|<< Back to Issues and Controversy: Measurement
of Crystalline Silica
Disclaimer: This is not DOL or OSHA controlled material and is provided here for reference only. We take no responsibility for the views, content, or accuracy of this information.
The purpose of this section is to consider the principle steps of the active standard analytical procedures emphasizing the function rather than the mechanics of the procedure. Each procedure will be examined; then they will be compared. Bulk sample treatment will be considered first. There are no certified procedures for bulk analysis of crystalline silica minerals, but there are many usable general procedures described in the literature.
The Analysis of Bulk Samples
The analysis of bulk samples for crystalline silica phases is essentially an extension of the usual quantitative procedures reported in the literature. There is no certified procedure established by any of the federal agencies because the samples to be analyzed have considerable variability in the associated phases, although a very elaborate procedure for analyzing soils which include quartz has been presented by Raab (1988). The NIOSH filter method has been used to analyze bulk materials, but it is not really suitable for the low concentrations as required by OSHA (0.1% quartz) (Blount, 1989) because the sample is not large enough to allow the detection and quantification of this amount. It should also be remembered that crystallite statistics strongly favor the bulk sample because of the larger number of crystallites in the sample.
An excellent review of quantitative procedures has been presented by Snyder and Bish (1989) and a thesis by Cline (1986). The references in these articles cover most of the available literature, and the paper discusses all the main techniques in use today. As mentioned above, there are essentially three methods which are in use: the absorption correction method, the internal standard method and the external standard method. There are also three ways to acquire and process the data in each method. Where applicable, individual peaks may be integrated for each phase, Alexander and Klug (1948). If individual peaks are not resolvable, then clusters of peaks may be used and the contribution of each phase to each cluster may be treated as a matrix, Copeland and Bragg (1958), Karlak and Burnett (1966), Chung (1974a and b, 1975), and Smith, et al. (1984). If the overlap is severe, the whole-pattern methods may be used, Wiles and Young (1981), Smith et al. (1987, 1988a and b), Hill and Howard (1987), Bish and Howard (1988), Madsen and Hill (1988), O'Connor and Rowan (1988) and Taylor (1991).
There are several methods mentioned in the literature that specifically involve the determination of quartz in bulk samples. The first description is by Clark and Reynolds (1936), and Klug et al. (1948) first apply modern theory to silica analysis. The reviews by McGlynn (1968) and Anderson (1975) cover much of the earlier literature. More recent studies will be indicated here. The analysis of quartz and other minerals in rocks is discussed by Pawloski (1985) and Bayliss (1986). Both use the external reference-intensity method. Fluerence et al. (1969) analyze raw materials and ceramics. Waersted (1986) has used the RIR method to determine quartz in calcium sulfate preparations. Kamarchik (1980) describes the determination of quartz, talc and clay in paints using the RIR method with TiO2 as the reference material. Davis et al. (1986) has analyzed coal samples using the RIR method coupled with absorption measurements to allow for the fact that not all the phases are determined and the coal component is amorphous. Emig and Smith (1989) have quantified quartz in agricultural dolostone using direct calibration methods because of the minimal interference from the other minerals in the rock. Corrections were required when mica was present. Carter et al. (1987) have quantified both quartz and cristobalite in bentonite products using an internal standard method with absorption corrections to allow for the presence of other phases. McKee et al. (19__) and Hamilton and Peletis (1991) determined quartz in perlite using spiking with quartz to prepare the calibration curves.
Whole pattern analysis has also been used for some quartz analyses, and as this technique becomes better known, it will be employed more frequently. Smith et al. (1989) used whole pattern fitting to analyze quartz and other minerals in sedimentary rocks. Jordan et al. (1990) applied the Rietveld method to analyzing quartz specimens for the crystalline fraction. This study applies specifically to the question of the presence of amorphous silica in quartz samples.
It is evident that there are many procedures which could be developed for the determination of quartz in bulk samples. An extension of the above techniques is to employ additional external information as constraints on the X-ray analysis. Renault (1987), Goehner (1982) and Garbauskas and Goehner (1982) and Smith et al. (1989) have used chemistry as a constraint on the reduction of the X-ray data. This addition leads to better reconciliation of the physical and chemical data which often causes doubts on the validity of the X-ray results when the two do not agree.
Quartz Analyses using Membrane Filters
NIOSH Method 7500, OSHA/SLC ID-142 July 1989 This method is designed to quantify quartz and cristobalite collected on membrane filters by air samplers or aerosol chambers. It is primarily used for monitoring crystalline silica in atmospheric dust but has been applied to bulk samples as well. Specific collecting conditions and apparatus are specified for the atmospheric samplers. The particles are collected on 37 mm diameter polyvinylchloride, PVC, membrane filters which are then transferred to the analytical laboratory. The PVC filters are then dissolved in tetrahydrofuran, THF, to release the particles which are then transferred to a 25 mm silver membrane filter. This step concentrates the particles to a smaller area and distributes them as an evenly-spaced particle layer for the X-ray diffractometer. This filter and its load are then transferred to the diffractometer for measurement. The desired amount of sample is 2 mg or less. The basis for this analysis have been discussed by Dollberg et al. (1980) and Abell et al. (1981a).
