Wit" 42.49:? 253.811" " -‘ ’ '1 \x l .kk’sl. u w- I h x4153.- ‘,. r «‘1: V '1‘; "1‘4“- ! ,- rewfr mil ,":5‘“ Yvu‘ _ :2" 4‘;- or.» ‘r.. < .. ,,,_, mnm ,. _ ~. ‘1“ . f'fik'i-‘r , a “ .i .51.! 11s? a j , at: f 9.", ‘<”& ). “MA A $45 C 4 7J6! b75507 lllllllllllllllllllll l LIBRARY” E Michigan State: University This is to certify that the thesis entitled CONTROLS 0N ELEMENT DISTRIBUTIONS: NEGAUNEE IRONjFORMATION, EMPIRE MINE, PALMER, MICHIGAN presented by Susan Elizabeth Crissman has been accepted towards fulfillment of the requirements for Master's Geology degree in Major professor November 7, 1988 Date *— 0-7639 MSUis an Affirmative Action/Equal Opportunity Institution MSU LIBRARIES “ RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. CONTROLS ON ELEMENT DISTRIBUTIONS: NEGAUNEE IRON-FORMATION, EMPIRE MINE, PALMER, MICHIGAN BY Susan Elizabeth Crissman A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Geological Sciences 1988 ABSTRACT CONTROLS ON ELEMENT DISTRIBUTIONS: NEGAUNEE IRON-FORMATION, EMPIRE MINE, PALMER, MICHIGAN BY Susan Elizabeth Crissman Core samples from the Proterozoic Negaunee Iron- Formation (NIF) and underlying Siamo Slate Formation at the Empire Mine were analyzed for magnetite, major elements, rare earth elements (REE), Cr, and Th. The REE and Cr con- centrations were determined for eight magnetite separates. The constant TioZ/A1203 ratio of 0.033, was an indication that the detrital sediments of the NIF and Siamo Formation were derived from a common source, probably the Archean basement gneiss. The formations can be discriminated by major and trace element correlations. The NIF sediments have higher concentrations of REE and Th, and a lower concentration of Cr relative to A1203 than the Siamo sediments. These differences are attributed to contrasting physicochemical conditions existing during deposition. The aluminosilicate minerals (chlorite, stilpnomelane, and biotite) control the trace element concentrations, including Cr, and do not fractionate the REE. When this phase is not significant, the heavy REE are preferentially incorporated in magnetite. To Mom and Dad iii ACKNOWLEDGEMENTS It is with sincere pleasure that I express my appreci— ation to the members of my committee, Professors D.T. Long and D.F. Sibley, for their help and advice. I am especially indebted to my principle advisor, Professor J.T. Wilband, who, in addition to offering critical advice, support, and encouragement, wrote many of the computer programs used in this investigation. My research relied heavily upon the work of Susan Tituskin, and I am grateful to her for permission to use her data. I wish to thank Tsu-Ming Han of Cleveland Cliffs Iron Co. for helpful discussions and for showing me around the mine. The staff of C.C.I. is also to be thanked for the use of the Satmagan. And, I very much appreciate the technical support from Loretta, Cathy and Mona in the Geology office, and especially Diane in the library. I owe my sanity to my family and friends who were always there with words of encouragement; especially Molly, Chris, Lou, Kay, Chris, LaVon, Brenda, Mary, Gene, Letla, Elina, and Manda. Finally, I might never have attempted this if it were not for the courage that Jim gave me. It would be better if he were here. iv TABLE OF CONTENTS LIST OF TABLESOIO-IIOO...-......COOCOOOOOOOCI. OOOOOOOO Vii LIST OF FIGURES..... ........ ............. ............ . viii INTRODUCTION ....... ......OOOOOOOOOOOOIIOIIO ........... 1 Purpose and Scope of Research..................... 1 Location.O0.0000...O0.000.000.000.0000.0.00.0.0... 2 GEOLOGIC AND TECTONIC SETTING................ ......... 5 Geologic Setting........................... ....... 5 Tectonic Setting .................................. 7 PREVIOUS WORK............. ...... ............. ......... 10 ANALYTICAL TECHNIQUESOCCOOOCOOOIOOOO OOOOOOOOOOOOOOOOOO ll MINERAIIOGYO......OOOOOOOOOOOO0.0I... 0000000000 O ....... 12 Mineralogy of the NIF at the Empire Mine .......... 12 Mineralogy of the Sample Groups Used in This Study......... ............................. 13 PETROCHEMISTRY... ..... . ............ ........ ........... 17 Chemical Data.. ..... ............... ............... 17 Carbonate... ............... ........ ............... 22 Chert....... ...... . ...... ............ ........... .. 25 Magnetite-rich Ore...... ................. ......... 25 Interbedded Clastic........... ........ ............ 25 Siamo Slate........................ ......... . ..... 26 GEOCHEMISTRY ..... . .................... ... ............. 27 Introduction ...................................... 27 Statistical Method: Factor Analysis ............... 27 Magnetite and Iron Versus REE.. ................... 33 Silica versus Magnetite and Iron .......... . ....... 36 Silica versus Carbonate ........ . .................. Silica versus REE...... ........................... Summary.GOOOOQOOOOOOOOCOIOIII.OIOIOOOOOOOO ........ CRYSTALLOCHEMICAL CONTROLS ON REE FRACTIONATION AND CI‘IROMIUM CONCENTRATIONOOO0.000......O... Introduction.................................. REE and Cr in Magnetite Separates........ ..... Effect of Magnetite on REE Fractionation...... Effect of Magnetite on Cr Concentration...... ..... IMPLICATIONS OF ELEMENT ASSOCIATIONS IN CLASTIC SEDIMENTS.. ...... .............. ................. Introduction............................. ........ Titanium and Aluminum as Source Indicators.... Distinguishing Characteristics of the NIF and Siamo Slate.........OQOOIOOI.IIIOOIOO... Discussion..... ........ ................... ........ Summary....................................... 0 SUMMARY AND CONCLUSIONS... ........ .................... APPENDIX: Q-MODE FACTOR LOADINGS...... ........... ..... LIST OF REFERENCES.............. ...................... vi 39 43 46 49 49 50 54 56 59 59 60 63 71 73 76 77 Table Table Table Table Table Table Table LIST OF TABLES Major Oxides (%) .................. ........... 