American Mineralogist, Volume 74, pages 85-100, 1989

Pyrometamorphic rocks associated with naturally burned coal beds,Powder River Basin, Wyoming*

Mrcn,Lrr, A. Cosc.l, Enrc J. EssrNBDepartment of Geological Sciences, University of Michigan, Ann Arbor, Michigan 48109, U.S.A.

JorrN W. Gnrssuli.qDepartment of Geology, University of New Mexico, Albuquerque, New Mexico 8713 l, U.S.A.

Wrr,r-r.q.vr B. SrnrlvroNsDepartment of Earth Sciences, University of New Orleans, New Orleans, Louisiana 70148, U.S.A.

DoN.q.Ln A. Co.c,rpsU.S. Geological Survey, Denver Federal Center, Denver, Colorado 80225, U.S.A.

AssrRAcr

Glass-bearing pyrometamorphic melt-rock, formed during natural burning of Tertiarycoal in the Powder River Basin, Wyoming, contains a variety of high-temperature igneousminerals, some with diverse and unusual compositions. The melt-rock or paralava, wastransformed from coal-ash and overlying Tertiary sedimentary rocks through local heatingto fusion temperatures. Paralava bulk compositions differ from overlying and surroundingbaked sedimentary rocks, or clinker, by marked depletions of Si, Al, and K and enrich-ments in Fe, Mg, and Ca. Paralava contains a variety of phenocryst and quench mineralsincluding fayalitic olivine, tridymite, cristobalite, sekaninaite (Fe end-member cordierite),anorthite, magnetite-hercynite-ulvtrspinel solid solutions, hematite-ilmenite solid solu-tions, pseudobrookite, melilite solid solutions, aluminous clinopyroxenes, a rhdnite-likephase (dorrite), titanian andradite, wollastonite, mullite, enstatite, potassium and bariumfeldspars, apatite, sahamalite, and nepheline. Glass compositions have been used to esti-mate minimum temperatures of paralava formation and are in the range 1020 to 1400 "C.It is suggested that varying degrees of disequilibrium melting, mixing, crystallization, vol-atilization, and liquid immiscibility are involved during paralava formation. Althoughrare in nature, the minerals and textures observed in paralava have affinities to morecommon igneous rocks and may provide insight into other high-temperature processes ingeologic systems.

INrnolucrroNr

Pyrometamorphism occurs at ultra-high temperatures(> 1000 "C) and low pressures (= I kbar) and often resultsin fusion of near-surface sediments. The Mottled Zone ofIsrael (Matthews and Gross, 1980) and the buchites inScotland (Tilley, 1924) arc well-known examples wheremany high-temperature minerals were first identified. Py-rometamorphism is probably most well known from re-ports of dry basaltic intrusions and their effects on in-cluded xenoliths of surrounding, usually sedimentary,country rocks (e.g., Tllley,19241- Agrell and Langley, 1958;Wyllie, 1961; Smith, 1969; Grapes, 1986). Less wellknown but equally widespread are pyrometamorphic(combustion metamorphic) rocks associated with bitu-minous sedimentary rocks (Mclintock, 1932 Bentor andKastner, 1976; Bentor et al., l98l; Bentor, 1984) and

* Contribution no. 450 from the Mineralogical Laboratory,Department of Geological Sciences, University of Michigan.

0003-004x/89/0102-0085$02.00 85

combusted coal beds (e.g., Fermor, l918; Baker, 1953;Whitworth, 1958; Church et al., 1979).

Although many areas contain pyrometamorphic rocksformed through natural coal combustion, few quantita-tive data have been obtained on these rocks, and lessinformation is available regarding their physical condi-tions of formation. Some data are available for rocks pro-duced by natural combustion of bituminous sedimentaryrocks of the Monterey Formation (Bentor et al., 1981).Based on variations in major- and trace-element com-positions, Bentor et al. (1981) made distinctions betweenlow-temperature melts (LTM), and higher-temperaturemelts (HTM). Although not well constrained, formationtemperatures were estimated at 1000 "C (LTM) to 1650'C (HTM). As discussed in this paper, some similaritiesexist between the combustion metamorphic rocks de-rived from bituminous sedimentary rocks of the Monte-rey Formation (Bentor et al., 198 l) and those formed bynatural coal combustion, but the detailed mineral andwhole-rock compositional data are distinct.

86 COSCA ET AL.: PYROMETAMORPHIC ROCKS

4 ?

Fig. l. Sample location map showing outline of the PowderRiver Basin.

Mineral formation associated with coal combustion isdependent on many physical and chemical variables in-cluding the original sediment bulk composition, temper-ature, degree of melting, and oxidation state. The degreeto which one or more of these factors affects mineral sta-bilities may be variable on the scale of a thin section.However, certain field and petrographic observationscombined vrith analytical data may constrain the relativeimportance of the various possible controls. The purposeof this study is to report mineral and whole-rock com-positional data for pyrometamorphic rocks associated withcombusted coal beds of the Powder River Basin, Wyo-ming (Fig. l), and to provide some constraints on theirformation conditions. The samples collected for this studyhave been roughly separated into rocks that were ther-mally altered but unmelted, termed clinker, and thosethat were fused, termed paralava (Fermor, l9l8; Ven-katesh, 1952). Although the term clinker is accepted inthe geological literature to include all rocks altered bycoal combustion, separation of rock types is important toavoid ambiguities and therefore the additional term para-lava is used for the purposes of this paper. Samples ofclinker and paralava were collected from outcrops in thePRB near Gillette, Wyoming (Fig. l). They were exam-ined petrographically, by electron-microprobe analysis(rure), X-ray fluorescence (xnr), and X-ray diffraction(xno) techniques.

RncroN.q.L GEoLocY

The Powder River Basin is an elongate basin coveringapproximately 37000 km2 between the Bighorn Moun-tains and the Black Hills and extends north into southern

Montana (Fig. l). Cretaceous and Tertiary sedimentaryrocks on the east side of the basin dip gently toward itsaxis in response to late Laramide orogenic activity (West,1964). Exposed through erosion and ignited by naturalprocesses, near-surface burning of Tertiary coal in thebasin has produced large volumes of thermally alteredsedimentary rocks (Bastin, 1905; Rogers, l9l8; May,1954; Sigsby, 1966; Bauer, 19721, Jackson, l98l; Budaiand Cummings, 1984; Coates and Naeser, 1984; Esseneet al., 19841 Cosca and Essene, 1985; Foit et al., 1987).

Coal fires were probably first initiated when coal wasexposed to the atmosphere through erosion and was eitherspontaneously combusted or ignited by lightning or prai-rie fires (Rogers, 1918). Indeed, spontaneous combustionof coal is a common phenomenon (Raask, 1985) and isoften observed in many of the open-pit coal mines withinthe Powder River Basin. Alternatively, C- and O-isotopedata from vegetation growing around a vent produced bypresent-day burning ofcoal in Canada indicate that groundwater is involved as an oxidant (Gleason and Kyser, 1984).Regardless of the cause in the Powder River Basin, onceignited, a burning "front" advances below the surface fol-lowing the coal bed, periodically vents, and operates as anatural blast furnace as long as adequate fuel and oxygenare available. Consequently, much of the basin's currentlandscape is in the form of mesas and buttes capped byerosion-resistant, bright red to orange clinker. From adistance, many of the outcrops exhibit sharply contrast-ing light-colored sedimentary rocks below the bright cap-ping of clinker. Fission-track dates from detrital zirconscontained in northeastern Powder River Basin clinker in-dicate that the burn zones become progressively older tothe east, ranging in age from 0.08 to 0.77 + 0.39 Ma(Coates and Naeser, 1984). Paleomagnetic reversal datareveal that some clinkers in the northem part of the basinformed more than 1.4 Ma ago (Jones et al., 1984). Igni-tion of coal in other basins has resulted in a semicontin-uous belt of clinker-bearing sediments stretching fromTexas to Canada (Rogers, 1918; Bentor, 1984).

Paralava and clinker of the Powder River Basin werederived from continental sedimentary rocks of the FortUnion and Wasatch Formations. The Fort Union For-mation is Paleocene in age and is a thick sequence (60G-900 m) of interbedded sandstones, shales, siltstones,limestones, and coal beds (Glass, 1976). The uppermost250 m of the Fort Union Formation is dominated by fine-grained sandstone, siltstone, shale, and coal. Coal bedswithin this formation are a major economic resource withsome in this area exceeding thicknesses of 25 m (Densonand Keefer, 1974). Conformably overlying the Fort UnionFormation is the Eocene Wasatch Formation (Glass,1976). The Wasatch Formation (300-700 m) crops outover much of the central part of the basin and containsnumerous siltstones, fine-grained sandstones, coals, andsome thin beds of ferruginous nodules. Coal beds withinthe Wasatch Formation are concentrated in the lowerportion and are generally l-5 m thick, but they locallymay attain thicknesses of 70 m (Glass, 1976).

r o 8

o M ' ( E 5 C / 7 Y I

II

G b*oi^

I< 0 ;d-.a

"tn"i

ra-+

1oz--{za

; *r,| \ /i ' oII

L _ _I

N E B R A S K A

I

E casPER

'":T'^lir,rto 50 too-

X I L O M E I E A S

4 5

4tr

COSCA ET AL.: PYROMETAMORPHIC ROCKS 8 7

OccunnpNcE oF pyRoMETAMoRpHrc RocK

The degree ofthermal alteration produced by burningcoal beds is variable, and a single outcrop (sometimesexceeding thicknesses of 30 m) may contain altered rockranging from slightly baked (clinker) to entirely fused(paralava). In close proximity to a heat source (gas ventor burnt coal bed), clinker becomes streaked and porce-laneous, and it may become partly or completely fusedto form paralava. Typical examples of paralava outcropinclude the following: (l) layers or lenses, presumed torepresent original sedimentary horizons; (2) small, com-monly spherical globules that may represent incipientmelting; and (3) vesicular, chimneylike pipes or dikes ofparalava often containing clasts ofunfused clinker.

