Provenance analysis of lower Paleozoic cratonic quartz arenites of the North American midcontinent region: U-Pb and Sm-Nd isotope geochemistry

Clark M. Johnson* Department of Geology and Geophysics, University of Wisconsin, Bryce L. Winter } 1215 West Dayton Street, Madison, Wisconsin 53706

ABSTRACT separates reflect mixing of the age groups and 1994), but few isotopic studies have focused on approximate relative proportions that are the quartz grains. Lower Paleozoic supermature quartz aren- identified from the single zircon results. The stratigraphy and physical sedimentology ites in Wisconsin and Michigan were derived All isotope data on quartz separates and U- of lower Paleozoic quartz arenites in the north- from several Proterozoic and Archean ter- Pb zircon data indicate that the detrital con- ern midcontinent region of North America have ranes. Single-grain, detrital zircon populations stituents (zircons and quartz framework been intensively studied; the provenance and from the Galesville and St. Peter Sandstones in grains) that compose the lower Paleozoic many aspects concerning the depositional his- Wisconsin yield very similar, closely concor- quartz arenites in Wisconsin were primarily tory of these extensive cratonic sheet sand- dant age distributions (Cambrian Galesville derived from the 2.7 Ga granite-greenstone stones have been the subject of considerable Sandstone: 1.1 Ga [n = 4], 1.4 Ga [n = 2], 1.8 Ga terrane of the southern Superior Province and debate since the early 1900s (Dake, 1921; Dott [n = 1], and 2.7 Ga [n = 2]; Ordovician St. Peter a 1.1 Ga terrane. The latter terrane is either and Byers, 1981; Dott et al., 1986). On the ba- Sandstone: 1.1 Ga [n = 2], 1.8 Ga [n = 1], and the silicic volcanic rocks associated with the sis of directional indicators of sediment trans- 2.7 Ga [n = 6]). In contrast, most of the nine de- Midcontinent rift system, or, more likely, the port (Potter and Pryor, 1961; Dott et al., 1986) trital zircons that were analyzed from the St. voluminous granitic rocks that are associated and the composition of heavy minerals (Tyler, Peter Sandstone in the Michigan basin have with the Grenville Province on the eastern 1936), it has been suggested that detrital grains high U contents (450–2500 ppm) and are margin of North America. The Middle Prot- that compose the cratonic sandstones in Wis- strongly discordant (60%–90%). Six zircons erozoic Grenville Province was the most im- consin were ultimately derived from Precam- from the Michigan basin that have ca. 1.0 Ga portant ultimate source of quartz and zircons brian felsic plutonic rocks that are exposed in 207Pb*/206Pb* ages define a regression line that to the St. Peter Sandstone in the Michigan the Lake Superior region. Precambrian bedrock has an upper intercept of about 1100 Ma and a basin; lesser amounts of material were con- of the northern midcontinent region of North lower intercept of 15 Ma; one zircon has a 2.7 tributed from the Archean Superior Province. America includes rocks that span >2.5 b.y. in Ga 207Pb*/206Pb* age. The zircon data indicate age (Fig. 1; e.g., Hoffman, 1989; Sims et al., that although a number of different terranes INTRODUCTION 1989), reflecting a diverse range of potential contributed detrital material to the Paleozoic sources for the Paleozoic cratonic sandstones in quartz arenites in Wisconsin, 1.1 and 2.7 Ga Supermature quartz arenites are common in Wisconsin and Michigan. These sources in- terranes were the dominant sources, and not the sedimentary record, particularly in the Prot- clude the >3.0 Ga southern gneiss terrane of the the local basement, which primarily consists of erozoic and lower Paleozoic sections of stable Superior Province; the ca. 2.7 Ga granite- the ≥2.7 Ga Marshfield terrane, the 1.8 Ga cratons, and usually are composed of >97% greenstone terrane of the Superior Province; Penokean orogen, and the 1.4 Ga Wolf River quartz and <1% heavy minerals (e.g., Tyler, the ≥2.7 Ga Archean inlier of the Marshfield batholith. A terrane that has a 1.1 Ga age is 1936; Dott et al., 1986). However, conventional gneiss terrane in central Wisconsin; the ca. probably the main source for the St. Peter petrographic (e.g., Dickinson and Suczek, 1979; 1.8–1.7 Ga Penokean orogen and postorogenic Sandstone in the Michigan basin. Folk, 1980) and bulk geochemical (e.g., Bhatia rocks; the ca. 1.4 Ga anorogenic granites of the Quartz separates were also analyzed for Pb- and Crook, 1986; Basu et al., 1990) approaches Wolf River batholith, eastern granite-rhyolite Pb and Sm-Nd isotope variations, and the data to provenance analysis of supermature quartz province, and parts of the Grenville Province; do not indicate significant source differences arenites generally yield few insights into these and the ca. 1.1 Ga Keweenawan rocks of the between the heavy mineral fraction (zircons) homogeneous rocks. The source ages of heavy Midcontinent rift system and the majority of and the quartz framework grains. Pb-Pb mineral components, such as zircons, have proven the Grenville Province (Fig. 1). isochrons and Sm-Nd isotope data for quartz to be an important aspect of provenance studies In this contribution we report the results of a (e.g., Ross and Parrish, 1991; Zhao et al., 1992; rare earth element (REE), Sm-Nd, Pb-Pb, and U- *E-mail: [email protected]. McLennan et al., 1993; Smith and Gehrels, Pb isotope study of quartz grains, as well as U-Pb

Data Repository item 9990 contains additional material related to this article.

GSA Bulletin; November 1999; v. 111; no. 11; p. 1723–1738; 11 figures; 6 tables.

1723 JOHNSON AND WINTER geochronology on single detrital zircons, from two lower Paleozoic supermature quartz sand- stone units in south-central Wisconsin (Cambrian Galesville Sandstone and Ordovician St. Peter Sandstone), and one unit in the east-central Michigan basin (Ordovician St. Peter Sandstone) (Figs. 1 and 2). Both zircons and quartz were an- alyzed in an attempt to address possible differ- ences in the sources of zircons and the primary quartz framework grains. The data highlight the great antiquity of most quartz and heavy minerals in the cratonic quartz arenites, and identify source terranes and possible sediment transport pathways.

GEOLOGIC SETTING

The lower Paleozoic stratigraphy of the North American midcontinent is characterized by 60– 90 m.y. (i.e., second order) cycles that consist of a basal quartz arenite and an upper carbonate unit, the former of which was deposited on wide- spread unconformities. These well-known, inten- sively studied strata were the basis for the devel- opment of the term orthoquartzite-carbonate Figure 1. Precambrian geologic basement map of the Great Lakes region (adapted from suite (Pettijohn, 1957), and the concept of uncon- Van Schmus, 1992). A major crustal boundary, the Great Lakes tectonic zone (GLTZ), divides formity-bounded lithologic sequences (Sloss, the southern Superior Province into a southern gneiss terrane (GT) of age 3.0–3.6 Ga and a 1963). These sandstones, which have exceptional northern granite-greenstone terrane (GGT) of age 2.7 Ga. The Niagara fault zone (NFZ) is the compositional and textural maturity, are com- southern boundary of the Superior Province and represents the suture zone with the magmatic posed of >97% unstrained, monocrystalline arcs of the ca. 1.8 Ga Penokean orogen. An inlier of 2.7 to >3.2 Ga Archean rocks in the Penokean quartz, and there is a distinct paucity of shale. orogen is defined as the Marshfield terrane (MT). Anorogenic plutons emplaced ca. 1.4 Ga per- They have a sheet-like geometry, and typically forate the Early Proterozoic crust throughout the midcontinent region; the Wolf River batholith have a thickness of 40–50 m over thousands of (WRB) in northeastern Wisconsin is a typical example. South of the Penokean orogen is the ca. square kilometers (Dott et al., 1986). 1.4 Ga eastern granite-rhyolite province (EGRP). The ca. 1.1 Ga Midcontinent rift system Most current sedimentologic models agree (MCRS) consists of an axial portion that is dominated by mafic igneous rocks (black), and postrift that the sheet geometry is largely the result of eo- flanking sedimentary basins (diagonal rules). The ca. 1.1 Ga Grenville Province (GP) comprises lian and fluvial processes that, following ex- the eastern margin of the North American craton. Additional data are from Sims et al. (1989). tended periods of subaerial erosion, distributed sand across large areas of the craton (Dott et al., 1986). Nonmarine deposition was followed by mented, and typically have only incipient quartz were a minor component in the ultimate source reworking and further deposition of sand in a overgrowths (<2 µm; Odom et al., 1976). Our en- region(s) (Tyler, 1936). shallow-marine environment. Paleocurrent indi- ergy-dispersive electron-microprobe analysis of The predominance of only the most durable cators from the subaerial facies of Cambrian and mineral inclusions in 38 quartz grains from these type of quartz (i.e., unstrained, monocrystalline) Ordovician quartz arenites are southwest di- two formations supports the detailed study of and composition of the framework heavy miner- rected, consistent with inferred deposition in the Tyler (1936), who physically separated and als in the detrital suites led Tyler (1936) to sug- paleo-trade wind belt. Paleocurrent indicators counted a statistically significant number of min- gest that the detrital grains have been through from the subaqueous facies are consistent with eral inclusions. Tyler’s results from the St. Peter multiple sedimentary cycles, and that the imme- other stratigraphic evidence that the Precambrian Sandstone indicate that zircon and apatite are diate source was a sedimentary terrane, rather craton of the Great Lakes region was a subtle dominant (70%–90% of heavy mineral inclu- than an igneous or metamorphic source. Possible topographic high, and therefore an ultimate sions), and titanite, rutile, ilmenite, garnet, and sedimentary sources for the Galesville Sandstone source of sediment throughout Paleozoic time pyrite are present in most samples. Biotite and and St. Peter Sandstone in Wisconsin, based on (Potter and Pryor, 1961). hornblende are important inclusions locally, and Tyler’s work, include the Cambrian Mt. Simon fluorite and kyanite are rarely present as mineral Sandstone (Wisconsin), the Upper Keweenawan Wisconsin Sandstones inclusions in quartz. The predominance of zircon (ca. 1.1 Ga) Hinkley Sandstone (Minnesota) and and apatite as inclusions suggests that most of the Bayfield Group sandstones (Wisconsin), and sev- The Cambrian Galesville Sandstone and the Or- quartz grains were derived from a terrane that eral of the Middle Proterozoic quartzites of the dovician St. Peter Sandstone in Wisconsin (Fig. 2) consisted largely of intermediate- to silicic-com- Baraboo interval (i.e., Sioux-Minnesota, Barron- are composed of quartz that is medium to coarse position plutonic rocks, whereas the presence of Wisconsin, Flambeau-Wisconsin). However, the grained, well sorted, and well rounded (Dott et al., only small quantities of kyanite and garnet inclu- majority of Middle Proterozoic quartzites in Wis- 1986). The quartz grains are very poorly ce- sions suggests that pelitic metamorphic rocks consin, including the Baraboo, McCaslin, and

