nl- INflCIfl‘OJ a.“ OW ’12 .. .. 4mg ‘I z. i‘ 2 )l- 01-. .\ ;.--‘“\‘ 83*“? 7";fl 2 r k. (n' - git - \- N 2. 2: 222; 22.2 ‘ 7‘56}? “r" :fia2f >23 22: :32 {2 - 4 ¢.IJI_§:”I ‘o'.’ J 1“"‘2‘ 'B'e‘lv 3" '\ wv—u ‘- W- ‘3' u}; as- '4 "EC”I‘T? g" _(r' I 54:11”: 1" $127,;- ’ 4:: "Il.\‘."“' .‘4/ o ‘4 ' . .- . ' ‘1 MI/fidé. I W”. f ;. . " III, ‘ ,. ' .7 ' '7 "I?” 1.5“ - ‘ -' -' ‘ ‘-*"x. . . .-’ '- .~ -2 ”fr" 4%“; ’Ic: 2,5555, . - 2 . o I: V. _ . .3 1:21 $41.52!. "2. . 7 :/.g;:' 2:32., 2 '....(. H.912?! “#1,- ‘J .42.". . .. .. 7 .2 .222:- .. _.,, 22-27,- " ‘ " ' ' y, I" f : r A?!" I) r' I I ‘1‘."- : . ' . '51 . ' -‘ . .. ‘-‘ , I ’- ’I"'J."If ' ‘ . "V1351! 3. ' 3"... v. . 2 ‘ 2,1,3?" ‘1’”, " ll“V 4/Jf;';§{{y./J/l’ /: 1 <‘: .‘o (2:? ml" ' ' r a If? 7 ”gag?” .-':I ,' , ri/ -' "VIC/.9 by” {.25er .’,:, [fr fipli'pr 5""fl'1‘wyil‘rbfi’ .-. 15722;"; ’2": ”Pi. {"1537}? ’95" .124: 212933923“ fl); 1/ I” 'I ; , ”a” - l . {W -.'§-L\-A PM». .- r v ..—2- (I ' W - In! 2-2: . , . . 2,2,", 1.2 22,24 . ‘1' r - ' .J .- .. .J ,4 . '-'%-'-"~'2' " .- .' 2- "2" WWI/2... , = . 2 4 , ~’/7/J"'/I.'r' ,2 (22,... I”- [9‘027'477 .‘I/V'LI’V'I, " r'yl” l‘ h', ' ‘ u' " ’ .- UWQ’I,‘ " ' .‘ "'3'.“ ./. _K'Ilg ’/1 I If!!!” amona- ~ Illlllllllllllll"Infill!"Ill!"lllllfllllllllfllllllll ? 0“ ‘5 b 5 7‘ 5 LIBRARY Michigan State University This is to certify that the thesis entitled A Model for the Origin of the Chemically Zoned Ammonia Tanks Member of the Timber Mountain Tuff Due to Crystal Fractionation presented by Timothy Patrick Rose has been accepted towards fulfillment of the requirements for Masters degree in Geology 9%er Major professor/ Date 37/7/77 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution MSU LIBRARIES n ‘ RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. A MODEL FOR THE ORIGIN OF THE CHEMICALLY ZONED AMMONIA TANKS MEMBER OF THE TIMBER MOUNTAIN TUFF DUE TO CRYSTAL FRACTIONATION By Timothy Patrick Rose A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Geological Sciences 1988 J 75/37 / ’ ABSTRACT A MODEL FOR THE ORIGIN OF THE CHEMICALLY ZONED AMMONIA TANKS MEMBER OF THE TIMBER MOUNTAIN TUFF DUE TO CRYSTAL FRACTIONATION By Timothy Patrick Rose The Ammonia Tanks Member of the Timber Mountain Tuff in southern Nevada is a large volume, compositionally zoned ash-flow sheet. Major and trace element chemical analyses of whole, glassy pumice fragments are used to reconstruct the chemical gradients inferred to be present in the pre- eruption chamber. The origin of these compositional trends are shown to be consistent with crystal-liquid fractionation. The crystal fractionation model predicts that the high- silica rhyolite (78% $102) of the Ammonia Tanks Member can be derived from the quartz latite (59% SiOz) by fractionating the minerals plagioclase, sanidine, clinopyroxene, biotite and magnetite. However, the common occurence of disequilibrium phenocrysts is not readily explained by this model. Therefore, the possibility that magma mixing acted in conjunction with crystal fractionation is also evaluated. The results of this combined model are consistent with both the chemical data and textural observations. ACKNOWLEDGEMENTS This thesis is dedicated to my parents, Robert and Dorothy. Without their support, both moral and financial, I would never have come this far. Special recognition also goes to my wife, Paula, who endured many long evenings of neglect, and also provided computer expertise during the preparation of this manuscript. Tom Vogel, my thesis advisor deserves many thanks for providing professional, financial and moral support. Tom has taught me a great deal, but I am especially indebted to him for showing me how to critically, and creatively evaluate scientific problems. My committee members, John Wilband and Dave Long have both contributed much to my education, and I am appreciative of the help and individual attention that each of them have given to me. I also thank William Cambray and Kazuya Fujita for the knowledge and support they provided. Very special thanks go to Jim Mills, who has helped in the preparation of this thesis in many more ways than could possibly be listed here. Jim has been an exceptional colleague and friend. I also wish to acknowledge Tim Flood and Ben Schuraytz for all the help that they gave me while learning the basics of ash-flow magmatism. ii This thesis was partially supported by the Basic Energy Science Program at Lawrence Livermore National Laboratory through a grant to Thomas A. Vogel. Support was also provided by the Department of Geological Sciences at Michigan State University through a fellowship from the Chevron corporation. iii TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES INTRODUCTION . . . . . . . . COMPOSITIONAL ZONATIONS IN ASH-FLOW SHEETS REGIONAL GEOLOGY . . . . . . . SAMPLING PROCEDURE . . . . . . ANALYTICAL PROCEDURE . . . . . MAJOR ELEMENT CHEMISTRY . . . . TRACE ELEMENT CHEMISTRY . . . . MINERALOGY . . . . . EVALUATION OF CRYSTAL FRACTIONATION EVALUATION OF MAGMA MIXING . . DISCUSSION AND CONCLUSIONS . . APPENDIX 1: SAMPLE LOCATIONS . . APPENDIX 2: SAMPLE PREPARATION TECHNIQUES APPENDIX 3: A METHOD FOR THE DETERMINATION TRACE PHASE FRACTIONATION PRECENTAGES APPENDIX 4: MAJOR ELEMENT MULTIPLE LINEAR REGRESSIONS CALCULATED WITH Y-INTERCEPTS EQUAL TO ZERO . . . . . . . . LIST OF REFERENCES . . . . . . iv OF 10 22 31 37 52 57 62 63 64 66 68 Table Table Table Table Table Table Table LIST OF TABLES Chemical analyses of pumice fragments . . Representative mineral analyses . . . . Major element multiple linear regressions . Mineral analyses used in regressions for crystal fractionation model . . . . Distribution coefficients . . . . . . Trace element models . . . . . . . Mineral analyses used in regressions for crystal fractionation + magma mixing model. 11 32 40 42 47 48 56 Figure Figure Figure Figure Figure LIST OF FIGURES Location of the southwestern Nevada volcanic field, including the Timber Mountain-Oasis Valley caldera complex . Variation of major element oxides versus silica. Units in weight percent . Enrichment factors for 17 minor and trace elements. (a) Element concentrations in the most silicic pumice fragment divided by concentrations in the most mafic pumice fragment. (b) Element concentrations in a pumice fragment of intermediate silica composition divided by concentrations in the most mafic pumice fragment . . . . Chondrite-normalized REE abundances (log scale) versus atomic number . . . Variation of trace element abundances (ppm) versus silica (wt%) . . . . . vi 17 23 26 28 INTRODUCTION In partially crystallized magma systems, the process of crystal fractionation is an efficient means of magmatic differentiation because of the potential it has for generating liquids of contrasting composition from the parent liquid (Bowen, 1928). Recent studies have shown that the evolution of some shallow, silicic magma systems can be modeled using crystal fractionation as the dominant magmatic differentiation mechanism (Michael, 1983; Cameron, 1983; Miller and Middlefehldt, 1984; Wolff and Storey, 1984). The Ammonia Tanks Member of the Timber Mountain Tuff is a large volume, compositionally zoned ash-flow sheet in southern Nye County, Nevada that varies in composition from a high-silica rhyolite to a quartz latite (Byers, et al., 1976b). Pumice fragments within a given handsample from the Ammonia Tanks Member may show as much compositional variation as that in the entire ash-flow sheet. This observation can lead to the hypothesis that the rhyolitic and latitic magmas are genetically related. This investigation uses chemical and mineralogical data obtained from individual glassy pumice samples of the Ammonia Tanks to test the possibility that the high-silica rhyolite was derived from the quartz latite as a result of crystal fractionation. COMPOSITIONAL ZONATIONS IN ASH-FLOW SHEETS It is generally acknowledged that ash—flow eruptions result from the rapid, partial evacuation of an upper crustal magma chamber (Williams, 1941; Smith, 1960). Ash-flow sheets commonly exhibit stratigraphic variations in both chemistry and mineralogy (Smith and Bailey, 1966; Lipman, et a1, 1966; Hildreth, 1979). These variations are inferred to represent an inverted record of the compositional and thermal zonations that existed in the magma chamber prior to eruption (cf. Lipman, et al., 1966; Smith, 1979; Hildreth, 1981; Mahood, 1981; Baker and McBirney, 1985). Although major element compositional variations within single ash-flow sheets are often small, trace element gradients over the same stratigraphic interval can show a variance of greater than three orders of magnitude (cf. Hildreth, 1977, 1979; Smith, 1979). Concomitant with these compositional changes, Fe-Ti oxide temperatures also change systematically with stratigraphic height in ash-flow sheets (Lipman, 1971; Hildreth, 1979; Schuraytz, et al., 1986). In most eruptive episodes, the bulk chemistry of the ash-flow sheet shows a trend toward higher temperature, more mafic, phenocryst-rich compositions with increasing stratigraphic height. At the same time, the chemical variation among pumice fragments becomes increasingly heterogeneous. Schuraytz, et al., (1983, 1986) observed that the chemical variation among pumice fragments from 2 3 the top of an ash-flow sheet may be representative of the chemical variation among pumice fragments throughout the entire ash-flow sheet. Based on these observations, it is logical to assume that the most highly differentiated, low temperature, silicic magma is segregated into the roof zone of the magma chamber, and that zoned magma bodies become progressively hotter and more mafic with increasing depth. Models for the dynamics of magma withdrawal from layered magma reservoirs (Blake, 