Annals of Occupational Hygiene Advance Access originally published online on June 22, 2005
Annals of Occupational Hygiene 2005 49(8):661-671; doi:10.1093/annhyg/mei024
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© 2005 British Occupational Hygiene Society Published by Oxford University Press
Original Article |
Comparison of Direct (X-Ray Diffraction and Infrared Spectrophotometry) and Indirect (Infrared Spectrophotometry) Methods for the Analysis of 
-Quartz in Airborne Dusts
Institut National de Recherche et de Sécurité (INRS), Avenue de Bourgogne, BP27, 54500 Vand
uvre lès Nancy, France
* Author to whom correspondence should be addressed. Tel: +33 (0)3 83502023; fax: +33 (0)3 83502060; e-mail: edmond.kauffer{at}inrs.fr
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODOLOGYRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGEMENTSREFERENCES |
|---|
In this study, the
-quartz contents measured by different analytical techniques (X-ray diffraction, direct method; and infrared spectrophotometry, direct and indirect methods) were compared. The analyses were carried out on filters sampled in an industrial setting by means of a Dorr-Oliver cyclone. To verify the methodology used, filters loaded with pure -quartz were also analysed. By and large, the agreement between the two direct methods was close on average, but on the basis of a comparison of the individual results, considerable differences exist. In absolute value, the mean relative deviation between the two techniques was <25% in only 47.8% of the cases. The results obtained by the indirect method (infrared) were on average 13% lower than the results obtained by the two direct methods with a more important difference (23%) for samples where calcite was identified by X-ray diffraction in comparison with those where it was not (8%). This underestimation, which was not owing to dust losses during preparation, is probably explained by the elimination of organic compounds during dust calcinations or by the transformation of mineral compounds. The indirect method introduces additional sample handling operations with more risk of material loss. When the quantity of calcined material was <0.4 mg, the weighing operations necessary to correct any losses of material resulted in considerable variability. In terms of overall uncertainty, it would be better in this case not to carry out correction and to employ an operating mode favouring the recovery of a maximum of material while accepting a bias of about 5–7%.
Keywords: direct and indirect methods • infrared • quartz • X-ray diffraction
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODOLOGYRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Methods to quantitatively analyse silica by X-ray diffraction or infrared spectrophotometry are widely used in numerous countries. Although they may be different in their detail, direct methods, where the analysis is carried out directly on the sampling filter (HSE, 1987
, 1988, 1994), can be distinguished from indirect methods requiring prior treatment of the filter (HSE, 1984; Heidermanns and Lichtenstein, 1993; NIOSH 1994a,b). X-ray diffraction allows differentiation of the three most widely encountered forms of crystalline silica (quartz, cristobalite and tridymite). Infrared spectrophotometry is generally recognized as being one of the least costly but also the least specific techniques. Colorometric methods are also used. It was, however, noted in a study examining the data of proficiency tests carried out in the United States (Proficiency Analytical Testing Program) that the interlaboratory variability of this method was considerably greater than the two others (Eller et al., 1999), and that there was a positive bias for low loading levels and a negative bias for high loading levels. As the evaluation of the quartz concentration in the atmosphere of workplaces is an essential part of risk assessment, these methods are periodically reviewed in the literature (Madsen et al., 1995; Miles, 1999).
Direct and indirect methods are often in competition in industrial hygiene to measure the airborne concentration of a pollutant in workplaces. In this respect, for the measurement of the airborne concentration of asbestos fibres by transmission electron microscopy, direct methods are used in Great Britain and in the United States (HSL, 1998; NIOSH, 1994c), whereas an indirect method is employed in France (AFNOR, 1996). In this specific case, the indirect method, on account of the dilutions that it allows during sample treatment, permits long-duration sampling to be carried out and thus, a better estimation of the mean concentration. However, the preparation method must be rigorously controlled to avoid any modification of fibre size distribution (Kauffer et al., 1996). With respect to the quantitative analysis of quartz, direct analytical methods keep sample handling operations to a minimum, but as the deposit of dust on the sampling filter in not uniform, the analytical conditions must be rigorously defined. In particular, the standards used to produce the calibration curves must be obtained under the same conditions as the filters undergoing analysis, and if the sampling operations and the analyses are not carried out by the same laboratory, in-depth dialogue between the two parties is necessary to ensure the coherence of practices. Indirect methods are not subject to this constraint. They have the advantage of allowing sample treatment to limit the risk of interference. However, the methodology must be strictly controlled to limit the risk of loss during preparation of the sample.