1724 Geological Society of America Bulletin, November 1999 LOWER PALEOZOIC CRATONIC QUARTZ ARENITES, NORTH AMERICAN MIDCONTINENT

zoic quartzites in Wisconsin, these procedures have the greatest likelihood of producing concor- dant U/Pb analyses (Van Wyck, 1995). Single zir- cons were then hand-picked for analysis on the basis of clarity and the absence of cracks and in- clusions in an attempt to analyze only those zir- cons that would give concordant ages. In order to analyze zircons reflecting all of the possible source terranes, the complete range of colors and morphological types was selected from each sample. Nonetheless, it remains possible that our sampling is biased, because only the largest zir- cons were analyzed in order to ensure sufficient radiogenic lead. Each zircon was cleaned two

times with warm ~7 M HNO 3 for ~30 min, fol- lowed by rinsing two times with H2O. After spik- ing with a 205Pb–235U tracer, each zircon was dis- solved with HF in Teflon microcapsules within a Teflon-lined acid-digestion bomb, evaporated to dryness, and then redissolved in HCl (Parrish, Figure 2. Generalized map (adapted from Dott et al., 1986) showing the outcrop areas of the 1987). U and Pb were separated using the meth- Galesville Sandstone and the St. Peter Sandstone in Wisconsin, and the isopachs (contour inter- ods of Parrish et al. (1987), and then analyzed on val = 100 m) of the St. Peter Sandstone in the Michigan basin. Stars mark general localities in this a Micromass Sector 54 mass spectrometer using study. Precise locations for sample collection sites are given in Data Repository Tables DR-1 and single-collector analysis and a Daly detector. DR-2.1 The outcrop in the northern peninsula of Michigan is that of the Munising Formation, which is presumed to be correlative with the Galesville Sandstone. The generalized lower Paleo- Quartz Grains zoic stratigraphy of Wisconsin is also shown. Samples for U-Pb analysis of quartz were col- lected from: (1) five different outcrops of the Waterloo Quartzites, are unlikely to be important potassium feldspar that are usually present at the Galesville Sandstone near Lone Rock, Wiscon- sand contributors because of the strained nature edge of the mounted grain, indicating that feldspar sin, and the Wisconsin Dells (one sample from of the quartz grains and the absence of tourma- occurs as diagenetic overgrowths. each locality), (2) six different outcrops of the St. line (e.g., Sims et al., 1993). The Cambrian Jor- Peter Sandstone in the vicinity of Madison, Wis- dan Sandstone (Wisconsin) and the Keweenawan ANALYTICAL TECHNIQUES consin (nine individual samples), and (3) two Freda Sandstone (Oronto Group, Wisconsin) are cores taken at ~3600 m depth in the east-central also deemed unlikely direct sources because of Detrital Zircons Michigan basin (19 individual samples) (Fig. 2). their relatively high garnet content (Tyler, 1936). The sand grains were fully disaggregated in a We collected ~70 kg of rock from outcrops of clean mortar and pestle and then sieved to obtain Michigan Basin Units the Ordovician St. Peter Sandstone near Madi- single, medium sand-sized (300–500 µm) son, Wisconsin, and from the Cambrian grains. This process produced samples of virtu- The lithology of the St. Peter Sandstone in the Galesville Sandstone in the vicinity of the Wis- ally 100% pure quartz with minimal detrital Michigan basin is more variable, vertically and lat- consin Dells (Fig. 2; see Table DR-11). In addi- heavy mineral contamination, because the trace erally, than it is in Wisconsin (e.g., Winter et al., tion, ~20 kg of the St. Peter Sandstone was ob- constituents (e.g., feldspar and heavy minerals) 1995). Samples for this study were obtained from tained from a core drilled in the east-central in these sandstones are generally <150 µm (e.g., a depth of ~3600 m in the east-central Michigan portion of the Michigan basin (Fig. 2; Table DR- Odom et al., 1976). The quartz grains analyzed basin, just south of Saginaw Bay (Fig. 2). In this 2 [see footnote 1]). Samples were crushed and from Wisconsin were very well rounded and part of the Michigan basin (Fig. 2), the St. Peter zircons separated using standard shaker table, contained no quartz overgrowths visible under a Sandstone is ~200 m thick and is composed of ma- heavy liquid, and magnetic techniques. The non- binocular microscope, whereas many samples ture quartz arenite that is interstratified with seven magnetic zircons were separated into size frac- from the Michigan basin contained visible carbonate units (1–6 m thick). The sandstone is tions using disposable nylon screens. After air quartz overgrowths. principally composed of quartz that is medium to abrading the largest size fractions with pyrite for For U/Pb or Pb-Pb isotope analysis, quartz coarse grained, well sorted, well rounded, and ~20 hr (Krogh, 1982), the zircons were cleaned grains were ultrasonically cleaned in ~5 M HNO3 moderately to tightly cemented by quartz over- repeatedly in warm ~5 M HNO 3. On the basis of for ~2 hr, rinsed with H2O, leached in warm 5 M growths (e.g., Graham et al., 1996). Energy- our experience with detrital zircons in Protero- HNO3 for ~10 hr, and then vigorously rinsed dispersive electron-microprobe analyses of min- multiple times with H2O. Quartz grains that were eral inclusions in 21 quartz grains from the Michi- 1GSA Data Repository item 9990, sample loca- anomalously colored or contained visible inclu- gan basin identify the presence of zircon, rutile, il- tion data tables, is available on the Web at sions under a binocular microscope were subse- menite, and apatite. In addition, it is important to http://www.geosociety.org/pubs/drpint.htm. Re- quently removed by hand-picking. After a final quests may also be sent to Documents Secretary, note that 9 of the 21 samples analyzed contain rel- GSA, P.O. Box 9140, Boulder, CO 80301; e-mail: rinse with warm HNO3 and multiple H2O rinses, atively large domains (50–200 µm across) of [email protected]. quartz grains were dissolved in Teflon vials with

Geological Society of America Bulletin, November 1999 1725 JOHNSON AND WINTER

TABLE 1. U-Pb DATA FOR SINGLE DETRITAL ZIRCONS FROM LOWER PALEOZOIC QUARTZ ARENITES, NORTHERN MIDCONTINENT Sample† Mass§ U 206Pb* 206Pb/204Pb 206Pb*/238U 207Pb*/235U Correlation 207Pb*/206Pb* 207Pb/206Pb Percent (µg) (ppm) (ppm) raw coefficient age discordancy (Ma) number St. Peter Sandstone, Michigan basin (field sample MI-SP, locality 13) MI-SP-A c 8 2508 35 1306 0.0161 ± 0.50% 0.1604 ± 0.51% 0.96 0.0724 ± 0.14% 998.3 ± 5.7 90 MI-SP-B spi 32 594 11 1850 0.0209 ± 0.40% 0.2590 ± 0.40% 0.97 0.0898 ± 0.09% 1420.9 ± 3.5 92 MI-SP-C spi 103 56 11 4660 0.2371 ± 0.15% 2.4478 ± 0.15% 0.95 0.0749 ± 0.05% 1065.5 ± 2.0 –32 MI-SP-D sy 14 662 125 4699 0.2190 ± 0.17% 5.2705 ± 0.18% 0.98 0.1745 ± 0.04% 2601.7 ± 1.2 56 MI-SP-E c 103 52 10 2407 0.2154 ± 0.28% 2.3010 ± 0.29% 0.97 0.0775 ± 0.07% 1133.3 ± 3.0 –12 MI-SP-F dp 20 2144 13 821 0.0072 ± 1.02% 0.1814 ± 1.01% 1.00 0.1827 ± 0.09% 2677.4 ± 3.0 99 MI-SP-G spi 40 444 5 416 0.0137 ± 1.23% 0.1361 ± 1.28% 0.95 0.0723 ± 0.39% 995 ± 16 92 MI-SP-H spi 294 38 2 1496 0.0466 ± 0.59% 0.5119 ± 0.59% 0.97 0.0797 ± 0.14% 1190.2 ± 5.3 77 MI-SP-I c 30 500 21 2242 0.0498 ± 0.41% 0.4954 ± 0.42% 0.97 0.0722 ± 0.10% 990.8 ± 4.0 70

St. Peter Sandstone, Wisconsin (field sample WI-SP, locality 3) WI-SP-A db 17 124 46 643 0.4238 ± 0.55% 11.0989 ± 0.55% 0.99 0.1900 ± 0.09% 2741.8 ± 3.0 20 WI-SP-B mb 23 38 6 132 0.1668 ± 2.88% 1.6939 ± 3.22% 0.90 0.0737 ± 1.38% 1032 ± 56 4 WI-SP-C c 14 86 36 879 0.4805 ± 0.84% 12.0206 ± 0.83% 1.00 0.1814 ± 0.08% 2666.0 ± 2.6 6 WI-SP-D c 38 75 21 514 0.3154 ± 0.52% 4.8532 ± 0.57% 0.93 0.1116 ± 0.21% 1825.5 ± 7.5 4 WI-SP-E c 27 17 8 144 0.4826 ± 2.07% 12.2801 ± 2.08% 0.98 0.1846 ± 0.41% 2694 ± 14 7 WI-SP-F db 20 127 56 1025 0.5054 ± 0.39% 13.1463 ± 0.39% 0.99 0.1887 ± 0.06% 2730.6 ± 1.9 4 WI-SP-G sb 20 40 7 142 0.1744 ± 3.03% 1.8014 ± 3.25% 0.93 0.0749 ± 1.21% 1066 ± 48 3 WI-SP-H dp 23 49 22 302 0.4922 ± 0.87% 12.6348 ± 0.89% 0.98 0.1862 ± 0.19% 2708.7 ± 6.3 6 WI-SP-I sp 14 58 27 250 0.5129 ± 1.15% 12.8624 ± 1.16% 0.98 0.1819 ± 0.24% 2670.0 ± 7.8 0

Galesville Sandstone, Wisconsin (field sample WI-G, locality 8) WI-G2 my 18 25 3 146 0.1578 ± 6.35% 1.5867 ± 6.30% 0.97 0.0729 ± 1.47% 1012 ± 59 7 WI-G3 sp 20 42 18 347 0.4802 ± 1.94% 11.6912 ± 1.91% 1.00 0.1766 ± 0.17% 2621.1 ± 5.8 4 WI-G4 sp/b 21 45 18 385 0.4652 ± 1.80% 11.8213 ± 1.77% 1.00 0.1843 ± 0.16% 2692.1 ± 5.4 10 WI-G5 c 21 138 21 1327 0.1780 ± 0.61% 1.8440 ± 0.62% 0.97 0.0751 ± 0.15% 1072.2 ± 5.9 2 WI-G6 c 28 87 17 794 0.2222 ± 0.86% 2.6175 ± 0.86% 0.98 0.0854 ± 0.18% 1325.2 ± 7.0 3 WI-G7 c 15 119 25 767 0.2433 ± 1.23% 3.0691 ± 1.24% 0.98 0.0915 ± 0.25% 1456.6 ± 9.6 4 WI-G8 my 32 41 6 447 0.1750 ± 2.34% 1.8030 ± 1.99% 0.97 0.0747 ± 0.53% 1062 ± 21 2 WI-G9 my 11 100 15 400 0.1763 ± 2.34% 1.8436 ± 2.34% 0.98 0.0758 ± 0.52% 1090 ± 21 4 WI-G10 c 24 49 13 817 0.3194 ± 0.65% 4.8777 ± 0.66% 0.98 0.1108 ± 0.13% 1812.1 ± 4.7 2

Notes: Isotope compositions have been adjusted for the following: (1) blank Pb (15–50 pg, 206Pb/204Pb = 18.8 ± 0.2, 207Pb/204Pb = 15.4 ± 0.2, 208Pb/204Pb = 37.6 ± 0.5) and U (<9 pg); (2) fractionation of +0.12% ± 0.05% per amu for Pb (based on 14 analyses of NBS-981 on Daly detector) and +0.05% ± 0.04% per amu for U (based on 7 analyses of U500 on Daly detector); (3) Pb contribution from 205Pb-235U spike; (4) initial Pb isotope composition from Stacey and Kramers model (1975), assuming uncertainties of 1.5% for 206Pb/204Pb, 0.4% for 207Pb/204Pb, and 2.0% for 206Pb/204Pb. Uncertainties for isotope ratios reported above are 1σ (SD) in percent; uncertainties for ages are 2σ (SE) in Ma. All data computations were made using the programs of Ludwig (1991a, 1991b). Locations of sample collection sites are given in Tables DR-1 and DR-2 (see text footnote 1). *Radiogenic. †Zircon appearance: s—slight, m—medium, d—dark, c—colorless, b—brown, p—purple, pi—pink, y—yellow. §Mass calculated from measured dimensions, assuming density of 4.69 g/cm3. #Percent discordancy is calculated using the 206Pb*/238U age relative to the 207Pb*/206Pb* age. concentrated HF on a hot plate for ~12 hr. An INTERPRETATION OF DATA FOR sandstones are relatively low (~20–140 ppm; aliquot of some samples was spiked with a mixed SINGLE DETRITAL ZIRCONS Table 1), and are within the range measured for 208Pb–235U tracer to determine concentrations. zircons from Archean intrusive rocks of the HF dissolution was followed by treatment with Detrital zircon U-Pb isotope data indicate that southwestern Superior Province (e.g., Davis and hot 6M HCl, to ensure that all inclusions would a variety of sources contributed to the sandstones, Edwards, 1986; Corfu, 1988), or the Middle be dissolved; no solid residue was visible. Pb and ranging from 2.7 to 1.1 Ga in age. The single zir- Proterozoic Grenville Province (e.g., McLelland U were separated using a small volume of anion con data provide important constraints for inter- et al., 1993; Owens et al., 1994). The concor- exchange resin in HBr and HNO3, respectively, preting the Pb-Pb and Sm-Nd data of the bulk dance of most of the analyses indicates that the then analyzed on a Micromass Sector 54 mass quartz samples discussed later. 207Pb*/206Pb* ages can be interpreted as crystal- spectrometer using static multicollection. lization ages with a high degree of confidence. For REE and Sm-Nd isotope analysis, large Wisconsin Most of the zircons were naturally well rounded samples (~500 mg) of cleaned (as in preceding) and contained few facets. All zircons that had a quartz from the Galesville Sandstone and St. Peter Nine zircons that have different colors and purple tint are Archean in age, although not all Sandstone in Wisconsin and Michigan were dis- morphology from the Cambrian Galesville Sand- Archean zircons are purple; in general, there is solved in Teflon vials with concentrated HF on a stone in Wisconsin yield nearly concordant ages little correlation between the physical character hot plate for ~12 hr. HF dissolution was followed of ca. 1050 Ma (n = 4), ca. 1400 Ma (n = 2), ca. of the zircons and their ages. by treatment with hot 6M HCl (see preceding), 1800 Ma (n = 1), and ca. 2700 Ma (n = 2) (Table The distribution of ages for detrital zircons and no solid residue was observed. The solutions 1; Fig. 3). Eight of nine zircons representing var- from the two Wisconsin sandstones is remark- were analyzed for REE contents (by isotope dilu- ious populations from the Ordovician St. Peter ably similar, although ca. 1.4 Ga zircons are not tion) and Nd isotope ratios. The analytical meth- Sandstone in Wisconsin yield nearly concordant represented in the nine grains analyzed from the ods and standard Nd isotope ratios were discussed ages of ca. 1050 Ma (n = 2), ca. 1800 Ma (n = 1), St. Peter Sandstone in Wisconsin. Tyler’s (1936) in detail in Johnson and Thompson (1991) and and ca. 2700 Ma (n = 6) (Table 1; Fig. 3). Ura- interpretation that the St. Peter Sandstone in Wis- VanWyck (1995). nium contents for the zircons from both of these consin was derived largely from the erosional re-