1981; Spera, 1984; Blake and Ivey, 1986) predict evacuation along subspherical isochrons that tap successively deeper levels in the magma chamber with increasing time. This is consistent with the observed maximum in pumice heterogeneity at the top of the ash-flow sheet since the uppermost part of an ash-flow sheet is believed to represent an instantaneous sampling of magma from all parts of the erupted magma body. Examination of the compositional and mineralogical variations within pumice fragments from an ash-flow sheet provides the means to reconstruct the variations that existed in the magma chamber prior to eruption. Given this variation, it is then possible to evaluate which magma differentiation processes may have been responsible for the compositional evolution of the magma system. This investigation is directed toward the quantitative evaluation of crystal fractionation as a differentiation mechanism to explain the compositional variations observed 4 in the Ammonia Tanks ash-flow sheet. REGIONAL GEOLOGY The southwestern Nevada volcanic field, in southern Nye County, Nevada was erupted during the late Miocene and early Pliocene, approximately 16 to 9 million years ago (Byers, et al., 1976b). This volcanic field produced at least nine large volume, high-silica ash-flow sheets, numerous smaller rhyolitic tuffs and lava flows, and subordinate amounts of basaltic volcanism, creating an extensive volcanic plateau that once covered an area of more than 11,000 km2 (Christiansen, et al., 1977). This activity was contemporaneous with widespread extensional normal faulting that occurred throughout the Great Basin from Eocene to Pliocene time (Lipman, et al., 1966; Christiansen and Lipman, 1972). A minimum of six calderas have been identified in the southwestern Nevada volcanic field; these are, from oldest to youngest, the Sleeping Butte, Silent Canyon, Claim Canyon, Timber Mountain-Oasis Valley, Black Mountain and Stonewall Mountain calderas (Fig. 1) (Christiansen, et al., 1977). All of these calderas, with the exception of Black Mountain and Stonewall Mountain, are nested together in a tight cluster, possibly indicating a single magmagenetic source (Christiansen, et al., 1977). Recent work by Flood (1987) on the Paintbrush Tuff strongly supports a cyclic evolution model. GoIdIIcId STONEWALI. MOUNTAIN CALDERA COMPLEX _,,.,o. a . SILENT CANYON BLACK MOUNTAIN CALDERA Southwestern CALDERA \ .. if: \\ Nevada 8 a l : Volcanic 1 i tween MOUNTAIN- , ,: OASIS VALLEY ,: : .5 ‘ \‘~ Beatly \ .o' /" 0 99/6 1. cause FLAT- _,,.,0. % PROSPECTOR PASS _ e CALDERA coupLsxm 4:, m to L . . . . ‘.°" °t 1:. 4 .Pahrump I Figure 1. Location of the southwestern Nevada volcanic field, including the Timber Mountain-Oasis Valley caldera complex (after Carr, et al., 1984. and Noble, et al., 1984). 6 The lower Pliocene, alkali-calcic Timber Mountain Tuff is defined by Byers, et al., (1976b) to include all quartz- bearing ash-flow sheets and all interbedded ash-fall tuff erupted from the Timber Mountain caldera center around 11 million years ago. The tuff includes, in chronological order, the large volume Rainier Mesa and AmmOnia Tanks Members, and the intracaldera tuffs of Buttonhook Wash and Crooked Canyon. Subsidence of the Timber Mountain caldera was initiated by the eruption of the 1200 km3 Rainier Mesa Member of the Timber Mountain Tuff approximately 11.3 million years ago (Christiansen, et al., 1977). This voluminous, compositionally zoned ash-flow sheet consists of an extensive high-silica rhyolite overlain with a partial cooling break by a considerably thinner quartz latitic tuff (Byers, et al., 1976b). Following a hiatus of roughly 0.2 million years, eruption of the 900 km3 Ammonia Tanks Member resulted in further collapse of the Timber Mountain caldera. Twelve separate K—Ar determinations on minerals from the Ammonia Tanks Member yielded an average age of 11.1 million years (Kistler, 1968; Marvin and others, 1970). The Ammonia Tanks Member is relatively complex in composition as compared with the Rainier Mesa Member, and is probably a composite sheet as defined by Smith (1960). Byers, et al., (1976a) divides the Ammonia Tanks Member into upper and lower cooling units, and these units have been further subdivided by Byers, et al., (1976b) on the 7 basis of composition into quartz latite and high-silica rhyolite. According to Byers, et al., (1976b), both the upper and lower cooling units are reversely zoned; that is, in both cases, quartz latitic magma was erupted first, followed later by rhyolitic magmas (Byers, et al., 1976b). However, the results of this study show that portions of the Ammonia Tanks ash—flow sheet are normally zoned with respect to composition. Soon after the Ammonia Tanks eruptions ended, the main cauldron block was uplifted as a resurgent dome. The maximum uplift is at least 1200 m., but this is partially compensated by about 600 m. of displacement along graben faults (Christiansen, et al., 1977). SAMPLING PROCEDURE Chemical data used in this study were obtained from the analysis of individual glassy pumice fragments. Glassy samples were used because they are believed to closely represent a discrete parcel of instantaneously quenched magma. That is, with the exception of volatile losses or secondary hydration, glassy pumice fragments (glass + phenocrysts) closely approximate the original composition of the magma. The possibility of chemical modifications due to secondary ground water hydration cannot be ruled out (Aramaki and Lipman, 1965; Lipman, 1965; Noble, 1965, 1967). Pumices fragments that showed evidence of vapor phase alteration, devitrification or intense weathering were not sampled. All pumice samples were collected during the summer of 1986. Sampling locations were based on published U.S.G.S. geologic 7 1/2 minute quadrangle maps (Ekrin, et al., 1966; Noble, et al., 1967). The chief sampling criterion was that good stratigraphic control could be maintained. Preferred stratigraphic sections were thoSe that contained contacts with upper or lower stratigraphic units. Byers, et al., (1976b) observed that the four compositional subunits of the Ammonia Tanks Member (two cooling units, each divided into quartz latitic and rhyolitic components) are not everywhere present outside the area of cauldron subsidence, nor are they precisely correlative over the entire extent of the Ammonia Tanks ash-flow sheet. Although all four units are present on the Timber Mountain resergent dome, field investigations in this area indicated that nearly all of this tuff is devitrified. Therefore none of the stratigraphic sections sampled in this study are representative of all four compositional subunits. Our sampling focused on obtaining the maximum number of visually different glassy pumice fragments at a given section. Most of the pumice fragments were collected from the non—welded base and top of the sections. In places, entire sections consisted of non-welded tuff. Under these circumstances, samples were collected at equally spaced intervals (usually 10-20 ft.) between the base and the top 9 of the section. Greatest heterogeneity occurs at the top of any given section. Of the 15 sections of Ammonia Tanks collected, three were selected for sample preparation. The sections used in the study were selected based on the number of samples taken, the observed heterogeneity among pumice fragments, and the stratigraphic control on the section. However, one of these sections (located at latitude 37°15’32.5"N, longitude 116°20’51.4"W, on the Dead Horse Flat 7 1/2 minute quadrangle map, 0.4 km. N of S. Silent Canyon) was later eliminated because the mineralogy and trace element chemistry of samples taken from the section was significantly different from that of samples taken from the other two sections. Although the section in question was mapped as the lower unit of the Ammonia Tanks Member, because it is chemically similar to the Rainier Mesa Member, we now believe that it is the Rainier Mesa Member. ANALYTICAL PROCEDURE Twenty-seven glassy pumice fragments were analyzed in this study. Each sample used for analysis was first course crushed and leached in a solution containing sodium acetate and glacial acetic acid in order to remOve any secondary carbonate minerals (see Appendix 2 for procedure). The samples were then powdered by hand in an agate mortar. For each sample, ten major and twenty trace elements were determined. Major elements plus Ba, Rb, Sr, Zr, Nb, Y, 10 Cr, Ni, Cu and Zn were analyzed by a Rigaku X—Ray fluorescence spectrometer (XRF) at Michigan State University. XRF analyses were carried out on glass wafers made according to the procedure described by Hagan (1982) (see Appendix 2). The remaining trace elements, Sc, Hf, Th, La, Ce, Sm, Eu, Tb, Yb and Lu were done by instrumental neutron activation analysis (INAA), also at Michigan State University. Concentrations were determined by linear least squares regression on U.S.G.S. standards. Eight samples representing the entire range of observed silica content among pumice were selected for detailed mineralogical analysis. Heavy mineral phenocrysts from a -60 to f140 sieve size fraction were separated in tribromomethane (bromoform) and mounted in epoxy, along with hand picked phenocrysts of quartz, plagioclase, sanidine, pyroxene, biotite, magnetite and ilmenite. In addition, thin sections were cut from each of the eight samples for petrographic and microprobe analyses. All mineral and glass analyses were made at Lawrence Livermore National Laboratory using a JEOL 733 electron microprobe. MAJOR ELEMENT CHEMISTRY The major element data, reported as weight percent oxides, is presented in Table 1. Because of secondary hydration of the glass, all values have been recalculated to 100 wt%. This allows for comparisons to be made among samples. Figure 2 illustrates the variation of the major 11 TABLE 1. CHEMICAL ANALYSES OF PUMICE FRAGMENTS SAMPLE ID AT4-14 AT4-16 AT4-53 AT4-55 AT4-69 WT% Si02 77.70 65.12 76.75 69.30 77.37 Ti02 0.14 0.70 0.17 0.37 0.13 A1203 12.20 17.72 12.55 16.31 12.18 FeO 0.67 2.89 0.74 1.50 0.69 MnO 0.08 0.16 0.07 0.09 0.08 MgO 0.00 0.86 0.00 0.33 0.03 CaO 0.43 2.20 0.42 1.17 0.38 NazO 3.76 4.77 3.03 4.62 2.98 K20 5.00 5.41 6.24 6.25 6.13 P205 0.01 0.16 0.01 0.06 0.01 TOTALS 99.99 99.99 99.98 100.00 99.98 PPM Zn 37 193 24 40 24 Rb 233 133 201 130 220 Sr 9 382 12 192 10 Y 29 27 28 26 28 Zr 119 697 159 316 125 Nb 41 24 37 27 40 Ba 296 2478 169 852 131 So 1.9 7.0 1.3 3.5 1.6 Hf 5.2 15.6 5.7 8.8 5.1 Th 34 17 28 21 33 La 35 200 48 103 32 Ce 69 295 86 172 64 Sm 6.2 7.1 6.6 6.7 6.0 Eu 0.3 3.3 0.4 1.3 0.3 Tb 0.7 0.6 0.6 0.6 0.7 Yb 4.3 2.5 3.7 3.2 3.7 Lu 0.4 0.4 0.4 0.3 0.4 TABLE 1. SAMPLE ID WT% SiOz T102 A1203 FeO MnO MgO CaO N82 0 K2 0 P2 05 TOTALS PPM Zn Rb Sr Zr Nb Ba Sc Hf Th La Ce Sm Eu Tb Yb Lu (Continued) AT4-75 63.77 0.78 17.56 3.28 0.12 1.67 2.62 4.37 5.48 0.34 99.99 68 87 491 26 801 18 3820 6.9 .16.? 