Fly Ash Classification – Old and New Ideas John M. Fox1 1BASF Corporation, 23700 Chagrin Blvd., Cleveland, OH 44122 CONFERENCE: 2017 World of Coal Ash – (www.worldofcoalash.org) KEYWORDS: fly ash, composition, classification FLY ASH CLASSIFICATION A key component of North American fly ash classification has long been major element chemistry by weight percentage: SiO2 + Al2O3 + Fe2O3 in the U.S.; CaO in Canada. The Canadian Standards Association (CSA) and the American Society for Testing and Materials (ASTM) recognize two classes of fly ash: Class F, normally produced from high-rank coals, and Class C, normally produced from low-rank coals. Coal rank has traditionally been an aspect of fly ash classification, particularly the division between low and high rank. In the search for an improved classification that provides meaningful information about the use of fly ash in concrete, the current classifications and the validity of using coal rank in classification are questioned. The transition from the current ASTM C 618 specification in the U.S. to a fly ash classification based on CaO (lime), is a necessary and logical step if a more useful classification is to be adopted. Iron oxide is essentially a nonreactive component of fly ash and is not expected to be correlated to fly ash performance in concrete.1,2 The SiO2 – Al2O3 – CaO (SAC) ternary diagram accurately predicts fly ash phase mineralogy and is a very useful tool for understanding geochemical variations between fly ashes from differing coal ranks and geological settings. Fly ash quality measures have been established based on major element chemistry. Recent studies have shown that the network ratio of fly ash glass is a rigorous indicator of fly ash performance in concrete. A simple geochemical variation diagram provides a measure of fly ash alkalinity that correlates relatively closely with the network ratio. Fly ash classification based on weight percent CaO is a far better approach than the traditional ASTM C 618 standard. The Canadian Standards Association (CSA) specification of low-, intermediate-, and high-CaO fly ashes may be a better approach to fly ash classification than the more simplified concept of splitting fly ashes into Class F and Class C at 18% CaO.

2017 World of Coal Ash (WOCA) Conference in Lexington, KY - May 9-11, 2017http://www.flyash.info/

COAL RANK Coal rank, as defined by ASTM D 388, is a concept rather than a property and is not measured but is assessed by physical and chemical properties which change during the process of coalification.3 Rank is the degree of transformation from vegetation to coal. Coalification increases with rank in the sequence peat – lignite – subbituminous – bituminous - semi-anthracite – anthracite – meta-anthracite. High rank coals include bituminous and anthracite. Lignite, also termed “brown coal,” and subbituminous coals are the low rank coals. During an initial biochemical phase diagenetic changes occur in the peat swamp until bacterial activity ceases at burial depths of greater than 10 meters and the hard brown coal stage is reached. Further increasing rank is the onset of the geochemical or “metamorphic” stage of coalification, leading to significant changes in organic material and volatiles as carbon content increases and hydrogen and oxygen decrease.4 Coalification is a highly complex chemical process. Metal elements such as Ca, Mg, Fe, and K originating primarily from coal swamp plant matter, with less input from pore solutions originating from rock alteration and weathering, are primarily bound in the form of humates. Humic acids, as well as functional, carboxyl, and carbonyl groups, decrease with increasing coal rank. When high rank is reached, bituminous and anthracite coals contain essentially no carboxyl and carbonyl groups and humic acid complexes are rare.5 Before conversion to bituminous coal, the low rank coals contain metal-organic compounds such as polymerized humic substances and macromolecules of essentially aromatic and aliphatic habit. Much of the alkali and alkaline earth metal content of lignites is held organically as chelated complexes. Water in the low rank coal acts as a solvent, transmitter and donner of elements. Each coal basin has a different geochemistry that determines the amount of Ca, Na, Mg, and K held in the organic molecules and pore fluids.3, 6 As coal rank transitions from subbituminous to bituminous coal a natural boundary is crossed between low rank “brown” coal and high rank hard or “black” coal which is accompanied by a significant increase in vitrinite reflectance. The conversion of subbituminous to bituminous coal marks a disappearance of humic acid and a significant loss of moisture, organic and volatile compounds, and alkali metals that were in solution and tied up in organic molecules. Expressed fluids and gasses take with them the alkali and alkaline earth metals that are typically more abundant in the low rank coals, leaving high rank coal with a characteristically low content of Ca, Na, and Mg and a characteristically high vitrinite reflectance.4,7,8 It is remarkable how variable alkali and lime contents are between low rank coal basins. Although some chemical variations stem from differing coal types; humic or woody coals and sapropelic or non-woody coals9, the geology and geochemistry is of prime importance. Variations in the geochemistry and geological setting of coal basins can be visualized on the SAC ternary diagram and with a traditional alkalis-silica variation