1726 Geological Society of America Bulletin, November 1999 LOWER PALEOZOIC CRATONIC QUARTZ ARENITES, NORTH AMERICAN MIDCONTINENT

cate that most of the zircons from the Michigan basin reflect derivation from a ca. 1100 Ma source terrane, which included distinctive high- U zircons.

INTERPRETATION OF DATA FOR DETRITAL QUARTZ GRAINS

The age and sources of quartz grains in sedi- mentary rocks has been a long-standing problem. In the following we show how U-Pb and Sm-Nd isotope systematics of bulk quartz separates can be explained through sedimentary mixing processes, using the age constraints provided by the detrital zircons.

Results of U-Pb and REE Analyses

As shown by previous studies, natural quartz can contain significant U, Pb, and REE contents (Götze and Lewis, 1994; Hemming et al., 1994). Lead and U contents of 11 quartz samples ana- lyzed in this study range from 0.10 to 1.1 ppm Figure 3. U-Pb concordia diagram for single detrital zircons from lower Paleozoic quartz and 0.065 to 0.26 ppm, respectively (Tables 2 arenites of the northern midcontinent region of North America. Likely source terranes for the and 3). The quartz samples from the Ordovician zircon age groups are indicated (see Fig. 1). Most U-Pb ages from Wisconsin are nearly concor- St. Peter Sandstone (n = 11) and the Cambrian dant, in marked contrast to the generally discordant ages from the Michigan basin. Most of the Galesville Sandstone (n = 12) in Wisconsin have highly discordant analyses from the Michigan basin have 207Pb*/206Pb* ages of ca. 1.1 Ga (see similarly radiogenic but variable Pb isotope ra- Table 1). Error ellipses are much smaller than the plotted symbols. Inset shows detail for dis- tios (206Pb/204Pb = 19.9–35.2 and 21.3–58.9, re- cordant ca. 1.1 Ga zircons from the Michigan basin. spectively), whereas quartz from the Michigan basin (n = 21) has relatively nonradiogenic and more restricted Pb isotope ratios (206Pb/204Pb = working of the Galesville Sandstone, based on cordant (Table 1; Fig. 3, inset), which is in 16.8–32.3; all but two are between 16.8 and virtually identical heavy mineral suites, is sup- striking contrast to the U-Pb systematics of zir- 20.7) (Table 2; Fig. 4). The 206Pb/204Pb ratios for ported by our results, which indicate similar age cons from this formation in Wisconsin. Many all quartz samples analyzed in this study are less distributions of the detrital zircons in the two of the zircons from the Michigan basin have radiogenic than those measured for the Early units. The ages determined for detrital zircons markedly higher U contents (all but three have Proterozoic Pokegama Quartzite (Minnesota) from both Wisconsin sandstones include most all U contents of ~450–2500 ppm) than those from studied by Hemming et al. (1994), and we inter- of the ages known for Precambrian basement Wisconsin, and the most discordant zircons are pret this to reflect a larger proportion of nonradi- rocks exposed in the Great Lakes region (Fig. 3), generally those that have the highest U con- ogenic mineral inclusions or overgrowths in the although there is a marked clustering of ages at tents. The contrast in concordancy in the zir- quartz grains studied here. REE contents of the 1.1 Ga (typical of volcanic rocks of the Midcon- cons from Wisconsin and Michigan is not due three quartz samples from this study (Nd = tinent rift system and the Grenville Province) and to analytical procedures, because both suites 0.7–2.1 ppm; Table 4) overlap those measured 2.7 Ga (typical of southern Superior granite- were analyzed simultaneously, and results of for whole-rock samples of the Pokegama greenstone terrane). Zircons of ca. 1800 Ma age zircon standard analyses were consistent Quartzite (Hemming et al., 1994; Fig. 5A). are surprisingly rare, given the extensive expo- throughout the study (Table 1). Zircons that sures of Penokean orogenic rocks in central and have comparatively high U contents are found Location of U, Pb, and REEs in northern Wisconsin (Fig. 1) and the large abun- in some highly evolved granitic rocks of the Quartz Grains dance of 1.8 Ga detrital zircons in Baraboo inter- Grenville Province (e.g., Chiarenzelli and val quartzites (Van Wyck, 1995; Medaris et al., McLelland, 1993). Although highly discordant 208Pb/204Pb–206Pb/204Pb Systematics. Our Pb 1996). Early Archean zircons from the 3.0–3.6 zircons make it difficult to confidently deter- isotope data and REE patterns for quartz grains are Ga Superior gneiss terranes, as well as older mine the crystallization ages, regression of six best explained by microinclusions of common ac- rocks from the nearby Marshfield terrane zircons that have 207Pb*/206Pb* ages of ca. 1 Ga, cessory and/or heavy minerals such as zircon, ti- (2.7–>3.2 Ga) are absent from the detrital zircon including the two that are reversely discordant, tanite, apatite, and monazite. Zircon, titanite, and population (Figs. 1 and 3). yields an upper intercept of 1104 ± 118 Ma and apatite are the most likely U- and REE-rich acces- a lower intercept of 14 ± 24 Ma (MSWD [mean sory (heavy) minerals of those identified by Tyler Michigan Basin square of weighted deviates] = 8.7). One zircon (1936) as inclusions in quartz grains. However, has a 207Pb*/206Pb* age of ca. 2600 Ma and is 208Pb/204Pb–206Pb/204Pb isotope variations of the All nine zircons analyzed from the St. Peter ~60% discordant (MI-SP5), and is certainly an quartz grains require that a mineral that has a high Sandstone in the Michigan basin are highly dis- Archean zircon. We interpret these data to indi- Th/U ratio, such as monazite, is present as a minor

Geological Society of America Bulletin, November 1999 1727 TABLE 2. U AND Pb DATA FOR DETRITAL QUARTZ FROM LOWER PALEOZOIC QUARTZ ARENITES, NORTHERN MIDCONTINENT Sample* Locality Mass U Pb 206Pb/204Pb 207Pb/204Pb 208Pb/204Pb (mg) (ppm) (ppm) St. Peter Sandstone, Wisconsin WI-SP-1 1 92.2 0.154 0.948 27.468 17.084 41.524 WI-SP-2 1 103.2 0.065 0.102 29.058 17.245 40.876 WI-SP-3 2 106.6 0.164 0.948 22.292 16.110 40.314 WI-SP-4 2 127.3 0.200 0.230 35.221 18.269 45.071 WI-SP-5 3 123.9 0.196 0.842 24.933 16.590 42.057 WI-SP-6 3 93.1 0.158 0.218 33.839 18.170 48.269 WI-SP-7, M-2.5 4 66.3 N.D. N.D. 19.886 15.802 39.595 WI-SP-7, NM-1 4 81.3 N.D. N.D. 23.183 16.430 43.447 WI-SP-10, NM-1 5 65.1 N.D. N.D. 21.538 16.083 41.491 WI-SP-10, M-1 5 72.9 N.D. N.D. 21.383 16.122 41.439 WI-SP-11 6 100.6 N.D. N.D. 30.217 17.768 42.207

Galesville Sandstone, Wisconsin GALE-D1, NM-1 7 89.3 N.D. N.D. 26.767 17.082 44.839 GALE-D1, M-2.5 7 94.0 N.D. N.D. 23.573 16.409 43.659 GALE-D1, M-5 7 77.7 N.D. N.D. 23.702 16.384 41.604 GALE-D6, NM-1 8 75.6 N.D. N.D. 41.360 19.676 49.536 GALE-D6, M-2.5 8 97.2 N.D. N.D. 21.308 16.048 39.772 GALE-D4 9 97.3 N.D. N.D. 25.559 16.694 43.868 GALE-D2, NM-1 10 99.2 N.D. N.D. 22.611 16.305 40.471 GALE-D2, M-5 10 46.9 N.D. N.D. 23.731 16.565 39.687 GALE-LR1, NM-1 11 99.3 N.D. N.D. 29.857 17.446 45.720 GALE-LR1, M-1 11 83.0 N.D. N.D. 42.030 19.127 55.398 GALE-LR1, M-2.5 11 88.7 N.D. N.D. 47.937 20.988 57.582 GALE-LR1, M-5 11 49.5 N.D. N.D. 58.945 21.957 55.137

St. Peter Sandstone, Michigan basin WI0944.3A 12 120.8 N.D. N.D. 17.826 15.583 36.672 W10944.3B 12 118.5 N.D. N.D. 17.877 15.558 36.985 W10950.2 12 153.2 N.D. N.D. 27.052 16.904 46.664 W10951.3 12 94.3 N.D. N.D. 18.106 15.600 37.226 W10951.7 12 114.0 N.D. N.D. 20.699 16.048 39.944 W10991.2 12 336.8 0.272 1.11 19.737 15.820 38.903 W11002.9 12 207.9 N.D. N.D. 17.816 15.540 37.470 SW10985.7 13 106.5 N.D. N.D. 19.185 15.759 39.154 SW10994.5 13 112.9 N.D. N.D. 19.939 15.944 41.071 SW10999.5 13 91.4 N.D. N.D. 17.885 15.597 37.839 SW11000 13 97.2 N.D. N.D. 19.100 15.607 38.117 SW11022.5A 13 111.6 N.D. N.D. 17.917 15.631 38.728 SW11022.5B 13 81.8 N.D. N.D. 17.671 15.534 36.958 SW11048.5A 13 114.8 N.D. N.D. 16.831 15.434 36.560 SW11048.5B 13 108.8 N.D. N.D. 16.751 15.429 36.515 SW11054.4 13 131.7 N.D. N.D. 17.912 15.567 38.885 SW11060.1 13 122.1 N.D. N.D. 17.895 15.581 37.376 SW11063.4 13 144.2 N.D. N.D. 32.362 17.901 58.189 Notes: Total procedural blanks ranged from 70 to 200 pg Pb (206Pb/204Pb = 18.5 ± 0.5, 207Pb/204Pb = 15.1 ± 0.4, 208Pb/204Pb = 37.1 ± 0.5) and <25 pg U, which resulted in negligible blank corrections. Mass fractionation correction for Pb was +0.10% ± 0.03% per amu as determined by 14 analyses of NBS-981 on Faraday collectors. Mass fractionation corrections for U were not applied for Faraday analyses, based on eight analyses of the U500 standard, which produced the accepted ratio ( 235U/238U = 0.9997). In run, 2σ uncertainties were <0.1%, but a minimum 0.2% error was assigned in isochron regression calculations to be conservative. Precise localities of sample collection sites are given in Tables DR-1 and DR-2 (see text footnote 1). Note that separate samples at localities 3, 8, and 13 were used for single zircon analyses (Table 1). Data for sample SW11068 (Shell Whyte core, Michigan basin), which was used for three-step dissolution experiment, are reported in Table 3. N.D. = not determined. *Samples separated on the basis of magnetic susceptibility are indicated: M = magnetic fraction. NM = non-magnetic fraction. The number indicates the tilt angle in degrees.