There are a few comparisons of techniques for airborne crystalline silica analysis in the literature. At the request of NIOSH (National Institute for Occupational Safety and Health), two analytical methods employing X-ray diffraction or infrared spectrophotometry were tested in interlaboratory trials (SRI, 1983). For each method, some 15 laboratories were involved, each receiving 8 filters loaded with quartz, possibly associated with matrices, such as calcite, kaolin, talc, iron oxide or coal dust. For each method, the relative error measured was, in total, close to 20% with contributions almost equal from the intralaboratory and interlaboratory errors. For these trials, the quartz loading varied from 50 to 200 µg depending on the filters. Pickard et al. (1985) also conducted a comparison of methods employing infrared spectrophotometry and X-ray diffraction, both used directly on a series of several hundred filters sampled in different industrial sectors. The agreement between the two techniques was in general close, but the authors emphasize that considerable differences can exist for certain filters. Lorberau et al. (1990) demonstrated a close agreement between the results obtained by X-ray diffraction between the direct methods and the indirect method of NIOSH (1994a). The results were, however, obtained from only one dust containing a high proportion of quartz, which limits the risks of interference. More recently, 30 samples coming from 3 coal mines were analysed in 5 European laboratories on the basis of their usual work method employing either infrared spectrophotometry, sometimes with Fourier transform, or X-ray diffraction (Addison, 1991). The author concludes that although both methods can be used, X-ray diffraction is probably less sensitive to potential interference and to particle size. Additional information is also available from the compilation of the results of the laboratories analysing -quartz in the Workplace Analysis Scheme for Proficiency. In this study Stacey et al. (2003) have shown that there was no significant difference between the average result of the laboratories using direct on-filter infrared analysis and the average result of the laboratories using direct on-filter X-ray diffraction analysis. Finally, the study of Verma et al. (1992), which focussed on the comparison of two sampling and analytical methods, only one of which was direct, demonstrated the difficulty in interpreting the results.
The aim of this study was to compare the quartz content measured by different analytical techniques (X-ray diffraction direct method, and infrared spectrophotometry direct and indirect methods), all the filters analysed having been sampled by means of a Dorr-Oliver cyclone.
|
METHODOLOGY |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODOLOGYRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Samples analysed
Among the filters regularly analysed in the laboratory by X-ray diffraction, 241 were selected for this study, those in which quartz had been detected being favoured. In every case, it was the respirable fraction of the dust that was sampled by means of the Dorr-Oliver cyclone. The sampling sites were concerned with very diverse activities (extraction and mining activities, manufacture of products likely to contain silica, cutting and finishing of building materials, foundries, manufacture of diverse equipment and apparatus, civil engineering and construction, etc.).
In order to verify the methodology used, 74 filters loaded with pure quartz and sampled in the same conditions were analysed.
Analytical methods
For the same filter, three methods were used for the analysis.
- Direct method by X-ray diffraction in accordance with AFNOR Standard NF X 43-296 (AFNOR, 1995a).
- Direct method by infrared spectrophotometry, the principle of which is described in document MDHS 37 (HSE, 1987).
- Indirect method by infrared spectrophotometry in accordance with AFNOR Standard XP X 43-243 (AFNOR, 1998).
The principle is the following.- The filters were calcined at 600°C for 3 h in platinum crucibles. Depending on the calcination residue, one of two analyses was carried out.
- Analysis of the dust collected after filtration on a Metricel® filter (vinyl/acrylic copolymer) with a pore diameter of 0.45 µm when the quantity of ash was <0.5 mg.
- Analysis of the dust in KBr pellets when the quantity of ash was >0.5 mg.
- Analysis of the dust collected after filtration on a Metricel® filter (vinyl/acrylic copolymer) with a pore diameter of 0.45 µm when the quantity of ash was <0.5 mg.