14 12 ATS-2 69.64 0.44 16.00 1.58 0.11 0.63 0.94 4.28 6.31 0.06 99.99 53 138 88 28 438 25 366 4.0 11.2 22 149 241 10.9 1.5 ONO cn~aq ATS-3 77.96 0.13 12.12 0.58 0.08 0.00 0.36 3.25 5.49 0.01 99.98 ATS-4 70.66 0.31 14.88 1.30 0.09 0.94 1.83 3.77 6.14 0.08 100.00 44 153 179 28 295 23 622 2.7 8.9 22 90 49 ONOHmi—t OIQMON AT5-8 77.30 0.16 12.53 0.59 0.08 0.00 0.38 3.35 5.60 0.01 100.00 219 TABLE 1. SAMPLE ID WT% $102 T102 A1203 Fe0 MnO Mgo CaO N820 K20 P205 TOTALS PPM Zn Rb Sr Zr Nb Ba Sc Hf Th La Ce Sm Eu Tb Yb Lu (Continued) ATS-9 70.41 0.41 15.50 1.67 0.11 0.37 1.02 4.10 6.34 0.06 99.99 54 140 134 28 378 26 691 4.2 11.6 23 116 196 10.1 1.3 ONO 0'th 13 ATS-25 77.46 0.14 12.47 0.65 0.08 0.02 0.38 3.03 5.77 0.01 100.01 29 223 14 29 133 34 90 1.4 ATS-29 77.24 0.15 12.60 0.56 0.08 0.00 0.40 3.33 5.63 0.01 100.00 228 ATS-3O 65.63 0.71 17.50 2.48 0.11 0.87 1.69 4.63 6.19 0.19 100.00 103 96 261 27 818 18 2128 6.1 16.1 17 ATS-31 62.85 0.84 18.06 3.72 0.12 1.51 2.94 4.90 4.71 0.33 99.98 125 100 574 27 1044 15 3616 6.3 19.1 13 148 235 3.4 ONO aims] TABLE 1. (Continued) SAMPLE ID AT5-35 WT% 8102 72.96 T102 0.29 A1203 14.52 FeO 1.24 MnO 0.09 MgO 0.25 CaO 0.75 Na20 3.70 K20 6.15 P205 0.04 TOTALS 99.99 PPM Zn 42 Rb 164 Sr 88 Y 28 Zr 247 Nb 28 Ba 394 So 2.3 Hf 8.0 Th 24 La 71 Ce 130 Sm 8.6 Eu 0.6 Tb 0.7 Yb 2.7 Lu 0.5 14 ATS-37 77.69 0.14 12.20 0.70 0.08 0.00 0.36 3.19 5.62 0.01 99.99 ATS-56 77.67 0.14 12.21 0.56 0.06 0.00 0.31 3.05 5.99 0.00 99.99 ATS-58 69.25 0.39 16.10 1.46 0.10 0.33 1.14 4.29 6.86 0.07 99.99 38 122 138 27 393 24 709 3.9 11.0 20 ATS—63 62.11 0.90 18.02 3.83 0.12 1.55 3.25 4.77 5.08 0.35 99.98 95 75 615 27 1017 12 3990 6.9 18.9 12 138 228 10.3 3.5 ONO nth-h“! TABLE 1. SAMPLE ID WT% $102 T102 A1203 FeO MnO MgO CaO NazO K2 0 P2 05 TOTALS PPM Zn Rb Sr Zr Nb Ba Sc Hf Th La Ce Sm Eu Tb Yb Lu (Continued) ATS—65 69.51 0.45 15.95 1.65 0.11 0.28 0.95 4.24 6.78 0.07 99.99 46 117 85 27 480 24 499 4.2 12.7 21 146 265 10.5 1.6 ONO (”Nd 15 ATS-68 73.79 0.24 14.11 1.04 0.08 0.11 0.74 3.80 6.05 0.03 99.99 34 157 83 28 220 30 373 1.3 7.0 53 0 ONOOOH OINNUION ATS-69 61.31 1.00 18.00 4.43 0.13 1.75 3.40 4.62 4.96 0.40 100.00 100 72 654 26 1154 14 4073 7.3 19.6 12 134 9.9 ONOOO ub-bslub AT5-71 66.95 0.56 17.10 1.90 0.10 0.53 1.35 4.41 6.97 0.11 99.98 47 89 164 26 525 21 1168 6.4 12.3 20 233 393 13.1 2.6 ONO (II-5‘) AT5—72 64.00 0.71 18.19 3.12 0.11 1.15 2.61 4.41 5.46 0.24 100.00 78 80 524 26 951 14 3560 6.3 17.8 13 135 9.6 ONO“ QUIQO 16 TABLE 1. (Continued) SAMPLE ID ATS-75 ATS-79 WT% 8102 59.44 76.79 T102 1.15 0.13 A1203 18.23 12.47 FeO 5.05 0.65 MnO 0.13 0.07 MgO 2.24 0.22 C80 4.21 0.48 N820 4.74 2.81 K20 4.33 6.37 P205 0.49 0.01 TOTALS 100.01 100.00 PPM Zn 96 28 Rb 68 214 Sr 819 15 Y 26 29 Zr 1150 127 Nb 10 34 Ba 4703 24 So 8.1 1.0 Hf 18.6 4.9 Th 11 31 La 128 30 Ce 217 56 Sm 8.9 5.2 Eu 3.3 0.1 Tb 0.7 0.8 Yb 2.4 3.0 Lu 0.3 0.5 NOTE: All major element analyses have been normalized to 100% 19 0 r I 0 l r I l r l l 000 17 - q” o - oz... .0 . 2' 0 00 13 - %- 11 J I 1 L 1__l l J l 59 63 67 71 75 79 2 5E r I I f I I I I I 2.0 *- d L 0 a 0 o 1 5 _ 00 - m - z 1 o - o o m o 0.5 r- 0 O - .- go 0 0 q 0 I I ,L I I I ICD I a; 59 63 67 71 75 79 6 I I I I I I I I r 0 q 0 o 4 b 00 1 OJ "' go -( U- o 2 - o 00 '- "' o 00 .1 Q: 0 I4 4 I I L I L I 59 63 67 . 71 75 79 8102 Figure 2. Variation of major element oxides versus silica. Units in weight percent 6.0 f I I I f I I I T 5.0 ' 0 " 9 0° “to o _ O 0 0 O m~4 0 P O "‘ 3.0 - g)?- 2.0 I I L I I J I I I 59 63 67 71 75 79 8.0 I I I I I I I I l 7.0 - c Q; d r .. d) :«6.0 - O % 00 8%- . M a- 5.0 - co 0‘ b o d o 4.0 J I J_ I L L l I L 59 63 67 71 75 79 5.0 I I I I I I I I I 5? - o o 3.0 r 00 . m (m c.) 0 .. - 0 o o 1.0 - $0 00 1 I J L I I I J I a” 59 63 67 . 71 75 79 8102 Figure 2. (continued) Variation of major element oxides versus silica. Units in weight percent. 1 25 I I 1 I I I I I I 0 1.00 - o « 0 0 c5.0.75 - $500 ~ 0” 0 r- 0.50 L 5 80 0.25 P 0 00 - Elm o I l I I I I L _I I 59 63 67 71 75 79 0018 I I I I I I I I I P 0 q 0.14 - .. (D o o - coo . C o o 00 Z 0.10 l- o o - o o o 0 0.06 - @- IAI L I I _L I l I 59 63 67 71 75 79 005 I I I I I I I I I 0.4 - o .. 0 o o 50.3 b .l O... o 0. 0.2 "' o .J 0 0.1 - 0 - e5" 0 0 4 I I I J I Io I w 59 53 57 . 71 75 79 8102 Figure 2. (continued) Variation of major element oxides versus silica. Units in weight percent. 20 elements plotted against silica. Major element oxide variation diagrams provide a guide to identifying where various crystal phases may have been fractionating from the melt. Inflections in the chemical trends represented on these diagrams may indicate that a mineral phase had begun to crystallize at that composition. Si02 was chosen as the abscissa on these diagrams because of it’s large compositional range. Furthermore, silica has a non- overlapping range of values for many of the minerals used in the fractionation model. This makes it easier to identify specific mineral control on any of the individual variation diagrams because different minerals will not plot at the same point (Wright, 1974). Most of the major element variation diagrams, with the exception of Na20 and K20, exhibit curvilinear trends with little scatter. Scatter on the Na20 and K20 diagrams may indicate secondary hydration processes. Well defined changes in slope occur in the A1202, FeO, MgO, CaO, K20 and P202 plots v.s. Si02, possibly indicating the fractionation of minerals containing these elements as important constituents. The twenty-seven glassy whole pumice fragments analyzed in this study exhibit a range in S102 from 59.4 wt% to 78 wt%. Previous whole rock analyses of the Ammonia Tanks Member are reported to range from 66 wt% to 77 wt% (Byers, et al., 1976b). Silica gaps appear between 67% and 69.25% Si02, between 70.7% and 73% Si02, and again between 73.8% 21 and 76.8% Si02. It may be that these discrete compo- sitional breaks indicate the presence of sharp compo- sitional interfaces that existed in the magma chamber prior to eruption. 0n the other hand, it is possible that some or all of the apparent compositional gaps reflect sampling bias. The variation of major elements with stratigraphic height for the sampling locations used in this study is well illustrated by the Pahute Mesa Road Section #5 (see Appendix 1 for location). Twenty-one of the twenty—seven glassy pumice fragments analyzed were taken from this section. At the base of the section, pumice compositions range from 78 wt% to 69.6 wt% Si02, whereas approximately 30 ft. above the base, the range is 77.7 wt% to 62.8 wt% Si02. At the top of the section, roughly 100 ft. above the base, pumice compositions vary from 77.7 wt% to 59.4 wt% Si02. Mafic pumice fragments become increasingly abundant higher in the section, whereas high-silica rhyolite pumice fragments are ubiquitous throughout. Similar trends were observed among glassy pumice fragments taken from the other stratigraphic section used in this study. These observations indicate that the portions of the Ammonia Tanks ash-flow sheet used in this study are normally zoned with respect to composition. TRACE ELEMENT CHEMISTRY Whereas major element concentrations in ash-flow sheets typically exhibit less than one order of magnitude in variation, trace element variations can exceed three orders of magnitude (Hildreth, 1979). Magma differentiation mechanisms are known to have a large effect on the distribution of trace elements in a magma chamber. The careful evaluation of trace element distributions in eruptive sequences may be used to infer which magma differentiation processes may have been operating in the magma chamber. Trace element concentrations for samples analyzed in this study are presented in Table 1. Figure 3 shows element enrichment factor diagrams, similar to those used by Hildreth (1979) except that points are plotted rather than vertical bars (see Wolff and Storey, 1984). Each point represents the ratio of element concentrations in two different pumice fragments. In each case, the numerator represents the sample that is more highly evolved with respect to silica. It is inferred that these higher silica samples represent magma that was erupted from the highly fractionated top of the magma chamber (Hildreth, 1979; Smith, 1979). This inference is supported by the observed enrichment of incompatible elements and depletion of compatible elements in high silica samples. Figure 3a is a plot of the concentrations of seventeen minor and trace elements in the most silicic pumice 22 23 IIjIIIjjIIIIIIIII LA 10" l 77.96% Si02 1 0" 59.44% Si02 10" @ IJJIIIJIIJIIJJIJJ mmmuCakEuTbeLu YHfchPufi' P ITIIIIIIIIIIIITI 0 10 , C ‘l . 