TABLE 3. U AND Pb DATA FOR QUARTZ SAMPLE SW11068 (LOCALITY 13), ST. PETER SANDSTONE, MICHIGAN BASIN

Sample Mass % of U U % of total Pb Pb % of total 206Pb/204Pb 207Pb/204Pb 208Pb/204Pb (mg) total mass (ppm) (ng) U (ppm) (ng) Pb One-step, total dissolution A 415.0 N.D. 0.260 108 N.D. 0.511 212 N.D. 24.18 ± 0.06 16.480 ± 0.09 39.15 ± 0.12 B 20.7 N.D. 0.141 2.92 N.D. 1.05 21.7 N.D. 18.89 ± 0.13 15.741 ± 0.16 40.14 ± 0.20 C 55.9 N.D. 0.200 11.2 N.D. 0.308 17.2 N.D. 19.79 ± 0.21 15.79 ± 0.22 38.64 ± 0.19

Three-step, sequential dissolution D1 34.8 13 0.232 8.08 16 1.95 67.8 74 17.24 ± 0.07 15.464 ± 0.10 36.53 ± 0.13 D2 127.3 48 0.150 19.1 39 0.132 16.8 18 33.01 ± 0.69 17.317 ± 0.30 41.35 ± 0.26 D3 101.3 39 0.216 21.9 45 0.071 7.19 8 98.7 ± 6.0 26.24 ± 3.3 149.9 ± 2.1 Total 263.4 100 0.186 49.1 100 0.349 91.8 100 26.47 16.63 46.22

E1 36.2 14 0.167 6.05 26 1.86 67.2 76 16.95 ± 0.08 15.420 ± 0.11 36.49 ± 0.13 E2 112.6 43 0.090 10.1 44 0.101 11.4 13 32.2 ± 1.2 17.515 ± 0.61 58.8 ± 1.2 E3 115.0 43 0.060 6.9 30 0.085 9.78 11 60.5 ± 4.3 22.94 ± 2.4 136.5 ± 4.5 Total 263.8 100 0.087 23.1 100 0.335 88.4 100 23.78 16.54 50.51 Note: In-run uncertainties are 2σ (in percent). See Table 2 for analytical details. Total values for three-step, sequential dissolution reflect mass-weighted sums of the individual steps. N.D. = not determined.

1728 Geological Society of America Bulletin, November 1999 LOWER PALEOZOIC CRATONIC QUARTZ ARENITES, NORTH AMERICAN MIDCONTINENT

REE enrichment (CeN/SmN=2.5 to 4.0; N refers to chondrite normalization) and negative Eu anom- 0.5 alies (Eu/Eu*= EuN /[SmN*GdN] = 0.5–0.7) of the quartz grains are consistent with derivation from granitic or rhyolitic rocks, rather than from more mafic (plutonic), quartz-bearing lithologies. The REE patterns are not consistent with zircon being the sole mineral inclusion in the quartz grains, nor with a simple mixture of, e.g., apatite and zircon. Permissible combinations include zir- con ± apatite and major contributions from titanite (~50% of the inclusion population) and/or minor contributions from monazite (~5% of the inclu- sions) (Fig. 5B), the latter of which is indicated by the 208Pb/204Pb–206Pb/204Pb variations (see pre- ceding; Fig. 4). The absolute REE contents of the quartz grains are well explained by ~1 part in 1000 concentration of accessory (heavy) minerals as microinclusions (Fig. 5B), which is consistent with the U and Pb abundances of the quartz grains relative to those expected for the accessory miner- als. Although garnet inclusions may in part ex- Figure 4. 208Pb/204Pb–206Pb/204Pb variations for quartz grains from the Galesville and St. Peter plain the relatively high Th/U ratios required by Sandstones, which constrain the Th/U ratios of the accessory (heavy) mineral inclusions in detri- the 208Pb/204Pb–206Pb/204Pb variations (e.g., tal quartz. Relative Th/U ratios for common accessory minerals are also shown, as calculated Mezger et al., 1989), garnet cannot be a signifi- from U-Pb studies that analyzed multiple coexisting minerals (Williams et al., 1983; Davis and cant inclusion based on the REE patterns, because Edwards, 1986; Corfu, 1988). The 208Pb/204Pb–206Pb/204Pb variations indicate that zircon cannot garnet has extreme enrichments of heavy REEs. be the primary accessory (heavy) mineral inclusion in the quartz grains, and that variable The REE patterns also confirm the similarity of amounts of monazite (mon) are required (percentages of accessory mineral population are the two Wisconsin sandstones (CeN/YbN = shown). Error ellipses are much smaller than the plotted symbols. 5.2–6.7) in contrast to the St. Peter Sandstone

from the Michigan basin (CeN/YbN = 3.5), which further substantiates different sources of quartz component in the inclusion mineral assemblage amounts that are required to explain the Pb isotope for these two geographic regions. (Fig. 4). It is striking that 208Pb/204Pb–206Pb/204Pb data may be missed by these observations. The Leaching Experiments. Sequential HF- variations for quartz from the Cambrian and Or- most likely terranes to contribute monazite are ca. leaching experiments on a Michigan basin sam- dovician Wisconsin sandstones indicate a rela- 1.4 Ga anorogenic granites of the Wolf River ple were undertaken to isolate the radiogenic Pb tively consistent proportion of monazite (2%–5% batholith (e.g.,Anderson and Cullers, 1978), or the component in the relatively nonradiogenic bulk of the accessory mineral population; Fig. 4), evolved, high-alkali granites of the Grenville quartz separates (Tables 2 and 3). The leaching whereas the 208Pb/204Pb–206Pb/204Pb variations for Province (e.g., McLelland et al., 1993). experiments indicate that the Pb isotope compo- the Michigan basin samples indicate a higher pro- REE Systematics. Chondrite-normalized REE sitions of the bulk quartz separates most likely re- portion of monazite (~10% of accessory minerals; patterns of the quartz grains broadly reflect those flect a mixture between high U/Pb (high-µ) ac- Fig. 4), suggestive of derivation from distinct of potential Archean and Proterozoic source ter- cessory mineral inclusions and low-µ inclusions source terranes. Although monazite was not ob- ranes, although REE contents of the quartz are or overgrowths; in the Michigan basin samples, served in our electron microprobe analysis, or by 10–100 times lower than those of bulk igneous the nonradiogenic component is likely to be K- Tyler’s (1936) petrographic work, the small rocks from these terranes (Fig. 5A). The light feldspar, consistent with petrographic observa-

TABLE 4. RARE EARTH ELEMENT AND Sm-Nd ISOTOPE DATA FOR QUARTZ GRAINS 147 144 143 144 ε ε § # wt Ce Nd Sm Eu Gd Dy Er Yb Sm/ Nd Nd/ Nd Nd (0)* Nd (t) TDM (mg) WI-SP-3 (locality 2) 575.4 2.46 0.900 0.145 0.023 0.111 0.120 0.085 0.094 0.09699 0.511379 ± 9 –24.6 –18.6 2.24 GALE-D2 (locality 10) 496.2 4.17 2.13 0.382 0.061 0.346 0.319 0.207 0.205 0.10823 0.511433 ± 8 –23.5 –17.7 2.40 SW11068 (locality 13) 516.1 1.93 0.706 0.115 0.026 N.D. 0.153 0.130 0.140 0.09799 0.511689 ± 6 –18.5 –12.7 1.87 Notes: WI-SP-3 and SW11068 are from the St. Peter Sandstone in Wisconsin and the Michigan basin, respectively. GALE-D2 is from the Galesville Sandstone in Wisconsin. Sample locations are given in Tables DR-1 and DR-2 (see text footnote 1). REE concentrations in ppm. Uncertainties of isotope ratios are 2σ errors. N.D. = not determined. ε 143 144 *Present-day Nd (0) were calculated using 0.512635 for the Nd/ Nd ratio of present-day CHUR, which reflects the averages of eight analyses of the BCR-1 standard during the study (±0.000020 2SD); the Nd isotope composition of the BCR-1 standard has been shown to be equal to that of CHUR (Wasserburg et al., 1981). §ε ε Nd values at the time of deposition [ Nd(t)] were calculated using 515 Ma for the Galesville Sandstone and 465 Ma for the St. Peter Sandstone depositional ages. # 143 144 Depleted mantle model ages (TDM) were calculated by assuming that the mantle evolved linearly to its present-day composition of ( Nd/ Nd)DM = 0.51315 and 147 144 ( Sm/ Nd)DM = 0.217.

Geological Society of America Bulletin, November 1999 1729 JOHNSON AND WINTER tions. Two large (~260 mg) aliquots (D and E, Table 3) of quartz grains from a single Michigan basin quartz sample (SW11068) were subjected to a sequential three-step leaching in warm HF; ~14% of the total mass of sample material was dissolved during the first stage of dissolution and ~43% of the total mass was dissolved in both the second and third stages (Table 3). Total (one step) dissolution of three aliquots of SW11068 (A, B, and C in Table 3) yielded U and Pb contents rang- ing from 0.14 to 0.26 ppm and 1.05 to 0.31 ppm, respectively (Table 3); reintegration of the U and Pb contents for each leach stage of aliquots D and E yields nearly identical total concentrations (Table 3). Of the total Pb in quartz samples D and E, ~75% is contained in the first leachates (~1.9 ppm; Table 3), even though this stage dissolved only 14% of the sample by mass. The second and third stage leachates each contain ~13% of the to- tal Pb (~0.1 ppm, Table 3). In summary, the most soluble fractions of the samples have the least ra- diogenic Pb isotope ratios and highest Pb con- tents, whereas successive leachates have progres- sively more radiogenic Pb isotope ratios and lower Pb contents. The leaching experiments suggest that potas- sium feldspar, which was identified as mi- croovergrowths from energy-dispersive electron microprobe analysis (see preceding), is the likely source of nonradiogenic Pb in the most soluble fraction. Lead in the second- and third-stage leachates was progressively derived from phases that have relatively high-µ values and low Pb contents, such as zircon or other accessory (heavy) minerals that are comparatively resistant to HF dissolution. A high-µ component that has ~0.1 ppm Pb is consistent with the proportion of accessory mineral inclusions in quartz that is sug- gested by the REE data (~1 part in 1000), as well Figure 5. Chondrite-normalized rare earth element (REE) patterns for quartz grains of the as measured Pb contents of single detrital zir- Galesville and St. Peter Sandstones. (A) Comparison of quartz grains with possible source ter- cons. Regression of the entire data set yields a ranes that may be represented by the Archean southern Superior Province (Shirey and Hanson, well-correlated 206Pb/204Pb–ppm Pb trend of [Pb] 1986), the Proterozoic Pokegama Quartzite (Hemming et al., 1994), and Proterozoic orogenic = [20*log(206Pb/204Pb)–26.5]–1; this correlation rocks from the Penokean orogen and post-Penokean and anorogenic Middle Proterozoic rocks will be used in constraining mixing models for of the northern midcontinent (Anderson and Cullers, 1978; Anderson et al., 1980; Van Wyck the Pb-Pb isochrons for the quartz grains, dis- and Johnson, 1997). Note the order of magnitude scale change from the left (data for quartz cussed in the following. grains and Pokegama Quartzite) to the right (Archean and Proterozoic rocks). (B) Comparison of quartz grains with possible accessory (heavy) mineral assemblages. Total inclusion density is Pb-Pb Isotope Systematics of Quartz Grains assumed to be 0.1%. The lower boundary of the shaded fields represents the REE pattern for quartz, if 100% of the inclusions are zircon. The upper boundary of the diagonally ruled field Wisconsin Sandstones. Analyses of quartz represents an inclusion assemblage of 50% zircon and 50% apatite. The upper boundary of the (n = 11) from the Ordovician St. Peter Sandstone stippled field represents an inclusion assemblage of 50% zircon and 50% titanite. The upper- that is exposed in Wisconsin define a linear array most boundary of the fields shown represents an inclusion assemblage of 95% zircon and 5% that yields a Pb-Pb age of 2539 ± 140 Ma monazite. Inclusions of zircon ± apatite with 50% titanite and/or 5% monazite can explain the (MSWD = 62), and 12 analyses of quartz from light REE/heavy REE enrichment that is observed in the quartz grains. Note that the important the Galesville Sandstone in Wisconsin define an property of the REEs for quartz is the relative REE patterns, rather than the absolute values, identical Pb-Pb age of 2512 ± 140 Ma (MSWD = which will vary depending upon inclusion density. Note the two order of magnitude scale change 324). The similarity in ages supports the interpre- between the left (data for quartz grains) and right (model accessory/heavy mineral assemblage) tation that the Galesville was the source for the axes. Relative REE partitioning data were taken from granites that have REE data for coexist- St. Peter Sandstone, or that they were derived ing minerals (Gromet and Silver, 1983; Fujimaki, 1986; Sawka, 1988; Sawka and Chappell, from similar sources. Regression of the data from 1988; Paterson et al., 1992; Kingsbury et al., 1993; Wark and Miller, 1993).

