Rare earth elements,
and sea water
JOSEPH L. GRAF. JR. Oremco.
N Y. 10016. U S.A.
formations probably formed as chemical sediments when an Fe-rtch solutton entered the sedimentary environment. The REE patterns of such sediments are affected by the amount and type of prectpttatmg and detrttal phases. the REE pattern of the non source solutton. the REE pattern of the sea water mto which the solution flows. and the mixing which takes place between input solutton and sea water. What IS known about the different types of iron formation suggests that all of these factors may vary from one type to another Therefore, variations m REE patterns cannot be attributed solely to changes in sea water REE patterns. Compartson of the REE patterns of Archean and Ordovician iron formations of the Algoma type shows that similar iron formations have similar REE patterns regardless of age. This implies stmilar genetic histortes.
been recently published by GRAF (1977), and to test the use of iron formation REE patterns as Indicators THE OBSERVATION that metalliferous sediments associof sea water REE patterns by comparing the REE ated with the East Pacific Rise have REE patterns patterns of Archean iron formations with those of very similar to that of sea water (DYMOND et (I/., 1973. younger iron formations which are essentially equivand others) suggests the possibility of using older alent. The Ordovician iron formations in New Brunsequivalents of these rocks to study the REE patterns wick (GRAF, 1977; DAVIES, 1972) have characteristics of sea water of the same age. Studies by ROBERTSON very similar to those of Archean, Algoma-type iron and FLEET (1976) and CORLISS et al. (1972) have formations (GROSS, 1965; GOODWIN, 1964). They are shown that similar metalliferous sediments in the banded, magnetite-quartz-carbonate rocks assoctated Troodos Massif. Cyprus also have sea water-like Ce with felsic volcanic rocks. If the REE pattern of sea depletion, suggesting that the REE distribution in water changed and if such change was reflected by Mesozoic sea water was roughly similar to that m the REE patterns of iron formations. the patterns of modern sea water. Projecting this reasoning much the New Brunswick iron formations should be sigmfifurther back in time is hampered by the poor presercantly different from those of Archean iron formavation of the upper portions of unequivocal oceanic tions. crust (ophiolites?) which contam metalhferous sediments. In a recent paper FRYER (1977a) used REE ANALYTICAL RESULTS data from several types of iron-rich chemical sediment Iron-rtch chemical sediments were analyzed for REE by to infer changes in the REE pattern of sea water wtth Instrumental neutron activation analysts (INAA) The anatime. He interpreted from these data that Archean lyttcal procedure (GRAF. 1975. appendix 5) was essentially sea water was enriched in Eu and Ce relative to that described III GORDON et ol (1968) except standard rock powders (AGVI, G2. Wll were used as momtors and modern sea water and that the relative Ce depletton samples were counted only twice, 5-10 and 15-30 days so characteristic of modern sea water began during after trradtation A comparison of the analyttcal results Proterozoic time. for the two most commonly used standard rocks with the The purpose of this paper is to present additional published ‘best v,alues’ for these standards (FLAYA~AN. REE data from iron formations. some of which have 1973) is gtven m Table 1 INTRODUCTION
of analyses with published G-2 and AGVI
G-2 This Work
La 92.2 f 11.99
Ce 152.44 * 10.7 Sm 6.67 f 1.44 Eu Tb Yb Lu
1.44 + 0.196 0.543 f 0.128 1.192 f 0.42 0.128 f 0.023
AGVI Flanagan (1973)
-- This Work
34.2 f I.54 63.4 * 3.83
7.3 1.5 0.54 0.88 0.11
are In parts analyses.