- To take into account any losses during preparation, the quantity of quartz measured was multiplied by a corrective factor R equal to,
where Mc2 – Mc1 represents the mass of ash in the crucible and M2 – M1 represents the mass used to prepare the KBr pellet or the mass deposited on the Metricel® filter after filtration.
- The crucibles were weighed either on a 1/100 mg balance (Sartorius R160D) or on a 1/1000 mg balance (Mettler Toledo Model AX26). No calcination blank was used except for the industrial dust calcination series with a 1/1000 mg balance.
The filters were weighed on a 1/1000 mg balance (Mettler Toledo Model MX5). In this case, as the quantity of material expected was small, three blanks were used when weighing the filters or crucibles.
- The filters were calcined at 600°C for 3 h in platinum crucibles. Depending on the calcination residue, one of two analyses was carried out.
- As an addition to the indirect method, a few analyses were carried out by mixing the dust and KBr directly in the crucible employed for calcination. In this case, no correction of any losses was made (R = 1).
Analytical conditions—X-ray diffraction
The diffractometer used was the Philips model X'Pert-MPD, equipped with a PW 1775 automatic sample changer.
The utilization parameters were
(i) long, fine-focus Cu anode tube with a power of 2.2 kW used at 40 kV, 50 mA, (ii) programmable divergence and anti-scatter slits, (iii) programmable analytical slit: 0.3 mm, (iv) primary Soller slits: 0.04 Rad, (v) secondary graphite monochromator PW 3123/00, (vi) proportional detector PW 3011/10 and (vii) rotating sample system.
The qualitative analysis was carried out on a diffraction spectrum recorded over a broad range (5–75° 2
, 0.03° step size and 1 s count per step).
The quantitative analysis was carried out on X-rays lines 100 (d = 4.26 Å), 101 (d = 3.343 Å), 112 (d = 1.817 Å) and 211 (d = 1.541 Å). Peak integration was carried out on an angular domain that depended on the position of the background noise of each side. For a width of 1.4° 2, the counting time was 700 s. (0.002° step size and 1 s count per step), the background noise being counted for an equivalent time.
Analytical conditions—infrared spectrophotometry
The analysis of the samples was carried out using a MAGNA 560 ESP type Fourier transform spectrophotometer (IRTF) manufactured by Nicolet, controlled by a computer and version 6.0 of the OMNIC acquisition software. The apparatus was equipped with an EVER-GLO type source, a DTGS (deuterated triglycine sulphate) detector and a Ge/CsI (germanium/cesium iodide) separator. This equipment allowed the acquisition of spectrums in the medium and far infrared.
Each spectral acquisition in absorbance mode was conducted with the following parameters:
(i) number of scans: 32, (ii) resolution: 4 cm–1, (iii) Apodization: Happ-Genzel, and (iv) phase correction: Mertz.
The intensity of the spectral bands was measured at a wave number of 798 and 780 cm–1 ± 0.001 of absorbance, using for each band a baseline of between 819 and 788 cm–1. An analogue procedure was applied to the bands located at 695 cm–1 (baseline between 721 and 672 cm–1), at 398 cm–1 (baseline between 418 and 388 cm–1) and at 374 cm–1 (baseline between 388 and 353 cm–1).
Calibration
For the direct methods, the calibration curves were obtained from quartz sampled in dust generation chambers. The respirable fraction of the dust was sampled with a Dorr-Oliver cyclone. For the analyses carried out by infrared spectrophotometry, indirect method, the standard preparations of KBr or the deposits on the Metricel® filter were produced from a sample of quartz of respirable size. This sample was obtained by sampling the quartz on polycarbonate filters, selection of the respirable fraction again being ensured by the Dorr-Oliver cyclone. The quartz deposit was then recovered in water by shaking the filters then by evaporation. This approach guaranteed that all the calibrations were produced from the same size fraction. This is particularly important insofar as it is well known that the response, both in X-ray diffraction and in infrared spectrophotometry, is highly dependent on the size distribution of the quartz dust (Bhaskar et al., 1994).
In every case, the quartz used for the calibrations was the reference QUIN1, supplied in the 1970s by the company Moulin des Prés. This quartz was in a previous study compared with the reference sample of quartz- from the National Institute of Standards and Technology (Kauffer et al., 2002). But as the aim of the present study was to compare the analytical techniques and that the same standard (QUIN1) was used throughout for the different calibrations, no correction for crystallinity was made.