The observed rock types and features associated witha combusted coal bed are schematically illustrated in Fig-ure 2. Below the coal ash, the rocks consist of a thicksequence of unaltered, well-sorted siltstones and sand-stones, sandy conglomerates, and occasional layers offer-ruginous nodules. In many instances, a sharp contact isobserved separating underlying unaltered sedimentaryrocks from thermally metamorphosed equivalents; thecontact has been interpreted to be recording a paleewatertable (Baker, 1953;Budai and Cummings, 1984). If so, awell-defined record of paleo-water tables, temporallyconstrained by fission-track or paleomagnetic methods,may prove useful for reconstructing paleohydrologic en-vironments. Alternatively, this contact may simply rep-resent a boundary across which little heat is transferred(Bustin and Mathews, 1982), implying that the majorityof heat was dissipated either laterally or upward by ad-vection, with only a small component of heat transferreddownward by conduction.

A regular, gradational pattern of partially to completelycombusted coal, coal-ash, paralava, and baked, nonmelt-ed clinker occurs above the unaltered sedimentary rocks(Fig. 2). The residual coal layer, ifpresent, is generally afew centimeters to two meters thick and grades from blackto gray to white upward in section; it has been referredto as coke (Bustin and Mathews,1982; Merrit, 1985). Thecoke grades upward into a distinctive white ash layer,approximately 15-25 cm thick, containing thin layers ofbanded and porcelaneous shale near the upper contact ofthe coal bed. Where observed, the contact at the top ofthe ash zone is sharp and underlies a 15-20 cm thick,highly vesicular, light brown to green paralava with abun-dant glass. The glass-rich paralava may have been de-rived from fusion of residual coal ash and/orimmediatelyoverlying sediment and indicates that intense heating oc-curred within, or centimeters above, the coal beds. Theparalava grades upward into a thick, laterally continuoussequence ofbaked and reddened clinker. Clinker occursin a variety of colors including various shades of yellow,green, red, purple, black, gray, and bleached white. Relictbedding is generally preserved, although local heating andmelting may obscure or totally obliterate original struc-tures.

CHII,INEY STRUCTURE

FRACTURED CLINKER

FUSED STDIiIEI{TARY LAYER

INCIPIEl{T I{ELTII'IG

CoAL-ASH 201{E

PORCELAI{EOUS LAYERS

RE!II{4]{T COAL LALYER

UNALTERED SEDIIi.IEI{TARYLAYERS

Fig. 2. Schematic cross-section through a typical clinker- andparalava-bearing outcrop in the Powder River Basin. The chim-ney structure is composed ofblack and vesiculated paralava con-taining clasts ofclinkers. The coal-ash zone contains glassy, ve-sicular paralava.

Erosion-resistant chimney structures (pipes) are pres-ent to some extent in nearly every clinker outcrop andmay reach 5-15 m in height. The chimneys usually con-tain angular clasts of clinker up to 0.5 m in diameterwithin a massive matrix of black, vesicular paralava.Many slump and drip textures contained in the paralavaand flow structures developed around clinker clasts re-semble pahoehoe lava and indicate relatively low viscos-ities. Paralava similar to that in the chimneys is devel-oped in subvertical fractures within clinker that probablyacted as conduits for hot gases or channels along whichparalavawas either formed or flowed. Breccias are locallydeveloped in rnany chimneys, possibly reflecting violentcombustion events. Similar chimneys have been ob-served in paralava outcrops in Australia (Baker, 1953),southern Utah (W. P. Nash, pers. comm.), and also fromthe combustion metamorphic rocks of the Monterey For-mation (Bentor et a1., l98l). The chimneys have beeninterpreted to develop by gravitational collapse into thecavities left by the combusted coal bed (Baker, 1953).The black paralava contained in the chimneys is the mostabundant type of paralava observed in outcrops of thePowder River Basin. Its abundance, phenocryst minerals,and flow textures suggest that this parulava has formedalmost entirely from a liquid. Similar conclusions werereached by Bentor et al. (1981) for the massive HTMrocks of the Monterey Formation.

Axalvtrc.Lr, METHoDS

EMPA

Quantative mineral and glass analyses were performed usinga Cameca cAMEBAx microprobe. For most crystalline phases, a15-kV accelerating potential, 10-nA sample cunent, and a fo-

p.1 t\' 1 ^ \ <

88 COSCA ET AL.: PYROMETAMORPHIC ROCKS

Treue 1. Representative mineral compositions from Powder River Basin paralava

Sample:Phase:

t-14Sp

AM-7 DR-7Mel

DR-7Pg

DR-8cpx

83E-38ASp

c-3 c-30Cpx Mul

83EU-2ol

sio,Tio,Alro3Fe.O.*CrrO.FeO-MnOl\4gOZnO

Na.OK.o

Total

SiTiAIFe3*'CrFe,*.MnMgZnCaNaK

n.o1 . 6 1

40.0420 09021

37 240240.94n.d.n.d.n.o.n d .

100.37

0.00.0381.4680.4700.0050.9690.006o 0440 00 00.00 .0

0 1 31 0 9 810 8733 910.69

42 810 1 30 . 1 70.06n.o.n.d.n .d .

99.75

0.0050.2970 4600.9170 0201 2870 0040 0090 0020.00 00 0

n o .n.o.

19 2057.84n o .1 .780 4 6

20 93n.o.n.dn.dn.d

r 00.21

0.00.00.6841 .3160.00.0450.0120.9430.00.00.00 .0

29 840.13

23.233.60n.d.1 .230.0s3 . 1 1n d .

38.241 . 1 9n.o.

100.62

1.3720.005'l 2590 . 1 1 30 00.0430.0020.2130.01.8840.1 060.0

43.17n.d.

35.841.08n.o.n d .0.01n.o.n o .

19.720.080.03

99.93

2.O070.01.9640.0380.00.00.00.00.00.9830.0080 001

30.321 . 1 0

18 9521 69

n.o.0.480.082.97n.d.

23.750 .14n.d

99.48

1.2080.0330.890U . O J U

0.00 .0160.0030 1760 01 .0140.0110.0

37.633.46

18.237.40n-o.0.460 0 38.83n.o.

24 280 . 1 6n d .

100.48

1.4040.0970.8120.2080.00.0130.0010.4910.00.9720.0120.0

29 72o.42

61.487.28n d .n.o.n.o.0.57n.o.n.o.0.03n.d.

99.50

1.6360.0173.9890 3020.00.00 .0o.0470.00 00.0030.0

30.96n.o.0.40n.o.n.o.

64.260.663.94n.o.0.04n.dn.d-

100 26

1 .0130.00 .0150.00.01 7(O

0.0180.1920.00 0010.00.0

Sample:Phase:

83EM-5opx

DR-8 C-35Dr Ne

c - 1 1Mel

KM-5Hem

83E-27 83E-38APsbr Ol

DR-8 L-8Wo Bfs

sio,Tio,Al203FerO"*Cr.O.FeO.MnOMgoBaOCaONaroKrO

Total

SiTiAIFe3**CrFe,*'MnMgBA

NAK

co 34

0.122.07n d .n o .9.82n o .

31.21n o .0.88n.o.0.09

100.73

0 9840 0020 0420 00 00.1 430.00.8090.00 .0160.00.002

11.340.59

z J . l I

39.270.102910.22c . c u

n.o.13 660.050.01

99.36

1 6320.0644.3614.2500.0130.3500.0261 .1790.02.1070.0130 0

44 85n.d.

34 951 1 8n.dn.d.n.d.0.04n.o.3.65

12.203.65

100.52

1.0310.0o 9470 0210 00 00.00.0010.00.0900.5440.1 07

43.340.04

10 901.24n.o.n.dn.o.7.O7n.o.

32.694.210.04

99.s3

1.9210.0010.5690.0420.00.00.00 4670 01.5530 3610 002

0.040.23l 2 E

97.32n.o.n.o.0.820 . 1 8n.on.o.n.d.n.d.

99.94

0 0010.004o.0421.9280.00.00 .0180.0070.00.00.00.0

0 .1021.455.78

71 .24U,JO

n.o.0.24n 2 R

n d .n.d.n.o.n.o.

99 52

0.0030.6230.2632.0690.1 100.00.0080.0200.00.00.00 .0

31.070.00.39n.d.n d .

b J . / J

u.b/4.39n d .0.01n.o.n.o.