diagram. Coal rank is a key in understanding coal and how to meaningfully classify fly ash. FLY ASH QUALITY MEASURES Fly ash quality measures have been established based on major element chemistry and mineralogy.10 The cementing efficiency of fly ash in concrete has been shown to be related to the physical and chemical properties of the fly ash. Fly ash cementing efficiency factors have been defined on the basis of cement replacement level, concrete strength, and water to cementitious materials ratio; however, based solely on these constraints the overall cementing efficiency of fly ash is not adequately represented.11 Comprehensive measures of fly ash quality in terms of pozzolanic reactivity commonly include compressive strength testing of mortar cubes in conjunction with chemical analysis.12 Recent in-depth studies have shown that the network ratio of fly ash glass is a rigorous indicator of fly ash performance in concrete. Detailed studies of seven typical fly ashes representing typical North American sources have shown a strong correlation between glass transition temperature and the amorphous X-ray diffraction (XRD) peak position. Using a combination of major element composition, XRD, and Scanning electron microscopy (SEM) a parameter of ash performance in cementitious systems has been developed that is termed the network ratio (Nr). Fly ash glass compositions that yield a high Nr value contain a higher lime content and are more reactive in concrete.13 Although a powerful tool for understanding fly ash reactivity potential, determination of Nr values requires access to XRD and SEM and the calculation is somewhat detailed. Geochemical indicies of fly ash that are correlated to the Nr value may be useful. ALKALI INDICIES A fundamental geochemical tool in petrology is the Harker diagram with weight percent alkalis on the ordinate. Although primarily used as a measure of alkalis in igneous rocks, this simple variation diagram is more widely useful. The Harker diagram of weight percent silica plotted against weight percent alkalis from LeMaitre, showing over 25,000 igneous rock analyses, can be used to visualize the “main sequence” of compositions.14 (Figure 1) End points of weight percent SiO2 values having the greatest frequency define a linear codabular function across the Daly gap that they frame. Departure of weight percent alkali values from the codabular function Y = 0.2231 X – 8.556 gives the most general expression for alkalinity as codabular variation (cv).

Figure 1. Harker diagram generalized from LeMaitre, 1976. Black points define endpoints of the codabular function. (Courtesy of Elizabeth Fox) Plotting the fly ash compositional data of Oey, et. al., 201513 in terms of Nr and cv shows a relatively close correspondence. (Figure 2)

Figure 2. A plot of Nr vs. cv values shows they vary directly and are well correlated. Building upon the work of Shand,15 an aluminum saturation index (ASI) has been used to classify fly ashes. The ASI is Al2O3/(CaO + K2O + Na2O) on a molar basis.16 The relationship between the Nr values and the ASI of Zen, 1986 for the fly ashes studied by Oey, et. al., 2015 is shown in Figure 3.

Figure 3. A plot of Nr vs. ASI values shows they are approximately correlated. A modified ASI index given by Al/(Na + K + 2Ca) has also been used to characterize fly ash chemically.17 The relationship between the Nr values and the ASI of Sisson, et. al., 2005 for the fly ashes studied by Oey, et. al., 2015 is shown in Figure 4.