Interferences and presentation of results
For the analyses carried out on the filters sampled in various industrial settings, the main source of error is the presence of interference. Close overlap of the diffraction peaks or infrared absorption bands from quartz and the interfering phase will lead to overestimation of the amount of quartz present. Partial overlap of more widely separated peaks or bands may lead to overestimation of the background and consequently to underestimation of the amount of quartz present. For the analyses carried out by infrared spectrophotometry, the general rule for reporting results was to choose the mean of the absorption bands figures a priori not subject to interference, whereas for the analyses carried out by X-ray diffraction, a peak was chosen from among those a priori not subject to interference, favouring, whenever possible, the most intense quartz diffraction peak.
Calculation of the uncertainty of factor R
In the indirect method, when the quantity of calcined material was <0.5 mg, the ash was deposited by liquid filtration on a Metricel® filter (see Analytical methods section). In this case, to take into account any losses during preparation, the results of the analysis were corrected by a factor R equal to the ratio of the quantity of dust in the crucible (Mc) after calcination to the quantity of dust on the filter after filtration (Mf). Three calcination and filtration blanks were used for each series.
To characterize the variations in mass of the blank crucibles or filters, it was possible, by following Standard NF ISO 15767, to calculate standard deviations 
c and f (AFNOR, 2004). These are the square roots of the mean of the variances associated with the variations in mass of the different series of blank crucibles or filters. The variances 
and 
associated with the determination of Mc and Mf are then expressed by
 |
 |
If it is assumed that the variations in mass of the blank crucibles and filters are not correlated, which is the case for our data (R2 = 0.0795), the variance of ratio R is expressed by
 |
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODOLOGYRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Table 1 gives, for the different methods studied, the number of filters analysed, both for the filters sampled in industry and for those loaded only with quartz dust. For the filters sampled in an industrial setting, the mean, minimum and maximum mass of material collected on the sampling filter are also indicated for each series. For the preparations produced by deposit on the Metricel® filter, the quantity of ash present in the crucible was always <0.5 mg; however, during the analyses carried out by mixing the dust in KBr, the mass of ash in the crucible was not always >0.5 mg. This is, in part, owing to the difficulty in obtaining highly-loaded filters when the respirable fraction of dust was sampled by means of the Dorr-Oliver cyclone (flow rate 1.7 l min–1), and also because of the wish to explore the method on lower quantities of dust. Higher quantities of material could have been obtained by taking the samples with the CIP 10 apparatus (flow rate 10 l min–1), the use of which is foreseen in Standard XP X43-243 (AFNOR, 1995b
), but in this case, the dust being collected on foam, direct analysis would not have been possible. For the filters loaded with quartz dust, the choice of the indirect method was made independent of the quantity of ash present in the crucible.
|
Figures 1
–4 present the results obtained with the direct methods. To avoid favouring either of the two techniques (X-ray diffraction or infrared spectrophotometry), the quantity of quartz measured by each technique was compared with the mean of the results obtained by each of them (Mdir). It is this value that was always reported on the x-axis. The results obtained by X-ray diffraction (Figs 1 and 2) or by infrared spectrophotometry (Figs 3 and 4) were reported on the y-axis. The dust deposited on the filters corresponds to the samples taken in industrial settings for Figs 1 and 3, and to the quartz dust samples for Figs 2 and 4.
|
|
|
|
The quantity of quartz measured in infrared spectrophotometry by the indirect method as a function of Mdir is represented in Figs 5
–14. For Figs 5 and 6, the deposit of the calcination ash was done on a Metricel® filter. For Figs 7–12, the ash was mixed with KBr to carry out the analysis. Measures were taken to correct possible losses during preparation (Figs 7 –10), the crucibles being weighed on a 1/1000 mg balance (Figs 7 and 8) or on a 1/100 mg balance (Figs 9 and 10). For Figs 11 and 12, no measure was taken to correct possible losses. Figures 13 and 14 group all the results obtained with the indirect methods, with the exception of those for which no correction was made. The dust deposited on the filters corresponds to the samples taken in industrial settings for Figs 5, 7, 9, 11 and 13, and to quartz dust samples for Figs 6, 8, 10, 12 and 14. Table 2 gives the parameters of the regression lines for all the figures presented.