1 66.95% SQ; ' ‘ 59.44% Si02 I I I J [D 10" ILII‘IIIIIIIIIIIHI ”mmuuamnnwvmabusp Figure 3. Enrichment factors for 17 minor and trace elements. (a) Element concentrations in the most silicic pumice fragment divided by concentrations in the most mafic pumice fragment. (5) Element concentrations in a pumice fragment of intermediate silica composition divided by concentrations in the most mafic pumice fragment. 24 fragment (AT5-3; 77.96% Si02) divided by their concentrations in the most mafic pumice fragment (ATS-75; 59.44% Si02). The elements are arranged along the x-axis according to the following scheme (from left to right): highly incompatible trace elements (Nb, Rb, Th), REES (including Y because of its similar behavior), compatible trace elements, in order of increasing compatibility in the high silica pumice (Hf, Sc, Zr, Ba, Sr), and lastly, P (to illustrate the control of apatite fractionation). It can be seen that Ba, Sr, P and Eu are highly depleted in the silicic sample, while Rb, Nb and Th are markedly enriched in this sample. The HREEs and Y are also enriched in the silicic sample, but the LREEs become progressively less enriched in the silicic sample as atomic number decreases. Sc, Hf and Zr are more enriched in the mafic sample. Figure 3b is a plot of the ratio of element concentrations in a pumice fragment of intermediate silica composition (ATS-71; 66.95% Si02) divided by the concentrations in the most mafic pumice fragment (AT5-75). A number of differences between this plot and the plot just described (figure 3a) are apparent. One important distinction is the behavior of the LREEs. La, Ce and Sm all exhibit incompatible trends in figure 3b, but compatible trends in figure 3a. This tendency toward increased compatibility of the LREEs in higher silica samples can be accounted for by the fractionation of a 25 LREE-bearing phase (see discussion below). Similarly, the increasing compatibility of Eu, Hf, Zr, Ba, Sr and P in figure 3a relative to figure 3b is attributable to crystal fractionation of plagioclase (Eu,Sr), sanidine (Ba,Sr), zircon (Hf,Zr) and apatite (P). Strongly incompatible elements such as Nb, Rb and Th are more enriched in high-silica samples (figure 3a) than in those of intermediate silica compositions (figure 3b). Figure 4 illustrates the chemical variation of selected REE’s for all analyses of the Ammonia Tanks Member. The ordinate values represent the ratio of the elemental concentration in the sample to the average concentration of that element in chondrites given by Haskin, et al., (1968). Although most of the elements show a fairly continuous distribution in composition, three compositional groups can be distinguished for the element europium. The well defined uppermost group (n=9) consists of mafic samples containing (67% Si02, and >2.5 ppm Eu. Samples in the middle group (n=6) were found to contain between 69% and 70.6% Si02 and 1-2 ppm Eu while those in the lowermost group (n=12) contain >73% Si02 and (0.7% Eu. (The compositional break between the middle and lower groups occurs at a normalized concentration of about 10). It is notable that both of the compositional breaks on the REE plot coincide with silica gaps that were noted in the major element chemistry discussion. The large negative Eu anomaly Sample/Chond. 26 I 0106'" j I T r1111. T TIIUUI'I 10 I I l lllll l J I ljllll l _LJ lJllll Lane SIEu In YbLu Atomic Number Figure 4. Chondrite-normalized REE abundances (log scale) versus atomic number. 27 seen in the high-silica rhyolite probably reflects the fractionation of plagioclase. In figure 5, various trace elements are plotted against Si02. The elements Ba, Zr, Sr, Sc and Eu all decrease with increasing Si02. These trends are consistent with crystal fractionation of plagioclase, sanidine, clinopyroxene and zircon. Conversely, Rb, Nb and Th behave as incompatible elements, increasing with Si02. Ce behaves incompatibly up until 67% Si02, after which it drops off rapidly in concentration. This behavior is best explained by the observation that the LREE-bearing phase chevkinite is present in all samples containing >67% Si02. La exhibits behavior identical to Ce. The leveling off of the Ce trend above 71% S102 may indicate a decrease in chevkinite fractionation, or it may indicate another phase (opx?) was actively partitioning the LREE’s between 69% and 71% Si02. Inflections in the Ba, Sr and A1203 trends at around 65% Si02 probably mark the initiation of sanidine fractionation. More significantly, the inflections at 67% 8102 on the Ba, Zr, Rb, Sr, Sc and Eu plots are coincident with inflections seen on major element plots of FeO, MgO, CaO, K20 and P205 v.s. 8102. These inflections may be related to the fractionation of mineral phases, or they may indicate the presence of a compositional discontinuity in the magma chamber associated with the silica gap observed between 67% and 69% Si02. Well defined changes 900 I I T I I I I I D - 700 r 00 - l- 0 c: L 500 L‘ g " (D .. .. 300 *- 1 i- 0 _ 0 100 - “3,8 00 1 I I I_ I I I I m 59 63 67 71 75 79 1200 b r I I I I I I r d - 00 .. 1000 b 0 _J 600 - o - L. .. .. N 600 '- 0 -‘ 400 *- §o ‘ 0 .. 200 : 0 00 .. __I I _L I L4 I L h— 59 63 67 71 75 79 5000 0 f T I r I I I T i— -i 4000 - 00 - - 0% a 3000 r " m _ j CD 2000 *- n 1000 - 0 1 .. $9 0 .. 0 I 1‘ I I L I I L w 59 53 67 . 71 75 79 8102 Figure 5. Variation of trace element abundances (ppm) versus silica (wt %) I q .. 000° .4 0500- .1 U) - @0 0 )- q 3.0 o 450 I I T I I I I I T )- 0 0 cl 350 - . h o d ,n 250 L 0 - u (DOOQ 959 t” 00 - 150 - o o - . r- 0 "l 50 .. a... J I I L L l I I L 59 63 67 71 75 79 4-0 I r I I I I I I I 00 ' o 00 3.0 P cab .J - 0 . :1 LL] 2.0 " ' go I 1.0 - 0 ~ .. 00 .- 0 I I I I I I J Ii 59 53 57 . 71 75 79 8102 Figure 5. (Continued) Variation of trace element abundances (ppm) versus silica (wt %) 250 I I I I I I I I 200 *- . I- o q a __ 0 a: 150 0 fig 100 b 0 o .. E°° 50 l J 1_JI 1 I J L_I 45 I I I I I I I I I .. (9.4 0 35 - fiq #- 0 - n 25 _ co 0 - Z 0 @o .. CD a 0 0 15 - c - 00 0 0 .- 5 L 1 _l l I l L l _L 59 63 67 71 75 79 35 I I I I I I r I g 30 - 0 o~ 0 25 - - .C 0 00 i— go 20 - o '- (53 15 - Q -' o 000 10 L l _L l_ L J 1 J L 59 53 57 0 71 75 79 8102 Figure 5. (Continued) Variation of trace element abundances (ppm) versus silica (wt %) 31 in slope also occur on the incompatible trace element plots (Rb, Nb, and Th) for Si02 compositions )76%. These inflections correspond to the gap in silica composition that occurs between roughly 73.8% and 76.8% Si02, possibly indicating the presence of a compositional discontinuity at a high level in the magma chamber. MINERALOGY Brief discussions of the variations in the chemistry of phenocrysts in whole rock samples of the Timber Mountain Tuff are given by Byers, et al., (1976b) and Broxton, et al., (1985). The mineralogy of only glassy pumice samples will be discussed here. Representative major element analyses of phenocrysts from eight glassy pumice samples are presented in Table 2. All mineral analyses were made at Lawrence Livermore National Laboratory using a JEOL 733 electron microprobe. Systematic variations in phenocryst assemblages, chemistry and modal percentages are observed in Ammonia Tanks pumice fragments as 8102 increases, reflecting the overall compositional evolution of the magma chamber. Plagioclase phenocrysts from 0.5 to 4 mm long occur in all glassy pumice samples and are the dominant mineral phase in samples containing (67% Si02. They are typically euhedral to subhedral and commonly exhibit resorption textures, particularly in samples containing (69% Si02. Many of the phenocrysts are broken. Plagioclase becomes 32 TABLE 2. REPRESENTATIVE MINERAL ANALYSES (Weight percent oxides) PLAGIOCLASE Sample ATS-35 ATS-69 ATS-63 AT5-69 ATS-75 Mineral ID 27 48 39 ‘50 91 Si02 63.95 58.75 58.11 54.20 49.10 A1202 22.59 24.34 26.09 29.03 32.91 Fe0 0.26 1.12 0.37 0.42 0.29 MgO 0.00 0.00 0.01 0.04 0.03 CaO 2.77 5.02 7.07 9.78 14.47 N320 8.71 7.57 6.67 5.63 2.90 K20 1.72 1.86 1.25 0.61 0.19 Ba0 0.00 0.30 0.32 0.19 0.00 TOTALS 100.00 98.96 99.89 99.90 99.89 SANIDINE KNEBELITE Sample ATS-35 ATS-71 AT5-71 ATS-3 ATS-3 Mineral ID 28 64 63 04 05 Si02 65.48 64.08 63.38 30.93 30.63 A1203 19.29 20.37 21.62 0.00 0.04 FeO 0.09 0.25 0.35 49.13 45.44 MgO 0.00 0.00 0.00 0.00 0.00 CaO 0.38 0.90 2.49 0.02 0.10 Na20 5.16 5.78 7.66 0.00 0.03 K20 9.33 8.19 4.35 0.00 0.02 Ba0 0.25 0.26 0.07 0.04 0.03 T102 0000 0010 MnO 19.86 23.60 Ni0 0.02 0.01 TOTALS 99.98 99.83 99.92 100.00 100.00 33 TABLE 2. (Continued) BIOTITE HORNBLENDE Sample ATS-3 ATS-35 ATS-72 ATS-75 Mineral ID 03 25 74 90 8102 41.96 40.58 37.23 42.33 A1203 13.12 13.67 15.67 11.73 FeO 12.73 15.06 14.05 11.67 M30 18.90 16.56 14.92 14.40 C80 0.19 0.28 0.01 11.64 N820 0041 0058 0075 2063 K20 8.55 8.00 7.47 1.21 T102 3.19 4.59 6.99 3.99 MnO 0.90 0.43 0.28 0.36 P205 0.00 0.00 0.00 0.00 380 0.04 0.24 2.63 0.05 NiO 0.00 0.00 0.00 0.00 TOTALS 99.99 99.99 100.00 100.01 CLINOPYROXENE ORTHOPYROXENE Sample ATS-58 ATS-71 ATS-75 ATS-58 Mineral ID 10 58 86 13 8102 52.66 52.66 51.86 52.56 FeO 9.08 9.21 7.94 18.90 M30 12.69 13.72 15.90 23.83 C80 21.12 20.18 20.31 1.18 N820 0.69 0.73 0.65 0.05 K20 0.02 0.00 0.00 0.02 T102 0020 0039 0047 0019 MnO 1.91 1.54 0.69 2.31 p205 0000 0000 0000 0000 380 0.44 0.08 0.04 0.00 N10 0.00 0.02 0.00 0.20 TOTALS 99.99 100.00 100.00 100.00 TABLE 2. Sample Mineral ID Crzoa A1203 MgO FeO T102 MnO TOTALS Sample Mineral ID CP203 A1202 MgO FeO T102 MnO TOTALS NOTE: ATS-3 08 0.03 0.64 0.71 90.57 4.92 3.12 99.99 ATS-3 09 0.00 0.06 1.56 44.77 46.91 6.71 100.01 All analyses 34 (Continued) MAGNETITE ATS-58 ATS-63 31 42 0.05 0.00 1.45 2.85 1.42 3.76 85.03 80.58 8.98 11.81 3.07 1.01 100.00 100.01 ILMENITE ATS-35 21 