1730 Geological Society of America Bulletin, November 1999 LOWER PALEOZOIC CRATONIC QUARTZ ARENITES, NORTH AMERICAN MIDCONTINENT

both formations together (n = 33) yields a Pb-Pb age of 2519 ± 85 Ma (MSWD = 486), which in- tersects the average crustal Pb curve (Stacey and Kramers, 1975) at 4 and 2521 Ma (Fig. 6). The greater scatter about the regression line by the four quartz samples that have the highest 206Pb/204Pb ratios (Figure 6) does not add signif- icant uncertainty to the Pb-Pb age determination. Michigan Basin. Analyses of quartz (n = 27) from the St. Peter Sandstone in the Michigan basin define a linear Pb-Pb array that yields an age of 2242 ± 93 Ma (MSWD = 446, Fig. 7). Elimination of the leachate samples (n = 21) yields a Pb-Pb age of 2396 ± 78 Ma (MSWD = 18). Unlike the data for the Wisconsin samples, the regressed Pb-Pb data for Michigan do not in- tersect the zero point on the Stacey-Kramers growth curve. Assuming that the first leach com- positions closely approximate the initial isotope compositions for the isochron, these composi- tions are significantly more radiogenic than the Stacey-Kramers curve at 2.2 Ga.

Figure 6. 206Pb/204Pb–207Pb/204Pb variations for quartz from the Ordovician St. Peter Sand- Interpreting Quartz Pb-Pb Ages stone and the Cambrian Galesville Sandstone in Wisconsin (see Fig. 2 for localities). Dashed lines show the 95% confidence limit error envelope for the data. The Pb isotope growth curve Hemming et al. (1994) showed that the Pb iso- for average continental crust (S-K; Stacey and Kramers, 1975) is marked at 500 m.y. intervals. tope composition of quartz grains directly sepa- Error ellipses are much smaller than the plotted symbols. rated from Archean plutonic rocks is sufficiently variable and radiogenic to yield reasonably pre- cise (±100 Ma) Pb-Pb ages that are consistent with crystallization ages determined by more con- ventional means. Hemming et al. (1994) analyzed quartz grains from three different plutons repre- senting a variety of chemical compositions and different origins (i.e., mantle- vs. crustal-derived), but all from the same major crust-formation inter- val within the 2.7 Ga Superior Province. Using sample sizes generally between 5 and 40 mg, Hemming et al. (1994) demonstrated that quartz grains from different plutons have different ranges of 206Pb/204Pb ratios, and together they define a Pb-Pb age of 2632 ± 64 Ma, which is in excellent agreement with the known crystallization ages of these rocks. Well-constrained Pb-Pb isochrons may also be determined on detrital quartz grains that were de- rived from source terranes of restricted age range. Hemming et al. (1994) showed that clear (igneous) detrital quartz in the Early Proterozoic Pokegama Quartzite (Minnesota) yields a Pb-Pb age of 2647 ± 16, which is identical to that of known igneous source terranes. In addition, separation of milky Figure 7. 206Pb/204Pb–207Pb/204Pb variations for quartz from the Ordovician St. Peter Sandstone (vein) quartz produced slightly younger at ~3600 m depth in the Michigan basin (see Fig. 2 for locality). Dashed lines show the 95% confi- 207Pb*/206Pb* ages of ca. 2.4 Ga (Hemming et al., dence limit error envelope for the data. Error ellipses for samples processed by total dissolution 1994), suggesting a different source. However, Pb- are much smaller than the plotted symbols. Data obtained from three-step sequential HF leaching Pb isochrons are expected to be more complicated are plotted with true error ellipses (sample SW11068). The Pb isotope growth curve for average for sandstones having depositional ages that are continental crust shown in the inset (S-K; Stacey and Kramers, 1975) is marked at 250 m.y. inter- much younger than those of their source terranes. vals. Third stage HF leaches are interpreted to largely record the isotopic compositions of included Unlike the Pokegama Quartzite, zircon data for the accessory minerals (ACC MINS). Wisconsin and Michigan cratonic sandstones (see

Geological Society of America Bulletin, November 1999 1731 JOHNSON AND WINTER preceding) indicate that detrital constituents were TABLE 5A. MIXING MODELS FOR QUARTZ Pb-Pb ISOCHRONS FOR WISCONSIN SAMPLES derived from multiple igneous terranes, spanning Measured Stage 1 Stage 2 data First cycle seds Second cycle seds an age range of 1600 m.y. In the following we (1000 points) (100 points) (First 25 points) model the effects of mixing on Pb-Pb regressions, (Fig. 8A) (Fig. 8B) which are generally applicable to multicyclic sed- Age (Ma): 2519 2527 2508 2522 Error: 85 16 20 41 imentary rocks that involve source terranes of Y Int:: 12.55 12.42 12.49 12.45 widely different ages. Error: 0.19 0.42 0.06 0.09

Fraction of 1.1 Ga component: 0 to 0.5 0 to 0.5 0 to 0.5 Two-Component Mixing Models for Quartz Pb-Pb Isotope Data Avg 206Pb/204Pb: 29.4 43.1 28.8 1 SD: 9.6 17.4 5.5 Single detrital zircons provide the first-order Notes: Mixing calculated assuming random sampling along 2.7 Ga and 1.1 Ga isochrons that are concordant with the Stacey-Kramers average crust curve (Stacey and Kramers, 1975). Maximum 206Pb/204Pb for the constraints for the range of ages of source ter- endmember isochrons set at 80. Intermediate components such as 1.8–1.9 Ga Penokean crust are not used ranes to multicyclic sediments. However, it is im- because single zircon ages indicate that these are minor components. For the fraction of 1.1 Ga component random values within the noted interval were calculated. Scaling of Pb contents follows regression of measured practical to use single zircons to rigorously deter- data: ppm Pb = 1/[20*log(206Pb/204Pb) – 26.5); r2 = 0.8. K-feldspar overgrowths assumed to contain 10 ppm Pb mine the relative proportions of source terranes, (e.g., Patterson and Tatsumoto, 1964), 206Pb/204Pb = 16.751, 207Pb/204Pb = 15.429 (equal to least radiogenic sam- which would require hundreds of analyses per ple, SW11048.5B). sample to be statistically significant. Instead, we use the range of source ages from the zircon data as a primary constraint, and calculate the relative TABLE 5B. MIXING MODELS FOR QUARTZ Pb-Pb ISOCHRONS FOR MICHIGAN SAMPLES Measured Stage 1 Stage 2 Stage 3 proportions of source terranes using the Pb-Pb data First cycle seds Second cycle seds Addition of isotope compositions of large bulk samples. (1000 points) (100 points) (First 25 points) feldspar overgrowths The 207Pb/204Pb–206Pb/204Pb regression ages (Fig. 8C) (Fig. 8D) (100 points) (First 25 points) calculated for quartz separates from Wisconsin (Fig. 8E) Age (Ma): 2242 2193 2229 2265 2213 2188 and the Michigan basin fall between those of the Error: 93 51 52 96 60 118 likely end-member source ages of 2.7 and 1.1 Ga Y Int:: 13.10 13.05 12.99 12.91 13.07 13.11 (as inferred by the detrital zircon ages). The large Error: 0.32 0.99 0.11 0.16 0.07 0.07 sample sizes we have used should be sufficient to Fraction of 1.1 Ga component: 0 to 1 0 to 1 0 to 1 represent the relative proportions of the different sources and the effects of multiple mixing events. Percent feldspar overgrowths: 0 to 5 0 to 5 We have modeled the regression ages that are Avg 206Pb/204Pb: 25.0 42.9 29.5 24.4 produced through two-component mixing in- 1 SD: 17.0 17.0 5.5 3.3 volving 2.7 Ga and 1.1 Ga sources; the relatively Note: See footnote for Table 5A for details of models. minor amount of 1.8–1.4 Ga zircons in the rocks indicates that these components can be ignored in the mixing calculations. The source isochrons are average mixture of 25% 1.1 Ga and 75% 2.7 Ga impossibly tightly and do not reflect the spread in assumed to have initial Pb isotope compositions sources (obtained through random variation be- the measured data. on the Stacey-Kramers average crust curve tween 0% and 50% 1.1 Ga component) produces The supermature quartz arenites studied here (Stacey and Kramers, 1975), and range to a max- the 2500–2600 Ma regression age for the Wis- represent multicyclic sedimentation, as argued by imum 206Pb/204Pb ratio of 80. Stage 1 of mixing consin quartz arenites (Table 5A), although the many other workers. We can estimate the effect (Table 5) involves 1000 random samplings from average 206Pb/204Pb ratio and spread along the of multicyclic sedimentation using a second along each of the source isochrons, and calcula- mixture isochron exceeds that measured in the stage (stage 2 of Table 5, A and B) of mixing, ob- tion of the resulting Pb isotope compositions of samples (Table 5A; Fig. 8A). A greater propor- tained by taking 100 10-point averages of the the 1000 mixtures. In addition, the relative frac- tion of a 1.1 Ga component is required to produce mixtures calculated in stage 1, weighted for their tion of 2.7 Ga and 1.1 Ga components is allowed the younger age of the Michigan basin sandstone, Pb contents. These results more closely match to randomly vary between set limits. We have al- and a ca. 2200 Ma regression age may be ob- the measured data, and illustrate the collapse of lowed the concentration ratio of the two compo- tained using an average mixture of 50% 1.1 Ga the mixtures to a greater number of nonradio- nents, (Pb)2.7Ga/(Pb)1.1Ga, to vary from 1 to 2, ac- and 50% 2.7 Ga sources (obtained through ran- genic isotope compositions (Fig. 8B for Wiscon- counting for the generally greater radiogenic Pb dom variations between 0% and 100% 1.1 Ga sin and Fig. 8D for the Michigan basin). Relative contents of the older component. In addition, Pb component), as shown in Table 5B and Figure to stage 1 mixing, second-stage mixing magnifies contents of the end members are varied as a func- 8C. The interplay between scaling of Pb contents the effect of scaling the Pb concentration as a tion of the 206Pb/204Pb ratio of the randomly along the source isochrons and the relative pro- function of the 206Pb/204Pb ratio of the source end generated end member, matching the observed portions derived from the respective end-member members, and illustrates how the large number of 206Pb/204Pb–ppm Pb variations measured in the sources is shown by comparison of Figure 8, A measured nonradiogenic compositions for the quartz grains (see above); this approach realisti- and C, which differ only in the relative proportion quartz grain samples can be generated (cf. Fig. 6 cally reflects the higher Pb contents that are mea- of quartz derived from 1.1 Ga and 2.7 Ga source and Fig. 8, B and D). Regression of the first 25 sured (and expected) in the relatively nonradi- terranes in the mixtures. If the mixture isochrons compositions calculated for stage 2 mixing, to ogenic quartz samples. are calculated using fixed relative proportions of more closely match the number of measured In the stage 1 mixing calculations, which the sources, rather than random variation within samples in Figures 6 and 7, results in excellent might approximate first-cycle sedimentation, an the limits noted here, the calculated data cluster agreement with the age and average 206Pb/204Pb