The X-ray measurement involves the integration of specific peaks for quartz, cristobalite and silver. The process may be run in automated mode (Abell et al., 1978) both to collect the data from each sample following specific steps and interrogate the information for deciding how to continue with the sample. This program will also process many samples in sequence. The silver peak is scanned to determine the intensity attenuation compared to a clean filter. The attenuation effect allows the thickness of the sample layer to be calculated, and if the thickness is acceptable, at least three peaks for quartz (and/or cristobalite) are scanned. The strongest peak from quartz is measured first to set the count time for all the peaks.
The integrated intensities are converted to weights of quartz or cristobalite using calibration curves from prepared filters with weighed amounts of silica deposited. Standard samples are interspersed with the analytical samples to monitor instrument drift. The weight fractions are determined from the total weight of the sample determined from pre- and post-weighings of the filter during preparation. The detection limits are reported to be 5 æg quartz and 10 µg (*) cristobalite for qualitative determinations and 10 æg quartz and 30 µg (*) cristobalite for quantitative determinations.
Bulk samples are analyzed by crushing to pass 325 mesh and then taking an aliquot for further processing. When the size range is right, the material is transferred to the silver filter in the same manner as for the calibration procedures. It is now treated in the standard fashion.
HSE Method MDHS 5 1/2, March 1988 This method is reported only for the determination of quartz in airborne dusts, but the principles could apply to the determination of cristobalite with appropriate changes. The method uses the 25 mm diameter PVC or PVC-acrylonitrile copolymer filter in the collection device which is transferred directly to the X-ray diffractometer. Sampling conditions are prescribed. The amount of sample is determined by pre- and post-weighings of the filter. Dust loadings should be kept below 2 mg to prevent loss of dust from the surface of the filter cake during handling and to keep the intensity response versus weight of quartz in the linear range. If the sample is too large, or if there are interfering phases present, an absorption correction may be necessary according to equation 1. The four principle peaks of quartz are monitored depending on the interfering phases which are present as determined by a fast diffractometer scan.
Calibration is performed by preparing standard samples in an aerosol chamber by creating a cloud of airborne quartz dust with an air jet then letting the large particles settle. Four samplers are attached to the cloud chamber. The suspended particles are then drawn through the samplers for prescribed times. The amount of quartz deposited is determined by weight. Progressively longer sampling times create a series of calibration samples. These samples are then measured in the same way that the analytical samples are to be measured. The diffractometer system is monitored by a drift sample such as aluminum.
Dow Chemical Company This method (Henslee and Guerra, 1977) uses PVC filters both to collect the sample and to support it in the X-ray diffractometer. A 25 mm diameter orifice (Gebhardt, 1975) is used in the sampler to restrict the effective part of the filter, and the effective 25 mm circle is cut from the 37 mm filter and mounted on a circular holder that fits the diffractometer spinner. This approach eliminates the need to transfer the sample.
Calibration samples are prepared by crushing quartz to pass a 400 mesh sieve and then settling the particles in ethanol to separate the larger particles and create a sample with a mean size of 5µm. A weighed amount of this sample was dispersed in water and then diluted to a fixed concentration. Aliquots of this suspension were then diluted further to specific concentrations and the suspension then filtered through a PVC filter. The actual amount deposited was determined by pre- and post-deposition weights. The intensity calibration curve was linear over the range 0 - 100 µg/ml. Addition of 20% iron oxides did not affect the magnitude or linearity of the intensity response per unit weight of quartz showing that the iron oxide did not mask the quartz.
U. S. Bureau of Mines, IR-1021, 1975 This method (Freedman, 1972; Thatcher, 1975) collects the dust on 37 mm PVC filters and then transfers the particles to silver membranes by a procedure very close to the NIOSH method. A 1.5% solution of parlodion in amyl acetate may be used to cement heavy dust deposits to the silver filter. The attenuation of the silver peak is used to determine the sample absorption and correct the intensity measurement.
NIOSH Method 7501 for Amorphous Silica As mentioned above, the determination of the amorphous content of a sample is not possible without special processing. Method 7501 uses the property that most amorphous forms of silica will convert to cristobalite with heat treatment. The X-ray analysis steps are similar to Method 7500, but the sample is ashed and fired to 1100 or 1500°C depending on the type of amorphous silica suspected prior to final transfer to the silver filter. The sample is then analyzed for cristobalite. The method depends on the quantitative conversion of the amorphous form to cristobalite without reacting with anything else in the sample.
