Link to USGS Home Page

U.S. Geological Survey Bulletin 1823

Edited by D.W. Golightly and F.O. Simon


By Jean S. Kane


Methods for the determination of eight elements in coal ash by atomic absorption spectrometry are described. Results from analyses of two National Bureau of Standards (NBS) standard reference materials, coal fly ashes NBS 1633 and l633a, and four quality-control samples are reported. Possible sources of error in each method are discussed, and steps in the procedure critical to accuracy and precision are identified. Accuracy of the methods is established by comparisons of measured concentrations for elements in NBS standards 1633 and l633a with certified values and with other values reported in the literature.


The elements Cd, Cu, Li, Mg, Mn, Na, Pb, and Zn are determined in coal ash by methods based on atomic absorption spectrometry (AAS). These methods are described, and results are reported from the analyses of two National Bureau of Standards (NBS) standard reference materials, coal ashes NBS 1633 and 1633a, and four U.S. Geological Survey in-house, quality-control samples. Comparisons of elemental concentrations determined for NBS 1633 and NBS 1633a to certified concentrations and to other values reported in the literature establish the accuracy of the methods.


Dissolution of Samples

A coal-ash sample, 500 1 mg, is weighed into a 50-mL TeflonTM beaker to which 10 mL of hydrochloric acid, 3 mL of perchloric acid, and 10 mL of hydrofluoric acid are added. Each of these acids is concentrated and of reagent-grade quality. The mixture of acids and sample is heated in an uncovered beaker for 1 h on a hot plate having a surface temperature of 200 C. Then, the beaker is removed from the hot plate and allowed to cool for 5 min, and 2 mL of nitric acid is added. The beaker is returned to the hot plate and heated until dense fumes of perchloric acid are formed and the acid volume has been reduced to 3 mL. The beaker is again removed from the hot plate and allowed to cool for 5 min, and 2 mL of nitric acid plus 5 mL of hydrochloric acid are added. The beaker is returned to the hot plate, and the mixture within is evaporated overnight to complete dryness. Finally, the beaker and its contents are cooled for 5 min, 2.5 mL hydrochloric acid is added, and the wall of the beaker is washed with approximately 5 mL of distilled water to dissolve the residue. Gentle warming of this solution for 5 to 10 min completes the dissolution and produces a clear solution that is transferred to a 50-mL linear polyethylene volumetric flask. This flask is filled to the volume mark with water after the solution reaches room temperature. For this approach, the concentration of an element in a sample of coal ash equals the concentration of the element in the sample solution times 100.

Determinations of Cu, Li, Pb, Mn, and Zn are made directly on each final solution. However, determinations of Na and Mg are made only after diluting each solution 1 to 100 with a matrix-modifier solution containing lanthanum at a concentration of 1,000 m g/mL. The cadmium determination requires extraction of the element with dithizone into xylene.

Cadmium Extraction

Typically, 12 samples, 2 blanks, and 4 calibration standards are extracted as a set. The concentrations of the calibration standards should range from 1 to 15 ng/mL in solutions freshly prepared by diluting a stock solution that contains 100 m g/mL of cadmium.

Pipet 5 mL of the sample solution, blank, or calibration solution, described later, into a 60-mL separatory funnel. Add 0.50 mL of 20 percent (w/v) hydroxylamine hydrochloride, 5 mL of 10 percent (w/v) sodium potassium tartrate, and 5 mL of 10 percent (w/v) sodium hydroxide. Then, add 15 mL 0.10 percent (w/v) dithizone in xylene, and mix for 5 min by an air bubbler or by a mechanical shaker. Allow the layers to separate, and discard the aqueous layer.

The xylene (upper) layer varies in color from pale straw to pink to a deep, almost-black purple, depending on the amounts of Zn, Pb, and Mn that coextract with the Cd. The coextracted elements are not extracted from the xylene layer in subsequent steps. The aqueous layer should be distinctively orange.