18 Magnetite and *Normalized Mineralogy (%) ..... 19 Trace Elements (ppm) ......................... 21 Sample Group Chemical Summary ........ ........ 23 Factor Analysis Summary ................ ...... ‘29 Magnetite Separate (M) and Whole Rock (T) Trace Element Data (ppm) ............... ....... 51 Characteristics of Best-Fit Lines ............ 65 Figure Figure Figure Figure Figure Figure Figure Figure Figure LIST OF FIGURES Generalized Regional Geology of the Marq‘lette TroughOOOOOOOOII......OOOQOOQOOOOOO 4 Samples plotted as loadings on pairs of factors defined by Q-mode factor analysis.......... ....... .................... 30 Correlation diagrams for samples with factor loadings that range from high on magnetite and total Fe to high on total REE.. 34 Correlation diagrams for samples with factor loadings that range from high on $102 to high on total iron and magnetite..... 38 Correlation diagrams for samples with factor loadings that range from high on SiO2 to high on carbonate ....... . ............ 40 Correlation diagrams for samples with factor loadings that range from high on Sio2 to high on total REE........... ......... 44 Chondrite normalized REE abundances in magnetite separate and whole-rock samples.... 52 Correlation diagrams including all samples... 61 3102 vs. La/Th for the elastic sediments ..... 67 viii INTRODUCTION Purpose and Scope of Research The approximately 2.0 Ga old Negaunee Iron-Formation (NIF) at the Empire Mine in Palmer, Michigan provides an excellent, albeit complex, spectrum of rock types to investigate elemental distribution and mineralogic variations of a typical "Superior—type" banded iron- formation. This orebody, similar in many respects to the almost contemporaneous Sokoman Iron-Formation of Labrador (Dimroth and Chauvel, 1973; Fryer, 1977a; Lesher, 1978) and the Brockman Iron—Formation of Australia (Trendall, 1973; Morris, 1980), is considered to have been precipitated dominantly as a chert-siderite (Han, in Gair, 1975), although hematite may have been a primary iron mineral in other parts of the NIF (Han, 1971, 1982). While metamorphism of the NIF at the Empire Mine is minimal (chlorite facies) there are considerable diagenetic changes (Han, 1962). Tituskin (1983) used whole-rock selected trace element data including the rare earth elements (REE), as well as microprobe data to evaluate element distribution patterns in the Neguanee Iron—Formation and the underlying Siamo Slate Formation. She broadly classified these samples, 2 based mostly on thin section work, into the clastics of the Siamo Slate, the interbedded clastics of the NIF, and the chemical precipitates of the NIF. The latter were divided into chert-rich, magnetite-rich, and carbonate-rich types. Without major element data, evaluation of mineralogical controls could not be quantitatively constrained. This investigation used the same samples, and provides whole-rock major oxide, loss on ignition, and magnetite separate data to refine the sample group classification model of Tituskin. Major element data were used here in normative schemes, to constrain the iron species assignment between carbonate (e.g. Fe2+ with loss on ignition), magnetite (direct measurement), hematite, and iron silicates (characterized by high A1203). Correlations between Tituskin’s trace element data and major oxide data from this study provided a means to evaluate the inter- and intra- group element distribution variations in terms of mineralogical and physicochemical controls. Trace element data from magnetite separates were used to evaluate the effect of this major mineral on the trace element distribution in the NIF. Location The samples used in this investigation were collected from the Empire Mine, operated by the Cleveland-Cliffs Iron Company. The mine is located about 1 kilometer northwest of the village of Palmer in Section 19 (T47N, R26W) of 3 Marquette County, Michigan (Figure 1). All samples are from drill cores. The reader is referred to Figure 3 from Tituskin, 1983, for the drill hole locations. /’ \ \ \ G) \ Humboldt a 3 (‘3 M 49oUETr \ E TQOUGH \ \ (WOTEROZOI _ \ Enc Cannon \ \ O - 'Troc, \S Notrnan or) / m9 / / . Standard 0 a)“; 3“ \O in / c4 \ \ (‘7’; 0’O\ ?"\ ‘J a €$\\ \ . \ / \ \ ' / E x PL ANATION \ ‘ / \ ‘_ / a Neqounee lion - Formation N Isograds (o'let James. 1955) — Metamorph-c zones (otvev House. I979) O . 26 ”t ‘0 Myles ‘ ‘ }—.L_T_.‘L—fi._-;._._ ‘ I 0 vsxm 1 Index Moo ’ Figure 1. Generalized regional geology of the Marquette Trough (from Haase, 1982). GEOLOGIC and TECTONIC SETTING Geologic Setting The Negaunee Iron-Formation is part of a sequence of middle Precambrian metasedimentary rocks which form the Marquette Range Supergroup located on the south shore of Lake Superior (Cannon and Gair, 1970). These rocks were deposited in a sedimentary- tectonic basin and are presently contained in the westerly trending, west-plunging Marquette synclinorium. Within this structure the NIF outcrops in the Marquette Trough which extends to the west for approximately 70 km from the city of Marquette; the Republic Syncline, located 12 km south of the western end of the Marquette Trough; and a belt trending south for about 20 km from the western end of the Republic Syncline (Haase, 1979). The Empire Mine lies stratigraphically in the lower portion of the NIF where it dips 30 to 40 degrees to the west-northwest. The Supergroup sequence at the Empire Mine, from oldest to youngest, consists of the Enchantment Lake Formation, Kona Dolomite, and Wewe Slate of the Chocolay Group; the Ajibik Quartzite, Siamo Slate, and Negaunee Iron-Formation of the Menominee Group; and the Goodrich Quartzite of the Baraga Group (Gair, 1975). Each group represents a transgressive sedimentary sequence (Cannon and Gair, 1970; 6 Ueng and