0.00 0.00 1.64 51.94 43.05 3.36 99.99 ATS-72 82 0.06 1.98 2.30 74.30 20.03 1.31 99.98 have been normalized to 100% 35 progressively more sodic with increasing silica content. The mole fraction of anorthite varies between Anna and Anas in the rhyolite and high-silica rhyolite, whereas in the quartz latite the variation is from An21 to An22. The average plagioclase composition is around Anaa. Zoned crystals with more calcium—rich cores are common, particularly at whole pumice compositions below 69% Si02. Sanidine first appears in samples containing 61% Si02, and is the predominant phenocryst phase in samples with >67% Si02. Sanidine crystals are euhedral to subhedral, ranging in size from 0.5 to 4 mm. Resorption textures and fracturing are common, but less prominent than in plagioclase. The compositional variability throughout the ash-flow sheet is small, ranging from Orig to Or52, with a variation in the celsian mole fraction ranging from Cno to qu. Sodic overgrowths on more calcium-rich plagioclase cores may contain up to 24 mole percent 0r. Quartz occurs only in the high-silica rhyolite, where it is fairly abundant as euhedral phenocrysts ranging from 2 to 4 mm in length. Resorption textures were not noted, but some grains are fractured. Magnetite is present in all samples, ranging from 0.1-0.5 mm euhedral grains in the rhyolite and high-silica rhyolite, to large clumps of subhedral grains up to 1 mm in size in the quartz latite. Ilmenite is at least an order of magnitude less abundant than magnetite. In the quartz latite, magnetite and ilmenite often occur with, and are 36 included in the ferromagnesium minerals, particularly pyroxenes. In contrast, the ilmenite and magnetite in the more rhyolitic samples occur in smaller, more isolated groups of grains, or individually, often associated with zircon, chevkinite and apatite. Magnetite and ilmenite are increasingly enriched in MnO with increasing silica. Biotite is the most abundant ferromagnesium silicate in all samples, occuring as euhedral, 0.5 to 3 mm phenocrysts. It contains up to 3.5% BaO in the more mafic samples, and becomes progressively more enriched in MgO with increasing Si02. MgO/MgO+FeO in biotite varies from 0.52 to 0.60. Clinopyroxene is found in all samples examined containing (73% Si02. It occurs as blocky, euhedral to subhedral phenocrysts, generally <1 mm in length. Mg-rich orthopyroxene is found in pumice samples containing between 69% and 73% Si02. It is found associated with opaque oxides and trace phases, typically as euhedral, yellowish- green prismatic crystals, 1 mm or less in size. Orthopyroxene compositions range from Ens? to Enos, and Clinopyroxene ranges from WoaaEnsana to W044En22Fsl2. Many Clinopyroxene crystals are slightly zoned, typically with more Mg-rich cores and Fe-rich edges. Trace amounts of hornblende occur in sample AT5-75 (59.44% Si02) as 1 mm partially resorbed crystals intergrown with plagioclase. They are not believed to be cognate to the Ammonia Tanks. Trace amounts of the minerals apatite, zircon, sphene, chevkinite and knebelite are also found in the Ammonia Tanks 37 ash-flow sheet. Apatite is most abundant in the quartz latite occuring a stubby, six—sided euhedral crystals up to 0.5 mm in size. It is often found as poikilitic grains in silicate phases. Apatite gradually becomes smaller and more acicular as the whole pumice silica composition increases. Zircon is present in all thin sections as 0.1 mm, euhedral crystals. Zircon tends to occur in biotite phenocrysts in the quartz latite, but is more often associated with magnetite and chevkinite in rhyolitic samples. Sphene is found only in the high-silica rhyolite as small ((1 mm), isolated, wedge-shaped crystals. Chevkinite, a LREE-rich silicate mineral, is present in all samples examined containing >67% Si02. It occurs as dark reddish-brown, subequant grains up to 0.1 mm in size, associated with the opaque Fe-Ti oxides. Knebelite, the Mn-rich analogue of fayalite, occurs as minute, needle-like crystals in the high—silica rhyolite. Knebelite normally occurs in rocks associated with Fe-Mn ore deposits or their metamorphic equivalents (Deer, et al., 1962). To the author’s knowledge, this is the first reported occurence of this mineral in ash-flow sheets. EVALUATION OF CRYSTAL FRACTIONATION Crystal fractionation can be quantitatively evaluated using two types of tests. Major element multiple linear regressions are used to determine the percentages of mineral phases that must be fractionated from the parent 38 magma to produce the derivitive magma. Trace element data is then used to test the validity of the major element regression. Maximum and minimum mineral/liquid distribution coefficients (Kn values) are used to calculate a predicted range of trace element concentrations in the derivitive liquid based on the amount of crystal fractionation predicted by the major element regression. If the model is valid, then the trace element concentrations observed in the derivitive magma should fall within the predicted range. The major element multiple linear regressions were initiated by choosing the least evolved pumice sample (ATS-75, 59.44% Si02) and regressing it against a slightly more evolved sample (ATS-63, 62.11% 8102) plus all potential combinations of major phenocryst phases. If individual phase chemistries showed a range in composition, then more than one composition was tested in the regression. Once a reasonable regression model was determined for the first step, then the former derivitive magma (AT5—63 in step 1) became the new parent magma, and the process was repeated with a new, more silicic derivitive magma. A total of six major element multiple linear regressions were used to model the entire compositional evolution exhibited by Ammonia Tanks pumice fragments (see Table 3). In general, the best fit model with the smallest sum of the squares of the residuals was chosen 39 for each step. One exception was made with the first step in the regression analysis. The model for step 1 that had the lowest residuals (r2 = 0.085) also predicted large amounts of hornblende fractionation. This seemed geologically unreasonable for three reasons: (1) the parent magma (ATS—75) contains only trace amounts of hornblende, and these crystals typically exhibit resorption textures in thin section; (2) ATS-75 is the only thin section examined from the entire Ammonia Tanks suite that contained hornblende; (3) Clinopyroxene fractionation occurs in steps 2-5 in the regression model, but not step 1. The fractionation of hornblende before Clinopyroxene seems unlikely. Therefore, Clinopyroxene was substituted for hornblende in step 1 of the major element regressions. Although the sum of the squares of the residuals are larger (r2 = 0.1576) than if hornblende were used, they are still considerably less than 1.0. Furthermore, the trace element data (see discussion below) fit the Clinopyroxene fractionation model much better than the hornblende model. Feldspar fractionation dominates each step in the major element regression model. Plagioclase is predicted to be the major fractionating phase between 59% and 67% Si02 whereas sanidine becomes dominant above 67% Si02. Clinopyroxene, biotite and magnetite make up the balance of the fractionating phases. The phenocryst compositions used in each step of the model are give in Table 4. 40 TABLE 3. MAJOR ELEMENT MULTIPLE LINEAR REGRESSIONS STEP 1 PARENT = ATS-75 (59.44% 8102) DAUGHTER = ATS-63 (62.11% 8102) ATS-75 : 0.0869 + 0.848 ATS-63 + 0.0923 PLAG + 0.0316 CPX + 0.0175 MT R2 = 1.00 SS : 0.1576 STEP 2 PARENT = ATS-63 (62.11% 8102) DAUGHTER ATS-72 (64.00% S102) ATS-63 : 0.0890 + 0.896 ATS-72 + 0.0614 PLAG + 0.0212 CPX + 0.0096 MT R2 = 1.00 SS : 0.1321 STEP 3 PARENT : ATS-72 (64.00% 8102) DAUGHTER = ATS-71 (66.95% 8102) ATS—72 = -0.0091 + 0.764 ATS—71 + 0.193 PLAG + 0.0285 CPX + 0.0161 MT R2 = 1.00 SS = 0.1814 STEP 4 PARENT = ATS-71 (66.95% 8102) DAUGHTER - ATS-58 (69.25% S102) ATS-71 = 0.0225 + 0.719 ATS-58 + 0.22 KSP + 0.0396 PLAG + 0.00888 CPX + 0.00896 MT R2 = 1.00 SS : 0.07865 41 TABLE 3. (Continued) STEP 5 PARENT = ATS-58 (69.25% 8102) DAUGHTER = ATS-35 (72.96% 8102) ATS-58 : 0.0547 + 0.672 ATS-35 + 0.294 KSP + 0.0109 PLAG + 0.0181 CPX + 0.0144 MT R2 : 1.00 SS : 0.004290 STEP 6 PARENT = ATS-35 (72.96% 8102) DAUGHTER = AT5-3 (77.96% 8102) ATS-35 : -0.0067 + 0.696 ATS-3 + 0.0503 PLAG + 0.231 KSP + 0.0176 BIOT + 0.00607 MT R2 = 1.00 SS = 0.06226 NOTE: These regressions were also calculated with a zero intercept (see Appendix 4). This removes the Bo coefficient on the right hand side of the equation and results in somewhat greater sums of the squares of the residuals (SS). 