100.18

1 .0150.00 .0150.00.01 .7410.0150.00.00.00.00.0

50.550.300.24n.o.n.o.0.43o.27n o .n o .

47 680 0 60.08

99.61

0.9820.0040.0050.00.00.0070.0040.00.0n ooq0.0020.002

39 820.38

27.331 . 1 3n d .n.o.0.020.06

26.732. ' t21 .450.83

oo eR

2.2160.0161.7920.0470.00.00 .0130.0520.5830.1260.1560.059

Note: Sp: spinel solid solutions; Mel: melilite; Pg: plagioclase; Cp: clinopyroxene; Mul: mullite; Ol: olivine; Opx: orthopyroxene; Dr: dorrite; Ne:nepheline; Hem: hematite; Psbr: pseudobrookite; Bfs: Ba-feldspar. Atomic fractions calculated on the basis of cations; n.d. : not determined.

- Calculated from normalized formula.

cused electron beam were standard operating conditions. Alkalivolatilization from the glasses necessitated a different set ofop-erating conditions. Conditions were optimized with a focusedelectron beam rastered over a 1.5-pm2 area with an acceleratingpotential of 12 kV and sample current of 5-10 nA; alkali vola-tilization was minimal under these conditions. Well-character-ized natural and synthetic standards were used and were period-ically checked during analytical sessions and corrected for anyinstrumental drift.

XRF

The xnr analyses were conducted individually by two ofus inseparate laboratories at New Orleans (W.B.S.) and at Michigan(M.A.C.). Because of the unusual compositions of these rocks,some xRF analyses were problematic in that the unknowns (para-lava) have a distinct and unusual composition compared to theordinary rock standards. Hence, the large mass absorption cor-rections yield erratic totals. Major-element analyses at New Or-leans performed by the pressed pellet method gave widely vari-

COSCA ET AL.: PYROMETAMORPHIC ROCKS 89

Fig. 3. Quenched mineral textures occurring in paralava. (A)Glass containing feathery clinopyroxene and a partially melteddetrital quartz grain (upper right) (scale bar -- 25 pm). (B) Skel-etal framework of hematite contained in a glass matrix (scale bar: 50 irm). (C) Symmetrically branching, quenched frameworkof clinopyroxene within a glass matrix. Note partially fused de-

trital quartz grains (scale bar : 200 pm). (D) Enlargement of adetrital quartz grain from C exhibiting a glass rind gradationalinto the matrix glass (scale bar : 50 pm). (E) and (F) Skeletalcrystals of olivine associated with glass and spinel solid solutions(scale bar : 250 pm). All photographs taken in plane-polarizedlight.

90 COSCA ET AL.: PYROMETAMORPHIC ROCKS

Fig. 4. (A) Cyclic twinning in cordierite (photograph takenunder crossed polars with accessory plate) (scale bar: 100 pm).(B) Plane-polarized-light photograph of coexisting clinopyrox-ene (Cpx), melilite (Mel), and plagioclase (Pg) (scale bar : 500irm). (C) Backscattered-electron (BSE) photograph of paralavaglass (Gl) containing quenched crystals of clinopyroxene (Cpx)

and potassium feldspar (Kfs). Note the boxwork exsolution ofmagnetite from a hercynite host (scale bar : 100 pm). (D) BSEphotograph of a fused clinker fragment (Cl) forming fine-grainedoxides and silicates within a matrix of crystallite-bearing glass(Gl) (see enlargement in C). Also shown are two-phase mixturesof magnetite and hercynite solid solutions (scale bar : 100 pm).

COSCA ET AL.: PYROMETAMORPHIC ROCKS 9 l

able results, probably the result of strong mineral effects. Samplesand additional standards prepared as fused disks were employedat Michigan and yielded somewhat better precision through an-aly'tical totals (excluding LOI) occasionally fell outside the range98-l12o/0. The occasionally significant -LOI is primarily de-rived from calcite that is a common vug-filling mineral. The+LOI is obtained from oxidation ofthe iron oxides during thefiring process.

XRDThe X-ray diffraction data were collected on a Phillips powder

X-ray diffractometer. Most samples were prepared as powders;however, some hand-picked mineral grains were also analyzedas single grains or agglomerates by Gandolfi techniques for pos-itive identification.

MrxBnar,s rN PowDER Rrvrn B.q.srN pARALAvA

A variety of crystalline phases and textures are en-countered in paralava. In general, minerals within para-lava are coarse enough (>2 pm) to quantitatively analyzeby euee, and representative analyses are presented in Ta-ble l. Minerals from many paralava samples were ana-lyzed both petrographically and chemically, and xRD wasused to distinguish different polymorphs. Some phases,especially those occurring as acicular needles, were toosmall for quantitative analysis, and a combination of pet-rographic, xno, and qualitative ros techniques was usedfor mineral identification. In addition to the larger detri-tal grains, unmelted clinker is composed of fine-grained(< I pm) metamorphic minerals (Clark and Peacor, 1987)that require scanning transmission electron-microscope(srrrvr) analysis for proper characterization.

Abundant features from paralava such as spinifex andhopper crystals, radial mineral clusters surrounded byglass, and crystals aligned around vesicles (Fig. 3) are con-vincing evidence that these minerals crystall ized(quenched) directly from a liquid. In some instances, tex-tural and chemical relationships at the thin-section scaleprovide strong evidence for local mixing of two or moreseparate liquids (Cosca et al., 1988). Rapid cooling inparalava may have been produced by the sudden openingof fractures or slumping of overburden, allowing the in-filtration of cooler air. Although factors such as the num-ber of nuclei, viscosity, temperature, oxygen fugacity, su-perliquidus heating, ald bulk composition may have someaffect on cooling rates, these textures are qualitativelyconsistent with those produced experimentally in basaltscooled at rates of approximately 30 "C/h (Lofgren et al.,1974). Other textures consisting ofcoarser-grained phe-nocrysts, similar to many volcanic rocks, are suggestiveof somewhat slower cooling from a liquid (Fig. 4B).

Phenocrysts and/or quench crystals identified by opti-

cal petrography, xno and supe./ssvr include olivine, en-statite, clinopyroxene, a rhonite-like phase (dorrite), cris-tobalite, tridymite, mullite, anorthite, nepheline, garnet,potassium feldspar, barium feldspar, gehlenite-akerman-ite-sodium melilite solid solutions, sekaninaite (Fe end-member cordierite), apatite, spinel-magnetite-hercynite-ulv6spinel, hematite-ilmenite, and pseudobrookite solidsolutions, wollastonite, xenotime, sahamalite, and someunidentified Cu- and Fe-bearing sulfides. Low quartz andzircon were observed in some paralava, presumably sur-viving as detrital phases even at high temperatures. Someofthe phases reported here have been identified in para-lava from other parts of the Powder River Basin (Bauer,1972; Jackson, l98l; Hooper and Foit, 1986; Foit et al.,1987) and from paralava occulTences elsewhere (e.g.,Baker, 1953; Whitworth, 1958; Bustin and Mathews,1982; Hooper, 1982). Not surprisingly, a similar collec-tion of minerals (olivine, cordierite, plagioclase, pyrox-ene, spinel, and glass) has been reported from slag pro-duced as a byproduct of thermally treated oil shale(Phemister, 1942). Mullite and spinel have also been re-ported from slag occurring on walls of pulverized coal-fired boilers (Raask, 1985). However, very few chemicaldata have been previously obtained for phases from nat-ural or synthetic slags.

Pyroxene

Clinopyroxene is a very common mineral in paralava.It occurs both as coarse-grained, equant phenocrysts andas feathery or acicular crystals (Fig. 3). The latter texturesmay have formed as crystals were quenched from a liquid(Figs. 3 and 4). Clinopyroxenes range in composition fromdiopside to near end-member esseneite, CaFe3*AlSiO6(Table l), indicating extensive substitution of Fe3t for Mgin the Ml site coupled with Al -f Fe3* substitution for Siin the tetrahedral site (Cosca et al., 1985; Hooper andFoit. 1986: Cosca and Peacor, 1987; Foit et al., 1987).At l-bar pressure, oxygen fugacities approaching the he-matite-magnetite (H-M) buffer are required to stabilizethe most Fe3"-rich pyroxenes (Cosca and Peacor, 1987).Orthopyroxene (Xo" : 0.13) is rare, occurring as fineneedles in glass. The general lack oforthopyroxene is con-sistent with a highly oxidizing environment.

Olivine

Olivine (fayalite) occurs as spinifex-like crystals up to3 cm in length and also as subhedral skeletal grains l0-15 mm across (Figs. 3E, 3F). The spinifex grains exhibitstrong pleochroism (X yellow green; Y: yellow orange;Z:lightgreen) and are fairly uniform in composition (X',: 0.034.09). Olivine coexists with cordierite, cristobal-ite, and magrretite-hercynite-ulv6spinel solid solutions.

(E) BSE photograph of paralava glass (Gl) containing crystals ofcristobalite (Crist) displaying shrinkage cracks, penetrating crys-tals of mullite (Mull) and Fe-Ti oxides Oright) (scale bar: 10

&m). (F) Vesicle in paralava lined with hematite (scale bar: 100pm).