Larue, 1988, in press). The complete series is summarized by Gair (1975) and Bayley and James (1973). The Goodrich Quartzite is essentially conformable with the NIF, but is separated locally from it by an erosional disconformity (Gair, 1975). The contact between the NIF and the underlying Siamo Slate is conformable, and might be gradational over as much as 30 meters. The Siamo contact with the underlying Ajibik Quartzite is also conformable. The entire Supergroup represents an accumulation of sediments and volcanics which might reach 8 km in thickness (Gair, 1975). At the Empire Mine, the NIF alone reaches a thickness of about 1 km (Boyum, 1964). Trendall (1968) estimated the depositional rate of iron-formations to be on the order of 150 meters per million years. At this rate, NIF deposition occurred over a period of about 6.5 million years. The basement rock (described by Gair, 1975) consists of the lower Precambrian Compeau Creek gneiss and the younger Palmer gneiss with numerous quartz veins and intrusions of Archean and Proterozoic dikes. The Compeau Creek gneiss forms the bulk of the lower Precambrian basement rock in the area, and includes remnants of older rocks, possibly mafic volcanic flows and graywacke (Gair, 1975). Upper Precambrian (Keweenawan) diabase dikes cut some units of the older Precambrian rocks. At the Empire Mine, the NIF has been offset by several faults and is cut by at least 12 mafic dikes (Han, in Gair, 1975). 7 Van Schmus and Woolsey (1975), on the basis of whole rock and mineral Rb/Sr isotopic studies, determined that the Menominee Group sediments were deposited approximately 1.9 to 2.0 Ga before present. Goldich (1973) concluded that iron-formation deposition was at least 1.9 Ga ago. These dates are supported by the work of Aldrich et a1. (1965) on correlative rocks from the Iron Mountain District 100 km to the south. More precise dates cannot be obtained because of the disruption of the Rb-Sr system caused by the subsequent metamorphism and deformation during the Penokean orogeny (Haase, 1979). The NIF was regionally metamorphosed from lower green- schist (chlorite facies) in the east to middle amphibolite facies in the west during the Penokean orogeny (James, 1955; Haase, 1979). This occurred 1.9 +/- .05 Ga before present (Aldrich, 1965; Van Schmus and Woolsey, 1975). The Empire Mine is located in the southeast limb of the synclinorium where metamorphism is minimal. Tectonic Setting The tectonic evolution of the Animikie basin in which the NIF was deposited has been debated. A summary of two opposing interpretations, vertical remobilization versus subduction, follows. Sims et a1. (1980) describe the Great Lakes tectonic zone which separates the relatively stable rocks of the Superior province from a more mobile terrane to the south. 8 This zone was initiated about 2.7 Ga ago when the late Archean greenstone-granite terrane to the north was joined to the older Archean gneiss terrane to the south. It passes through what is now the Marquette trough. According to Sims et a1., crustal foundering along this suture zone in the early Proterozoic initiated the structural basins in which sediments, including the NIF, were deposited. The sediments were deformed during the Penokean orogeny, and subsequent extension accounts for the mafic intrusions. Along this zone, there is no conclusive evidence that subduction occurred during the compressional stage associated with the Penokean orogeny. Sims et a1. conclude that thermal processes resulted in the alternating contraction and expansion of the mobile gneiss basement relative to the more stable greenstone-granite terrane with which it was coupled. Ueng and Larue (1988, in press) describe the structural evolution of an early Proterozoic suture zone which is oriented in an east-west direction and lies about 50 km to the south of where Sims et al. (1980) place the Great Lakes tectonic zone. This zone resulted from a northern passive margin assemblage being subducted below and accreted to a southern magmatic arc complex during the Penokean orogeny. In the plate tectonic paradigm, the NIF is a part of the passive margin assemblage. Formation of the structural troughs which accumulated the thicker strata, was probably accompanied by syn-sedimentary rifting. Cambray (1977) 9 also subscribes to the plate tectonic model in which "the Chocolay Group represents an early shelf facies, the Menominee Group coincided with the rifting which provided trenches for the iron formations and the Baraga Group was deposited during a phase of subsidence common to Atlantic type margins". The rifting caused structural weaknesses within the basin along which shortening occurred during the Penokean orogeny. PREVIOUS WORK Because of its economic importance, the Empire Mine has been the subject of several investigations. Han (1962, 1971, 1982) examined the major replacement textures and structures associated with the enrichment and metamorphism of the NIF ores. Han (in Gair, 1975) and Gair (1975) describe in detail the lithology, stratigraphy and petrology of the ore deposits at the Empire Mine and the surrounding area. The metamorphic history of the NIF including the Empire Mine has been detailed by James (1955) and Haase (1979). Tituskin (1983) examined the trace element distribution in selected samples from the Empire Mine. 10 ANALYTICAL TECHNIQUES Forty-five drill core samples from the Empire Mine were used in this study. Six samples represent the underlying Siamo Slate, while the remaining 39 samples are character- istic of the different lithologies of the iron-formation at the mine. The major oxides were determined by x-ray fluorescence from fused glass discs prepared from a mixture of nine grams of lithium tetra-borate flux and one gram of sample powder. Titrametric methods were used to determine the ferrous iron content. Loss on ignition was measured after heating the samples to 10500 C for 45 minutes. The per- cent magnetite in the samples was determined by magnetic susceptibility using a Satmagan belonging to Cleveland Cliffs Iron Co.. The Satmagan (Saturation Magnetization Analyzer) is a magnetic balance in which the sample is weighed in gravitational and magnetic fields. The ratio of the two weighings is linearly proportional to the amount of magnetic material in the sample, and is accurate to 0.1 percent. The silicate minerals were identified in five samples (28, 35, 39, 41 and 45) by means of X-ray powder diffraction techniques. Magnetite separates were obtained from eight samples using a hand magnet and by density separation. The trace element contents of the magnetite separates were determined by instrumental neutron activation analysis. 