5.34 * 0.31 1.721 l 0.123 0.743 * 0.135 1.72 f 0.377 0.283 * 0.11
5.9 1.7 0.7 1.7 0.28
given are standard
J L. GRAF JR. Table
Sample JG346 JG3417 JG36D JG770 JG77BM JG77BNM JG37G JGl JGSOA N095B NO50 N050M NO5OQ
N022B HRlH HRlQH HRl3M HR13Q HR15 HR27
2. REE (ppm) in selected
16.77 13.91 3.08 3.75 -0.51 18.8 18.38 3.69 4.55 -0.82 3.13 -.37 9.24 3.44(31) 1.86 13.88 __ 1.9 35.83 82.96 6.92 85.67 218.92 12.99 0.47 -0.06 6.22 7.68(24) 1.36 7.28 -1.1 0.96 1.17(16) 0.15 2.09 -0.55 0.37(18) -0.10 0.41(16) 0.86(64) 0.05 2.60(17) -0.32 1.43(17) 2.99(26) 0.27 25.82(17) 50.34 6.84 1.48(17) 2.50 0.18
0.9 0.42 1.0 ___
0.21 0.12 0.29 0.13 0.14
23 44 29 50 52
0.02 0.12 0.26 0.03 0.05 -_ --
13.5 32 48 3 29 49 4
1.36 3.44 0.6 3.1 4.46 3.52 1.48 0.24 0.26 2.52 1.61 1.86 1.35
0.19 0.11 0.85 0.03
50 9 23 2
1.75 2.19 2.54 1.67 6.55 2.81 1.66 3.43 0.15(10) 2.19 2.05 0.21 0.12(17) o.oa(l4) 0.06(9) 0.11(21) 0.53(7) 1.65(7) 0.07(11)
0.42(31) _0.55 0.67 0.02(84) -_ 0.77(40) 0.05(54) -_-__ 0.15 1.65 --
0.32 1.66 0.54 _-1.49 3.36 4.36 0.22(35) 0.91 1.01 0.15 0.10(53) --0.82 0.73 5.89 0.20
4.5 -_ 5.0 __ __ 1.9 __ 12.0 16.9 __ 5.7 __ 7.6 -__
1.28 0.34 2.00 0.24 0.39
11.3 7.4 13.7
All concentrattons were determined by INAA. Fe concentrations m percent. Approxrmate analyttcal uncertamttes are as follows: La and Sm. lo”,,: Eu. 5”,,: Tb, 30”,,; Yb and Lu. 25”,,. and Ce. IO”, in low Fe samples and very high m Fe-rich samples. Where the uncertainties exceed these values. the actual uncertainties are given in parentheses.
Results for New Brunswick iron formations and massive sulfides were reported in GRAF (1977: Table 2) and some are included here in Table 2. Also included in Table 2 are analyses of two Fe-Mn sedtments from New Brunswick: three Archean iron formations from the Abttibi greenstone belt. Canada: and five samples from the Hamersley Basin. Australia. Sample locations and descriptions are given in Appendix 1. Eu Magnetite iron formations from New Brunswick (Table 2. Fig. 1. this study and GRAF. 1977). the two Archean iron formations in this study (Table 2 and Fig. I). and the Archean iron formations reported in FRYER (1977a) all have positive Eu anomalies. Figure 1 shows that the REE patterns of the Archean and Ordovician. Algoma-type iron
formattons in this study are practically identical. Fe-Mn sediments which occur in a roughly similar stratigraphic positton to that of the New Brunswick iron formations have EuiSm ratios similar to that of shales (Table 2 and Fig. 2). The massive sulfide band (N022B) whtch is located up dtp from a Zn-Cu massive sulfide deposit (Amulet ‘C’) in the Noranda area. Quebec, also has no Eu anomaly. Three of SIX Hamersley samples have postttve Eu anomalies (Table 2, Fig. 3). Interestmgly. m both HR13 and HRl the iron oxrde-rtch bands have lower Eu anomalies (no anomaly m HRl3M) than the more cherty material. However. the other chert sample (HR27) IS not anomalous. The sttlpnomelane-rtch shale (HR15) has a pattern very simtlar to that of shales even though tt has a high Fe content (23 wt’,, by INAA).
Y) a, %_ -0 c 0
; E ," 2 :
Fig. 1. REE patterns of Archean and Ordovician type tron formattons: 3-JG346. 4-N050W. 6-N095B.
patterns of Fe-Mn sediments Brunswick. I-JGSOA. 2-JGI.