Other Reported Methods There are several other reports of the use of filter methods for the quantification of quartz. Most of these methods describe the use of the filter method for bulk analysis or analysis where sufficient dust is available for direct processing. Some allow the addition of an internal standard by co-dispersing the phases before depositing on the filter. The methods include reports by Murray and Merkl (19__), Malik and Viswanathan (19__), Clayton Environmental Consultants (19__), Bumsted et al. (19__), and Davis and Johnson (1982a and b). Other methods are reported in papers by Schliephake (1963), Schmelzer (1951, 1955), Plowman (1978), O'Connor and Joklevic (1981) and Kudo (1982).
Several authors have been specifically interested in the determination of cristobalite. The NIOSH Method 7500 does include cristobalite in the procedure and most of the other filter methods could be easily modified to include cristobalite. Kupka (1967), and Stefanov (1972) describe methods to quantify cristobalite in rocks. Carter et al. (1980) quantifies cristobalite in bentonite clays. Janko et al. (1989) determine the cristobalite content of dusts in a ceramic plant where it occurs in mullite products and in the atmospheres in the plant.
The determination of cristobalite is not as straightforward as the determination of quartz because its structure is more variable than quartz. The true structure is based on a large unit cell which is a supercell of the simple cubic arrangement usually used to describe cristobalite. The structure lends itself to several types of disorder including distortions which create small domains and stacking faults which lead to some tridymite character. The effect of these defects is to cause the peaks to broaden and shift and change in intensity. Usually the intensity change of the principal peak (101) is less affected than the weak peaks, so quantification is little affected when the peak is integrated by the usual techniques.
Comparisons of Filter Methods
Selection of the Filter Material There is considerable controversy concerning the most appropriate filter for collecting and supporting the dust samples. The organic filters allow the recovery of the particles where the procedure calls for the reconcentration of the particles because the filter is soluble in tetrahydrofuran or other solvent. Where the filters are large, 37 mm, the particles usually require processing to redistribute the particles over a smaller area for effective coverage by the X-ray beam. PVC, polyvinylchloride; PVCA, polyvinylchloride acrylonitrile; Nucleopore polycarbonate, PC; and MCE, mixed cellulose ester have been used for filters. These filters may also be used for direct analysis. MCE filters has the advantage in the X-ray beam of being of low absorption and low scattered background, thus allowing weak diffracted peaks from low particle loadings to be detected. The silver filters and the MCE filters are reported to produce a lower background in the diffraction pattern than the PVC filters. Evaluations of filters have been done by Mark (1974), Henslee and Guerra (1977), Altree-Williams et al. (1977), Chung (1978), Dobreva et al. (1982), Davis and Johnson (1982a and b), Foster and Walker (1984) and Knight (1984).
The silver filters are used both for collection when the sample is to be analyzed directly and as the substrate when the particles are transferred from an organic filter. One advantage of the silver filters is that they hold the particles better than the organic filters because of their structure, but particles are drawn into the interior of the filter where they are masked from the X-ray beam by the surrounding silver. Thus, a correction factor must be applied for this masking. Another advantage is the crystallinity of the silver which allows the intensity of its diffraction peak to be monitored. If the sample is thicker than one particle layer, the attenuation of the silver intensity provides the sample absorption coefficient. Where the PVC or MCE filters are used for direct analysis, they can be placed over a crystalline support like a silver, aluminum or zinc sample holder to provide the diffracted beam needed to measure the sample attenuation. Greases may be used to hold the filters in place, but the grease may contribute to the diffraction pattern and to the attenuation of the substrate peak intensity.
Direct-on-Filter Methods versus Transfer Methods Several laboratories have preferred to use direct-on-filter methods because of the convenience of simple processing, more rapid analysis and a concommitment improvement in the costs involved. The standard NIOSH Method 7500 is very time consuming, but it is considered to be the accepted legal technique, so all other methods must be compared with it. A very recent evaluation of the direct-on-filter methods was performed by Lorberau et al. (19__). The results of this evaluation showed that the agreement between the NIOSH Method 7500 and the MCE filter method agreed well, but that there was a slight bias in the comparison of the NIOSH method and the silver filter. This bias was thought to be due to a mismatch in particle size between the samples tested, but it might also be due to masking by the silver in the filter.