|
|
|
|
|
|
|
|
|
|
|
In the specific case where the dust to be analysed was deposited on a Metricel® filter, Table 3 (for industrial dust) gives the parameters of the regression lines linking the quantity of quartz measured by the indirect method to the mean of the results obtained with the direct methods. The calculations were made by correcting possible losses during preparation, but also by making no correction (R = 1). The bias between the two calculation methods (with and without correction) is also indicated as a function of the mass of calcination residue.
|
The values of
Mc and Mf determined from 16 calcination series and 11 filtration series, respectively are indicated in Table 4. This table also shows the uncertainty R of the determination of correction factor R. For this calculation, it was assumed that masses Mc and Mf were identical (case where possible losses during preparation were low).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODOLOGYRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGEMENTSREFERENCES |
|---|
X-ray diffraction and infrared direct methods
On the whole, the agreement between the two direct methods is close, as Figs 1 and 3 show. If the mean of the two methods is taken as references, a slight overestimation of
2% can however be observed for the analyses carried out using X-ray diffraction. An underestimation of equivalent magnitude can also be noted for the analyses carried out using infrared spectrophotometry. Although the difference is small, it is significant, as the confidence of the slopes shows (Table 2). It is possible that the difference is owing to different practices of the analysts in charge of the two techniques with respect to the choice of the result. Indeed (see Interferences and presentation of Results section), in one case (X-ray diffraction), the result rendered corresponds to the choice of one diffraction peak, whereas in the other (infrared) it is the mean of the results of absorption bands not subject to interference which led to the result. No such difference is observed for the pure quartz samples (Figs 2 and 4 and Table 2), which strengthens the view that different practice by the two analysts may be the reason for the observed difference between X-ray diffraction and infrared spectrophotometry for industrial samples. These results are in agreement with those published by Pickard et al. (1985), which focussed on the analysis of air samples sampled in a foundry or during the production of bricks and tiles or pottery. However, if the results are compared on an individual basis, considerable differences can exist between the two techniques. The absolute value of the differences between the results obtained by X-ray diffraction and by infrared divided by the mean of the results obtained by the two techniques and expressed as percentage are equal to 41.3, 23.5 and 18.9%, respectively for quartz content of <0.05 mg, between 0.05 and 0.1 mg or >0.1 mg. When the quartz content is <0.05 mg the mean relative deviation between the two techniques is <25% in only 47.8% of the cases. It is clear that from this point of view the situation can only deteriorate in the future; with an equal sampling time and at a constant flow rate, owing to the improvement of the occupational hygiene conditions in the workplaces, the analyst has less and less material to carry out the analysis. In this respect, for the filters received by the laboratory, the median quantity of dust dropped from 0.4 to 0.2 mg over the period 1986–1998 (samples of an average duration of 310 min carried out at a flow rate of 1.7 l min–1 with the Dorr-Oliver cyclone).
Direct and indirect infrared methods
Figures 5 and 6 present the results obtained with the indirect method (case where, after calcination, the dust was transferred to a Metricel® filter) compared with the mean of the results obtained with the two direct methods. An underestimation of the results obtained for the analyses carried out on the industrial dust can be noted (Fig. 5), this underestimation not being observed for the analyses carried out on the quartz dust (Fig. 6). The slope of the regression line is significantly different from unity for the analysis of the dust sampled in industrial settings, which is not the case when the dust analysed was quartz dust (Table 2).
When the analysis of the dust recovered after calcination was carried out on a KBr pellet, an underestimation of the quartz concentration was also noted compared with the mean of the results of the two direct methods for the analyses carried out on the industrial dust (Figs 7 and 9). This underestimation was not observed for the analysis of the quartz dust (Figs 8 and 10). These observations were confirmed by analysis of the slopes of the regression lines (Table 2). According to Standard XP X43-243 (AFNOR, 1998), a correction must be made to take account of the fact that the quantity of dust mixed with KBr to produce the pellet intended for analysis can be appreciably less than the calcination residue determined by weighing the crucibles. It appears that the use of a 1/1000 mg balance for these weighing operations considerably reduces the dispersion of the points around the regression line (Figs 7 and 8) compared with the use of a 1/100 mg balance (Figs 9 and 10). However, taking into account blank crucibles (data not presented in Table 2) only marginally changes the results (variation in the slope of the regression line of 0.908–0.902), which indicates that the correction is negligible compared with the mass weighed.