Rare earth elements, iron formations and sea water
DISCUSSION Iron formations
oj sea water R EE
The REE pattern of an iron formation does not appear to be a useful indicator of the REE pattern of sea water of the same age. Most iron formations probably formed by chemical sedimentation as a result of an unusually high influx of Fe (often as an Fe-rich solution) into the sedimentary environment. As discussed in GRAF (1977). the REE pattern of the chemical precipitate fraction of iron formations is related to the following three factors:
1. The REE pattern of the iron source material (probably some iron-bearing solution). 2. The REE pattern of the water into which the solution flows. In some cases this will be sea water of that time. 3. The degree of mixing which takes place between the above two solutions.
3. REE patterns of Hamersley iron formations: 7-HR15. 8-HR13M. 9-HR27. lo-HR13Q.
Although strong reservations must be placed upon the results of Ce analyses by INAA, certam of them deserve mention. The Fe-Mn sediments from New Brunswick do not show sea water-hke Ce depletion but have Ce/Sm ratios similar to that of shales. A non-magnetic concentrate from one New Brunswick iron formation (JG77BNM. Table 2. Fig. 4) is depleted in Ce suggesting some influence of sea water REE. As a general rule the New Brunswick iron formations have Ce/Sm ratios below that of shales. Most ratios are lower than that of NO5OQ in this study and than those of Archean Iron formations reported in FRYER (1977a) which are only shghtly below the shale value. The three cherty Hamersley samples have Ce/Sm
ratios close to that of shales.
Fig. 4. REE patterns of magnetic and non-magnetic concentrates from New Brunswick sample JG77B and magnetite and quartz bands from Adams Mine Sample N050: 11-JG77BNM, 12-N050M. 13-JG77BM. 14-N050Q.
The whole rock REE pattern is also affected by the amount and type of detrital phases deposited along with the chemical precipitate. It is quite probable that some or all of these factors varied from one iron formation type to another or even from one iron formation to another of the same type. In addition, conditions during deposition of an iron formation may change with time and conditions may vary from one part to another within the same iron formation. Therefore, the REE pattern of sea water cannot be singled out as the sole explanation for variations in the REE patterns of iron formations and the effect of sea water REE may be completely masked by changes in the other factors. Negative Ce anomalies are good evidence of the effect of sea water with a REE pattern similar to modern sea water. However, for the reasons given above, the lack of such anomalies in an iron formation does not necessarily imply that sea water of the same age was not depleted in Ce. REE data are most common for the followmg types of Fe-rich sediment, each of which has its own geologic-sedimentary setting and its own set of factors which affect its REE pattern: 1. Algoma-type iron firmatlons: felsic volcanism is very Important; the formations are a really somewhat restricted and are most common in Archean greenstone belts. 2. Superior-type iron formations: the role of VOIcanism is uncertain but most formations are not intimately associated with volcanic rocks; the formations have large areal extents and evidence is present for both quiescent and turbulent conditions during deposition. Superior-type Iron formations are most common in Proterozoic sedimentary sequences. 3. Metalliferous deep-sea sediments: ocean ridge basaltic volcanism is important: metalliferous deepsea sediments show good evidence of precipitation in a sea water-dominated fluid; the deposits are widespread on the modern sea floor and are found in some ophiolitic rock sequences.