After discarding the aqueous layer, wash the xylene phase twice with 5 mL of 0.10 percent (w/v) ammonium hydroxide, agitate the mixture for 1 min, and discard the aqueous wash solution each time.

Back extract the cadmium into 5 percent (v/v) hydrochloric acid by adding 10 mL of 5 percent hydrochloric acid to the xylene phase in each funnel, and then mix each solution for 10 min with either an air bubbler or a mechanical shaker. Rinse the stem of each separatory funnel with distilled water, and dry each stem with a KimwipeTM. Then, collect the aqueous phase, which is needed for the AAS measurement. The concentration of cadmium in the coal-ash sample is equal to the concentration of cadmium measured in the stripped solution times 200.

Calibration Solutions

Solutions for calibration of the atomic absorption spectrometer are prepared to contain Cu, Li, Mn, and Zn in 5 percent (v/v) hydrochloric acid at concentrations ranging from 0.10 to 20 m g/mL by serial dilutions of single-element stock solutions in which each element has a concentration of 1,000 m g/mL. Four to five solutions are prepared to calibrate over each decade of concentration.

Solutions of lead, which have the same range of concentrations as the previous multiple-element solutions, are matched to the matrix of a typical coal ash by adding an aliquot of a concentrated iron solution to give a final iron concentration of 5,000 m g/mL. The use of solutions that contain the principal elements aluminum, calcium, and iron provide no improvement in matrix matching over the iron solution alone.

Concentrations of magnesium and sodium in calibration solutions range from 0.05 to 2.0 m g/mL. These calibration solutions contain 0.10 percent lanthanum to minimize errors originating from matrix suppression of the magnesium absorption and from ionization effects on sodium.

Calibration solutions for cadmium are taken through the extraction procedure previously described. Without this extraction, background correction by use of a deuterium lamp is inadequate in the determination of cadmium.


All measurements by flame AAS are made by either a Perkin-ElmerR model 603 or model 5000 atomic absorption spectrometer that has background correction based on a deuterium arc lamp. Measurements of cadmium by electrothermal-atomization AAS are accomplished with a Perkin-ElmerR model 603 atomic absorption spectrometer that has deuterium-lamp background correction, a graphite atomizer (model 500 or model 2100), a model AS40 autosampler, and a model 56 strip chart recorder. Solution volumes of 20 m L are injected into nonpyrolytic graphite tubes that are purged by argon. Each hollow cathode lamp used in these measurements is operated at the current recommended by the manufacturer.

Instrumental operating parameters for each of the elements determined are listed in table 19. Although the operation of a fuel-lean (blue) flame and the use of an observation height of five units on the vernier of the burner head are common to all measurements reported here, the accuracy of these measurements is relatively insensitive to variations in flame stoichiometry and observation height. Thus, an inability to reset exactly these parameters on a day-to-day basis should be of no consequence.

Detection limits, sensitivities, and approximate upper limits of linearity for calibration curves are listed in table 20.

Detection limits for the cited elements are a simple function of dilution and thus can be improved by as much as a factor of 5 by decreasing the extent to which the dissolved sample is diluted. For cadmium, a change in the extraction factor can improve the detection limit by up to a factor of 10. For most coal ashes, the sensitivities for individual elements need to be reduced much more frequently than the detection limits need to be improved. Such a reduction in sensitivity can be accomplished either through serial dilutions of the sample solution until the analyte concentration is within the linear calibration range or through a reduction of the effective length of the flame that is accomplished by rotation of the burner head. Only positions at 30 , 60 , and 90 can be reproducibly set on the Perkin-ElmerR model 603 spectrometer, but the model 5000 permits the continuous variation of the burner position through 90 . For copper, manganese, and lead, the option of using less-sensitive spectral lines for measurements is available. Alternate operating parameters that can produce reduced sensitivity are listed along with detection limits, sensitivities, and calibration ranges in table 21. Also, alternative approaches are available through standard methods described in American Society for Testing and Materials (1984a, b) publications.