For one series of filters, previously calcined dust and KBr were mixed directly in the calcination crucible in order to minimize losses. It was then no longer possible to make a correction. The results are presented in Figs 11 (industrial dust) and 12 (quartz dust), and in Table 2. It can be seen that these results are lower than those obtained when a correction was made. The slopes of the regression lines passing through the origin vary from 0.872 to 0.804 for the industrial dust and from 1.046 to 0.978 for the quartz dust (Table 2). This approach, therefore, introduces a bias of 7%.
Let us now consider all the results obtained with the indirect methods, with the exception of those for which no correction was made (mixing the dust and KBr directly in the crucible). An underestimation of 13% can be noted compared with the mean of the results obtained by the two direct methods studied (Fig. 13 and Table 2). As such, an underestimation was not observed for the analyses carried out on the quartz dust (Fig. 14 and Table 2); it is probable that this underestimation is not the result of any loss of material during preparation. Calcination certainly allows the elimination of interference (owing to the disappearance of organic compound dust or to the transformation of mineral compounds) that would not be identified on account of the low quantities of dust generally available for analysis, but can also in certain cases and probably to a certain degree introduce new interference as a result of the transformation of certain minerals. Moreover, as mentioned in method 7500 of NIOSH (NIOSH, 1994a), if the samples contain a significant amount of calcite (>20% of total dust loading), silica may be lost owing to formation of CaSiO3. This may have been the case in our study as the underestimation is more important (23%) for sample where calcite was identified by X-ray diffraction than for samples where it was not (8%). Removal of calcite may be done by an acid test as described in method 7500 of NIOSH. This was not done in the present study as the standard XP X 43-243 (AFNOR, 1998) does not mention this possibility. To appreciate the real interest of the indirect method compared with the direct methods, it would be necessary to carry out comparisons on synthetic mixtures including various matrices, the quartz content of which is therefore known. But, in this case to avoid differences, following cyclone elutriation, in the proportion of quartz in the initial mixture compared with their proportion in the collected dust, filters should probably be prepared by liquid filtration. Figure 15 shows the relationship for the dust sampled in industry between the mass of dust in the crucible after calcination and the mass of material on the sampling filter. It can be seen that calcination reduces the initial quantity of material by 30% on average. This of course could have an impact on the dust analysis.
|
There are a few references in the literature where direct and indirect methods have been compared for the quantitative analysis of quartz.
In the work of Loberau et al. (1990), two direct methods to determine the airborne concentration of quartz by X-ray diffraction were compared with the NIOSH 7500 method (1994a). For the two direct methods, the sampling filters were silver filters or mixed cellulose ester filters, whereas PVC filters were used for the samples intended to be analysed with the 7500 method. In the latter case, as it is an indirect method, the dust was recovered after dissolving the PVC filter in tetrahydrofurane. Overall, the agreement between the two direct methods and the indirect method was close. In 95 of 100 cases, the differences between the direct and indirect methods were <25% when using the mixed cellulose ester filters and 30% when using the silver filters. Considerable differences do however exist between that study and the present work. In the former, the quantity of quartz to be analysed was between 0.2 and 1.2 mg, which is considerable compared to our domain of exploration (0–0.2 mg). In addition, the dust, which came from a mine in Colorado, was identical in nature for all the filters, and quartz was also the majority constituent of this dust (59%). It is clear that high loading reduces the variability of the analyses and that the dominant character of the presence of quartz limits the risk of interference.
The study of Verma et al. (1992) consisted in comparing the results obtained by a direct method (analysis on silver filter by X-ray diffraction) with those obtained by NIOSH method 7602 (1994b) (analysis of calcined dust by infrared spectrophotometry). Here again, the nature of the dust sampled was quite homogeneous as all the samples came from the same gold mine. The authors did, however, find that the two methods (direct compared with indirect) had a positive bias of 30% which could not be explained entirely either by the difference in the efficiency of the filtering media used or by the difference in the standards. One possible problem linked to the presence of interference was raised.