11 MINERALOGY Mineralogy of the NIF at the Empire Mine The mineralogy of the Negaunee Iron-Formation is complex, and is characterized by many local variations. The NIF is a Superior-type iron-formation, as classified by Gross (1965), and consists of chemically deposited banded or laminated iron-rich rocks which are locally interbedded with mechanically deposited sediments. Han (1971) defined the former as "iron-formation" and the latter as "clastic" sediment. The principal minerals at the Empire Mine are magnetite, siderite, hematite, ankerite, chert, quartz, and iron silicates (Tituskin, 1983; Han, in Gair, 1975). Layers containing carbonate, magnetite, iron silicate, or chert, alone or in mixtures, can be associated with one another in virtually all possible combinations. According to Han (1971), the two principal ferrous iron silicates are minnesotaite and stilpnomelane which formed as a result of silication. (Silication is the diagenetic or metamorphic growth of a silicate mineral by the combination of silica and suitable cations.) The ferric iron silicates are less common than the ferrous iron silicates, and are abundant only very locally. They include crocidolite, which is a diagenetic mineral, and riebeckite and acmite, which are 12 l3 metamorphic minerals. Chlorite is a minor iron silicate, except locally in beds of graywacke, or in argillaceous beds in the transitional zone between the iron-formation and the Siamo Slate (Gair, 1975). The formation at the Empire Mine precipitated dominantly as chert and siderite (Han, 1962, 1971; in Gair, 1975). Han (1971) concluded from textural and structural features that the magnetite ore body was enriched through banding replacement (as opposed to enrichment through porphyroblast growth as noted in the adjacent martitic ore reserve of the Tilden property). He described the chief replacement processes as magnetitization, silication, sideritization, ankeritization, and silicification. Major diagenetic alterations include chert to magnetite, and siderite to magnetite and ankerite. Magnetite that replaced cherty layers was apparently derived from initially more ferruginous layers adjacent to the cherty layers (Han, in Gair, 1975). Some siderite formed through replacement of chert, chert-silicate, silicate-chert, or magnetite (Han, 1971). Post-metamorphic oxidation occurred locally in zones of structural weakness, and converted magnetite to martite, and the carbonates to hematite or goethite (Han, in Gair, 1975) Mineralogy of the Sample Groups Used in This Studv The importance of mineral types has been mentioned by Tituskin (1983), Haase (1979), and Han (1962, in Gair, 14 1975), with regard to certain element associations reported in the iron-formation. Because of the complex mineral associations, a prerequisite to a discussion of the chemistry of the rocks requires a classification of the samples into groups based mostly on their mineralogy. These groups have classically been defined as iron oxide- rich, carbonate-rich, chert-rich, or elastic-rich zones. In this section, the general sample classification scheme of Tituskin is described with respect to the mineral associations in these groups. Tituskin (1983) grouped the 45 samples from the Empire Mine based on the dominant mineralogy as determined by hand sample and thin section examination: 12 samples of primarily carbonate minerals, 7 chert—rich samples, 12 magnetite-rich ore samples, 8 samples from clastic interbeds, and 6 samples from the underlying Siamo Slate. The following mineralogical characterizations of these five groups were taken from her descriptions. The carbonate samples consisted of alternating laminae of both meso- and microbands of chert and carbonate, and contained at least 70% carbonate and less than 5% magnetite. Iron silicates were present locally in minor amounts. Chert-rich samples were predominantly quartz, with small amounts of magnetite. Carbonate-rich laminae and minor iron silicates were present locally. 15 Samples of magnetite-rich ore were comprised of magnetite coexisting with chert and less than 10% carbonate. This group also contained some hematite and iron silicates. Major clastic lenses were sampled, and contained 60 to 70% detrital quartz, with minor detrital magnetite. The two primary matrices associated with the elastic material were (a) carbonate with magnetite, and (b) iron silicates with minor magnetite. stilpnomelane was observed to comprise up to 25% of one sample. Hematite and Chlorite were also present. Samples from the upper unit of the Siamo Formation were classified as graywacke, or coarse grained clastic sediment. The graywacke was massive, and consisted of 45 to 60% detrital quartz with minor plagioclase, magnetite, and mafic interclasts of Chlorite, biotite, and possibly epidote. The matrix was primarily Chlorite, with minor biotite, muscovite and sericite. The elastic lenses were comprised of 70 to 80% detrital quartz, 5 to 10% mafic interclasts, approximately 1% muscovite, and