The accuracy of analysis for the AAS methods described here is verified by comparisons of measurements on NBS standard reference materials 1633 (table 22) and 1633a (table 23) with certified concentrations and with

concentrations determined by instrumental neutron activation analysis (Ondov and others, 1975; Rowe and Steinnes, 1977). U.S. Geological Survey "inhouse" standards, designated ASH1, S1, S2, and S3 in table 23, are used as references to provide long-term accuracy and to conserve supplies of NBS standard reference materials. The precision of each method is expressed as the standard deviation for 10 or more replicate determinations. Factors that are important to the maintenance of accuracy and precision are treated in the following discussions on individual elements.


In general, determinations of manganese are biased low for measurements on Mn 279.42 nm if spectral background corrections are based on the use of a deuterium arc lamp. Iron lines at 279.42, 279.47, and 279.50 nm, which occur within the spectrometer bandpass for the Mn 279.42-nm line, absorb the continuum radiation from the deuterium lamp and thus produce an overcorrection for background (Zander, 1976). Because of the close proximity of the iron lines to Mn 279.42 nm, reduction of the spectrometer slit width produces no improvement. Also, at minimal slit widths, analyte absorption of the continuum radiation produces overcorrection for background. Furthermore, matrix matching cannot be used to eliminate this error because of the variability of concentrations of iron in coal ashes. For best accuracy, a background correction should not be made for manganese.


As the concentration of iron increases from 0 to 1,000 m g/mL, the lead atomic absorption signal is sharply suppressed. Further increments in iron concentration up to 4,000 to 5,000 m g/mL produce little further suppression of lead absorption, as illustrated in figure 12. Typically, matrix matching can be used to correct for this suppression over most of the concentration range encountered in the analysis of coal ash. Lead calibration solutions, each containing iron at a concentraton of 1,000 m g/mL, are well suited for this purpose. However, this approach overcorrects when iron occurs at concentrations below 5 percent in the ash and is inadequate for iron concentrations above 40 to 50 percent in the ash. In the latter case, removal of the iron from solution by extraction with concentrated hydrochloric acid into methylisobutylketone is necessary before measuring lead absorbance. The method of standard additions is unsatisfactory in correcting for the suppression of lead absorption signal because of the greatly reduced sensitivity for lead absorption measurements. Furthermore, the extraction of lead by diethylammonium diethyldithiocarbamate prior to measurement is unsuccessful because of the coextraction of iron by this reagent.

Although most AAS methods use the Pb 283.3-nm line, the more sensitive 217.0-nm line is recommended for detectability that is better by a factor of 2.5. Background correction at this wavelength is essential. The Sb 217.59-nm line potentially can cause overcorrection for background (Vajda, 1981), but Pb-to-Sb concentration ratios in coal ash are so high that this error is rarely significant.


The concentrations of zinc in NBS 1633 determined by AAS are in excellent agreement with the NBS certificate value. Background correction is required at the Zn 213.86-nm line. Positive bias can occur for zinc determinations in coal ashes having high iron concentrations because of the direct spectral overlap of iron 213.859 nm on zinc 213.856 nm. For a concentration of 10 percent iron in the ash, the error is equivalent to 15 ppm of zinc. For typical coal ashes, which have zinc concentrations between 200 and 1300 ppm, the relative error is less than 5 percent. Precision of measurements, which generally is 5 percent, is illustrated by the plot in figure 13 of zinc concentrations measured for duplicate dissolutions of various samples. Data for the other seven elements are quite similar and thus are not shown here.


Measurements of copper in NBS 1633 by AAS show the best accuracy and precision of the eight elements determined. Background corrections are not needed.