Figure 4. A plot of Nr vs. modified ASI values shows they are approximately correlated. The cv value is easily calculated from a normal major element analysis and can be used as an index to classify fly ashes. Since it is a simple computation using only silica and alkalis, it can be prone to elevated values arising from constant sum effects, particularly in high rank coal ashes containing relatively high iron contents. The ASI value is also a useful chemical index of fly ash that correlates approximately with the Nr. The modified ASI index offers no better correlation with Nr values than the ASI index. The cv value is a useful alkali index of fly ash chemistry that correlates with the Nr values at least as well as the ASI and modified ASI values, and may be preferred to them. Average cv values for fly ashes of all ranks and coal basins worldwide are given in Tables 1 and 2. Typical cv values for fly ashes from bituminous and anthracite coals are low or negative numbers, unless there is elevated iron content. Low-rank coal fly ash cv values are generally higher than those of the high-rank coals; however, a wide variation in geochemistry exists between low-rank coal basins and even vertically within seams in a single basin.22 Where cv values of low-rank coal fly ashes are similar to those of bituminous coal fly ash values, as in the example of the southern U.S. lignites, coal ashes are sold as Class F and, although they contain more lime than is typical for fly ashes from bituminous coals, they are similar in performance to high-rank coal ashes and have no self-cementing properties.

Table 1. Average cv Values for High-Rank Coals

High-Rank Coals Location Average cv

Anthracite Worldwide 0.1 Bituminous South, Central America 1.5 Bituminous Europe, Russia, Israel 0.1 Bituminous Africa -1.2 Bituminous China, Korea -1.9 Bituminous Japan -2.5 Bituminous Canada 0.6 Bituminous Western U.S.A. -0.3 Bituminous Eastern, Central U.S.A. 0.4 Bituminous Spain 0.9 Bituminous India -3.7 Bituminous Australia -2.6 Bituminous U.K. 1.7

Table 2. Average cv Values for Low-Rank Coals

Low-Rank Coals Location Average cv

Subbituminous Western U.S.A. -0.9 Subbituminous Powder River Basin, U.S.A. 3.4 Subbituminous Spain 0.2 Subbituminous Alberta, Canada -0.4

Lignite Texas, Louisiana, U.S.A. -1.8 Lignite Saskatchewan, Canada 4.9 Lignite MT, ND, MN, U.S.A. 7.3 Lignite Bulgaria 0.7 Lignite Germany 2.2 Lignite Spain 1.1 Lignite Thailand 4.0 Lignite Australia 11.5 Lignite Turkey 1.8 Lignite Greece 0.5

SAC TERNARY As more than 95% of inorganic components of coal consist of the minerals quartz, kaolinite, illite, calcite, and pyrite, coal ash is composed primarily of SiO2, Al2O3, CaO, and Fe2O3 with less than 5% of other elements.5 The SiO2-Al2O3-CaO ternary phase diagram (SAC) (Figure 5) is a good predictor of fly ash mineralogy after thermal treatment at 1000 degrees centigrade.2 High-rank coal ashes in the mullite field remain largely stable. With increased lime content the glass crystallizes to anorthite, as shown by the example of lignites from Texas and Louisiana in America. If 5% to 10% MgO is considered in an additional dimension of the phase diagram, the crystallization of diopside is expected and observed.19 At still greater CaO contents, fly ash glass is rapidly crystallized to melilite on heating, as in the example of subbituminous coal ashes from the Powder River Basin in America. When MgO is considered, the melilite stability field expands at the expense of feldspar and pyroxene fields.

Figure 5. The SAC diagram showing phase boundaries is from Dunstan18.

Figures 6 to 12 are SAC plots of coal ashes shown by rank and location in which significant amounts of data exist in the literature. Lignite coal ash geochemistry varies widely and would be too complex if shown on only one SAC diagram. The complexity of lignite ash variability is due both to variations in geochemistry between coal basins, as well as vertical variability within coal seams, as shown in Figure 13.

Figure 6. High-rank coal fly ashes from worldwide sources are plotted on the SAC ternary. (Courtesy of Elizabeth Fox) .