42 TABLE 4. MINERAL ANALYSES USED IN REGRESSIONS FOR CRYSTAL FRACTIONATION MODEL (mineral analyses are from the parent magma) PLAGIOCLASE Sample ATS—75 ATS-63 ATS—72 AT5—71 ATS-58 Step 1 2 3 - 4 5 Mineral ID 92 38 77 61 15 S102 55.17 58.15 58.84 57.77 61.28 A1203 28.86 26.26 25.33 26.51 23.70 FeO 0.20 0.32 0.84 0.40 0.38 MgO 0.05 0.01 0.00 0.03 0.00 CaO 9.65 6.67 6.57 6.84 4.68 N820 5.25 6.78 6.77 6.94 8.34 K20 0.54 1.27 1.38 1.01 1.40 BaO 0.18 0.34 0.25 0.39 0.10 TOTALS 99.90 99.80 99.98 99.89 99.88 PLAGIOCLASE SANIDINE Sample ATS-35 AT5—71 AT5-58 AT5-35 Step 6 4 5 6 Mineral ID 27 65 14 28 8102 63.95 65.21 64.34 65.48 A1203 22.59 19.89 20.01 19.29 FeO 0.26 0.24 0.27 0.09 MgO 0.00 0.00 0.00 0.00 CaO 2.77 0.64 0.52 0.38 Na20 8.71 4.84 5.77 5.16 K20 1.72 9.17 8.59 9.33 BaO 0.00 0.00 0.38 0.25 TOTALS 100.00 99.99 99.88 99.98 43 TABLE 4. (Continued) CLINOPYROXENE Sample ATS—75 AT5-63 ATS-72 ATS—71 Step 1 2 3 4 Mineral ID 86 35 72 57 8102 51.86 52.29 53.52 52.95 A1203 2.14 2.13 2.40 2.06 FeO 7.94 7.52 7.89 7.86 MgO 15.90 15.49 15.45 15.02 CaO 20.31 20.76 18.67 20.16 N820 0.65 0.58 0.59 0.64 K20 0.00 0.01 0.00 0.00 T102 0.47 0.58 0.52 0.48 MnO 0.69 0.55 0.93 0.80 P205 0.00 0.00 0.00 0.00 BaO 0.04 0.09 0.03 0.00 N10 0.00 0.00 0.00 0.03 TOTALS 100.00 100.00 100.00 100.00 MAGNETITE sample ATS-75 ATS—63 ATS-72 ATS-71 Step 1 2 3 4 Mineral ID 97 42 83 69 Cr203 0.04 0.00 0.03 0.09 A1202 3.61 2.85 2.43 1.47 MgO 3.25 3.76 2.77 1.66 FeO 84.20 80.58 81.55 79.54 T102 8.05 11.81 11.75 15.35 MnO 0.85 1.01 1.47 1.90 TOTALS 100.00 100.01 100.00 100.01 ATS-58 12 52.54 1.17 8.26 13.64 21.94 0.76 0.00 0.28 .35 .00 .05 .00 OOOH 99.99 ATS—58 32 0.03 1.01 1.27 87.65 7.75 2.29 100.00 TABLE 4. (Continued) BIOTITE Sample ATS-35 Step 6 Mineral ID 25 3102 40.58 A1203 13.67 FeO 15.06 MgO 16.56 CaO 0.28 N820 0.58 K20 8.00 T102 4.59 MnO 0.43 BaO 0.24 TOTAL 99.99 Sample Step Mineral ID Cr203 A12 03 MgO FeO T102 MnO TOTAL MAGNETITE ATS-35 6 20 0.02 0.82 0.73 90.14 6.46 1.83 100.00 NOTE: All analyses have been normalized to 100% 45 As a second test, these major element regressions were evaluated using trace element data. Ideally, mineral/ liquid distribution coefficients for trace elements used in the model should be calculated for the system of interest. However, this is not always possible because it requires handpicking enough phenocryst and pure glass separates for the chemical analysis of each. Time limitations prevented the determination of distribution coefficients for the Ammonia Tanks. The alternative is to use distribution coefficients that have already been determined for other systems of similar chemistry and inferred P-T conditions. Unfortunately, distribution coefficients often vary significantly from system to system, even for systems of similar chemistry. Therefore every effort was made to examine all of the recent literature that provided compilations of distribution coefficients for intermediate- to high-silica systems to insure that the maximum permissible range in Kn values would be utilized. Distribution coefficient data was compiled (Table 5) from Arth and Hanson (1975), Arth (1976), Henderson (1982), Mahood and Hildreth (1983), Hyndman (1985), and Nash and Crecraft (1985). The trace elements selected to test the major element regression model are Rb, Sr, Ba, Sc, Ce, Yb and Th. In general, incompatible trace elements exhibiting a wide range in composition, with small measured uncertainties in concentration are best suited for the modeling process. 46 The availability of distribution coefficient data was the limiting factor in the element selection process. As a result, some of the trace elements used in the model exhibit narrow compositional ranges while others are compatible in one or more mineral phases. The results of the trace element tests are shown in Table 6. The predicted range of element concentrations are Rayleigh fractionation values. As stated before, for this data to be consistent with the crystal fractionation trends predicted by the major element regressions, the observed trace element concentrations should fall within the fractionation window predicted by the maximum and minimum trace element distribution coefficients. Some of the lack of agreement between the calculated and observed trace element data shown in Table 6 can be explained by taking into account trace mineral phases that strongly partition certain elements. For instance, the mineral chevkinite, which is present in all samples containing >67% S102, readily accepts the REE’s and Th into it’s crystal lattice, thereby causing the melt to become depleted in these elements. If chevkinite is excluded from the fractionation model, it would cause the predicted concentrations of Ce, Yb and Th to be higher than they would be if the model included chevkinite. This is precisely what is observed in steps 4, 5, and 6 of the model; the fractionation windows all overpredict the observed element concentrations. By applying the TABLE 5. DISTRIBUTION PLAGIOCLASE max min Sc 0.20 0.01 Rb 0.46 0.016 Sr 33 1.45 Ba 3.3 0.3 Ce 0.4 0.11 Yb 0.3 0.049 Th 0.09 0.01 BIOTITE max min Sc 20.0 4.9 Rb 5.3 2.24 Sr 0.53 0.12 Ba 36 3.7 Ce 11.0 0.037 Yb 3.0 0.179 Th 2.0 0.27 Data compiled from Arth and Hanson (1975), Henderson (1982), Hyndman (1985), 47 COEFFICIENTS SANIDINE max min 0.059 0.01 2.4 0.11 26 3.6 24 1.0 0.095 0.02 0.04 0.005 0.03 0.0095 MAGNETITE max min 15.6 1.5 0 0 0.08 0.05 22.9 0.28 2.2 0.2 5.04 0.04 Mahood and Hildreth (1983), and Nash and Crecraft (1985). CLINOPYROXENE max min 172 18 0.09 0.032 0.516 2.3 0.131 21.6 0.36 11.6 1.14 6.56 0.01 Arth (1976), TABLE 6. TRACE STEP PARENT DAUGHTER Sc predicted observed Rb predicted observed Sr predicted observed Ba predicted observed Ce predicted observed Yb predicted observed Th predicted observed 75-79 75 1 35-811 615 2 3641-5288 3990 x 75-245 228 * 1.7-2.7 2.4 x 9.6-13.3 12.1 t * Indicates good fit 48 ELEMENT MODELS 2 ATS-63 ATS-72 0.14-5.0 6.3 80-82 80 1 80-610 524 * 3373-4298 3560 8 120-247 238 * 2.0-3.2 2.4 x 1206-1704 13.3 x 3 ATS-72 ATS-71 0.02-4.5 6.4 95-104 89 0.5-496 164 x 2090-4352 1168 93-300 393 200-302 2.4 * 1206-1704 19.6 TABL STEP PARE DAUG Sc Rb Sr Ba Ce Yb Th E 6. (Continued) NT HTER predicted observed predicted observed predicted observed predicted observed predicted observed predicted observed predicted observed ATS-71 AT5-58 1. 7 .4 2 .2 3 x 65-120 122 0.06-83 138 2.8-1230 709 3 328—534 222.89 .2 2.8— 2 23.7-27.0 20.1 Indicates good fit 49 5 ATS-58 ATS-35 0.1-3.9 2.3 * 77-177 164 * 0.01-56 88 0.2-742 394 t 134-329 129.90 3.0—3.9 2 7 2307-3002 24.3 * 6 AT5-35 ATS-3 109-300 1.3 106—218 216 * 0.01-43 7 t 0.3-390 74 * 120-184 58.48 50 Rayleigh fractionation law (see Cox, et al., 1979, p.346) it was possible to estimate the amount of chevkinite crystallization needed to bring the predicted element concentrations into agreement with the observed concentrations (see appendix 3). It was found that by fractionating only 0.2% chevkinite, the predicted Ce, Yb and Th concentrations can all be made to agree with the observed concentrations of these elements in the melt. This amount of chevkinite fractionation does not seem unreasonable based on the observed abundance of chevkinite in thin section, and it generally accounts for the discrepancy between the predicted and observed concentrations of Ce, Yb and Th in steps 4, 5, and 6 cf the model. Fractionation of the minerals apatite and zircon can be used to account for the chemical variations exhibited by the elements Zr, Hf and P in a similar manner. Several factors other than the fractionation of trace phases may have also adversely affected the results of the trace element model. A potentially significant source of error arises from the fact that the distribution coefficients used in the trace element tests were not calculated for the Ammonia Tanks magma system. As previously discussed, the use of inappropriate distribution coefficients can result in a lack of agreement between the predicted and observed trace element concentrations. 51 As an example, consider the partitioning of strontium into plagioclase and sanidine. The distribution coefficients used in the model (see Table 5) indicate that Sr is highly compatible in the feldspar minerals. However, in steps 4 and 5 of the trace element model, the predicted range for Sr concentrations is much less than the observed Sr concentrations. There are several possible interpretations for this. It may be that the Kp’s correctly predict the the behavior of Sr, and the model simply fails for these steps. Alternatively, it is possible that the distribution coefficients overpredicted the partitioning of Sr into the feldspars. By using smaller distribution coefficients, the predicted and observed concentrations might have concurred. Another potential explanation is that not all of the crystals formed in the parent liquid were actually separated from the residual liquid. In other words, it could be that fewer crystals were separated from the liquid than was predicted by the major element regression. Analytical error is another factor that may have affected the results of the model. Trace elements that are present in low concentrations, or those that exhibit narrow compositional ranges may exhibit discrepancies between the predicted and observed concentrations due to analytical error. This problem will be compounded by the use of inappropriate distribution coefficients. 0f the elements used in the trace element model, Sc and Yb would 52 be the most susceptible to this type of error because they are present in concentrations that are near the detectibility limit. Finally, an important factor that must be considered is the possibility that other magma differentiation processes acted in conjunction with fractional crystallization. Several independent lines of evidence indicated that magma mixing may have taken place between the time of eruption of the Rainier Mesa Member and the Ammonia Tanks Member. Since magma mixing can easily be evaluated, it is logical that the possibility of magma mixing in the Ammonia Tanks system should be explored before drawing any conclusions based exclusively on the crystal fractionation model. EVALUATION OF MAGMA MIXING The potential for generating compositional variations due to magma mixing has been demonstrated for a number of silicic systems (cf. Reid, et al., 1983; Vogel, et al., 1984). In ash flow sheets, the presence of disequilibrium phenocryst populations within individual pumice fragments is a strong indication that magma mixing may have occurred (Vogel, et al., 1987). Disequilibrium phenocryst textures were observed to some extent in all thin sections of Ammonia Tanks pumice fragments. These textures are strongly developed among the low-silica pumice fragments, particularly in samples that contain between 59% and 69% 53 S102. Plagioclase grains exhibit the most pronounced disequilibrium textures at all silica compositions. In pumice fragments containing (69% 8102, many plagioclase grains are deeply embayed, and sieve-like textures are widely developed. Resorption textures are also common in Clinopyroxene grains. Biotite and the opaque Fe-Ti oxides appear to be affected far less than the feldspars and pyroxene. In pumice fragments containing >69% 8102, only the feldspars exhibit disequilibrium textures, primarily in the form of minor embayments along grain boundaries. Not surprisingly, microprobe analyses show that the feldspar and Clinopyroxene grains exhibit the greatest compositional heterogeneities among phenocrysts within any given glassy pumice fragment. It is notable that banded pumice were sometimes found in the Ammonia Tanks Member, particularly in the stratigraphic horizons where mafic pumice were predominant. Most of these banded pumice were composed of phenocryst- rich orangish-brown glass that was streaked with phenocryst-poor white glass. While this is not prima facie evidience of magma mixing, it does indicate some commingling of heterogeneous magmas. However, it is uncertain if the commingling occured in the magma chamber, or during eruption. As previously discussed, apparent gaps are observed in the silica composition of the Ammonia Tanks ash-flow sheet. 