Chlorite. Because major element data from Q-mode analyses essentially reinforced her groupings, the general class- ification scheme of Tituskin (1983) was used in this investigation. Based mostly on the magnetite determin- ation, and supported by the major element abundances, the assignment of samples to the groups has been modified to better fit the mineralogical characterizations outlined by 16 Tituskin. For this study, the groups representing the precipitated sediments are comprised of eleven carbonate-rich samples with 10 - 26% loss an ignition (assumed to be C02), three chert-rich samples with greater than 80% $102, and seventeen magnetite-rich ore samples with 11 - 50% magnetite. The NIF clastic and Siamo Slate sample groups remain the same. The data in Tables 1 to 4 are presented according to this classification scheme. PETROCHEMISTRY Chemical Data The major oxides are reported in Table 1. Because the rocks are carbonate-rich, loss on ignition (L.O.I.) is considered to represent an approximation of the CO2 in the samples. It is unknown why, in some samples, the sum of the oxides exceeds 100%. The amount by which the percent exceeded 100% correlated directly with the L.O.I., so it was thought that some of the sample might have sputtered out of the crucible during heating. However, when L.O.I. was measured again with caps on the crucibles, the results were identical to the nearest tenth of a percent. Nazo was not reported because those data were not always reproducible. The samples in Table 1 are numbered from 1 to 45 for simplicity. Numbers 1-11 are carbonate—rich, 12-14 are chert-rich, 15-31 are magnetite-rich ore, 32-39 are interbedded clastic, and 40-45 are the underlying Siamo Slate samples. The second identification number refers to diamond drill hole locations and depths from which the core samples were collected (see Figure 3, Tituskin, 1983). The percent magnetite in the samples is reported in Table 2. Table 2 also lists the normative mineralogy as determined from the major element data. The Fe3+ not in 17 18 0.00 00.~ 0~.0 ~0.0 v0.N H0.0 0—.0 00.~ «0.0 0n.~ 00.0 ~0.05 0~0= 0v m.~0— 00.n no.0 no.0 n~.0 0N.o vv.n 0A.~ v0.0 00.- 00.0 50.~0 000: vv 0.~0n 00.v ~0.0 no.0 00.n 00.0 -.0 00.0 nn.0~ mv.~ 0~.0~ 00.~0 0.500: nv n.00~ 00.“ —«.0 00.0 «0.0 no.0 0~.0 5v.~ “0.0 No.0 50.v n~.00 0.000: Nv 0.n0a 0n.5 . ~0.0 no.0 00.H 00.0 v~.0 vv.0 00.~N 0N.~ Nn.v~ 0~.~0 mn~0u av n.noa ~0.~ n~.0 "0.0 00.« ~0.0 50.0 n0.~ 00.0 n0.0 0~.0 00.~0 tm:=5m .ao_Eozu daotu odaEam 5~.n5 nn.~ 0—.n ««.o 5«.p 0~.o~ n5.0 om.~ mm.n «5.F on.nm o~.— nn.~ 05.m o—.0 ««.5o o~.o n«.0 «0.0 FF.0 00.—« n5.— -.« 5n.0 ~5.p om: .« 0.905 m0.«o «5.5 «0.55 00.F~ ~n.m ~5.0 0~.5n «—.0. o~.5~ n~.— O»; oo.~0— 0—.~5 o5.-F —m.0n —o.0 5F.55 5«.5 no.0 No.0— Fo.n oo.«0 on.m «n.o n—.«5 00.0 on.oo— «n.~ 0~.~ 00.0— 0m.0 m~.on 5~.0 0«.0 «0.0 5w.0 —~.0 no~.< ~«.- 0n.mp on.- —~.~— ~5.o— x<>.mwou >wo.—m n~.n¢ x.mwOU >w0.5m mm.«m x.mw00 00.0 >u0.»m «o.m« z.mw00 m—.m >w0.5m «5.00 z.uwou «5.0 >wo.5m no.n« z.mwOU ~0.~ >w0.»m —~.— z.;w°0 ~0.n >uo.5w 0m.~ z.uw00 05.n >uo.hm m«.n z.;w°u «0.0 >m0.hm 0m.0 z.0000 -.0 >wo.km 00.~ z 017 O!) Figure 3. high on total REE. 34 Total REE 215 180 145 110 Correlation diagrams for samples with factor loadings that range from high on magnetite and total Fe Numbers refer to sample numbers. to b v r . on" 0 0n " can an :4 010 00" on ‘ 3 00:) 7 I L A 4 a 12 16 A120, . v v d 0:: 0n ‘ on 0 - 020 u . 030 u . con 17 0 A son 3’ A A I 4 12 16 A120 35 and Cr, lead to the conclusion that the mineral in which these elements are associated is an aluminosilicate. Any of the aluminum-bearing iron silicate minerals in the iron- formation could account for these correlations. Aluminum has a positive correlation with the REE along essentially the same sequence as with T102 and with the remaining Feo (Figure 3d). It is therefore proposed that in this suite of samples, the REE are primarily incorporated in the aluminosilicate minerals. In these samples, as the abundance of magnetite increased, the abundance of aluminosilicate minerals, as represented by A1203, decreased. Gruner (1944) said that most of the A1203 found in the iron-formations of the Cuynna and Mesabi ranges seemed to be in the mineral stilpnomelane. He concluded that magnesium and aluminum were essential to the structure, and that without these elements, quartz and magnetite probably would have resulted. The chemical data from this suite of samples, and the observations of Han (in Gair, 1975) support Gruner’s conclusion. Han noted that layers of iron silicate appear to replace chert along boundaries between layers rich in chert and magnetite. LaBerge (1964) found that stilpnomelane is typically associated with magnetite in iron-formations of the Lake Superior region. To identify the aluminosilicate minerals present, five samples with high aluminum content (samples 28, 35, 39, 41 and 45) were analyzed using X-ray diffraction techniques. 36 The 12 A stilpnomelane peak was only detected in the ore sample, 28. All samples had 7 A chlorite peaks, and all except 39 had 14 3 chlorite peaks. Biotite was also detected in all samples, and the peaks were especially strong in samples 39 and 45. These samples also had more K20 than the other samples analyzed. The five samples had from 9.5 to 14.32% A1203. Haase (1979) identified the minerals chamosite and ripidolite (7 A and 14 A chlorites respectively) in samples from the Empire Mine. Based on electron microprobe analyses, the chlorite contains between 16.3 and 20.4% A1203 (Haase, 1979, Table 2). This translates to between 47 and 58% chlorite in the sample with 9.5% A1203, and between 70 and 88% chlorite in the sample with 14.32% A1203. If the aluminum is also in biotite (A1203 = 13.81%) and stilpnomelane (A1203 = 6.37 to 7.00%), the portion of the sample which is aluminosilicate is even higher (A1203 