J. L. GRAF JR.
There IS good geological and geochemical evidence that deep sea, metalliferous sediments were precipitated in sea water. Some can be observed forming today (ZELENOV,1964). However, there IS no evidence that Algoma-type iron formations and all Superiortype iron formations precipitated from a solution dominated by sea water of the same age. It is not unreasonable to expect the REE pattern of Archean ‘sea water’ to be different from that of modern sea water. In the Archean greenstone belts broad basins, geochemical If not topographical. were present (GCK)DWIN. 1973) and felsic volcanism must have had a much stronger influence on water chemistry than it does today. However, when REE patterns of New Brunswick iron formations are compared to those of similar Archean iron formations, no significant differences are observed. Because of differences in conditions of formatlon, a comparison of two types of iron formation will probably not show differences whether the sea water REE pattern changed or not. For example. if deep sea sediment data were compared with data from New Brunswick iron formations, one conclusion regarding Eu m the Ordovician would result. Comparing the same data with data from New Brunswick Fe-Mn sediments would result in an entirely different conclusion about the same element in the same water. Eu Anotntrlies In GRAF (1977) it was suggested that the Eu anomahes m the New Brunswick iron formations and massive sulfides resulted from the fact that the hydrothermal solutions presumed to be responsible for their formation were themselves anomalous. It was shown that the REE patterns of the sediments could be related to water-rock interactions within a hydrothermal system and that interaction between a solution and a felsic, feldspar porphyry could produce a positive Eu anomaly in the solution. The similarity between REE patterns of Archean and Ordovician, Algoma-type iron formations strongly suggests that the processes and the conditions of their formation were very similar. This in turn implies that both were formed by hydrothermal solutions which interacted with felsic volcanic rocks. The Eu anomalies found m some of the Hamersley samples are somewhat unusual for Superior-type iron formations. However, there is no consistent pattern to the anomalous samples. Of two iron oxide-rich samples (HRlH and HR13M) only HRlH is anomalous. Likewise of three cherty samples two (HRlQH and HR13Q) are anomalous while one (HR27) is not. FRYER (1977a) reported analyses of nine Proterozoic iron formations and of them only one (No. 10 from the Mesabi Range, USA) could be considered to be weakly anomalous. FRYER (1977b) also reported two analyses of Hamersley iron formations which had Eu/Sm ratios somewhat but possibly not significantly higher than the chondritic ratio. Eu anomalies are apparently localized features in Proterozoic iron for-
mations. Possibly, local, more strongly reducing conditions can account for the weak Eu anomalies. The stronger anomalies observed in some of the Hamersley samples may suggest somewhat different conditions of formation than those of other Proterozoic iron formations. In this regard. it is Interesting to note the felsic volcanic rocks which occur in the Hamersley section (TRENDALL.1973). REE conrenrs
The data presented here and m GRAF (1977) when compared to the data presented in FRYER (1977a) show that whereas some Precambrian iron formations have REE concentrations much lower than those in the New Brunswick iron formations, others have nearly identical REE concentrations. In addition, the concentrations in the New Brunswick Fe-Mn sediments are significantly higher than those in magnetite iron formations and are of the same magnitude as those of metalliferous sediments. It seems. therefore, that the conditions of formation are more important than the age in determining REE concentrations with metalhferous, deep sea and intervolcanic sediments normally having higher concentrations than magnetite-hematite iron formations. Knowledge of the types and relative amounts of the various components within iron formations is essential for an understandmg of the abundance data. Certain minerals such as aluminosilicates or apatite tend to have high to very high concentrations of REE while other minerals such as quartz tend toward very low concentrations. Magnetite appears to contain low to moderate REE concentrations and iron-oxide precipitates which are not able to scavenge large water volumes. as they do in deep sea metalliferous sediments, would probably also have moderate concentrations. Study of the New Brunswick Iron formations (GRAF. 1975. 1977) showed that concentrations of REE (except ELI) correlate positively with the amount of detrital material as represented by aluminosilicate minerals and negatively with the amount of oxide and sulfide minerals. This may in part explain the very low REE contents of some Precambrian iron formations (FRYER, 1977a). These rocks, being much more extensive than their younger counterparts, may consist almost entirely of chemical preclpltates. The analytical data tend to confirm this suggestion in that Precambrian samples rich in detrltal material (Soudan slate. FRYER, 1977a: and stilpnomelane shale, HR15. this study) have much higher concentrations of REE than do most other iron formations. REE concentrations in the non-magnetic concentrate from sample JG77B are approximately three times those in the magnetic concentrate (Table 1). However, the magnetite band in sample NO50 has REE concentrations almost ten times those in the quartz band. The banded Hamersley samples show little difference between bands with concentrations in the iron oxide bands generally somewhat higher.