92 COSCA ET AL.: PYROMETAMORPHIC ROCKS

Co"MeSi"O, Co"At"SiO,

Fig. 5. Compositional projection of melilites from paralavashown within the system Ca.MgSirO, (akermanite)-CarAlrSiO,(gehlenite)-CaNaAlSi.O, (soda melilite). Symbols correspond toindividual point analyses ofdifferent samples (solid: core; open:nm).

Many of the olivines contain spinel solid solutions asinclusions.

SiO, polymorphs

Paralava may contain any one or all three of the SiO,polymorphs quartz, tridymite, and cristobalite. Three SiO.polymorphs were observed to coexist in one paralava de-rived from sandstone, where detrital (low) quartz, tridy-mite, and cristobalite were identified by xno. More oftenonly one polymorph is observed, usually cristobalite, oc-curring as needles in complex interpenetrating textures.Backscattered-electron images illustrate that some cris-tobalites contain inclusions of mullite and iron oxidesand shrinkage cracks formed during cooling that reflectdifferential thermal contraction (Fig. a). Quantitativeanalyses indicate that a small combined percentage (0.1-0.3 wto/o) of Al, Fe, and Ti is incorporated into the SiO,structure during the transformation of quartz to cristo-balite or tridymite, as with tridymite in silica bricks(Schneider and Majdic, 1984, 1985; Seifert-Kraus andSchneider, 1 984).

Melilite

Melilite is a common phenocryst in silica-undersatu-rated paralava occurring as large (10-20 mm) euhedralphenocrysts with aluminous pyroxenes, dorrite, spinelsolid solutions, and garnet. Many grains display tetrago-nal outlines, some are twinned, and many are opticallyand chemically zoned. Compositions are variable and ex-ist over a wide range of akermanite-gehlenite-sodiummelilite solid solutions (Ak,o_roGeh,o_ruSMr-r) (Fig. 5). Asmall amount (2-5o/o) of ferrigehlenite and ferriakerman-

ite is also present in most samples. No systematic chem-ical variation is evident in core vs. rim analyses, althoughmany of the rim analyses appear to have a higher con-centration of a sodium melilite component (Fig. 5). Thelarge grain size and chemical zoning may imply that themelilite formed under conditions of relatively slow cool-ing. Melilite is readily identified by its low birefringence,uniaxial symmetry, and euhedral shape.

Feldspars

At least three different feldspars have been identifiedin paralava, but the coarsest and most abundant is an-orthite (Annr_rrAb,_rOr,_r). It occurs as 1-5 mm long,polysynthetically twinned, lath-shaped grains associatedwith clinopyroxene, dorrite, spinel solid solutions, and,more rarely, with melilite. Some of the anorthite grainsare preferentially aligned, indicating they have flowed ina liquid. Other grains appear to have been forcibly dis-placed around vesicles.

Celsian, hyalophane, and potassium feldspar occur asfine fibers in some paralavas. The Ba feldspars appearrestricted to irregular pockets containing radiating crys-tals in glass formed along grain boundaries. These pock-ets also contain wollastonite and lesser ulvdspinel andresemble the barred structure of chondrules. Potassiumfeldspar is present in many samples as l- to l0-pm-long,acicular crystals usually contained in glass. Their size pre-cludes quantitative analysis, but only K, Al, and Si wereobserved by ros.

Dorrite

A newly described rhtinite-like mineral called dorrite(ideally Ca,(Mg,Fei*)(Alosir)Oro) with extensive substi-tutions ofboth octahedral and tetrahedrally coordinatedFe3* and Al3* occurs in paralava from near the coal-sed-imentary rock interface (Cosca et a1., 1988). Dorrite isrelated to rhonite and occurs only in silica-undersaturat-ed paralava usually in the presence of esseneite-rich cli-nopyroxene, melilite, magnetite, plagioclase, garnet withor without nepheline, apatite, and wollastonite. It is near-ly opaque in thin section, and many crystals contain acore of irregularly shaped magnetite, suggesting that thespinel is being replaced by dorrite. The occurrence ofdor-rite appears restricted to conditions of high temperatureandfo, and may be related to reactions involving pyrox-ene and magnetite (Cosca et al., 1988).

Cordierite

Cordierite occurs as euhedral grains 2-5 mm in lengthwith hexagonal cross sections. xRD patterns indicate thepresence of high cordierite (indialite) and chemical anal-yses indicate them to be Fe rich (X," : 0.70). The cor-dierite crystals are strongly pleochroic (O: violet; E: col-orless) and many display cyclic twinning (Fig. aA).Cordierite appears restricted to paralava containing theassemblage fayalite-cristobalite-spinel (magnetite-her-cynite), and it is often found in paralavas forming chim-ney structures.

. DR-A* oR-7. L - 8L L - t z. C-12I c-35f c- l lr c-3

COSCA ET AL.: PYROMETAMORPHIC ROCKS 93

Oxides

Spinel is widely distributed in paralava. Substitutionsinvolving Fe2+-Fe3+-Al-Ti account for a variety of inter-mediate spinel-magnetite-hercynite-ulv0spinel composi-tions. These substitutions are also reflected in solid so-lutions of hematite-ilmenite and pseudobrookite. Manyof the spinels and other oxides have unmixed upon cool-ing to produce complex intergrowths of magnetite-her-cynite-ulvdspinel (Figs. 4C, 4D), hematite, pseudobrook-ite, andlor ilmenite (Modreski and Herring, 1985). Theexsolution patterns indicate subsolvus cooling ofa once-homogeneous oxide phase formed at higher temperature.Crystal habits and textures may exhibit extreme variationwithin a single thin section, from euhedral inclusions tofelted or radiating dendritic masses. Many of the oxidesare concentrated along small cracks, lining vesicles or otheropenings that allow access ofoxygen and/or metalliferouscomplexes.

Wollastonite

Wollastonite occurs in accessory quantities as fineneedles up to l0 pm long that fluoresce bright blue underthe electron beam. Chemical analyses indicate that it isnearly pure (Wo.). It is distinguished from pseudowol-lastonite by its optical properties (length-slow).

Garnet

Titanian andradite is found in some silica-undersatu-rated paralavas. It contains up to 6.0 wto/o TiO, and isassociated with melilite, magnetite, clinopyroxene, anddorrite (Cosca et al., 1988). In thin section, garnets areanhedral in shape, yellow-green in color, and commonlydisplay signs of textural disequilibrium.

Apatite

Apatite is an abundant accessory phase typically foundas needles included in glass. Qualitative energy-disper-sive spectra indicate only the presence ofCa and P.

Mullite

Mullite occurs as prismatic crystals approximately l0-20 pm in length, with cristobalite and glass. Chemicalanalyses indicate the presence of up to 8 wto/o FerOr.Mullite is rare in the paralava collected for this study andis probably restricted to the more aluminous sedimentarylayers.

Nepheline

Nepheline (carnegieite?) is found in small amounts insome Si-undersaturated paralava and forms the only sig-nificant phase concentrating Na in these rocks. Chemicalanalyses display a small Fe3* content. xno analysis wasnot possible because of the small grain size and smallnumber of grains identified. This phase could be carne-gieite, a high-temperature polymorph of nepheline.

Sahamalite

Sahamalite ((Mg,Fe)r(Ce,Nd,Y)r(CO3)) occurs as me-tallic, spherical blebs approximately 04.5 mm in diam-eter. It is closely associated with melilite in Si-poor para-lava. Qualitative analysis of sahamalite separated fromsome samples confirmed the presence of Fe, Mg, Ce, Y,and Nd.

Xenotime

Xenotime is occasionally found in paralava as irregu-larly shaped grains and was identified by ros from theX-ray peaks of Y and P; no quantitative chemical resultswere obtained. This phase could be detrital or primary inorigin and does not appear restricted to any particularparalava composition.

PAnar-Av.c, cLASs coMPosrrroNs

Glass is a common constituent in many paralavas fromthe Powder River Basin and implies rapid quenching fromconditions of high temperature at near-atmospheric pres-sure. These conditions are geologically rare yet producea wide range of mineral and chemical diversity. Paralavaglasses usually contain a variety of phenocrysts and/orcrystals and represent liquid (melt) compositions quenchedon or above their liquidi. Because petrographic texturesare so similar to those produced artificially, these rocksoffer natural analogues to many laboratory experimentsconducted at I atm.