Prior to 1980, values for lithium concentrations in coal-ash standards were unavailable in the literature. Thus, accuracy of analysis was evaluated from concurrent measurements of lithium concentrations in several U.S. Geological Survey standard rocks. The dissimilarities of both matrices and concentration ranges for lithium cause problems in comparisons. For example, the lithium concentrations determined for standard rocks generally have relative standard deviations of 5 percent, which is only half that observed for coal ashes.

Solutions of coal ash naturally contain concentrations of total alkalis (Na + K) that approach 2,000 m /mL. Thus, additional ionization buffering is unnecessary for the determination of lithium. Nonetheless, the precision of this determination is poorer for the coal ash than that for silicate rocks and is poorer than that for determinations of copper or manganese in coal ash.


The use of glass volumetric flasks in determinations of sodium in solutions of NBS 1633 produces values that are consistently high by 10 to 100 percent. Replacement of all glassware with linear polyethylene containers eliminates this bias, enabling sodium values to agree within 7 percent of literature values. Also, the precision for the method is 8-percent relative standard deviation. The concentration of sodium in solutions from coal ash frequently exceeds the linear range of absorbance for the Na 589.0-nm line. However, detection of sodium often requires the use of this line in preference to the somewhat less sensitive doublet at 589.0-589.6 nm. Use of several calibration solutions that span a small concentration range enables accurate determinations of sodium in the nonlinear portion of the calibration curve.


The accurate determination of magnesium is quite dependent on the dissolution procedure. The use of hydrofluoric acid is necessary for the decomposition of siliceous materials, but the formation of insoluble MgF2 can lead to losses of magnesium if the hydrofluoric acid is not completely evaporated. The initial fuming of the sample in a mixture of nitric and perchloric acids must be followed by evaporation in aqua regia and by thorough drying of the resulting residue. Without this treatment, losses of magnesium can approach 50 percent.

Curvature of calibration curves for magnesium occurs above concentrations of 0.5 m g/mL. With sufficient calibration points, the useful range can be extended to 3 m g/mL. Concentrations of magnesium in solutions of coal ash are rarely high enough to allow the use of the 202.6-nm line, which is less sensitive (by a factor of 24) than the 285.6-nm line.


Several factors affect the overall accuracy and precision of the cadmium determination. Control of the chemical blank is the most important factor. Commonly, acids used for dissolving samples produce blanks having cadmium concentrations that greatly exceed those of cadmium in solutions of coal ash. Containers of such acids should be avoided in sample preparation.

Adsorption of cadmium on the walls of a separatory funnel or a volumetric flask from samples having very high concentrations of cadmium can contaminate the next sample placed in the same vessel. Thorough cleaning of glassware with 50 percent nitric acid, followed by repeated rinsing with distilled water, minimizes this type of error. Autosampler cups made of linear polyethylene must be used to prevent initial contamination from the cups because the slightly coextracted organic phase in the solution dissolves the polystyrene cups. The cups should be cleaned between uses, along with the glassware, to avoid cross contamination from one sample to the next. The pipet tip of the autosampler has been observed to cause cross contamination that results from inadequate rinsing if the sample-to-sample variation of cadmium concentration is 1,000-fold or more. This cross contamination decreases rapidly with repeated pipetings and is readily identified by making three or more successive absorption measurements at each autosampler position. Concentrations of cadmium in coal ashes are directly correlated with zinc concentrations. The latter can reach quite high levels in coals that are located near ore bodies (Ruch and others, 1974).

When cadmium concentrations in the extract are expected to be well above the normal calibration range for graphite-furnace AAS, that is, above 100 ng/mL, flame AAS can be used to minimize the possibility of carry-over contamination. As illustrated in table 24, high concentrations of cadmium are sometimes encountered in analyses of coal ash.