percentages from Haase, 1979, Table 2). Haase observed that stilpnomelane, biotite and chlorite are typically associated in samples from the Empire Mine. The importance of stilpnomelane as an REE sink might not be as great in the clastic sediments as it is in the precipitated facies. Silica Versus Iron and Magnetite Figure 2b shows the sequence of samples between factors 1 and 2 which defines a trend of increasing total iron and 37 magnetite with decreasing Si02 toward factor 1. Figure 4a is a plot of this sequence. If the REE follow iron or are concentrated in magnetite, evidence of such should be apparent in this suite of samples. As can be seen in Figure 2b, the sequence is discontinuous. Three groups of samples define distinct continuous sequences along the trend, and correlation plots show that the distribution of the elements is sometimes dictated by mineralogies which differ between these groups. The correlations from this sequence are evidence that the samples contain variable mixtures of chert, magnetite, aluminosilicate minerals and carbonate. While chert and the aluminosilicates increase in the direction of factor 2, chert becomes increasingly abundant relative to the aluminosilicates such that the overall relationship between the minerals is antipathetic. The carbonates increase with magnetite toward factor 1. Figure 4b shows that the three groups are distinctive in the plot of MgO and total iron. While the overall correlation is positive, the correlations within each group are negative. Conversely, the correlation of SiOZ and Mgo is generally negative, while correlations within the groups are positive. This can be explained if Mgo and SiO2 are associated in aluminosilicate minerals which are most abundant in the magnetite-rich samples. Within the individual groups, however, the magnetite and alumino- silicate minerals have an antipathetic relationship. —°"—r fit I u on a on 46- Oil on Q) on L 36- - Oz: v—1 Oil (0 u 0 26' - ,— 01! 16> . I: “00 5 1 1 1 L 1 32 42 52 62 72 82 92 8102 56 v 1 r v 270 C on 1! 0 46- < on on 0.) on LL 36- - 025 P1 02! (D 4.1 O 26* . *— 0|! 16> - OH on: 6 1 1 0.1 0 4 0 7 1.0 13 1. A1203 7 e . v v v n m 011 0 6 0 019 "° on 5 0” E 4 8: 20.012 3 (O __J 013 2 1 On 0 1 1 1 1 0.1 0.4 0.7 1.0 13 1 A1203 Figure 4. Correlation diagrams for samples with 38 56 b v v r d v ’30 17 on 45 - 4 NO 017 (1) on L1. 36 — - on r—“ 028 (O J—J O 26 - _ ,_ OI! 16 I? 0014 5 1 . 1 1 0 0 5 10 .S 2 0 2 5 M90 0.05 d . v v v o 2: ”an o 04 — " < o “On ON ... 0.03 - - ’— on on 0 02 - a On on on 0 01 ‘ ' ‘ 01 0.4 0.7 10 13 16 “.0. f x r f v I 120 2000 ~ on H O 0 ..-< 1500 ‘— \ L on t.) 1000 °" on 0” 21 00:: 025 oz: 500 J A 1 1 l 32 £2 52 62 72 82 92 8102 factor loadings that range from high on SiO2 to high on total iron and magnetite. Numbers refer to sample numbers. 39 Aluminum has positive correlations with MgO, magnetite, total iron, TiOz and Th. All of these elements plus magnetite have positive correlations with the REE except Tb and Lu. This is illustrated in Figures 4c to 4e by the correlations of A1203 with total iron, Tioz and La. The above correlations have been interpreted to mean that while Th and the REE, except Tb and Lu, are most abundant in the magnetite-rich samples of this suite, their concentrations are not dependent on the amount of magnetite in the samples. These trace elements are mainly associated with the aluminosilicate minerals, and a greater abundance of these minerals is associated with the magnetite-rich samples than with the chert. The Cr concentration appears to be independent of the minerals in this sample group. Silica has negative correlations with both Cr and T102, but a positive correlation with the Cr/Tioz ratio (Figure 4f). There are antipathetic correlations of magnetite, MgO, A1203 and iron with the Cr/TiOz ratio. This is because the abundance of Cr remained relatively constant in these samples and was only moderately affected by silica dilution. Silica versus Carbonate The sequence of samples between factors 2 and 4 (Figures 2c and 5a) ranges from high silica (or chert) to high FeO and L.O.I. (or siderite). In this section, and 40 28 I I I Y T l 1' I 0.91 a b c: 3 0.5 - 1 21- °' . AN 05 0! o c.) i 0.7 — . V 3 0: Lu on . 14 - < \ "1 011 LEI 0.6 _ ‘ 0. .... _1 7_ _ 0.5 - - 14 J 0 014 0 o 1 J 1 1 . HQ 0.‘ L 40 1 30 40 so 5.0 70 so 90 o 10 20 30 do 8102 F80 0.10 v v v 1 v 4 v T v I C d 110 0.03 - o - 1| 3L . E 0.06 - 01 4 E O. O. 8 014 E 2" ‘ 02 3 0.04 r~ 01 oz . c o: _1 1— 01 012 012 1- . 0.02 - 05 - &l 14 o I J I 1 I o I 1 I 1 o 0.02 0.04 0.06 0.03 0.10 0.12 o 0.5 1,0 1.5 2.0 P205 111203 Figure 5. Correlation diagrams for samples with factor loadings that range from high on Si02 to high on carbonate. Numbers refer to sample numbers. 41 related figures, L.O.I. is understood to mean C02. Most of the iron in these samples is in the reduced state. Feo and L.O.I. have a strong positive correlation along the major trend. Because of the low abundance of aluminum, which is an indication of the absence of a significant concentration of aluminosilicate minerals, carbonate controls on the REE abundance should be detectable in this suite of samples. The Eu anomaly (Eu/Eu*) is negative and approaches 1.0 as FeO and L.O.I. are increased and Sio2 is decreased (Figure 5b). If Eu is in the reduced state, it could substitute for Fe2+ in siderite. Sample 3 is an exception to this trend, having a more negative Eu anomaly than predicted from the amount of FeO. There are, then, other controls on the abundance of Eu in the samples. The positive correlation between Eu and Th might be an indication that the amount of Eu is not solely controlled by the redox conditions. Thorium does not exist in more than one oxidation state, and has no correlation with any of the other REE. From the strong positive correlation of P205 with Lu (Figure 5c) and