As explained in the ‘Analytical methods’ section, AFNOR Standard XP X43-243 (AFNOR, 1998) explicitly foresees correcting any losses during preparation by multiplying the quantity of quartz measured on the Metricel® filter by the ratio of the quantity of calcined dust to the quantity of filtered dust. The lower the calcination residue, the higher the relative uncertainty introduced by this correction. For calcination residues between 0.5 and 0.05 mg, it varies from 1.9 to 19.2% (Table 4). This correction is intended to remove any bias introduced by the preparation method. Indeed, if the calculations are done without making this correction, a bias appears. The comparison of the slopes of the regression lines (Table 3) shows that this is 5%. This bias does not seem to depend on the value of the calcination residue for quantities of dust of between 0.070 and 0.437 mg (Table 3). The overall uncertainty, taking into account errors introduced during preparation, is equal to the sum of the bias and the extended relative uncertainty expressed in percentage.
 |
Correction of any losses linked to preparation removes the bias but introduces a variability into the result; the lower the calcination residue the higher the variability. The absence of correction removes the variability introduced by the different weighing operations but introduces a bias. In terms of overall uncertainty, Table 4 shows that correction is interesting only when the calcination residues are greater than 0.4 mg. For this value, the extended relative uncertainty (2 x 0.024 x 100 = 4.8%) becomes slightly lower than the bias (5%).
The same reasoning applies when the analysis is carried out on dust in pellet form in KBr. Correction (ratio of the calcination residue to the quantity of dust used to make the KBr pellets only reduces the overall uncertainty for calcination residues of >0.4 mg. For lower values, it is better to make the mixture of dust and KBr directly in the crucible, but we have seen that a bias of 7% must then be accepted.
|
CONCLUSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODOLOGYRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGEMENTSREFERENCES |
|---|
The present study has not only allowed the comparison of two direct methods (analysis by X-ray diffraction and infrared spectrophotometry), but also the comparison of these methods with an indirect method (analysis by infrared spectrophotometry after calcination of the sampling filter).
By and large, the agreement between the two direct methods is on average close, but on the basis of an individual comparison of results considerable differences do exist. This high variability is associated with the low quantities of material available for the analyses. To overcome this, either the flow rate of the selector used to sample the respirable fraction of the dust should be increased considerably or much longer sampling times should be envisaged.
We have also shown that the results obtained with the indirect method were on average lower by 13% compared with the mean results obtained with the two direct methods. These differences are probably explained by the elimination of organic compounds or by the transformation of mineral matrices during calcination of the dust. To compare the accuracy of direct and indirect methods, it would be necessary to study synthetic samples with a known quartz content.
The indirect method introduces additional sample handling operations with the higher risk of loss of material. When the quantity of calcined material is less than 0.4 mg, the weighing operations necessary to correct any losses of material result in considerable variability. In this case, correction is best not carried out, and an operating mode favouring the recovery of a maximum of dust while accepting a bias of about 5–7%.
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODOLOGYRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGEMENTSREFERENCES |
|---|
The authors would like to thank Mrs Pierre for the care given to the execution of the weighing operations.
Received February 21, 2005; in final form May 4, 2005
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODOLOGYRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Addison J. (1991) Improvements of analysis of mineral components of coalmine dusts. Technical memorandum series. IOM report TM/91/10. pp. 1–49.
AFNOR (1995a) Norme NF X 43-296 Détermination par rayons X de la fraction conventionnelle alvéolaire de la silice cristalline—Echantillonnage sur membrane filtrante. 1er tirage 95-06 AFNOR 1995.
AFNOR (1995b) Norme NF X 43-295 Détermination par rayons X de la concentration de dépôt alvéolaire de silice cristalline—Echantillonnage par dispositif à coupelle rotative. 1er tirage 95-06 AFNOR 1995.
AFNOR (1996) Norme NF X 43-050 Détermination de la concentration en fibres d'amiante par microscopie électronique à transmission—Méthode indirecte. 1er tirage 96-01 AFNOR 1996.
AFNOR (1998) Norme XP X 43-243 Dosage par spectrométrie infra range à transformée de Fourier de la silice cristalline—Echantillonnage par dispositif à coupelle tournante ou sur membrane filtrante. 1er tirage 98-01 AFNOR 1998.