Brady and Greig (1939) conducted melting experi-ments with clinker from Arizona and concluded thatminimum temperatures of I I 17 oC must be reached be-fore melting begins. Similar experiments by Whitworth(1958) required temperatures in excess of 1350'C beforemelting was observed. Precise temperatures at whichparalavas form may be constrained by examination ofglass compositions. Powder River Basin glasses exhibit alimited range in silica content, with somewhat wider vari-ations occurring in Na, K, Ca, Al, and Fe contents (Table2). If low-pressure phase relations are known, assumingequilibrium melting conditions, a minimum liquidustemperature may be determined for such glasses by com-parison with l-atm experimental data for similar com-positions. Glass samples from basin paralavas analyzedby rvrre indicate minimum melt temperatures of 1020 to1400 "C (Table 2). These temperatures were determinedby projecting the major components of the glass onto theliquidus of appropriate experimentally deterrnined l-atmmelting diagrams (e.g., Schairer and Bowen, 1947;Schairer, 1950; Osborn and Muan, 1960). Because mostternary melting diagrams are constructed in terms ofmineral components on a weight percentage basis (e.g.,An-Ab-Or), paralava glass compositions were first com-puted in terms of their CIPW normative equivalents (inweight percent) and renormalized in terms of the threemost abundant components. For most glass composi-tions, the normative phases are closely approximated bythe ternary system CaAlrSirO'-KAlSi3O8-SiO, (Fig. 6). A

CoAl25i2Cs

A N O R ' H ' I E

L E U C I I E

94 COSCA ET AL.: PYROMETAMORPHIC ROCKS

KArS50"

Fig. 6. Norrnative compositions of paralava glass (filled cir-cles) projected onto the liquidus of the system CaAl.Si.Or-KAlSi3O8-SiO, (redrawn from Schairer and Bowen, 1947).

few ofthe glasses plot near or along the cotectic betweenanorthite and tridymite, whereas most glasses plotwholly within the anorthite field, indicating somewhathigher minimum temperatures. Glass compositions notpresented in Figure 6 are better represented by other sys-tems, usually KAlSirOr-NaAlSi3O8-SiO, or AlrOr-KAl-si308-sior.

For some glass compositions (e.g., 83EM5-A; DR-4A)the projection scheme described above is complicated bythe presence of a fourth major normative component.Because few experimental data are available for quater-nary systems approximating these glass compositions, anaverage minimum temperature determined from differ-ent ternary systems is presented in Table 2. One glasscomposition (KM-22) best represented in the systemAlrO3-KAlSi3Os-SiO, plots in a liquidus field with steepisotherms, and the minimum temperature estimated forthis glass is probably subject to somewhat larger system-atic error. The presence of additional glass components,primarily FerO. and PrOr, may reduce these temperatureestimates by ca. 20-40 "C.

Ifpresent, water, carbon dioxide, or other volatiles couldshift the liquidi of paralava to somewhat lower temper-atures relative to the anhydrous case. Although the effectof water on lowering the liquidus is minor at pressuresnear I bar, it should be considered, especially ifgroundwater was involved in the burning process. At high tem-peratures, however, any water would probably dissociateto H, and form either CO or COr, depending on oxygenfugacity. Furthermore, the lack of hydrous phases and agenerally small loss on ignition (LOI) (Table 3) is evi-dence against significant HrO in the paralavas examined.Therefore, despite the limitations in quanti$,ing volatileconcentrations, the water content is probably low, andthe calculated minimum temperatures are probably reli-

able. In general, corundum-normative compositionsyielded the highest minimum temperatures of formation(Table 2).

Individual residual detrital quartz grains often have adistinct melt developed along their margins, which gradesinto a matrix glass (Fig. 3D). Analyses indicate that theglass around the quartz is slightly enriched in silica rela-tive to the matrix glass (GR-1lA and -l lB; C58-A and-B, Table 2) and most likely represents a quenched (tran-sient) diffusion profile. xno studies of some detrital SiO,grains with glass rinds reveal the presence of low-quartz,tridymite, and cristobalite. Evidently, local heating andquenching in some cases is Ibst enough to prevent thecomplete conversion of the detrital quartz to tridymite orcristobalite. High quartz melts at approximately 1423 "C,whereas cristobalite melts at a significantly higher tem-perature of 1723 "C (Chase et al., 1985). However, thedifferential substitution of even minor additional com-ponents in the SiO, phases may markedly affect the phasediagrarn (Essene, 1982). Therefore, solid solutions com-bined with rapidly varying temperature may lead to themetastable formation and/or persistence of one or moreof the polymorphs.

Temperatures of melting determined from under-ground coal-gasification experiments in the Powder RiverBasin have been measured directly (> 1100 'C) and com-pared to estimates based completely on petrographic ob-servations (Craig et a1., 1983; Ethridge et al., 1983;Youngberg et al., 1983a, 1983b). These workers estimat-ed the temperatures of melting to be in the range of 1200to 1400 { based on the melting points of simple phasessuch as cristobalite, cordierite, mullite, plagioclase, andtridymite. One caveat to this approach is that fusion tem-peratures for a given phase may be greatly reduced byaddition of solid solutions or fluxing by a liquid. Forexample, the growth of cristobalite phenocrysts or themelting of cristobalite in paralavas (or lavas) by no meansrequires a temperature of 1723 "C (unless the liquid thatforms is pure SiOr). Therefore, the appearance or disap-pearance of a solid phase in multicomponent melts rnaylead to erroneously high estimates of temperature andshould not be used for geothermometry.

Wuor,n-nocK coMPosrrIoNS oFCLINKER AND PARALAVA

Samples of clinker that were examined from the Pow-der River Basin display variations of up to l5 wto/o amongits major oxides (SiO,, Al,O3, KrO, and CaO; Table 3).This variation probably reflects local heterogeneity relat-ed to small changes in the original sedimentary rock. Incontrast, paralava compositions are much more variableand, when compared to compositions of clinker from thesame rock, display marked enrichments in Fe, Mg, andCa and depletions in Si, Al, and K (e.g., 83EU-3A, -3B;DR4A-1, -2;Table 3). From the preliminary composi-tional data in Table 3 and the assumption that paralavais derived from fused clinker, it does not appear possibleto generate any of the paralava compositions fiom whole-

COSCA ET AL.: PYROMETAMORPHIC ROCKS

TABLE 2, Glass comoositions and CIPW norms of Powder River Basin paralavas

95

Sample: KM-s KM-22 GR. l1A GR-118 83EM-5A 83EM-58 G-11 G-7 83EU-3A

sio,Tio,AlrosFe.O.MgoCaONaroKroD A

F2

F:OTotal

o

orab

dienacnemtiaprufl

r,," fc)

83.030.188.380.440.040.380.226 0 00 0 50.150.06

98.81

s9.20| . J 3

35 891 8 80.530.00 .100.00.450 00 1 20 . 1 80 3 0

1 290

71 620 .14

13.471 . 1 80.410.300 1 68520 .130 .150.06

96.02

39 17428

52.431 .41U . J O

0.01 .060.01 2 30.00.310 1 50.32

1 400

69.460.53

'14.64

3.770.982.920 . 1 76 7 80.060 .150.06

99.40

35.632.25

40 311 4 5

I J . I J

0.02.450.03.750.00 . 1 4U . C J

0.30

1240

72 310.40

13.582.5'l0.742.560.217.270 0 3o.080.03

99.66

37 120.98

43.101.79

12.000.01 .850.02.520.00.070.400.16

1230

71 22o52

15.530.810.063.641.865.090 . 1 20 0 5o.02

98.88

33.880.75

30.4215.921 7 1 70.00 .150.00.820.0a.280.530.08

1090

76.370.61

12 350.50U . I J

1.051.45o-zt0.200 . 1 00.04

98.9942.761.99

37.4312.403.340.00.320 00.510.00.460.62o.17

1 100

75.240.04

11.702.770 1 6o . 1 70.359.760.12

n.o.n o -

100 31

35.470 5 4

57.50z . J c

0.060.00.400.02-760.00.280.040.0

1 040

74.651 .06

12.811 .580 -141 1 40.367.800 .17

n.o.n.o.99.71

40.652 . 1 2

46.233.064.560.00.350.01 .580 00.391 .060.0

1 080

75.640 0 6

12.781 . 2 10.063.23

n.d.

026n.o.n.o.100.46

41.710.0

42.470 0

13.490.320.00.01 .200.150.600.00 0

1250

Sample 83EU-38 GR8.A GR8-B AM-13 c-58A c-58B DR-4A 83E-1 3

sio,Tio,Al,o3FerO.MgoCaONa.OKrOPrOu

F:OTotal

orao

dien

nemtlaprUfl

r.. cc)

/ .J 55

n.d.13 .610 8 80.041 . 1 10.05

10.420 . 1 10 . 1 80 0 8

99 71

31.340.48

o t . / J

0.424.840 00 1 00 00.880.00 .120.00.09

1 080

72040 7 6

14.211.560.8s3.000.177.050.320.640.27

99.68

37.321.60

41 .711 4 4

12 830 02 . 1 20 01.560 00.350.760.30

1240

73.49

13 021 0 60 7 72.020.127.710.230-090.04

98.82

38 641 1 5

46.091.039.220.01 .940.01 0 70.0o.320.540.0

1210

71.830.14

16.061 5 00.065.210.115.680.79

n.o.n.o.