These observations are best explained by differences from one formation to another m the mineral components of the iron-poor fraction but may also result from different processes. The REE distribution in the analyzed fractions of the New Brunswick sample was controlled by partitioning among the phases while the bandmg in the Hamersley and Adams Mine samples suggests some form of precipitation zoning. CONCLUSIONS Comparison
of the New Brunswick iron formations iron formations and other iron-rich has led to the following conclusions:
Iron formations and sea water
1. There exist many different types of iron formation. each with its own geologic setting and genetic history. 2. There is no good evidence that all iron formations precipitated from a sol&on identical to their contemporaneous sea water. It seems likely that in many cases the solution was some mixture of an iron source (hydrothermal solution?) and sea water. The iron source and degree of mixing may be different from one iron formation type to another or even from one iron formation to another. 3. The REE patterns of iron formations cannot be used to show a change with time in the sea water REE pattern. Observed differences are bettter related to iron formation type. 4. Algoma-type iron formations of both Ordovician and Archean age have similar REE patterns. suggesting stmilar processes of formatlon. 5. Iron formations and hematltic, Fe-Mn sediments of the same age from New Brunswick have different REE patterns. 6. REE patterns of Proterozoic iron formations suggest local variations m conditions of formation 7. Absolute REE concentrations m iron formations are related to the mineral components of the iron formations and to processes of formation. il~~no~t/rtl~e,llenr\-~~Thls work was done durmg the course of thesis research at Yale Umverslty. While at Yale. 1 had the benefit of many useful dlscuaslons with Professors B. J. SKINNER. K K. TUREK~AN and D M RYE Prof J. B. CORLISS. Oregon State Umverslty. Introduced me to the study of rare earth elements and to Instrumental neutron actlvatlon analysis and some of his computer programs were used m the data reduction. The assistance of the Rhode Island Nuclear Science Center and of A. F. DIMEGLIO. director. and M. P DOYLE. assistant director. was mvaluable during the analytlcal work. Hamersley samples were collected by B. J SKINNER Comments of B. J. SKINNER and D M. RYF on the mitral manuscript are much appreciated as are the reviews of H. D. HOLLAND and B J FRYFR. TERRY WEC~NERtyped the manuscript
REFERENCES CORL~SS J. B., GRAF J. L. JR. SKINNER B. J. and HUTCHINSON R. W. (1972) Rare earth data for iron- and manganese-rich sediments associated with sulfide ore bodies of the Troodos Massif. Cyprus. Geol. Sot. Am. (Abstracts) 1972 Annual Meeting. Minneapolis. Mn.. 476.
L. (1972) The geology and geochemistry of the Austin Brook Area with spectal emphasis on the Austin Brook Iron formation. Unpublished Ph.D Thesis. Carleton Untverslty. DYMOND J.. CORL~SS J B. HEATH G R.. FIELD C. W., DASCH E J. and VII:H H H. (1973) Orrgm of metalhferous sediments from the Pacific Ocean Gerjl. .S(lc 4m Bull. 84. 3355.-3372. FI ANA(;AN F J. (1973) 1972 values for mternational geochemical reference samples. Grochrm Cmmochrnt Autr 37, Il89~1200 FRYER B J (1977a) Rare earth evidence m Iron-formatlons for changmg Precambrian oxldatlon states Grochml. Cn\mochrm. 4~1~141, 36 I-367. FRYER B. J. (1977b) Trace element gcochemlstry of the Sokoman Iron Formation. Ctrn J Etrr/h &I. 14, 1598-1610. GOODWIN A M (1964) Geochemlcal studies at the Helen iron Range Ec,on. G~,~~/.59. 684-718 GOODWIN A M. (1973) Archean iron formattons and tectome basins of the Canadian Shield &on. Ger~l 68, 915-933 GOODWIN A M.. RIDLER R. H. and ANNFLLS R N with contributions bq BRIC;CS D. N 1 NALI~R~TT A. J , SPI.NCE A and SPFNCF C D. 11972) Precambrian volcanism of the Noranda-KIrkland Lake-Timmms. Mlchiplcoten. and Mamamae Point areas. Quebec and Ontario. 