AFNOR (2004) Norme NF ISO 15767 Contrôle et caractérisation des erreurs de pesée des aérosols collectés. 1er tirage Mars 2004. AFNOR 2004.
Bhaskar GW, Jianliang L, Ligun XA (1994) comparative study of particle size dependency of IR and XRD methods for quartz analysis. Am Ind Hyg Assoc J; 55: 605–9.[Web of Science][Medline]
Eller PM, Amy FH, Song RS et al. (1999) Proficiency Analytical Testing (PAT) silica variability, 1990–1998. AIHA Journal; 60: 533–9.[CrossRef]
Heidermanns G, Lichtenstein N. (1993) März 1992 Analytishe Methoden Luftanalysen. Meth. Nr.1 Quarz. Lieferung, Ausgabe 1993. pp. 1–15.
HSE. (1984) Quartz in respiable airborne dusts. Laboratory method using infrared spectroscopy (KBr disc technique). MDHS 38. Health and Safety Executive.
HSE. (1987) Quartz in respirable airborne dusts. Laboratory method using infrared spectroscopy (Direct Method). MDHS 37. Health and Safety Executive.
HSE. (1988) Quartz in respirable airborne dusts. Laboratory method using X-ray diffraction (Direct Method). MDHS 51/2. Health and Safety Executive.
HSE. (1994) Cristobalite in respirable airborne dusts. Laboratory method using X-ray diffraction (Direct Method). MDHS 76. Health and Safety Executive.
HSL. (1998) Fibres in air. MDHS 87. Health and Safety Laboratory.
Kauffer E, Billon-Galland MA, Vigneron JC et al. (1996) Effect of preparation methods on the assessment of airborne concentrations of asbestos fibres by transmission electron microscopy. Ann Occup Hyg; 40: 321–30.
Kauffer E, Moulut JC, Masson A et al. (2002) Comparison by X-ray diffraction and infrared spectroscopy of two samples of alpha-quartz with the NIST SRM 1878a -quartz. Ann Occup Hyg; 46: 409–21.
Lorberau CD, Carsey TP, Fischbach TJ et al. (1990) Evaluation of direct-on filter methods for the determination of respirable quartz. Appl Occup Environ Hyg; 5: 27–35.
Madsen FA, Rose MC, Cee R. (1995) Review of quartz analytical methodologies: present and future needs. Appl Occup Environ Hyg; 10: 991–1002.
Miles WJ. (1999) Issues and controversy: the measurement of crystalline silica; review papers on analytical methods AIHA Journal; 60: 396–402.
NIOSH. (1994a) Silica, crystalline, by XRD. Method 7500. National Institute for Safety and Health.
NIOSH. (1994b) Silica, crystalline, by IR. Method 7602. National Institute for Safety and Health.
NIOSH. (1994c) Asbestos by TEM. Method 7402. National Institute for Safety and Health.
Pickard KJ, Walker RF, West NG. (1985) A comparison of X-ray diffraction and infrared spectrophotometric methods for the analysis of quartz in airborne dusts. Ann Occup Hyg; 29: 149–87.
SRI International. (1983) Collaborative tests of two methods for determining free silica in airborne dust. Fev 83. Menlo Park, CA: SRI International. pp. 1–156.
Stacey P, Tylee B, Bard D et al. (2003) The performance of laboratories analysing -quartz in the Workplace Analysis Scheme for Proficiency (WASP). Ann Occup Hyg; 47: 269–77.
Verma DK, Sebestyen A, Julian JA et al. (1992) A comparison of two methods of sampling and analysis for free crystalline silica -quartz. Appl Occup Environ Hyg; 7: 758–63.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  P. Stacey, E. Kauffer, J.-C. Moulut, C. Dion, M. Beauparlant, P. Fernandez, R. Key-Schwartz, B. Friede, and D. Wake An International Comparison of the Crystallinity of Calibration Materials for the Analysis of Respirable {alpha}-Quartz Using X-Ray Diffraction and a Comparison with Results from the Infrared KBr Disc Method Ann. Hyg., August 1, 2009; 53(6): 639 - 649. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||