100 99

38.93't 2133.240.92

23.000.00 . 1 50.01 .490.00 9 20.140.0

1 3 1 0

76.250.27

11.520.710-160.490 4 39.610 . 1 10.140.06

99.63

36.530.13

57.003.650.770.00.400.00.710.00250.270.28

1020

74 97o.22

1 1 . 6 80.800 . 1 10.480.799.930 1 80.270.11

99.10

32.730.0

59.1 I4.830.00.0o.241.680.220.540.210.0o.27

1 030 1 030 1 200

72.62 71.281.02 0.35

13.29 14.523.06 0.590.72 0.011 .86 1.220.55 0.038.05 12.250.05 0.220.05 0.370.02 0.16

101.22 100.41

33.35 22.510.46 0.0

47.00 72.064.59 0.258.64 3.290.0 0.01.77 0.020.0 0.03.02 0.590.0 0.s60.05 0.251.01 0 .120.10 0 35

Note: T^t^: minimum temperature; n d. : not determined.

sale melting of clinker. Although it is possible that clinkermay represent a more refractory composition than para-lava, evidence from both field and thin-section observa-tions indicate that given the appropriate time, clasts ofclinker melt when in contact with paralava. The moststriking compositional difference between the clinkers andparalavas a\alyzed for this study is the concentrations ofFe. The mechanism for Fe enrichment in paralavas is notentirely understood. The Fe may come from the break-down of detrital ferromagnesian phases in the melted sed-

imentary rocks, but chemical analyses of unfused clinkerin this investigation reveal low total Fe contents (Table3). Alternatively, Fe that was formerly in thin layers offerruginous nodules (Table 3), or perhaps in unburnedcoal, may now be concentrated in the clinker after havingbeen transferred by liquids or gases that filtered throughfractures in the clinker.

Liquid immiscibility has been proposed as a mecha-nism for the formation of silicate and phosphatic liquidsformed in combustion-metamorphic rocks associated with

96 COSCA ET AL.: PYROMETAMORPHIC ROCKS

Trel-e 3. Whole-rock compositions and CIPW norms of Powder River Basin clinker and paralava

Sample: 83EU-3ARock type: Clinker

83EU-38Paralava

KM-4Clinker

DR-9Clinker

DR4A.1Clinker

DR4A-2Paralava

DR2Fe nodule

sio,Tio,Alr03FerO"MnOMgoCaONa"OKrOP.ouHrOLOt-

Total

o

orAD

dienwomti lhemti

ru

74.310 8 3

16 393.620.020.79u . t o0 3 42.290.040.210.21

98.79

63.1513.3713.752930.540.02.000.00 .00.043.680.00.09082

48.950.60

1 1 . 3 138.080.261.301 . 1 40 0 11 . 6 10.260.96

-1.42103.52

39.198 . 1 29 5 40.08J . V /

0.03.250.00 00.56

38.1 I0.00.600.31

o J . o /

0.5210 052.640.065.09

15.050 . 1 32.250.290.020.06

26.780.0

13.37' t . 1 120 2927.480.0b.bb

0 00 1 32.6s1 . 1 20.670.0

68.880.74

16.254 .180.042.892 9 70 4 82.650 .160 .100.09

99.24

46.28L O l

15 854 . 1 1

13.850.07.290.00.00.094.230.00.370.70

66.980.86

21.853.200.011.45o.920.252.690.050.16

-1 0798.26

53 3217 3316.232.164.330.03.690 00.00.023.270.0o .120.87

50.040.71

21.776.940 . 1 32.61

14.00n.d.1 2 R

0.440.720.24

07 0n

1s.040.08.200.0

56.967.363.270.00.00.297.131.421 0 50.0

22.900.235.13

42.790.932.59

1 8 .190.090.448.090.220 . 1 1

101 .41

6.280.02.68o.78

12.6614.330.03 . 1 12.440.45

42.390.0

19.300.0

Note-'n.d. : not determined.. Includes HrO, COr, SO,

bituminous sedimentary rocks (Bentor et al., l98l). TheFe enrichments observed in paralava, when compared toclinker, may also be explained by liquid immiscibility.Comparison of experimental and natural occurrences ofbasalt formed at high temperature and conditions of highfo, indicate a field of liquid immiscibility that becomeslarger with increasedf, (Naslund, 1983). The position ofthis immiscibility field is outlined in Figure 7 togetherwith the position of whole-rock and glass compositionaldata from paralava. Paralava and glass compositions arestrikingly similar to natural and experimental immiscibleliquids, a fact which suggests that liquid immiscibilitymay be a viable mechanism for generating some paralavacompositions. If the Fe-rich paralava, which often formsflows and injections, represents an immiscible silicate liq-uid, then an Fe-depleted residue should remain else-where. One possible candidate for this residue is the para-lava at the coal-sedimentary rock interface; however,further work is necessary to test this hlpothesis. In mostgeologic environments where immiscible liquids havebeen observed, the immiscible phases are not greatly sep-arated. However, if liquid immiscibility is an effectivemechanism for generating the compositional variationobserved in paralavas, then a more efficient liquid sepa-ration phenomenon may be indicated.

OxrnlrroN-REDUCTToN RELATToNS

Oxidation (burning) ofcoal can readily reduce clinkerin a natural fashion that is analogous to the Fe-smeltingprocess. Indeed, small amounts of native iron have been

reported in clinkers from Edmonton, Canada (Tyrrell,1887) and South Australia (Baker, 1953). In the PowderRiver Basin, clinkers and paralavas are less reduced, withmany samples containing hematite, magnetite, and/orfayalite. Paralava and clinker containing magnetite and/or hematite acquire a high-intensity thermoremanentmagnetization that appears to faithfully record the am-bient geomagnetic field, allowing mapping of burn zones(Jones et al., 1982) and tracing of magnetic reversals inancient burns (Jones et al., 1984;' Cisowski and Fuller,1987). The demagnetization experiments of Jones et al.(1984) indicate that remanent magnetization is containedin titanomagnetite, hematite, and goethite, but that thepredominant magnetization in clinker resides in titano-magnetite. The highest magnetic susceptibilities in clink-er may develop from those sedimentary rocks that areleast permeable to oxygen (Hooper, 1984).

Mineral assemblages and compositions of paralavasmay be influenced partly by the composition and partialpressure of gas produced during coal combustion. As thegases are produced from burning coal, they may contin-ually mix with any additional fluids and/or gases present.Mixing of this sort may produce significant variation inthe composition of these gases, which may be calculatedby combining thermochemical data with chemical reac-tions in the system C-O-H. It is difficult to evaluate therole ofH-bearing gas species in these calculations becauseofa lack ofhydrous phases in paralavas. Although inter-action with ground water could conceivably introducesignificant HrO that may react to form H, and/or CHo

COSCA ET AL.: PYROMETAMORPHIC ROCKS 9'.l

FeO + FerO, *

+ MgO + MnO

Ti02+P

2

+ CoO

o,

Ir.rl t i . ''{tr'.

under more reducing conditions during coal combustion,we will assume a system dominated by C and O. Ulti-mately, direct gas sampling at active burning areas ormeasurement of fluid inclusions in the paralava pheno-crysts may lead to a better understanding of the gas com-position.

Examination of several paralava samples from a singleoutcrop reveals marked contrasts in mineral content thatmay reflect different oxidation states. For example, para-lava within the coal-ash layer described earlier may con-tain the phases pyroxene, melilite, anorthite, magnetite-ulvdspinel-hercynite solid solutions and glass, whereasparalava occurring in chimneys above the coal-ash zoneoften contain fayalite, tridymite, abundant magnetite-ul-vdspinel-hercynite solid solutions, cordierite, late hema-tite, and lesser glass. The contrasting mineral content re-quires steep gradients in /o,. Assuming that the coal-

l t t

sedimentary rock interface is in equilibrium with graphite(i.e., ar: 1), using tabulated AGo data for CO and CO,(Robinson et a1., 1982) the position of graphite stabilityin T-fo, space may be calculated by solving the followingsimultaneous reactions:

2 C + O , : 2 C O ( l )

and

C + Or : CO,

and

P,- : P.o t Pcoz. (3)

Figure 8 illustrates the position of this curve at surfacepressures. A paralava bearing the quartz + fayalite +magnetite (QFM) assemblage was collected from one toseveral meters above the coal-sedimentary rock inter-

\\\ I

It.o I

"\J' trJOtrrl

--

NorO + KrO + At2O3 sio2

Fig. 7. Field of liquid immiscibility for basalts determined experimentally under oxidizing conditions (1og,o fo,: l0-") (Naslund,1983); mt: magnetite; tr: tridymite; 2liq: two liquids. Paralava glasses (filled dots) and some coexisting paralava (hexagons)are shown in the left projection. Positions of experimental and natural coexisting immiscible liquids in basalts are shown on theright projection (data taken from Kuo et al., 1986).