54 These gaps occur between 67% and 69% 8102, between 70.7% and 73% $102, and again between 73.8% and 76.8% 8102, possibly indicating the presence of sharp compositional interfaces between magma bodies of contrasting composition. It was hypothesized that the magma represented by intermediate SiO2 compositions may have formed from mixing of the high-silica rhyolite (>76% 8102) and low— silica quartz latite (<67% SiO2) magmas. Magma mixing can be readily evaluated using element-element and ratio- ratio plots (Langmuir, et al., 1977). Data points representing mixed magmas should plot along a straight line on x-y element-element plots, and along a hyperbolic curve on x—y plots of element ratios (A/B v.s. C/D). A number of element-element plots were made using various trace elements, but strictly linear trends were rarely seen. Ratio—ratio plots of incompatible elements also failed to produce the hyperbolic curve expected if magma mixing had been the dominant magma differentiation process. However these tests did not preclude the possibility that magma mixing acted in conjunction with crystal fractionation. Magma mixing plus crystal fractionation can be evaluated using major element multiple linear regressions in much the same way as crystal fractionation alone is evaluated. The only difference is that the regression equation is now set up so that the hybrid magma is equal to the sum of the two mixed magmas minus the fractionated 55 phenocryst phases. Several different pumice fragments of intermediate silica composition were tested as potential candidates for a mixed + fractionated magma. The best result obtained was: ATS-35 = —0.0442 + 0.539 ATS-71 + 0.513 ATS-3 - 0.0372 KSP -0.00754 PLAG - 0.00266 CPX SS = 0.04616 R2 = 1.00 If the regression equation is restricted to mixing of end- member compositions (i.e. the most primitive and most evolved pumice fragment compositions) then the best result was: AT5-35 : 0.022 + 0.501 ATS-75 + 0.704 ATS-3 - 0.0584 CPX -O.135 PLAG - 0.014 MT SS = 0.2469 R2 = 1.00 These regressions were also calculated with a zero intercept for comparative purposes (results are given in Appendix 4). Phenocryst compositions used in these models are given in Table 7. These results are consistent with the interpretation that crystal fractionation along with magma mixing can explain the chemical and mineralogical trends observed in the Ammonia Tanks ash-flow sheet. 56 TABLE 7. MINERAL ANALYSES USED IN REGRESSIONS FOR CRYSTAL FRACTIONATION + MAGMA MIXING MODEL (mineral analyses are from the derivitive magma) A. Derivitive magma = ATS-35 (72.96% 8102) Mixed magmas : ATS-71 (66.95% 8102) ATS-3 (77.96% 8102) PLAGIOCLASE SANIDINE CLINOPYROXENE Mineral ID 27 28 22 8102 63.95 65.48 52.75 A1203 22.59 19.29 0.81 FeO 0.26 0.09 9.01 MgO 0.00 0.00 13.23 CaO 2.77 0.38 21.43 Na20 8.71 5.16 0.67 K20 1.72 9.33 0.02 BaO 0.00 0.25 0.09 TiOZ 0020 MnO 1.76 NiO 0.04 TOTALS. 100.00 99.98 100.01 B. Derivitive magma = AT5-35 Mixed magmas = ATS-75 (59.44% S102) ATS-3 (77.96% 8102) Note: The plagioclase composition used in this model is the same as the composition given above. CLINOPYROXENE MAGNETITE Mineral ID 23 Mineral ID 20 8102 52.50 Cr203 0.02 A1202 0.93 A1202 0.82 FeO 7.74 MgO 0.71 MgO 14.96 FeO 90.14 CaO 22.52 T102 6.46 Na20 0.45 MnO 1.83 K20 0.00 BaO 0.00 TOTAL 99.98 T102 0.20 MnO 0.70 N10 0.00 TOTAL 100.00 DISCUSSION AND CONCLUSIONS This study was aimed at determining whether the most evolved high-silica rhyolite present in the Ammonia Tanks Member of the Timber Mountain Tuff can be modeled as originating from the quartz latitic magma by fractional crystallization. The results of a series of six major element multiple linear regressions strongly supports the proposed model. The regressions predicted plagioclase and magnetite fractionation throughout the compositional evolution of the magma. Clinopyroxene fractionation was predicted for the first five steps in the model (up to 73% 8102), after which biotite fractionation was predicted. Sanidine fractionation was important in the last three steps of the model (above 67% S102). These predictions seem geologically reasonable based on petrographic analyses of the samples used in the model. Evaluation of the major element multiple linear regressions using trace element data produced largely favorable results (Table 6). Much of the lack of agreement between the predicted and observed trace element concentrations can be accounted for by the fractionation of mineral phases that were not included in the regression model. As previously discussed, fractionation of the mineral chevkinite can account for discrepancies between the predicted and observed concentrations of Ce, Yb and Th in steps 4, 5 and 6. Similarly, zircon and sanidine fractionation can account for inconsistencies in Th and 57 58 Ba concentrations, respectively, in step 3. The use of inappropriate distribution coefficients can also result in a lack of agreement between predicted and observed trace element concentrations. For example, the model can overpredict the partitioning of an element into certain mineral phases if the distribution coefficients used are too large. This may have been the case for Sc partitioning in Clinopyroxene and magnetite (steps 1, 2 and 3) and Sr partitioning in plagioclase and sanidine (steps 4 and 5). Discordant predicted and observed Rb concentrations (steps 3 and 4) may be due to the fact that Rb, like Na and K, is highly susceptible to remobilization due to the secondary hydration of pumice fragments. Rb is also strongly partitioned by biotite. Although biotite was observed to be present in all Ammonia Tanks samples examined, it was not included in the first five steps of the major element multiple linear regression model because it did not improve the statistical results. Thus, the fractionation of small amounts of biotite in these steps might also account for the discrepancy in Rb concentration. The presence of disequilibrium phenocrysts is not explained by the crystal fractionation model. Therefore, the possibility that the high-silica rhyolite and quartz latite magmas mixed to form a magma of intermediate composition was tested. Major element multiple linear 59 regression tests demonstrated that Ammonia Tanks magma of rhyolitic composition (73% 8102) can originate from a combination of magma mixing and crystal fractionation. The best fit mixing + fractionation regression model predicts the mixing of subequal proportions of quartz latite (ATS-71) and high-silica rhyolite (ATS-3) and the fractionation of 3.4% sanidine and (1% plagioclase and Clinopyroxene. Modal\phenocryst abundances in the derivitive magma (ATS-35) are roughly proportional to the quantities expected based on the calculated percentages of mixing and fractionation. Magma mixing most likely occurred due to the disruption of compositional zonations in the magma chamber resulting from the eruption of the Rainier Mesa Member of the Timber Mountain Tuff. Quartz latite and high-silica rhyolite pumice fragments of similar composition to those used in the mixing + fractionation model occur together at the top of the Rainier Mesa ash-flow sheet (Mills and Rose, 1986). This indicates that neither of these magma types were completely evacuated during the eruption of the Rainier Mesa Member. Therefore, it is inferred that the turbulence brought about by the Rainier Mesa eruption resulted in the mixing of at least some of the unerupted volumes of high-silica rhyolite and quartz latite to form a rhyolitic magma. Although these results demonstrate that magma mixing can act in conjunction with crystal fractionation to 60 produce the observed compositional trends, this does not prove that there is any correlation between the observed compositional gaps in silica and the mixing event. Changes in slope on some trace element plots v.s. 8102 can be correlated with some of the silica gaps; while many of these inflections can be attributed to crystal fractionation, there are notable exceptions. For instance, there is no ready explanation for the change in slope that occurs at 75% $102 on plots of highly incompatible elements (Rb, Nb, Th) v.s. S102 (see Figure 5). If these inflections are related to the proposed mixing event, then it may be that they are observed on incompatible element plots simply because of the minimal effect that crystal fractionation has on these