1732 Geological Society of America Bulletin, November 1999 LOWER PALEOZOIC CRATONIC QUARTZ ARENITES, NORTH AMERICAN MIDCONTINENT

Figure 8. Three-dimensional histograms for Pb-Pb isochron mixing (see summary in Table 5) between 2.7 Ga and 1.1 Ga source terranes (source isochrons are shown where visible). The Z scale (frequency) varies for each plot and is not shown for clarity. (A) 1000 point stage 1, first-cycle mixing for Wisconsin samples (Table 5A). Regression of 1000 calculated values yields an age of 2527 ± 16 Ma. Maximum fre- quency value is 40. (B) Second-stage mixing of A (see Table 5A) for Wisconsin samples, which represents 100, 10-point, concentration- weighted averages of the first-stage mixing. Regression of the 100 cal- culated values yields an age of 2508 ± 20 Ma. The maximum fre- quency value is 20. (C) 1000 point stage 1, first-cycle mixing for Michigan samples (Table 5B). Regression of the 1000 calculated values yields an age of 2193 ± 51 Ma. The maximum frequency value is 42. (D) Second-stage mixing of C (see Table 5B) for Michigan samples, which represents 100, 10-point, concentration-weighted averages of the first- stage mixing. Regression of the 100 calculated values yields an age of 2229 ± 52 Ma. The maximum frequency value is 18. (E) Third-stage mix- ing for Michigan samples, where 0%–5% K-feldspar overgrowths are added to the compositions of second-stage mixing shown in D (see Table 5B). Regression of the 100 calculated values yields an age of 2213 ± 60 Ma. Maximum frequency value is 30. ratio measured for the Wisconsin samples, as such a component is supported by the leaching for the Michigan basin samples are closely ap- well as in a reasonable match for the errors in experiments noted here (first-stage leach), and we proximated by this model (Table 5B). An impor- slope and intercept of the mixture isochron have modeled this effect by adding a K-feldspar tant aspect of this model is that the initial Pb iso- (Table 5A). component to the results of the stage 2 mixing by tope compositions of the mixture isochron no The petrographic observation of K-feldspar randomly varying the percentage of K-feldspar longer intersects the zero point on the Stacey- overgrowths in the Michigan basin samples leads component between 0% and 5% by mass (Table Kramers curve (despite the fact that the 2.7 Ga us to a third-stage model to explain the signifi- 5B). These calculations show how further col- and 1.1 Ga components are concordant with the cantly less radiogenic Pb isotope ratios of these lapse of the mixtures to nonradiogenic Pb isotope Stacey-Kramers curve), because the K-feldspar samples; this stage involves addition of a nonradi- compositions can occur (Fig. 8E). The age, aver- component is off the Stacey-Kramers curve, as in- ogenic diagenetic component. The presence of age 206Pb/204Pb ratio, and errors in the regression dicated by the first-stage leaches (Fig. 7).

Geological Society of America Bulletin, November 1999 1733 JOHNSON AND WINTER

ε Figure 9. Nd-age evolution trends for quartz grains (shown in heavy lines) from the Cambrian Galesville Sandstone in Wisconsin (G-WI), St. Peter Sandstone in Wisconsin (SP-WI), and the St. Peter Sandstone in the Michigan basin (SP-MI). Nd isotope evolution trends for possi- ble source terranes are shown in various patterns. The Nd isotope compositions for quartz grains can be obtained by mixing average 1.1 Ga ε Grenville crust with average 2.7 Ga crust from the southern Superior Province at the time of deposition. The comparatively high Nd value of the St. Peter Sandstone from the Michigan basin can be explained by 65:35 mixing of 1.1 Ga and 2.7 Ga sources, whereas the significantly ε lower Nd values of quartz from the Galesville and St. Peter Sandstones in Wisconsin suggest a 30:70 mixing of 1.1 Ga and 2.7 Ga sources. The Nd isotope composition of the 2.7 Ga source is likely to be similar to that of the Pokegama Quartzite from Minnesota (Hemming et al., 1994). Additional Nd isotope data from Ashwal and Wooden (1983), Ashwal et al. (1986), Shirey and Hanson (1986), Barovich et al. (1989), Marcan- tonio et al. (1990), Nicholson and Shirey (1990), Daly and McLelland (1991), Nelson (1991), Dickin and Higgins (1992), Emslie and Hegner (1993), McLelland et al. (1993), Owens et al. (1994), and Van Wyck and Johnson (1997).

Sm-Nd Isotope Systematics of Quartz Grains 70:30 mixing between 2.7 and 1.1 Ga sources, re- mately derived from either the proximal Ke- markably consistent with the mixing proportions weenawan volcanic rocks that are associated with The general mixing proportions that are indi- suggested by the Pb-Pb isochron for the quartz the Midcontinent rift system or from synorogenic cated by the detrital zircon U-Pb and quartz Pb-Pb grains. A dominant contribution from 2.7 Ga intrusions of the Grenville Province. Derivation data are supported by the Sm-Nd isotope data ob- sources is also indicated by the detrital zircon of significant volumes of quartz from Ke- tained on quartz grains from Wisconsin and the population. weenawan rhyolites, which contain 5%–15% ε Michigan basin. Although the Nd-time evolution The Sm-Nd isotope data for the Michigan phenocrysts of quartz and accessory zircon, trends for the quartz data might suggest a domi- basin sample indicate a greater proportion of a would require significant concentration in pre- nantly ca. 1.8 Ga source, such as the Penokean younger component, and the comparatively high early Paleozoic basins, inasmuch as rhyolites rep- ε orogenic rocks in Wisconsin, the near lack of ca. Nd value can be explained by ~35:65 mixing be- resent only ~10% of the exposed Keweenawan 1.8 Ga detrital zircons instead indicates that the tween 2.7 and 1.1 Ga sources. This proportion is volcanic rocks (Nicholson, 1992; Green and Sm-Nd isotope data are best interpreted as reflect- also consistent with the mixing proportions esti- Fritz, 1993); most of the estimated 1.5 x 106 km3 ing a mixture of old (ca. 2.7 Ga) and young (ca. mated by the Pb-Pb isochron for the quartz grains. of volcanic rocks associated with the Midconti- 1.1 Ga) sources (Fig. 9); variable contributions nent rift system are basaltic. In contrast, the from intermediate-age components are likely PROVENANCE OF LOWER PALEOZOIC Grenville Province of North America is an ~500- variants from a simple two-component mixing CRATONIC SANDSTONES km-wide orogenic belt, extending from the coast model. The Sm-Nd isotope data for quartz from of Labrador to northern Mexico, that consists of the Galesville Sandstone and St. Peter Sandstone All of the isotope data indicate that the quartz voluminous granitic plutons ranging in age from in Wisconsin are similar (Fig. 9), supporting the and heavy minerals were ultimately derived from 1.0 to 1.2 Ga (e.g., Rainbird et al., 1997), and this hypothesis that they were derived from similar Precambrian basement terranes. The granite- seems to be the most likely ultimate source for sources. Assuming relatively equal Nd contents greenstone terrane of the Superior Province is the 1.1 Ga constituents in the lower Paleozoic quartz ε between the two source regions, the Nd values of ultimate source of the major 2.7 Ga detrital com- arenites in Wisconsin and Michigan. the two Wisconsin samples can be produced by ponents, whereas the 1.1 Ga components are ulti- The comparatively minor amount of Middle

1734 Geological Society of America Bulletin, November 1999 LOWER PALEOZOIC CRATONIC QUARTZ ARENITES, NORTH AMERICAN MIDCONTINENT

are likely to be the immediate sources for the lower Paleozoic quartz arenites of this study. The postvolcanic Keweenawan Bayfield Group in northern Wisconsin is the most probable direct source of Keweenawan or Grenville age (1.1 Ga) detrital quartz and zircons to the Cambrian Galesville and Ordovician St. Peter Sandstones in Wisconsin. The Bayfield Group (Table 6) in- cludes texturally and compositionally superma- ture quartz arenites, which have been interpreted to have been derived from uplifted Archean granitic basement (i.e., the Superior Province) that was exposed after erosion of the Keweenawan volcanic sequence (Morey and Ojakangas, 1982; Ojakangas and Morey, 1982a). In contrast, quart- zose units (Oronto Group; Table 6) of the lower part of the postvolcanic Keweenawan sequence are relatively immature in terms of texture and composition, and reflect erosion of Midcontinent rift–related volcanic rocks (Ojakangas and Morey, 1982a); these rocks are unlikely contribu- tors to the Galesville and St. Peter Sandstones. The 1.1 Ga component in the St. Peter Sandstone of the Michigan basin could be directly derived from Grenville basement (Fig. 10), or from post- Keweenawan quartzose sediments, such as the Ja- cobsville Sandstone (Table 6). The immediate source for the 2.7 Ga compo- nent in the Galesville and St. Peter Sandstones remains unclear. Prevolcanic Keweenawan quartzose sandstones in Minnesota, Wisconsin, and Michigan (Table 6) could have contributed 2.7 Ga detrital components, on the basis of grain compositions and paleocurrent indicators (Ojakangas and Morey, 1982b). However, the widespread Middle Proterozoic quartzites that are exposed throughout Wisconsin and south- west Minnesota, such as those belonging to the Figure 10. Interpretive map of the northern midcontinent region of North America showing Baraboo interval of Dott (1983) (Table 6), can- sediment transport pathways for quartz grains and zircons analyzed in the Galesville Sandstone not have contributed substantial detrital compo- (Cambrian) and the St. Peter Sandstone (Ordovician). Sample localities in Wisconsin and nents to the Galesville and St. Peter Sandstones Michigan are marked with large stars. Major transport pathways are shown in heavy arrows. because the quartzites contain major 1.8 Ga Sources and ages (in Ga) for quartz and zircons are shown in zircon shae symbols. Map patterns Penokean components that are virtually absent for various terranes are from Figure 1. Shown below the map are proportions of 1.1, 1.4, 1.8, and from the Cambrian and Ordovician sandstones 2.7 Ga components for the Wisconsin and Michigan basin localities based on the age distribu- (Fig. 11). This conclusion is similar to that tion of single zircons, as well as the proportions of 1.1 and 2.7 Ga components that are inferred drawn by Tyler (1936), based on the contrast in from mixing models estimated from Pb-Pb and Sm-Nd isotope data for quartz grains (see text heavy mineral assemblages of the quartzites and for discussion). the lower Paleozoic sandstones. Early Protero- zoic quartzites (Table 6) that were deposited prior to the Penokean orogeny probably contain Proterozoic (ca. 1.4 Ga) zircons (and possibly the ca. 1.8 Ga Penokean orogenic belt and the 2.7 2.7 Ga components (Grout et al., 1951; Morey quartz grains) in cratonic sandstones from Wis- to >3.2 Ga Marshfield terrane; Fig. 1) provides 1973; Sims et al., 1981), and it is possible that consin were probably ultimately derived from the the best evidence that the detrital components these were reworked and eventually incorpo- Wolf River batholith; generally south-directed pa- have been transported a considerable distance rated in the lower Paleozoic sandstones. leocurrent indicators throughout the Paleozoic se- from their ultimate basement source terrane. Rainbird et al. (1997) envisioned that a very quence (Hamblin, 1961; Potter and Pryor, 1961) large river system emanated from the Grenville are less consistent with derivation from the Mid- Sedimentary Source Units orogenic highlands and delivered sediment to the dle Proterozoic eastern granite-rhyolite province northwest margin of the continent in Late Prot- to the south (Fig. 1). The striking paucity of zir- Proterozoic quartz-rich sandstones, which are erozoic time. Proximal and medial portions of cons derived from the major local basement (i.e., common in the Lake Superior region (Table 6), such a river system could have deposited detrital

Geological Society of America Bulletin, November 1999 1735 JOHNSON AND WINTER material from the Grenville Province in Late TABLE 6. PRECAMBRIAN QUARTZ-RICH SEDIMENTARY UNITS OF THE LAKE SUPERIOR REGION Proterozoic basins (rift or foreland) anywhere in Northwest Minnesota Wisconsin Michigan Southeast region region the midcontinent region. It is possible, therefore, Postvolcanic Bayfield Group that late Keweenawan quartz-rich sandstones Keweenawan Hinkley (150 m) Chequamegon Jacobsville were the direct sources for both 2.7 Ga and 1.1 Sandstone (1) Fond du Lac (150 m) (900 m) (600 m) Devils Island Ga detrital components to the Galesville and St. (90 m) Peter Sandstones (Fig. 10). Orienta (570 m)