(2)

98 COSCA ET AL.: PYROMETAMORPHIC ROCKS

N

o

o(-9oJ

20

- 4 0

500 900 1300 t700T ( " C )

Fig. 8. T-fo, plot of the following reactions: 4 Magnetite +O, : 6 Hematite (HM); 3 Fayalite * Oz : 2 Magnetite + 3Quartz (QFM); C + O, : CO,; and 2C + 02: 2CO, for con-ditions of P,", : Pco f P.o, and ac : l. Dot-dash line accountsfor reduced actiwity in the solid phases assuming ideal mixing.

face; this paralava places additional constraints on /or.Using tabulated free energies for fayalite (Fa), quartz, andmagnetite (Mt) (Robinson et al., 1982) and correcting forsolid solutions, the values of fo, at a given temperaturefor this reaction may be calculated by solving the follow-ing equation:

AG,: AGor + RZ ln(/".)(Xfl)' - nf h(Xf;X),. (4)

Figure 8 contains the position of this curve for the QFM-bearing paralava relative to pure QFM and relative toparalava in equilibrium with C. The gradients in/o, be-tween these paralavas therefore require variations of sev-eral logs units at high temperature. The composition ofthe gas phase associated with either of the above para-lavas, calculated with the assumptions Pn"ia : P.o * P.o,: P,o,: I bar, is represented in Figure 9. Paralava inequilibrium with graphite (i.e., at the coal-sedimentaryrock interface at I bar) has CO as the dominant gas speciesat temperatures above about 650 oC. Below that temper-ature, CO, is dominant. Pressure increases of severalhundred bars have little effect on these results.

Because the coal-ash paralavas indicate oxidation statesapproaching those of the hematite-magnetite buffer (Cos-ca and Peacor, 1987), they did not remain in equilibriumwith C. Remnant coal is often present, however, indicat-ing gradients of at least five orders of magnitude in f,(and a.) over a few centimeters to meters of the burningcoal bed.

No vertical gradients in fo, were detected from chem-ical analyses ofthe oxide phases at different elevations ina chimney structure. Oxides at various levels of the chim-ney consist of solid solutions of magnetite-hercynite-ul-

t o

p' o o r so . 5

500 800 l ooo 1200 1400 1600 1800

T ( 'C)

Fig. 9. Plot ofpartial pressures ofCO and CO, in the systemC-O calculated for a total pressure of I bar and for conditionsofeither graphite or QFM present.

vdspinel often with hematite-ilmenite and pseudobrook-ite lamellae. The complex exsolution and oxidationpatterns in the oxides may in part be due to changes in

6, with time. This interpretation is suggested by hematiteovergrowths on or exsolutions from many of the mag-netite solid solutions. Late hematite is also observed con-centrated around vesicles, but it appears unrelated to ver-tical position in the chimney (Fig. 4F). Factors locallyaffecting progressive oxidation may be locally quite vari-able, but presumably ground water, rock porosity, frac-ture densities, and fluids released from clay minerals couldall be important.

DrscussroN

When interpreting the chemical variations observed inparalava, it is difficult to evaluate the exact degree towhich melting or crystallization processes (or other pro-cesses) have been involved. However, certain field andpetrographic observations combined with analytical datamay help constrain their relative importance. For ex-ample, partial melting is certainly a mechanism at leastpartly responsible for the local melts formed within frac-tured clinker, although formation of other paralava, in-cluding that found in chimney structures or slaglike blocks,probably involved large-scale melting. Because the para-lava-forming liquids reached extremely high tempera-tures, mechanisms such as fractional melting and/or in-congruent volatilization, vapor transport, liquid mixing,or liquid immiscibility may have significantly altered theoriginal liquidus compositions. Such processes may ex-plain the unusual enrichments of Fe in paralava or howcertain paralava compositions are driven toward silicaundersaturation. Alternatively, field evidence from chim-neys and flow structures indicates transportation ofliquid(+ phenocrysts) that may have digested foreign sediment(clinker) enroute, or mixed with separate fluid(s) (Coscaet al., 1988). Other possibilities like fractional crystalli-zation and its effects on liquid composition may be im-

a -a nu v U O a - _

- G R A P H I T E P R E S E N T- _ Q F M

P t o t = | b o r

P = P + D' T o t ' C O ' C O .

COSCA ET AL.: PYROMETAMORPHIC ROCKS 99

portant especially when interpreting phenocryst-bearingsamples. Crystallization of high-temperature, refractoryphases such as spinel, melilite, or mullite could substan-tially alter any residual glass composition relative to theoriginal liquidus composition. Whether one process dom-inates another is currently difrcult to evaluate, but maybe better understood with a detailed evaluation of manyglass and bulk-rock analyses. It is clear, however, thatclinker and paralava have distinct compositional char-acteristics and involve some type of differential elementalremobilization.

AcrNowloocMENTS

We wrsh to acknowledge the Kerr McGee Co. for allowing access to theJacobs Ranch and Clovis Point Mines, and the Nerco Mining Co. foraccess to the Antelope Coal Mine. B. H Wilkinson is thanked for drawingour attention to combustion metamorphic outcrops in Utah. Kudos to C.E Henderson for maintaining the electron microprobe in excellent work-ing condition and to R. J. Arculus for helpful discussions and for instruc-tion with xrn analyses. Discussions with A. Afifi, A T. Anderson, and S.R Bohlen and comments by R. B Finkelman, W. J. Harrison, P. JModreski, W P Nash, and an anonymous reviewer are gratefully ac-knowledged. This study was supported by grants from the GeologicalSociety of America, Sigma Xi, and the Turner Fund of the University ofMichigan to M A C Some aspects ofthis research were additionally sup-ported by NSF grant EAR-84-08169 to E J.E The electron-microprobeanalyzer used in this work was acquired under Grant EAR-82- I 2764 fromthe NSF

RprnnnNcns crrnnAgrell, S O., and Langley, J.M (1958) The dolerite plug at Tierebullaigh,

near Cushendall, County Antrim Proceedings ofthe Royal Insh Acad-emy, 59, 92-128

Baker, G. (1953) Naturally fused coal ash from Leigh Creek, South Aus-tralia. Transactions ofthe Royal Society ofSouthern Australia, 76, l-20.

Bastin, E S. (1905) Note on baked clays and natural slags in eastern Wy-oming Journal of Geology, 13,408412.

Bauer, L.P. (1972) The alteration of sedimentary rocks due to burningcoal in the Powder River Basin, Campbell County, Wyoming. M.S.thesis, Texas Tech University, Lubbock, Texas.

Bentor, Y.K (1984) Combustion-metamorphic glasses. Joumal of Non-Crystalline Solids, 67, 443448.

Bentor, Y.K., and Kastner, M (1976) Combustion metamorphism insouthern Califomia. Science, 193, 486-488

Bentor, Y K., Kastner, M., Perlman, I., and Yellin, Y. (1981) Combustionmetamorphism of bituminous sediments and the formation of melts ofgranitic and sedimentary composition Geochimica et CosmochimicaActa.45.2229-2255

Brady, L.F., and Greig, J.W. (1939) Note on the temperarure attained ina burning coal seam. American Joumal of Science,237, 'l16-119.

Budai, C.M., and Cummings, M.L (1984) Geologic conrrols of coal burnsand oxidized zones in the Antelope Coal Field, Converse County, Wy-oming. In R L Houghton, Ed , Symposium on the geology of RockyMountain coal,84-1, p. 168-183 Geological Society ofNorth Dakota,Bismarck, North Dakota

Bustin, R.M., and Mathews, W.H. (1982) In situ gasification of coal, anatural example: History, petrology, and mechanics of combustion.Canadian Joumal ofEarth Sciences, 19. 514-523.

Chase, M W., Jr., Davies, C A , Downey, J.R , Jr., Frurip, D.J , McDonald,R A , and Syverud, A N. (1985) JANAF thermochemical tables (thirdedition) In D.R. Lide, Ed , Journal ofPhysical and Chemical ReferenceDara, 14, 1856 p.

Church, B.N, Matheson, A., and Hora, ZD. (1979) Combustion meta-morphism in the Hat Creek area, British Columbia. Canadian JoumalofEarth Sciences. 16. 1882-1887.

Cisowski, S M , and Fuller, M. (1987) The generation of magnetic anom-

alies by combustion metamorphism of sedimentary rock, and its sig-nificance to hydrocarbon exploration Geological Society of AmericaBulletin, 99, 2l-29

Clark, B.H., and Peacor, D.R. (1987) Mineralogic and textural transitionsin shales due to pyrometamorphism and incipient melting (abs.). EOS,68 .452 .

Coates, D A , and Naeser, C.w. (1984) Map showing fission-track ages ofclinker in the Rochelle Hills, southem Campbell County Wyoming.U.S. Geologrcal Survey Miscellaneous Investigations Series Map l-1462.

Cosca, M.A., and Essene, E J. (1985) Paralava chemistry and conditionsof formation, Powder River Basin, Wyoming (abs.) EOS, 66, 396.

Cosca, M A, and Peacor, D.R. (1987) Chemistry and structure of es-seneite, CaFeAlSiOu, a new pyroxene produced by pyrometamorphism.American Mineralogist, 7 2, | 48-l 56.

Cosca, M.A., Peacor, D.R., and Essene, E.J. (1985) The chemistry andstructure of a natural ferri-aluminum tschermakitic pyroxene produced

by pyrometamorphism. Geological Society of America Abstracts withPrograms, 17,553.

Cosca, M.A., Rouse, R.C, and Essene, E.J. (1988) Dorrite Ca,(Mg,Fe;*)-(Alosir)Oro, a new member of the aenigmatite group from a pyrome-tamorphic melt-rock American Mineralogist, 7 3, 1440-l 448.

Craig, G.N., Burns, L.K., Ethridge, F.G., Laughter, T.F, and Youngberg,A D (1983) Overburden characterization and post-bum study of theHoe Creek, Wyoming Underground Coal Gasification site and com-parison with the Hanna, Wyoming site U.S. Department of EnergyPublication DOE /LC/ 10496-Tl, I88 p.