24th Int Geol Congress. Montreal. Canada. GuIdebook for excursions A40-C40. GORDON G. E. RANDALL K .GOLES G G.. Co~~iss J B, BEESON M. H and OXLLY S. S (1968) Instrumental actlLatlon analysis of standard rocks with high-resolution X-ray detectors Grochm. Covnochrm. Acru 32, 364-396. GRAF J. L.. JR. (1975) Rare earth elements as hydrothermal tracers during the formatlon of massive sulfide deposits and associated iron formations in New Brunswick Unpublished Ph.D Thesis. Yale University. 226~ GRAF J. L. JR. (1977) Rare earth elements as hydrothermal tracers during the formation of masstve sulfide deposits In volcanic rocks. Econ. Grol 72. 527 548 GROSS G. A (1965) Geoloyj, of Iron Drpmrr,~ 01 Cmdu. Vol. I. Grnerul Groloy~~ trnd Etuluutmn of Iron Drpo\u\ Geol. Survey Canada Econ Geol. Report No 22 R~BFRTSON A H. F and FLEET A. J (1976) The orlgms of rare earths In metalllferous sedlments of the Toodos Massif. Cyprus. Eurfh Plonrt Sir Lclt. 28, 385 394 TRFNDALL A F (1973~ Precambrian Iron-formattons of Australia. Eton. Grol 68. 1023-1034 ZELENOV K. K (1964) Iron and manganese m cxhalatlons of the submarine Banu Wuhu volcano (Indonesia) D&l. A!& Nod. SSSR 155, 9496
APPENDIX Sump/e lot drwn\
JGI: Fe-Mn Sediment. outcrop south of Wedge Mine, NB. Extremely fine-gramed hematltic sediment with some detrltal material (AI,O, = 7”,,) JGSOA: Fe -Mn Sedtment. Outcrop. Tetogouche Falls. NB. Extremely line-grained hematitlc sediment with detrltal material (A1203 = 12”,,). N095B: Banded cherty Iron formation. Lucy Mine. Michlpicoten District. Ontario. 43”,, Quartz. 13”,, magnetite. 44”,, carbonate. trace of pyrite N050W: Banded Iron formation. Adams Mine. Ontario 53”,, magnetite, 4l”,, quartz. 3”,, carbonate. 3”,, green mica. trace 0r pyrite. N022B: Massive sulfide band, outcrop up dip from Amulet ‘C’ Deposit. Noranda District, Quebec. Roughly 80”” pyrite and 20”,, quartz with traces of chlorite. stlIpnome_ lane and carbonate. HRI: Banded Iron formation, Mt Sylvia Formation, North Whaleback Pit. Hamerslej Range, Australia.’ HRlH-band with roughly 80”,, hematite. 20”,, quartz
J. L. GRAFJR.
HRIQH-quartz band wtth thinner bands containing finely disseminated hematite and specularite grains. HR13: Banded iron formatton. BIFI. Dales Gorge Mine. Hamersley Range. Member. Colonial HRI3M-Band of nearly pure magnetite. HRI3Q-Fepoor band, largely quartz and carbonate with very fine opaques. Contains thm bands of shghtly concentrated opaques. HR15: Stilpnomelane shale. Jo&e Falls Member, Hamersley Range. Contains about 50”,, stilpnomelane. 30”,, opaques. and 20”,, quartz. HR27: Chert pod, Intersection of Red and Hancock Gorges. Hamersley Range. Very fine-grained quartz containing dark bands. From the bands many dark blue needles (rtebeckite?) extend out mto the quartz. JG3ti Banded iron formation near top of Brunswtck Mmmg and Smelting No. 6 iron formatron. l6”,, magnetite.
35”,, carbonate, 34”,, quartz. IS”,, white mtca. trace-hemattte, apatite. JG3417: Banded non formation near center of Brunswtck Mining and Smelting No. 6 tron formation. 72”,, magnettte. 5”,, hematite. 18”,, quartz. 2”” carbonate, 3”; white mica. JG36D: Banded iron formatton near top of Austm Brook tron formation. 36”,, magnetrte. 36”,, carbonates, 13”,, quartz, 8”,, chloride. 7”” white mtca. trace-apatite. JG77B: Banded non formation near bottom of Austm Brook tron formation. 729,, magnettte, 16”,, quartz. 7”” chlortte. Y,, carbonate, trace-apatite, hematite, sphalerrte. 77BM ts a magnettc concentrate and JG77BNM a nonmagnettc concentrate from this sample JG37G: Banded iron formatton. May be lateral equtvalent of Austm Brook tron formatton 62”” magnetite, 18”,, carbonate. 17”,, quartz. 3”,, chlorite. trace-apatite.