Tb, it could be inferred that apatite controls at least part of the REE abundance. There is, however, no correlation between P205 and Cao. One explanation might be that the correlation is masked by Cao in carbonate minerals. Alternatively, while apatite is stable during metamorphism, it can be dissolved in 42 supergene conditions of diagenesis and the phosphorus released to be precipitated with other minerals (Morris, 1985). Potassium also has positive correlations with Lu and Th, and both K20 and P205 might be associated in iron silicate minerals. Lanthanum, Sm and Eu do not have good positive correlations with any element, except for that between Eu and Th. There are negative correlations of these REE with magnetite, and only fair positive correlations with FeO, A1203, Ti02 and K20 with variable sequences in the trends. An association of these elements in alumino- silicate minerals is not conclusive. Aluminum has a strong positive correlation with Th (Figure 5d), and only a fair correlation with TiOz. As shown in the other sequences, the correlation between A1203 and Tioz is normally high in these samples. The low absolute amounts of these elements in this sequence probably accounts for the lack of a good correlation because they fall below the limit of accurate detect- ability. The lack of a good correlation between A1203 and the REE, contrary to what is seen in other sequences, is likewise explained. Based on the these observations, the REE in this suite of samples are not considered to be controlled by a single mineral phase. A study of limestones by Scherer and Seitz (1980) lead to the conclusion that carbonates tend to concentrate LREE during diagenesis. The data from these 43 samples do not support this conclusion. Trace amounts of apatite might be controlling the HREE abundance. Diagenetic processes probably involved mobilization of most of the REE to be incorporated in more compatible mineral phases not represented here. W If the REE distribution were controlled by a single mineral, it could be best defined in this suite of samples. The sequence of samples between factors 2 and 3 defines a trend of increasing total REE and decreasing silica toward factor 3 (Figures 2d and 6a). Sample 35 which has the highest REE content also has the most Feo, C02 (L.O.I.), total iron, and normative siderite and MgCO3. The sample with the lowest REE content, 17, has the most magnetite and normative MnCO3 and CaCO3. Sample 14, also with a low REE content, has the greatest amount of Fe203 and normative hematite. It is concluded from these observations that in this suite of samples, the REE content is not dependent on carbonate or iron-oxide minerals. The correlation plots are support for this conclusion, as total iron, magnetite, L.O.I., and normalized hematite have no correlations with the REE. Ferrous iron has no correlation with L.O.I., so it is not primarily associated with carbonate minerals in these samples. It has fair positive correlations with the REE, 44 am e . 4se . . , T a b o I! 60 175 - 50 LU 140 > 0)! ‘ LLJ CE 40 .C .—. 105 v - ,_ (U 30 on u l O 032 l >— 70» on . 20 4 Om \ 35 . 045 ‘ 10 0:5 0“ " "a. 0L ‘ A 1 094 0 on A . . , . J 55 65 75 as 95 0 ‘ 5 2.2 2.8 3 4 8102 M90 0.6 . . r . 21o . . . s c, C d u :50 0.5 - o 175 - . )9 04- < w uo- o. UJ 19 O: N H 0.3 ‘- . :r-U‘ 105 b _ t- 045 4.; O 0 2 - ‘9" . r— 70 - . ‘ on 0 1r 0“ . 35 L 045 . Ono L‘ ‘01: 0 o" I I A I O I A 1 0 3 s 9 12 15 0 3 6 9 12 15 A1203 A1203 0 7 . . , . fi 210 r r . a [e 0 f n N t 175 - . 0.5 a “0 ’ on - [I O m Hiw- . (J m 0 3 4.4 O o): r— 70 - . 0n 0‘5 on x: o I: 35 . 45 . 01° I 040 Om g4. ‘ ‘ I 0L l u A g 0 0.02 0.04 0 06 0 08 0.10 0 0,02 0,04 0 06 0 06 0.10 9 D ‘ 205 1 205 Figure 6. REE . Correlation diagrams for samples with factor loadings that range from high on $102 to high on total Numbers refer to sample numbers. 45 Th and TiOZ, falling nearly along the major trend, but there is no correlation between FeO and A1203. Magnesium has positive correlations with A1203, FeO, Tioz and the REE. Thorium has a better correlation with Mgo than with any of the other elements (Figure 6b). Titanium and A1203 have positive correlations with each other and with the REE (Figures 6c and 6d). These oxides also have positive correlations with Th and K20. It follows that 5102 has antipathetic correlations with A1203, Tioz, FeO, Th and the REE. The sequence of samples in these correlations is nearly the same as the major sequence. There is a weak positive correlation of A1203 with Cr which does not follow the major trend. There is some evidence of apatite in these samples. The strong positive correlation of P205 and CaO follows the major trend with the exception of one chert sample which has anomalously high abundances of these elements (Figure 6e). Both of these oxides have positive correlations with the Tioz/A1203 ratio along the same trend. They also have fair positive correlations with the REE, Th, A1203, T102, MgO, and Mno, with the exception of the chert sample which has relatively high amounts of P205 and CaO, but the lowest concentrations of the other elements (Figure 6f). This is an indication that there is some apatite associated with the alumino- silicate minerals which might affect the trace element concentration in those minerals. 