One of the main drawbacks of the use of direct-on-filter analysis is that the original distribution of particles is spread over the full 37 mm diameter filter. This fact is the primary reason for the concentration step incorporated in the transfer methods. Bradley (1967), Leroux and Powers (1969a and b), Crosby and Hamer (1971), Knight et al. (1971, 1972), Leroux et al. (1972, 1973), Knight (1975, 1986), Altree-Williams et al. (1977), Frevel and Roth (1982) and Kohyama (1985) used direct deposition on silver filters. Gebhardt (1975) modified the personnel samplers to concentrate the particles in the center of the filter by using a aperture with a 25 mm diameter. Others have used this same modification. Chung (1978) and Henslee and Guerra (1977) have examined by direct imaging the distribution of particles collected on filters and showed that personnel samplers do lead to uniform dispersions if run for sufficient time and if the atmosphere is not changing. There is some question whether liquid suspension filtration used to prepare calibration standards yields as uniform a deposit as do the personnel samplers. Spinning the sample during diffraction helps even out the irregularities which do occur. In theory, the particles should be most likely to attach to the filter where the flow rate is the most active, i.e. where no other particles are blocking the passages. This effect should tend to give a uniform distribution of particles as the loading increases.
All the methods of quantitative analysis require calibration standards and calibration procedures and tests. For the on-filter methods the particle sizes as well as the minerals must correspond to the samples under study. Obtaining such samples is not easy.
Procedures Because the on-filter methods do not usually use internal or external standards, the standard approach for calibration is to prepare filters loaded with known amounts of silica and establish calibration graphs of intensity response of the specific diffractometer versus weight of silica. The silica should be deposited on the filters in essentially the same manner as for the analyte samples. Elaborate aerosol chambers have been designed for creating uniform deposits by Davis and Johnson (1982b), Davis (1986) and Carsey (1987). Most of the standard procedures prescribe the methods to follow for making the reference standards.
Basically the technique is to disperse the calibrant in a suspension with either air or water and then dilute the suspension to the desired level. The diluted aliquot is then forced through the filter either by vacuum or by pressure. Weighing the filter before and after deposition confirms the loading. The main purpose of the selected procedure is to produce a deposit which is uniform and crystallographically random. Fortunately, the silica minerals show little tendency to orient except on very smooth filters (Edmonds et al. 1977).
For bulk samples, the preparation of calibration samples is usually the problem of thoroughly mixing a weighed set of ingredients. Both dry mixing and wet mixing may be used. Individual components should be ground to the desired particle size prior to mixing because size reduction in a mixture usually leads to unsatisfactory results. Except for purity, the materials to be used for the calibrations are probably not as critical as for the filter standards, especially the particle size distribution.
Quartz Very high purity quartz is easily obtained in highly perfect crystals. The question is whether this source is proper for use as a quantitative reference material. The difficulty is that the perfection of the crystallites even after crushing leads to significant extinction effects. Where the quartz in the analyte is natural and coarse grained, this reference is probably acceptable if the crystallite size is equivalent. However, if the quartz is a recrystallization product like chert or produced by reaction, the crystallites may be strained and imperfect altering the intensity response. Simply crushing coarse quartz will not produce the same effect. Thus, the selection of the proper quartz is not always simple. Kacsmar and Tomb (1984) have reviewed suggested materials for quartz calibration.
For respirable quartz, the problem of a suitable quartz is even more acute. Many studies have shown that the particle size distribution affects the intensity response. It is probably more important that the size range be similar to the analyte size range than the precise distribution. There is only one certified quartz standard available for respirable particle analysis. It is distributed by the Office of Standard Reference Data, OSRD, of the National Institutes for Standards and Technology, NIST, as SRM-1878. It has a range of 0.33 to 5.0 µm with an average size of 1.62 µm. It is certified as to purity at >95% quartz. The NIST SRM's are very expensive because of the effort required to certify them. The NIST/OSRD programs for certifying standards have been described by Hubbard (1982) and Dragoo (1986). Chung (1982) suggests Min-U-Sil (Pennsylvania Glass Sand Corp.) as a source of quartz. Currently, 1991, Min-U-Sil is produced in large quantities at many different places and is no longer sufficiently uniform to be an off-the-shelf standard.
Cristobalite OSRD also supplies a cristobalite as SRM-1879 certified as >98.0% crystalline. Chung (1982) reports that cristobalite can also be made in the laboratory by starting with a silica gel or Min-U-Sil and firing it in a platinum crucible at 1450C for up to 48 hours. The product is usually >99% converted to cristobalite. Diffraction patterns of cristobalite made in this manner usually show some crystallite size broadening indicating that the average crystallite size is less than 0.2 µm.
Tridymite There is no certified sample available for tridymite. Chung (1982) reports that tridymite may be made by fusing Min-U-Sil in NaCl at 1100°C for 72 hours and then washing the product in water to remove the NaCl. This tridymite must contain substituted Na to stabilize it, but it is suitable as an X-ray standard.
* These values reported in the original reference are in error. It is not clear which value is wrong, but the detection limit seems high.