Oronto Group CONCLUSIONS Solor Church Freda (600–1000 m) (1500–4000 m) Detrital zircon U-Pb ages and Sm-Nd and Pb- Prevolcanic Puckwunge Bessemer Bessemer Pb isotope data on quartz separates from quartz Keweenawan (~100 m) (~100 m) (~100 m) arenites of the Paleozoic northern midcontinent Sandstone (2) Nopeming region demonstrate the importance of Archean (~100 m) and Late Proterozoic terranes as the ultimate Mid-Proterozoic Sioux Baraboo sources for the detrital constituents. The data pre- quartzites (3) Barron Flambeau sented here demonstrate the small amount of de- McCaslin trital constituents in the Wisconsin Paleozoic Waterloo quartz arenites that was derived from the local Early-Proterozoic Baraga quartzites (4) Group basement, including the lack of >2.7 Ga material, Goodrich Penokean age rocks (ca. 1.8 Ga), or Middle Prot- Animikie Menominee erozoic quartzites (Baraboo-interval), which pro- Group Group vides strong support for a multicyclic origin for Pokegama Ajibik the Galesville and St. Peter Sandstones. This is in Palms contrast to the detrital components in the Middle Mille Lacs Choclay Proterozoic quartzites, which are dominated by Group Group local basement components (Fig. 11). Sunday Mesnard Sturgeon The dominant sedimentary pathways for the Note: (1) Daniels (1982); Kalliokoski (1982); Morey and Ojakangas (1982); Ojakangas and Morey (1982a); (2) detrital constituents of the St. Peter Sandstone in Ojakangas and Morey (1982b); (3) Dott (1983); Sims et al. (1993); (4) Grout et al. (1951); Hemming et al. (1994); Wisconsin are as follows. Morey (1973); Sims et al. (1981, 1993). 1. There was deposition of 2.7 Ga quartz de- rived from the southern Superior Province into late Keweenawan (ca. 1.1 Ga) basins, which also received 1.1 Ga material, either from local Ke- weenawan silicic volcanic rocks, or, more likely, from Grenville rocks to the east (Fig. 10). 2. Keweenawan quartz-rich sandstones were Figure 11. Histograms comparing the eroded and reworked during Cambrian time; this distribution of ages of single detrital zircon material was transported southward and de- populations from the Ordovician St. Peter posited as the Galesville Sandstone in Wisconsin Sandstone (Wisconsin and Michigan (Fig. 10). basin), Cambrian Galesville Sandstone 3. The Ordovician St. Peter Sandstone in Wis- (Wisconsin), and Baraboo Interval Middle consin may have been largely derived through re- Proterozoic quartzites (Wisconsin; data working of the Cambrian Galesville Sandstone, from Medaris et al., 1996; Van Wyck, 1995). which has a closely similar detrital zircon popula- MCR—Midcontinent rift; GP—Grenville tion, as well as Pb-Pb and Sm-Nd systematics, and Province; PEN—Penokean; SP—Granite- heavy mineral suites. In contrast, the St. Peter greenstone terrane (GGT) of the Superior Sandstone in the Michigan basin received a much Province; WR—Wolf River batholith; greater proportion of 1.1 Ga detrital zircons and GT—Gneiss terrane of the southern Supe- quartz from the nearby and extensive Grenville rior Province. The Middle Proterozoic Province (Fig. 10), as indicated by the U-Pb zircon Baraboo Interval quartzites have a very and Pb-Pb and Sm-Nd data on quartz separates. large local basement (Penokean) compo- Mixing models for Pb-Pb isochron arrays, as nent, whereas the Cambrian and Ordovi- well as Sm-Nd data, determined on quartz sepa- cian rocks have only minor components that rates, indicates that the relative proportions of were derived from the local basement (in- old and young quartz grains approximately re- cluding the Middle Proterozoic quartzites). flects those of the single detrital zircon popula- tions. This observation suggests that there was little relative fractionation of quartz and zircon during multiple sedimentary cycles, and hence

1736 Geological Society of America Bulletin, November 1999 LOWER PALEOZOIC CRATONIC QUARTZ ARENITES, NORTH AMERICAN MIDCONTINENT

nical Series, v. 6, 225 p. and subduction magmatism to evolution of the lithosphere: that little age bias exists between the framework Daly, J. S., and McLelland, J. M., 1991, Juvenile Middle Prot- Journal of Geophysical Research, v. 96, p. 13593–13608. grains and the heavy mineral fraction. Because erozoic crust in the Adirondack Highlands Grenville Kalliokoski, J., 1982, Jacobsville Sandstone, in Wold, R. J., and the statistical applicability of our single detrital Province, northeastern North America: Geology, v. 19, Hinze, W. J., eds., Geology and tectonics of the Lake Su- p. 119–122. perior basin: Geological Society of America Memoir 156, zircon data set is limited, the Pb-Pb isochrons Daniels, P.A., Jr., 1982, Upper Precambrian sedimentary rocks: p. 147–155. determined on bulk quartz separates better re- Oronto Group, Michigan-Wisconsin, in Wold, R. J., and Kingsbury, J. A., Miller, C. F., Wooden, J. L., and Harrison, flect the average source mixing proportions, Hinze, W. J., eds., Geology and tectonics of the Lake Su- T. M., 1993, Monazite paragenesis and U-Pb systematics perior basin: Geological Society of America Memoir 156, in rocks of the eastern Mojave Desert, California, U.S.A.: when the end-member ages are constrained us- p. 107–133. Implications for thermochronometry: Chemical Geology, ing single-zircon U-Pb data. Davis, D. W., and Edwards, G. R., 1986, Crustal evolution of v. 110, p. 147–167. Archean rocks in the Kakagi Lake area, Wabigoon sub- Krogh, T. E., 1982, Improved accuracy of U-Pb ages by cre- province, Ontario, as interpreted from high-precision U-Pb ation of more concordant systems using an air abrasion ACKNOWLEDGMENTS geochronology: Canadian Journal of Earth Sciences, v. 23, technique: Geochimica et Cosmochimica Acta, v. 46, p. 182–192. p. 637–649. Dickin, A. P., and Higgins, M. D., 1992, Sm/Nd evidence for a Ludwig, K., 1991a, PBDAT: A computer program for process- We thank Scott McLennan for motivating us major 1.5 Ga crust-forming event in the central Grenville ing Pb-U-Th isotope data: U.S. Geological Survey Open- to conduct single-detrital zircon analyses. We Province: Geology, v. 20, p. 137–140. File Report 88–542, 34 p. thank colleague Bob Dott for reviewing an early Dickinson, W. R., and Suczek, C. A., 1979, Plate tectonics and Ludwig, K., 1991b, ISOPLOT: A plotting and regression pro- sandstone compositions: American Association of Petro- gram for radiogenic isotope data: U.S. Geological Survey version of the manuscript and helpful discus- leum Geologists, v. 63, p. 82–86. Open-File Report 91–445, 39 p. sions. Journal reviews were provided by Scott Dott, R. H., Jr., 1983, The Proterozoic red quartzite enigma in Marcantonio, F., McNutt, R. H., Dickin, A. P., and Heaman, the north-central United States: Resolved by plate colli- L. M., 1990, Isotopic evidence for the crustal evolution of McLennan, Sydney Hemming, and Gerald sion?, in Medaris, L. G., Jr., ed., Early Proterozoic geol- the Frontenac Arch in the Grenville Province of Ontario, Ross. Nicholas Van Wyck and Karin Barovich ogy of the Great Lakes region: Geological Society of Canada: Chemical Geology, v. 83, p. 297–314. are thanked for assistance in the laboratory. This America Memoir 160, p. 67–84. McLelland, J. M., Daly, J. S., and Chiarenzelli, J., 1993, Sm-Nd Dott, R. H., Jr., and Byers, C. W., 1981, SEPM research con- and U-Pb isotopic evidence of juvenile crust in the Adiron- research was supported by a grant from the ference on modern shelf and ancient cratonic sedimenta- dack Lowlands and implications for the evolution of the Donors of the Petroleum Research Fund admin- tion: The orthoquartzite-carbonate suite revisited: Journal Adirondack Mts.: Journal of Geology, v. 101, p. 97–105. istered by the American Chemical Society of Sedimentary Petrology, v. 51, p. 329–347. McLennan, S. M., Hemming, S., McDaniel, D. K., and Hanson, Dott, R. H., Jr., Byers, C. W., Fielder, G. W., Stenzel, S. R., and G. L., 1993, Geochemical approaches to sedimentation, (PRF-25677AC8) and by National Science Winfree, K. E., 1986, Aeolian to marine transition in provenance, and tectonics, in Johnson, M. J., and Basu, A., Foundation grants EAR-910566 and EAR- Cambro-Ordovician cratonic sheet sandstones of the eds., Processes controlling the composition of clastic sedi- northern Mississippi valley, U.S.A.: Sedimentology, v. 33, ments: Geological Society of America Special Paper 284, 9304455 to Johnson. p. 345–367. p. 21–40. Emslie, R. F., and Hegner, E., 1993, Reconnaissance isotopic Medaris, L. G., Jr., Johnson, C. M., Schott, R. C., Dott, R. H., geochemistry of anorthosite-mangerite-charnockite-gran- Jr., and Van Wyck, N., 1996, The Baraboo Quartzite, Wis- ite (AMCG) complexes, Grenville Province, Canada: consin: Proterozoic deposition and deformation in the REFERENCES CITED Chemical Geology, v. 106, p. 279–298. Lake Superior region: Geological Society of America Ab- Folk, R. L., 1980, Petrology of sedimentary rocks: Austin, stracts with Programs, v. 28, no. 7, A-376. Anderson, J. L., and Cullers, R. L., 1978, Geochemistry and Texas, Hemphill’s, 184 p. Mezger, K., Hanson, G. N., and Bohlen, S. R., 1989, U-Pb sys- evolution of the Wolf River batholith, a late Precambrian Fujimaki, H., 1986, Partition coefficients of Hf, Zr, and REE tematics of garnet: Dating the growth of garnet in the Late rapakivi massif in northern Wisconsin, U.S.A.: Precam- between zircon, apatite, and liquid: Contributions to Min- Archean Pikwitonei granulite domain at Cauchon and brian Research, v. 7, p. 287–324. eralogy and Petrology, v. 94, p. 42–45. Natawahunan Lakes, Manitoba, Canada: Contributions to Anderson, J. L., Cullers, R. L., and Van Schmus, W. R., 1980, Götze, J., and Lewis, R., 1994, Distribution of REE and trace Mineralogy and Petrology, v. 101, p. 136–148. Anorogenic metaluminous and peraluminous granite plu- elements in size and mineral fractions of high-purity Morey, G. B., 1973, Stratigraphic framework of Middle Pre- tonism in the mid-Proterozoic of Wisconsin, USA: Con- quartz sands: Chemical Geology, v. 114, p. 43–57. cambrian rocks in Minnesota, in Young, G. M., ed., tributions to Mineralogy and Petrology, v. 74, p. 311–328. Graham, C., Valley, J., and Winter, B. L., 1996, Ion microprobe Huronian stratigraphy and sedimentation: Geological As- Ashwal, L. D., and Wooden, J. L., 1983, Sr and Nd isotope analysis of 18O/16O in authigenic and detrital quartz in the sociation of Canada Special Paper 12, p. 211–249. geochronology, geologic history, and origin of the Adiron- St. Peter Sandstone, Michigan basin and Wisconsin arch: Morey, G. B., and Ojakangas, R. W., 1982, Keweenawan sedi- dack Anorthosite: Geochimica et Cosmochimica Acta, Contrasting diagenetic histories: Geochimica et Cos- mentary rocks of eastern Minnesota and northwestern v. 47, p. 1875–1885. mochimica Acta, v. 60, p. 5101–5115. Wisconsin, in Wold, R. J., and Hinze, W. J., eds., Geology Ashwal, L. D., Wooden, J. L., and Emslie, R. F., 1986, Sr, Nd, Green, J. C., and Fritz, T. J., III, 1993, Extensive felsic lavas and and tectonics of the Lake Superior basin: Geological So- and Pb isotopes in Proterozoic intrusives astride the rheoignimbrites in the Keweenawan Midcontinent Rift ciety of America Memoir 156, p. 135–146. Grenville Front in Labrador: Implications for crustal con- plateau volcanics, Minnesota: Petrographic and field Nelson, B. K., 1991, Proterozoic continental growth near the tamination and basement mapping: Geochimica et Cos- recognition: Journal of Volcanology and Geothermal Re- Archean craton: The isotopic record from anorogenic mochimica Acta, v. 50, p. 2572–2585. search, v. 54, p. 177–196. granites in Wisconsin: Eos (Transactions, American Geo- Barovich, K. M., Patchett, P. J., Peterman, Z. E., and Sims, P. K., Gromet, L. P., and Silver, L. T., 1983, Rare earth element distri- physical Union), v. 72, p. 544. 1989, Nd isotopes and origin of the 1.9–1.7 Ga Penokean butions among minerals in a granodiorite and their petro- Nicholson, S. W., 1992, Geochemistry, petrography, and vol- continental crust of the Lake Superior region: Geological genetic implications: Geochimica et Cosmochimica Acta, canology of rhyolites of the Portage Lake Volcanics, Ke- Society of America Bulletin, v. 101, p. 333–338. v. 47, p. 925–939. weenaw Peninsula, Michigan: U.S. Geological Survey Basu, A. R., Sharma, M., and DeCelles, P. G., 1990, Nd, Sr-iso- Grout, F. F., Gruner, J. W., Schwartz, G. M., and Thiel, G. A., Bulletin 1970-B, 39 p. topic provenance and trace element geochemistry of 1951, Precambrian stratigraphy of Minnesota: Geological Nicholson, S. W., and Shirey, S. B., 1990, Midcontinent rift vol- Amazonian foreland basin fluvial sands, Bolivia and Society of America Bulletin, v. 62, p. 1017–1078. canism in the Lake Superior region: Sr, Nd, and Pb iso- Peru: Implications for ensialic Andean orogeny: Earth Hamblin, K. W., 1961, Paleogeographic evolution of the Lake topic evidence for a mantle plume: Journal of Geophysi- and Planetary Science Letters, v. 94, p. 1–21. Superior region from late Keweenawan to Late Cam- cal Research, v. 95, p. 10851–10868. Bhatia, M. R., and Crook, K. A. W., 1986, Trace element char- brian time: Geological Society of America Bulletin, Odom, I. E., Doe, T. W., and Dott, R. H., Jr., 1976, Nature of acteristics of graywackes and tectonic setting discrimina- v. 72, p. 1–18. feldspar-grain size relations in some quartz-rich sand- tion of sedimentary basins: Contributions to Mineralogy Hemming, S. R., McLennan, S. M., and Hanson, G. N., 1994, stones: Journal of Sedimentary Petrology, v. 46, and Petrology, v. 92, p. 181–193. Lead isotopes as a provenance tool for quartz: Examples p. 862–870. Chiarenzelli, J. R., and McLelland, J. M., 1993, Granulite fa- from plutons and quartzite, northeastern Minnesota, USA: Ojakangas, R. W., and Morey, G. B., 1982a, Keweenawan sed- cies metamorphism, paleo-isotherms and disturbance of Geochimica et Cosmochimica Acta, v. 58, p. 4455–4464. imentary rocks of the Lake Superior region: A summary, the U-Pb systematics of zircon in anorogenic plutonic Hoffman, P. F., 1989, Precambrian geology and tectonic history in Wold, R. J., and Hinze, W. J., eds., Geology and tec- rocks from the Adirondack Highlands: Journal of Meta- of North America, in Bally, A. W., and Palmer, A. R., eds., tonics of the Lake Superior basin: Geological Society of morphic Geology, v. 11, p. 59–70. The geology of North America—An overview: Boulder, America Memoir 156 p. 157–164. Corfu, F., 1988, Differential response of U-Pb systems in coex- Colorado, Geological Society of America, Geology of Ojakangas, R. W., and Morey, G. B., 1982b, Keweenawan pre- isting accessory minerals, Winnipeg River Subprovince, North America, v. A, p. 447–511. volcanic quartz sandstones and related rocks of the Lake Canadian Shield: Implications for Archean crustal growth Hurley, P. M., and Rand, J. R., 1969, Pre-drift continental nu- Superior region, in Wold, R. J., and Hinze, W. J., eds., Ge- and stabilization: Contributions to Mineralogy and Petrol- clei: Nature, v. 164, p. 1229–1242. ology and tectonics of the Lake Superior basin: Geologi- ogy, v. 98, p. 312–325. Johnson, C. M., and Thompson, R. A., 1991, Isotopic composi- cal Society of America Memoir 156, p. 85–96. Dake, C. L., 1921, The problem of the St. Peter Sandstone: tion of Oligocene mafic volcanic rocks in the northern Rio Owens, B. E., Dymek, R. F., Tucker, R. D., Brannon, J. C., and Missouri School of Mines and Metallurgy Bulletin, Tech- Grande rift: Evidence for contributions of ancient intraplate Podosek, F. A., 1994, Age and radiogenic isotopic com-