Figure 7 Worldwide subbituminous coal fly ashes are shown. (Courtesy of Elizabeth Fox)

Figure 8 Typical ranges of subbituminous coal fly ashes are shown. PRB is Powder River Basin, U.S.A. (Courtesy of Elizabeth Fox)

Figure 9. Worldwide lignite fly ash values are shown. (Courtesy of Elizabeth Fox)

Figure 10. Important lignite fly ash data from European locations are shown. (Courtesy of Elizabeth Fox)

Figure 11. Lignite fly ash data from Australia and Thailand are shown. (Courtesy of Elizabeth Fox)

Figure 12. North American lignite fly ash compositions are shown. (Courtesy of Elizabeth Fox)

Figure 13. Vertical chemical variation of fly ash from within one Turkish lignite seam is shown by these coal ash plots. Data from Ural.22 (Courtesy of Elizabeth Fox) CaO CLASSIFICATION The lime (CaO) content of fly ash is a much more important parameter than Fe2O3 for fly ash classification in terms of material properties and applications for use in concrete. Fly ash classification based on lime has been the standard in Canada for many years. Although it has changed over time, the current CSA specification limits are up to 15% CaO for Class F fly ashes, 15-20% CaO for Class CI fly ashes and above 20% CaO for Class C fly ashes. An earlier CSA specification considered the Class F fly ashes to contain 8% and less CaO, a more typical range for high-rank coal ashes than extending the range to 15% or higher. The lower limit of 20% CaO for Class C fly ashes is near the lower limit of tri-calcium aluminate (C3A) stability.2

Fly ashes classified as Class C by ASTM and CSA are expected to possess cementitious character in addition to being pozzolanic. Fly ash containing sufficient CaO to be self-hardening, a process primarily due to the formation of ettringite, monosulphoaluminate hydrate and C-S-H, depends primarily on the presence of significant amounts of C3A. Lowering the CaO limit for Class C ash below 20% may be too low a specification to ensure self-cementitious character. Measurement of fly ash CaO is commonly done by X-ray fluorescence analysis. Fly ash CaO content can also be closely approximated by XRD using the strong linear correlation between CaO and the amorphous diffraction maxima.20 Alternatively, fly ash lime content can be estimated by a simple calorimetry technique.21

Elimination of the intermediate lime class of fly ash by going to a Class F-Class C break at 18% CaO would move the broad class of fly ashes derived from low-rank coals, having properties between those derived from high-rank coal and low-rank coal ashes carrying cementitious properties, into a more arbitrary status. Perhaps the previous CSA specification of CaO breaks at 8% and 20% made the most sense of all. CONSLUSIONS

1 Fly ash classification based on lime content makes more sense than the current ASTM specification.

2 Coal rank is a useful aspect of understanding and classifying fly ashes. 3 The Nr parameter of fly ash can be approximated by cv values, which are

preferred over other alkalinity indices. 4 High-rank coal ashes are low in lime content, while low-rank coal ashes vary

more widely. 5 The previous CSA fly ash specification of three classes divided at 8% and 20%

lime may be as good as it gets. REFERENCES [1] Malhotra, V.M. and A.A. Ramezanianpour, 1994, Fly Ash in Concrete, MSL 94-45, CANMET, Canada Center for Mineral and Energy Technology, National Resources Canada, Ottawa, Ontario, Canada. [2] McCarthy, G.J., J.K. Solem, O.E. Manz, and D.J. Hassett, 1990, Use of a database of chemical mineralogical and physical properties of North American fly ash to study the nature of fly ash and its utilization as a mineral admixture in concrete, Material Research Symposium Proceedings, Vol. 178, Materials Research Society, pp. 3-33.