46 Based on these correlations, it is concluded that the REE in this suite of samples are primarily incorporated in the aluminosilicate minerals. Except for the two chert samples, the samples in this suite are from NIF and Siamo clastic sediments. Chlorite and biotite were identified from X-ray diffraction data as the principal alumino- silicate minerals in three of these clastic samples. There is probably some apatite associated with these minerals, but the influence of apatite on the trace element concentration is less than that of the aluminosilicate minerals. The composition of the aluminosilicate minerals varies among the samples, and Th appears to be concentrated in those which are magnesium-rich. Summary There is a prevailing association of A1203 and TiOz with the REE, which is evidence that the REE are preferentially incorporated in aluminosilicate minerals. Chlorite, biotite and stilpnomelane were the alumino- silicate minerals identified by means of X-ray diffraction analysis in four of the samples. A greater abundance of aluminosilicate minerals are associated with magnetite-rich samples than with carbonate or chert samples. This explains Tituskin's (1983) observation that the abundance of the REE is greater in the magnetite-rich ore facies than in the other precipitated facies. 47 Based on consistent REE patterns, Tituskin (1983) suggested that the primary mineralogies at the Empire Mine were more homogeneous than seen today, and that diagenetic and metamorphic processes have not resulted in significant variation of the original REE imprint. Because the REE pattern was consistent for all facies, regardless of mineralogy, she attributed it to a reflection of the seawater concentration of the REE. In this investigation, it was found that there were strong correlations of the REE with A1203 and Tioz, elements attributed to detrital phases. These aluminum—rich detrital phases were recrystallized during diagenesis to form chlorite, stilpnomelane and biotite (Han, in Gair,1975), and did not fractionate the REE. The abundance of aluminosilicate minerals controlled the total concentration of the REE in all facies. It is therefore proposed that the REE distribution pattern is a reflection of the source rock from which the detrital phases were derived rather than of the basin water. In this system, the REE do not follow major cations such as iron or magnesium consistently, and are not preferentially incorporated in magnetite or carbonate minerals. The species of aluminosilicate mineral might affect the concentration of Th, as this element appeared to be concentrated in the magnesium-rich aluminosilicate minerals in one suite of samples. There is probably some apatite associated with the aluminosilicate minerals, but 48 weak correlations of P205 with the REE are an indication that this mineral is not the principal REE sink. Chromium does not follow iron or magnetite, but is concentrated in aluminosilicate minerals associated with magnetite. CRYSTALLOCHEMICAL CONTROLS ON REE FRACTIONATION AND CHROMIUM CONCENTRATION Introduction Fryer (1977a) studied the REE distribution in rocks from the Sokoman Iron-Formation, and found that the enriched oxide facies rocks were enriched in the heavy REE. He suggested that the REE were mobilized when water and C02, released during diagenetic reactions, transported elements incompatible with the crystallizing structure as complexes. The HREE enrichment in diagenetically enriched oxide facies has also been observed in samples from the Brockman Iron-Formation in Australia and the Rapitan Iron~Formation in Canada (Fryer, 1977a). The purpose of this section is to determine whether or not a similar HREE enrichment has occurred in the enriched oxide facies rocks of the Negaunee Iron-Formation. Magnetite at the Empire Mine is a diagenetically iron- enriched oxide which formed by replacement of chert or, together with ankerite, siderite (Han, 1971). In this section, the REE distribution was examined in magnetite separates relative to that in the whole-rock samples from which they were derived. The concentration of Cr in the magnetite separates is also addressed. 49 50 REE and Cr in Magnetite Separates Magnetite was separated from eight samples: six ores, one slate, and one Clastic sample. Results of INAA of the magnetite separates and whole-rock samples are summarized in Table 6. Chondrite normalized REE abundances of magnetite separate - whole-rock pairs are plotted in Figures 7a to 7h. In all samples, magnetite is LREE depleted and HREE enriched relative to the whole rock. The extreme variation in LREE abundances among the magnetite separates is attributed to contaminants. Due to its replacement nature, magnetite is intimately associated with chert, carbonate and other minerals making complete mechanical separation impossible. Sample 20, which showed the least relative LREE depletion, contained significant carbonate contamin- ation. It also had the highest total REE concentration, and the lowest chromium concentration. Magnetite in sample 44 was coarse grained, and a nearly complete separation of contaminants was attained. This sample shows a greater relative depletion of the LREE than the other samples, and has the lowest total REE concentration. The HREE and Cr concentrations in the magnetite separates are higher than in the whole-rock samples. A positive correlation of Cr with magnetite, however, was not observed in the whole-rock samples. Positive correlations 51 No.0 00.0 mv.~ no.q om.— o~.F mm.o mo~.< x on.5 no.o~ «0.0 -.~F m~.o— o«.- ~o.m— no.mn ~p.o- wn.m~ ne.n oo.«F ow.o m~.—~ o~.~w om.~m Lu om.mp om.o 00.x —o.F o~.—— m—.F o—.n— Fm.~ NF.~— oe.p mm.n Po.« 2&3 oo.«m mo.~— Fo.o~ "v.0 o—.op ~5.F~ Fm.np 00.0 om.~m oe.mp 4<—o»-mwx ~o.o op.o no.0 o~.o mw.o ~—.o no.0 oo.o 00.0 o—.o mo.o q~.o v0.0 ~r.o no.0 or.o 3; auflo HCOEO. w UUQLH mm.o om.— mo.o mm.r no.0 mp.w co.o -.F m—.— o~.F 9» AFC xuox.a.o;3 pco Axo ouutadom ou_uocmmx no.0 m~.o m-.o nm.o 00.0 m~.o F~.o o~.o fl ~F.o o~.o mF.o m~.o mw.o m~.o -.o F~.o oq.o um.o no.0 co.— on.— 05.0 m~.F No.0 05.0 oo.o nh.o qr.— «m.o 0m.o oq.— wo.P Em mq.—— ~—.c 0n.op —m.m No.o~ mo.~ mm.~w «m.q oo.o en.~— n~.o Np.n mm.m Fm.o m—.m om.~ Nq.o— mm.w e~.~« 00.~ hm.o mn.p om.o mo.m mw.m F~.F mm.o— «n.m a; coo: m.momw m.ovmz m.po~o Nqopu nmu mm m.m~mo .5.— .o 032 ahv Arc AFC Axv any Axv Apv Axv Ahv Axv “by Axv Rho “xv Apv Axv we mm mm —n —m ow o~ ~— N— @— op wgax