Geological Society of America Bulletin, November 1999 1737 JOHNSON AND WINTER

position of a late- to post-tectonic anorthosite in the granodiorite: Implications for heat production distribu- Van Wyck, N., 1995, Oxygen and carbon isotopic constraints Grenville Province: The Labrieville massif, Quebec: tions in the Sierra Nevada batholith, California, U.S.A.: on the development of eclogites, Holsnoy, Norway and Lithos, v. 31, p. 189–206. Geochimica et Cosmochimica Acta, v. 52, p. 1131–1143. major and trace element, common Pb, Sm-Nd, and zircon Parrish, R. R., 1987, An improved microcapsule for zircon dis- Shirey, S. B., and Hanson, G. N., 1986, Mantle heterogeneity geochronology constraints on petrogenesis and tectonic solution in U-Pb geochronology: Chemical Geology, and crustal recycling in Archean granite-greenstone belts: setting of pre- and early Proterozoic rocks in Wisconsin v. 66, p. 99–102. Evidence from Nd isotopes and trace elements in the [Ph.D. thesis]: Madison, University of Wisconsin, 280 p. Parrish, R. R., Roddick, J. C., Loveridge, W. D., and Sullivan, Rainy Lake area, Superior Province, Ontario, Canada: Van Wyck, N., and Johnson, C. M., 1997, Common lead, Sm-Nd, R. W., 1987, Uranium-lead analytical techniques at the Geochimica et Cosmochimica Acta, v. 50, p. 2631–2651. and U-Pb constraints on petrogenesis, crustal architecture, geochronology laboratory, Geological Survey of Canada, Sims, P. K., Card, K. D., and Lumbers, S. B., 1981, Evolution and tectonic setting of the Penokean orogeny (Paleopro- in Radiogenic age and isotopic studies, Report 1: Geolog- of Early Proterozoic basins of the Great Lakes region, in terozoic) in Wisconsin: Geological Society of America ical Survey of Canada Paper 87–2, p. 3–7. Campbell, F. H. A., ed., Proterozoic basins of Canada: Bulletin, v. 109, p. 789–798. Paterson, B. A., Rogers, G., and Stephens, W. E., 1992, Evi- Geological Survey of Canada Paper 81–10, p. 379–397. Wark, D. A., and Miller, C. F., 1993, Accessory mineral behav- dence for inherited Sm-Nd isotopes in granitoid zircons: Sims, P. K., Van Schmus, W. R., Schulz, K. J., and Peterman, ior during differentiation of a granite suite: Monazite, Contributions to Mineralogy and Petrology, v. 111, Z. E., 1989, Tectono-stratigraphic evolution of the Early xenotine and zircon in the Sweetwater wash pluton, p. 378–390. Proterozoic Wisconsin magmatic terranes of the Penokean southeastern California, U.S.A.: Chemical Geology, Patterson, C., and Tatsumoto, M., 1964, The significance of orogen: Canadian Journal of Earth Sciences, v. 26, v. 110, p. 49–67. lead isotopes in detrital feldspar with respect to chemical p. 2145–2158. Wasserburg, G. J., Jacobsen, S. B., DePaolo, D. J., McCulloch, differentiation within the Earth’s mantle: Geochimica et Sims, P. K., Anderson, J. L., Bauer, R. L., Chandler, V. W., M. T., and Wen, T., 1981, Precise determinations of Cosmochimica Acta, v. 28, p. 1–22. Hanson, G. N., Kalliokoski, J., Morey, G. B., Mudrey, Sm/Nd ratios, Sm and Nd isotopic abundances in stan- Pettijohn, F. J., 1957, Sedimentary Rocks (second edition): M. G., Ojakangas, R. W., Peterman, Z. E., Schulz, K. J., dard solutions: Geochimica et Cosmochimica Acta, v. 45, New York, Harper, 718 p. Shirey, S. B., Smith, E. I., Southwick, D. L., and p. 2311–2332. Potter, P. E., and Pryor, W. A., 1961, Dispersal centers of Pale- Weiblen, P. W., 1993, The Lake Superior region and Williams, I. S., Compston, W., and Chappel, B. W., 1983, Zir- ozoic and later clastics of the Upper Mississippi Valley Trans-Hudson orogen, in Reed, J. C., Jr., Bickford, M. E., con and monazite U-Pb systems and the histories of I- and adjacent areas: Geological Society of America Bul- Houston, R. S., Link, P. K., Rankin, D. W., Sims, P. K., type magmas, Berridale batholith, Australia: Journal of letin, v. 72, p. 1195–1250. and Van Schmus, W. R., eds., Precambrian: Contermi- Petrology, v. 24, p. 76–97. Rainbird, R. H., McNicoll, V. J., Theriault, R. J., Heaman, nous U.S.: Boulder, Colorado, Geological Society of Winter, B. L., Valley, J. W., Simo, J. A., Nadon, G. C., and John- L. M., Abbott, J. G., Long, D. G. F., and Thorkelson, D. J., America, Geology of North America, v. C-2, p. 11–120. son, C. M., 1995, Hydraulic seals and their origin: Evi- 1997, Pan-continental river system draining Grenville Sloss, L. L., 1963, Sequences in the cratonic interior of North dence from the stable isotope geochemistry of dolomites orogen recorded by U-Pb and Sm-Nd geochronology of America: Geological Society of America Bulletin, v. 74, in the Middle Ordovician St. Peter Sandstone, Michigan Neoproterozoic quartzarenites and mudrocks, northwest p. 93–114. basin: American Association of Petroleum Geologists Canada: Journal of Geology, v. 105, p. 1–17. Smith, M., and Gehrels, G., 1994, Detrital zircon geochronol- Bulletin, v. 79, p. 30–48. Ross, G. M., and Parrish, R. R., 1991, Detrital zircon ogy and the provenance of the Harmony and Valmy Zhao, J. X., McCulloch, M. T., and Bennett, V. C., 1992, Sm- geochronology of metasedimentary rocks in the southern Formations, Roberts Mountain allochthon, Nevada: Geo- Nd and U-Pb zircon isotopic constraints on the prove- Omineca Belt, Canadian Cordillera: Canadian Journal of logical Society of America Bulletin, v. 106, p. 968–979. nance of sediments from the Amadeus basin, central Aus- Earth Sciences, v. 28, p. 1254–1270. Stacey, J. S., and Kramers, J. D., 1975, Approximation of ter- tralia: Evidence for REE fractionation: Geochimica et Sawka, W. N., 1988, REE and trace element variations in ac- restrial lead isotope evolution by a two-stage model: Earth Cosmochimica Acta, v. 56, p. 921–940. cessory minerals and hornblende from the strongly zoned and Planetary Science Letters, v. 26, p. 207–221. McMurry Meadows pluton, California: of the Royal So- Tyler, S. A., 1936, Heavy minerals of the St. Peter Sandstone in ciety of Edinburgh Transactions, Earth Sciences, v. 79, Wisconsin: Journal of Sedimentary Petrology, v. 6, p. 157–168. p. 55–84. MANUSCRIPT RECEIVED BY THE SOCIETY AUGUST 12, 1997 Sawka, W. N., and Chappell, B. W., 1988, Fractionation of ura- Van Schmus,W. R., 1992, Tectonic setting of the Midcontinent REVISED MANUSCRIPT RECEIVED DECEMBER 15, 1998 nium, thorium and rare earth elements in a vertically zoned Rift system: Tectonophysics, v. 213, p. 1–15. MANUSCRIPT ACCEPTED MARCH 9, 1999

Printed in U.S.A.

1738 Geological Society of America Bulletin, November 1999