[3] Diessel, Claus F.K., 1992 Coal-Bearing Depositional Systems, Springer-Verlag, Berlin, 721 pp. [4] Thomas, Larry P., 2002, Coal Geology, John Wiley and Sons, West Sussex, 384 pp. [5] Bouska, Vladimir, 1981, Geochemistry of Coal, Coal Science and Technology 1, Elsevier, Amsterdam, The Netherlands. [6] Miller, Robert N. and Peter H. Given, 1986, The association of major, minor and trace inorganic elements with lignites. I. Experimental approach and study of a North Dakota lignite, Geochimica et Cosmochimica Acta, Vol. 50, pp. 2033-2043. [7] Vassilev, Stanislav V., Kunihiro Kitano, and Christina G. Vassileva, 1996, Some relationships between coal rank and chemical and mineral composition, Fuel, Vol. 75, pp. 1537-1542. [8] Miller, Robert N. and Peter H. Given, 1986, The association of major, minor and trace inorganic elements with lignites. I. Experimental approach and study of a North Dakota lignite, Geochimica et Cosmochimica Acta, Vol. 50, pp. 2033-2043. [9] O’Keefe, Jennifer M.K., Achim Bechtel, Kimon Christanis, Shifeng Dai, William A. DiMichele, Cortland F. Eble, Joan S. Esterle, Maria Mastalerz, Anne L. Raymond, Burno V. Valentim, Nicola J. Wagner, Colin R. Ward, James C. Hower, 2013, On the fundamental difference between coal rank and coal type, International Journal of Coal Geology, Vol. 118, pp. 58-87. [10] Vassilev, Stanislav V., Christina Vassileva, David Baxter, and Lars K. Anderson, 2009, A new approach for the combined chemical and mineralogical classification of the inorganic matter in coal. 2. Potential applications of the classification systems, Fuel, Vol. 88, pp, 246-254. [11] Babu, K. Ganesh and G. Siva Nageswara Rao, 1996, Efficiency of fly ash in concrete with age, Cement and Concrete Research, Vol. 26, pp. 465-474. [12] Sarat Kumar Das, Yudhbir, 2006, A simplified model for prediction of pozzolanic characteristics of fly ash based on chemical composition, Cement and Concrete Research, Vol. 36, pp. 1827-1832. [13] Oey, Tandre, Cynthia Huang, Ryan Worley, Sarah Ho, Jason Timmons, Ka Lam Cheung, Aditya Kumar, Mathieu Bauchy, and Gaurav Sant, 2015, Linking Fly Ash Composition to Performance in Cementitious Systems, Proceedings, World of Coal Ash, Nashville, TN,18 pp.

[14] LeMaitre, R.W., 1976, Some problems of the projection of chemical data in mineralogical classifications, Contributions to Mineralogy and Petrology, Vol. 56, pp. 181-189. [15] Shand, S.J., 1949, Eruptive Rocks, Wiley, New York, 488 pp. [16] Zen, E-an, 1986, Aluminum enrichment in silicate melts by fractional crystallization: some mineralogic and petrographic constraints, Journal of Petrology, Vol 27, pp. 1095-1117. [17] Sisson, T.W., K. Ratajeski, W.B. Hankins, and A.F. Glazner, 2005, Voluminous granitic magmas from common basaltic sources, Contributions to Mineralogy and Petrology, Vol. 148, pp. 635-661. [18] Dunstan, Jr. E.R., 1980, A possible Method for identifying Fly Ashes that will Improve the Sulfate Resistance of Concrete, Cement, Concrete, and Aggregates, Vol. 2, pp. 20-30. [19] Kalmanovitch and Williamson, 1986, Crystallization of coal ash melts, in: K.S. Vorres, editor, Mineral Matter and Ash in Coal, ACS Symposium Series 301, pp. 234-299. [20] McCarthy, G.J., D.M. Johansen, S.J. Steinwand, and A. Thedchanamoorthy, 1988, X-Ray Diffraction Analysis of Fly Ash, in Barrett, et. al., Advances in X-ray Analysis, Vol. 31, pp. 331-341. [21] McKerrall, W.C. and W.B. Ledbetter, 1982, Variability and control of Class C fly ash, Cement, Concrete, and Aggregates, Vol. 4, pp. 87-93. [22] Ural, Suphi, 2005, Comparison of fly ash properties from Afsin-Elbistan coal basin, Turkey, Journal of Hazardous Materials, B119, pp. 85-92.