The extraction yield for cadmium varies as a function of both pH and extraction time. Extraction for 5 min at pH 8.5, followed by back-extraction for 10 min, assures constant yields of 95 5 percent for solutions prepared for calibration and for solutions of coal ashes. Variations in either pH or extraction time lead to irreproducible yields that deteriorate the precision of the method. Small changes in these two variables and the difficulty of exactly reproducing the chemical blank account for the 20-percent relative standard deviation of the method.

Losses of cadmium occur prior to atomization in the graphite furnace if the charring temperature exceeds 250 C. Time constants of the Perkin-ElmerR model 603 atomic absorption spectrometer and recorder are too long for accurate measurements of fast transient signals produced by rapid atomization. A sequence of temperatures for atomization that provides a heating rate matched to instrumental time constants is detailed in table 19. An increase in heating rate, either by adding programmable ramp features or by increasing the final atomization temperature above 2100 C leads to loss of the absorption signal.

Without the extraction of cadmium before AAS measurements, serious errors in corrections for background can occur from Ni 228.73-nm, Ni 228.84-nm, and Co 228.78-nm lines that are within the bandpass of the spectrometer (Zander, 1976). Moreover, iron produces a severe suppression of the cadmium atomic absorption signal, particularly for furnace atomization.

Results of analyses for NBS 1633a coal ash and for four U.S. Geological Survey quality-control samples are summarized in table 23. Accuracy for analysis of NBS 1633a, as evident from comparisons of results from AAS methods with the certified values, is excellent. Except for manganese and lead, the determined concentrations agree with NBS-certified values within one standard deviation. The reason for the larger disparity of manganese and lead values is not apparent. Measured concentrations of manganese and lead in NBS 1633 (table 22) agree with NBS-certified values well within one standard deviation.


American Society for Testing and Materials (ASTM), 1984a, D3682-78(1983) Standard test methods for major and minor elements in coal and coke ash by atomic absorption, in 1984 annual book of ASTM standards, petroleum products, lubricants, and fossil fuels, sect. 5, v. 05.05: Gaseous fuels, coal, and coke: Philadelphia, ASTM, p. 458-465.

_______ 1984b, D3683-78(1983) Standard test methods for trace elements in coal and coke ash by atomic absorption, in 1984 annual book of ASTM standards, petroleum products, lubricants, and fossil fuels, sect. 5, v. 05.05: Gaseous fuels, coal, and coke: Philadelphia, ASTM, p. 466-469.

Gladney, E.S., 1980, Elemental concentrations in National Bureau of Standards biological and environmental standard reference materials: Analytica Chimica Acta, v. 118, no. 21, p. 385-396.

Ondov, J.M., Zoller, W.H., Olmez, llhan, Aras, N.K., Gordon, G.E., Rancitelli, L.A., Abel, K.H., Filby, R H., Shah, K.R., and Ragaini, R.C., 1975, Elemental concentrations in the National Bureau of Standards' environmental coal and fly ash standard reference materials: Analytical Chemistry, v. 47, no. 7, p. 1102-1109.

Rowe, J.J., and Steinnes, E., 1977, Instrumental activation analysis of coal and fly ash with thermal and epithermal neutrons, Journal of Radioanalytical Chemistry, v. 37, no. 2, p. 849-856.

Ruch, R.R., Gluskoter, H.J., and Shimp, N.F., 1974, Occurrence and distribution of potentially volatile trace elements in coal: a final report: Illinois State Geological Survey, Environmental Geology Notes, no. 72.

U.S. Environmental Protection Agency (EPA), 1977, Potential pollutants in fossil fuels: Report no. EPA-R2-73-249, Esso Research and Engineering.

Vajda, Ferenc, 1981, Line absorption of matrix elements as a background correction error in atomic absorption spectrometry: Analytica Chimica Acta, v. 128, p. 31-43.

Zander, Andrew, 1976, Factors influencing accuracy in background corrected atomic absorption: American Laboratory, Nov., p. 11.

[ Contents | Coal Quality CD ]

Created by the EERT WWW Staff.

[an error occurred while processing this directive]