidity absorption is 30 times lower. The IOM sampling cassettes tested seem to be derived from several subsets with clearly distinguishable relative weight increases as a function of time. However, the cause for the differentiation into subsets was not found. To reduce the imprecision substantially below 0.05 mg, either metal sampling cassettes shall be used, or for plastic sampling cassette, specimen shall not be mixed between the subsets. A cover which is weighed together with the cassette has been designed and tested to ensure the integrity of the sample during transportation. A new flow adapter has been designed and tested for which no hands are needed to keep the flow meter attached to the IOM sampler. Acknowledgements—This project was supported by the Swedish Fund for Working Life Research, project number 95-0186. We wish to thank Anders Lindquist (National Institute for Working Life) for the parts he has manufactured which have allowed us to test whether the IOM sampler may be improved for workplace sampling.
REFERENCES American Conference of Governmental Industrial Hygienists. 1992–1993 Threshold Limit Values for Chemical Substances and Physical Agents and Biological Exposure Indices. Cincinnati (OH): American Conference of Governmental Industrial Hygienists; 1992. Comite´ Europe´ en de Normalisation. EN481 Workplace Atmospheres—Size Fraction Definitions for Measurement of Airborne Particles. Brussels: Comite´ Europe´ en de Normalisation; 1993. Hald A. Statistical Theory with Engineering Applications. New York: John Wiley and Sons, 1952. Hetland S. Flow Adapter in Deldrin for the PIDS. Norway: Norwegian National Institute of Occupational Health; 1996 (personal communication). International Organisation for Standardization. ISO7708 Air Quality—Particle Size Fraction Definitions for HealthRelated Sampling. Geneva: ISO; 1995. International Organisation for Standardization. ISO/CD15767 Workplace Atmospheres—Gravimetric Analysis of Aerosol Collection Media. Geneva: ISO; 1999. Kenny LC, Aitken R, Chalmers C, Fabrie`s JF, Gonzalez-Fernandez E, Kromhout H, Lide´ n G, Mark D, Riediger G, Prodi V. A collaborative European study of personal inhalable aerosol sampler performance. Annals of Occupational Hygiene 1997;41:135–53. Li S-N, Lundgren DA. Weighing accuracy of samples collected by IOM and CIS inhalable samplers. American Industrial Hygiene Association Journal 1999;60:235–6. Lide´ n G. Evaluation of the SKC personal respirable dust sampling cyclone. Applied Occupational and Environmental Hygiene 1993;8:178–90. Mark D. The use of dust-collecting cassettes in dust sampling. Annals of Occupational Hygiene 1990;34:281–91. Mark D, Vincent JH. A new personal sampler for airborne total dust in workplaces. Annals of Occupational Hygiene 1986;30:89–102. Smith JP, Bartley DL, Kennedy ER. Laboratory investigation of the mass stability of sampling cassettes from inhalable aerosol samplers. American Industrial Hygiene Association Journal 1998;59:582–5. Vaughan NP, Milligan BD, Ogden TL. Filter weighing reproducibility and the gravimetric detection limit. Annals of Occupational Hygiene 1989;33:331–7.