Denson, N M., and Keefer, W R (1974) Map of the Wyodak-Andersoncoal bed in the Gillette area, Campbell County, Wyoming. U.S. Geo-logical Survey Miscellaneous Investigations Series Map I-848-D, scalel : I 25 000.

Essene, E J ( I 982) Geologic thermometry and barometry. MineralogicalSociety of America Reviews in Mineralogy, 10, 153-206.

Essene, E J., Coates, D A, Geissman, J.W., and Simmons, W.B. (1984)Paralavas formed by burning ofcoal beds (abs.). EOS, 65, 300.

Ethridge, FG., Burns, L.K., Alexander, W.G., Craig, G N., and Young-berg, A D. (l 983) Overburden characterization and post-bum study ofthe Hoe Creek, Wyoming Underground Coal Gasification site andcomparison with the Hanna, Wyoming site. U.S. Department of EnergyPublication DOE/LC/ 10705-Tl, 185 p.

Fermor, L.L. (1918) Preliminary note on the buming of coal seams at theoutcrop. Transactions ofthe Mining, Geological, and Metallurgical In-stitute of India, 12, 50-63.

Foit, F.F, Jr., Hooper, R.L, and Rosenberg, P.E. (1987) An unusualpyroxene, melilite, and iron oxide mineral assemblage in a coal-firebuchite from Buffalo, Wyoming American Mineralogist, 72, 137-147.

Glass, G.B ( 1 976) Update on the Powder River coal basin. In R.B. Lau-don, Ed., Geology and energy resources of the Powder River Basin,28th Annual Field Conference Guidebook, p.209-220 Wyoming Geo-logical Association, Casper, Wyoming.

Gleason, J.D, and Kyser, T.K. (1984) Stable isotope compositions ofgases and vegetation near naturally burning coal. Nature, 307, 254-257

Grapes, R.H. (1986) Melting and thermal reconstitution of pelitic xe-noliths, Wehr Volcano, East Eifel, West Germany Journal of Petrol-ogy,27,343-396.

Hooper, R.L. (1982) Mineralogy of a coal bum near Kemmerer, Wyo-ming. M.S. thesis, washington State University, Pullman, Washington.

- (l 984) Factors affecting the magnetic susceptibility of baked rocksabove a burned coal seam. Geological Society of America Abstractswith Programs, 16, 543.

Hooper, R.L., and Foit, F.F. (1986) Pyroxenes and melilites with exten-sive tetrahedral Fe3* substitution from a coal-fire buchite near Buffalo,Wyoming (abs.). EOS, 67, 1270.

Jackson, K.C. (1981) Clinker genesis and the comparison of naturallyoccurring clinker to compounds in the system FeO-AlOySiOr. B.S.thesis, Beloit College, Beloit, Wisconsin.

Jones, A H., Geissman, J W., Coates, D.A., Naeser, C.W., and Heffem,E L (1982) Paleomagnetism of Powder River Basin clinker, Wyomingand Montana: Documentation of Bruhnes and Matuyama burning ofTertiary coals (abs.) EOS, 63, 304.

Jones, A H, Geissman, J.W., and Coates, D.A. (1984) Clinker deposits,

100 COSCA ET AL.: PYROMETAMORPHIC ROCKS

Powder River Basin, Wyoming and Montana: A new source of high- Schairer, J.F. (1950) The alkali-feldspar join in the system NaAlSiOo-fidelity paleomagnetic data for the Quaternary. Geophysical Research KAlSiOo-SiOr. Joumal of Geology, 58, 512-517.Letters, I 1, 123l-1234. Schairer, J.F, and Bowen, N L. (1947) The system leucite-anorthite-sili-

Kuo, L-C., Lee, J.H., Essene, E.J., and Peacor, D.R. (1986) Occurrence, ca. Geological Society ofFinland Bulletin, 20,67-87.chemistry, and origin of immiscible silicate glasses in a tholeiitic basalt: Schneider, H., and Majdic, A (1984) Iron incorporation in tridymite andA TEM/AEM study Contributions to Mineralogy and Petrology, 94, crislobalite Neues Jahrbuch fiir Mineralogie Monatshefte, 559-568.90-98 -(1985) Titanium incorporation in tridymite Neues Jahrbuch fiir

Lofgen, G., Donaldson, C.H., Williams, R J., Mullins, O., Jr., and Us- Mineralogie Monatshefte, 418426selman, T. (1974) Experimentally reproduced textures and mineral Seifert-Kraus, U., and Schneider, H. (1984) Cation distribution betweenchemistry of Apollo l5 quartz normative basalt. In W.A. Gose, Ed., cristobalite, tridymite, and coexisting glass phase in used silica bricks.Proceedings of the 5th Lunar and Planetary Science Conference, l, Ceramics International, 10, 135-142.549-569. Pergamon, New York. Sigsby, R.J (l 966) "Scoria" of North Dakota. Ph.D thesis, University

Matthews, A., and Gross, S. (1980) Petrologic evolution of the "Mottled of North Dakota, Grand Forks, North DakotaZone" (Hatrunm) metamorphic complex of Israel. Israel Joumal of Smith, D.G W (1969) Pyrometamorphism of phyllites by a dolerite plug.Earth Science, 29,93-106. Joumal ofPetrology, 10,20-55.

May, P.R. (1954) Clinker In North Dakota Geological Society Guide- Tilley, CE. (1924) Contact metamorphism in the Comrie area of thebook, Southwestern North Dakota Field Conference, 18-19 Perthshire Highland. Quarterly Journal of the Geological Society of

Mclintock, W.F.P. (1932) On the metamorphism produced by the com- London, 80,22-71.bustionofhydrocarbonsintheTeniarysedimentsofsouth-westPersia. Tyrrell,JB (1887)Naturallyreducediron AmericanJoumal ofScience,Mineralogical Magazine, 23,207-227 33,73.

Menit, R.D. (1985) Review of coking phenomena in relation to an oc- Venkatesh,V (1952)Developmentandgrowthofcordieriteinpara-lavacurrence ofprismatically fractured natural coke from the Castle Moun- American Mineralogist, 37, 83 l-848.tainMine,Matanuskacoalfield,Alaska. IntemationalJoumalofCoal West, W.E. (1964) The Powder River Basin. In E.R. Keller and R.E.Geology, 4, 281-298. Glaze, Eds., Highway geology of Wyoming, p. 20-28 Wyoming Geo-

Modreski, P.J., and Herring, J.R (1985) High-temperature hematite and logical Association, Casper, Wyoming.spinelphasesin i ron-r ichslagsproducedinshalesaboveburnedcoal Whitworth,H.F (1958)Theoccurenceofsomefusedsedimentaryrocksbeds. Geological Socrety of America Abstracts with Programs, 17, 666. at Ravensworth, N.S.W. Journal of the Royal Society of New South

Naslund, H R (1983) The effect of oxygen fugacity on liquid immiscibil- Wales, 92,204-210ity in iron-bearing silicate melts American Journal of Science, 283, Wyllie, P.J (1961) Fusion ofa Torridonian sandstone by a picrite sill in1034-1059. Soay (Hebrides). Joumal ofPetrology,2, l-37

Osborn, E.F, and Muan, A. (1960) Phase equilibrium diagrams of oxide Youngberg, A D, McClurg, J.E., and Schmitt, J.G. (1983a) Depositionalsystems. The American Ceramic Society, Columbus, Ohio enyironments, subsurface stratigraphy, and post-burn characterization

Phemister, J (1942) Note on fused spent shale from a retort at Pum- of the Paleocene-Eocene Hanna, Wyoming Underground Coal Gas-pherston, Midlothian. Transactions of the Geological Society of Glas- ification Site Hanna III experiment U S. Department of Energy Pub-gow, 20, 238-246 lication DOE/LC/RT-83-1, 120 p.

Raask, E (1985) Mineral rmpurities in coal combustion: Behavior, prob- -(1983b) Depositional environments, subsurface stratigraphy, andlems, and remedial measures, 484 p. Hemisphere Publishing, Wash- post-burn characterization of the Paleocene-Eocene Hanna Formationington, D.C a1 the Hanna, Wyoming Underground Coal Gasilication Site Hanna

Robinson, G.R., Jr., Haas, J.L., Jr., Schafer, CM, and Haselton, HT, II phase I experiment. US. Department of Energy Publication DOE/Jr. (1982) Thermodynamic and thermophysical properties of selected LClRl-83-2, 163 pphases in the MgO-SiOr-HrO-COr, CaO-AtO3-SiOr-HrO-COr, and Fe-FeO-FerOr-SiO, chemical systems with special emphasis on the prop-erties ofbasalts and their mineral component. U.S. Geological SurveyOpen-File Report 83-79, 429 p

Rogers, G S. (1918) Baked shale and slag formed by the burning ofcoal MnNuscnrrr RECETwD Apnrr 25, 1988beds. U.S Geological Survey Professional Paper 108-.4, l-10 Mrxuscnrpr ACcEnTED Ssmsusen 27, 1988