elements. A larger data base could be used to determine if the observed silica gaps really exist, or if they are an artifact of sampling bias. However, it seems likely that the observed inflections in chemical trends must ultimately be related to some magma differentiation process. Based on the results of this study, it appears that a combination of crystal fractionation with subordinate magma mixing provides a plausible explanation for these trends that is consistent with the data. To summarize, it has been demonstrated that the high-silica rhyolite of the Ammonia Tanks Member of 61 the Timber Mountain Tuff can be modeled as originating from the quartz latite as a result of fractional crystallization. However, to explain the presence of pervasive disequilibrium phenocrysts in these rocks, it was necessary to include an additional magma differentiation process in the model. Magma mixing was combined with crystal fractionation to give a model that is consistent with both the chemical data and the textural observations. Crystal fractionation is favored as the dominant magma differentiation mechanism because it can be used to model the chemical evolution of the Ammonia Tanks magma system by itself whereas magma mixing cannot. APPENDICES Samples AT4-14 to AT4-75 ATS-2 to ATS-79 APPENDIX 1 SAMPLE LOCATIONS Location NE 1/4, SE 1/4, S 1/2, Silent Butte 7.5 Min. Quad. Approximately 0.7 km north of Pahute Mesa Rd., near South Silent Canyon SE 1/4, SE 1/4, S 1/2, Silent Butte 7.5 Min. Quad. 1 km southeast of Pahute Mesa Exploratory Hole 1 along Pahute Mesa Rd. 62 APPENDIX 2 SAMPLE PREPARATION TECHNIQUES A. To leach secondary carbonate minerals from pumice samples. For each sample, prepare the following solution: (1) measure 41 g. of sodium acetate (2) add 13.5 ml. of glacial acetic acid (3) add enough distilled water to make 500 ml. The sample should be crushed to no larger than pea- sized fragments. Add the crushed sample to the solution and and boil for 30 minutes, stirring occasionally. Allow to settle, then flush four times with distilled water, centrifuging between rinses. B. To prepare glass wafers for XRF analysis Combine the following in a Pt-Au alloy crucible: (1) 9.0000 g. lithium tetraborate (2) 1.0000 g. finely crushed and leached sample (3) 0.160 g. ammonium nitrate Carefully homogenize the contents of the crucible. Heat this mixture to ~1100°C with a bunsen burner for 30 minutes while gently shaking the crucible. A homogeneous molten liquid should result. Pour the liquid into a hot Pt-Au alloy mold and allow to cool on a hot plate. 63 APPENDIX 3 A METHOD FOR THE DETERMINATION OF TRACE PHASE FRACTIONATION PERCENTAGES The Rayleigh fractionation law is given by Cox, et al., (1979) as CL/Co = F‘D’li (1) where CL conc. of element in derivitive liquid Co initial concentration of element in the parent magma F = proportion of original liquid remaining D = bulk dist coeff : 8wiKn. Wi = the wt. % of the 1‘“ mineral phase Kn. = the mineral/liquid partition coefficient for an element in the it“ mineral phase Taking logs of both sides of equation (1) and rearranging gives the expression log F + log (CL/Co) = D log F (2) which can be solved for D if F, 02 and Co are given. As an example of how to determine the percentage of chevkinite fractionation needed to bring the predicted and observed trace element concentrations into agreement, consider the partitioning of Ce in step 4 of the fractionation model. The observed Ce concentrations are: CL Co 222.89 ppm Ce (in ATS-58) 393.42 ppm Ce (in ATS-71) The amount of crystallizaion predicted by the major element regression in step 4 was 27.74%. Therefore, Using equation 2, D is readily determined to be 2.749. 64 Although Kn values were not available for chevkinite, Mahood and Hildreth (1983) give Kn’s for Ce, Yb and Th in the mineral allanite (Kn for Ce in allanite = 2063 i 30). Allanite and chevkinite are similar enough in structure that we can assume they will partition the REE’s and Th in a similar manner. This allows for reasonable estimates of the degree of chevkinite fractionation. For step 4, h we can write D 2 (wt % sanidine)(Kn Ce in sanidine) + (wt % plag) x (Kn Ce in plag) + (wt % cpx)(Kn Ce in cpx) + (wt % magnetite)(Kn Ce in magnetite) + (wt % chevkinite)(Kn Ce in Chevkinite) The wt %’s of fractionation mineral phases (predicted by the major element regression) are normalized to 100%. Letting x = the amount of chevkinite fractionation, and plugging in appropriate Kp’s gives the following expression D = 2.749 : (0.793-x)(0.02) + (0.1427-x)(0.11) + (0.032-x)(0.36) + (0.0323-x)(0.28) + 2063x. Solving for x gives 1.308x10‘3, or 0.13% chevkinite fractionation. Similar calculations were carried out for Yb and Th. The maximum amount of chevkinite fractionation predicted by this method for steps 4, 5 and 6 was 0.21%. 65 APPENDIX 4 MAJOR ELEMENT MULTIPLE LINEAR REGRESSIONS CALCULATED WITH Y-INTERCEPTS EQUAL TO ZERO A. CRYSTAL FRACTIONATION MODEL STEP 1 PARENT : ATS-75 (59.44% 8102) DAUGHTER = ATS-63 (62.11% 8102) ATS-75 = 0.844 ATS-63 + 0.0963 PLAG + 0.0338 CPX + 0.0186 MT R2 = 1.00 SS : 0.1943 STEP 2 PARENT = ATS-63 (62.11% 8102) DAUGHTER : AT5-72 (64.00% 8102) ATS-63 = 0.890 ATS-72 + 0.0677 PLAG + 0.0236 CPX + 0.0109 MT R2 : 1.00 SS : 0.1718 STEP 3 PARENT = ATS—72 (64.00% 8102) DAUGHTER : ATS-71 (66.95% 8102) ATS—72 = 0.765 ATS-71 + 0.192 PLAG + 0.0238 CPX + 0.0160 MT R2 = 1.00 88 = 0.1942 STEP 4 PARENT = ATS-71 (66.95% 8102) DAUGHTER = ATS-58 (69.25% 8102) ATS-71 = 0.711 ATS-58 + 0.229 KSP + 0.0393 PLAG + 0.0103 CPX + 0.00928 MT R2 = 1.00 SS = 0.08818 66 STEP 5 PARENT = AT5-58 (69.25% S102) DAUGHTER = ATS-35 (72.96% 3102) ATS-58 = 0.634 ATS-35 + 0.340 KSP + 0.0019 PLAG + 0.0187 CPX + 0.00502 MT R2 : 1.00 SS : 0.1000 STEP 6 PARENT = ATS-35 (72.96% 8102) DAUGHTER = ATS-3 (77.96% 8102) ATS-35 : 0.697 ATS-3 + 0.230 KSP + 0.0501 PLAG + 0.0172 BIOT + 0.00606 MT R2 = 1.00 SS = 0.06764 B. CRYSTAL FRACTIONATION + MAGMA MIXING MODEL 1. Best fit model (overall) DERIVITIVE MAGMA = ATS-35 (72.96% S102) MIXED MAGMAS : ATS-71 (66.95% 8102) ATS-3 (77.96% 8102) ATS-35 = 0.519 ATS-71 + 0.525 ATS-3 — 0.00442 PLAG - 0.0338 KSP - 0.00452 CPX R2 = 1.00 88 : 0.05489 2. Mixing of end-member compositions DERIVITIVE MAGMA : ATS-35 (72.96% S102) MIXED MAGMAS = ATS-75 (59.44% S102) ATS-3 (77.96% S102) ATS-35 : 0.508 ATS—75 + 0.702 ATS-3 - 0.138 PLAG - 0.0583 CPX - 0.0140 MT R2 = 1.00 SS = 0.2479 67 LI ST OF REFERENCES LIST OF REFERENCES Aramaki, S., and Lipman, P.W., 1965. Possible leaching of Na20 during hydration of volcanic glasses. Japan Acad. Proc., 41: 467-470. E Arth, J.G., 1976. Behavior of trace elements during magmatic processes: A summary of theoretical models and their applications. Jour. Research U.S. Geol. Surv., 4: 41-47. Arth, J.G., and Hanson, G.N., 1975. Geochemistry and origin of the early Precambrian crust of northeastern Minnesota. Geochem. Cosmochim. Acta, 39: 325-362. Baker, B.H. and McBirney, A.R., 1985. Liquid Fractionation. Part III: Geochemistry of zoned magmas and the compositional effects of liquid fractionation. J. Volcanol. Geotherm. Res., 24: 55-81. Blake, 8., 1981. Eruptions from zoned magma chambers. J. Geol. Soc. London, 138: 281-287. Blake, S. and Ivey, G.N., 1986. Magma-mixing and the dynamics of withdrawal from stratified reservoirs. J. Volcanol. Geotherm. Res., 27: 153-178. Bowen, N.L., 1928. The Evolution of the Igneous Rocks. Princeton University Press: 334 p. Broxton, D.E., Byers, F.M., Jr., Warren, R.G., and Scott, R., 1985. Trends in phenocryst chemistry in the Timber Mountain-Oasis Valley volcanic field, SW Nevada: Evidence for episodic injection of primitive magma into an evolving magma system (abstract). Geol. Soc. Am. Abstr. Programs, 17: 345. Byers, F.M., Jr., Carr, W.J., Christiansen, R.L., Lipman, P.W., Orkild, P.P. and Quinlivan, W.D., 1976a. Geologic map of the Timber Mountain caldera area, Nye County, Nevada: U.S. Geol. Surv. Miscellaneous Geologic Investigations Map I-891. 68 69 Byers, F.M., Jr., Carr, W.J., Orkild, P.P., Quinlivan, W.D. and Sargent, K.A., 1976b. Volcanic suites and related cauldrons of Timber Mountain-Oasis Valley caldera complex, southern Nevada. U.S. Geological Survey Professional Paper 919, 70p. Cameron, R.L., 1983. Bishop Tuff revisited: New rare-earth element data consistent with crystal fractionation. Science, 224: 1338-1340. Carr, W.J., Byers, F.M., Jr., and Orkild, P.P., 1984. Stratigraphic and volcano-tectonic relations of Crater Flat Tuff and some older volcanic units, Nye County, Nevada. U.S. Geol. Surv. Open-File report 84-114: 42 pp. Christiansen, R.L. and Lipman, P.W., 1972. Cenozoic volcanism and plate-tectonic evolution of the Western United States. II. Late Cenozoic: Phil. Trans. R. Soc. London, A 271: 249-284. Christiansen, R.L., Lipman, P.W., Carr, W.J., Byers, F.M., Jr., Orkild, P.P. and Sargent, K.A., 1977. Timber Mountain-Oasis Valley caldera complex of southern Nevada. Geol. Soc. Am. Bull., 88: 943-959. Cox, K.G., Bell, J.D. and Pankhurst, R.J., 1979. The Interpretation of Igneous Rocks. George, Allen and Unwin, London: 450 p. Deer, W.A., Howie, R.A., and Zussman, J., 1962. Rock— Forming Minerals, v.1, Ortho- and Ring Silicates. Longmans, Green and Co., London: 333p. Ekrin, E.B., Anderson, R.E., Orkild, P.P., and Hinrichs, E.N., 1966. Geology of Silent Butte quadrangle, Nye County, Nevada: U.S. Geol. Survey Geol. Quad. Map GQ-493. Flood, T.P., 1987. Cyclic evolution of a magmatic system: The Paintbrush Tuff, SW Nevada volcanic field [Ph.D. thesis]. Michigan State University, 147 p. Hagan, R.C., 1982. X-ray fluorescence analysis: Major elements in silicate minerals. Los Alamos-9400-MS: 1-13. Haskin, L.A., Haskin, M.A., Frey, F.A., and Wildeman, T.R., 1968. Relative and absolute terrestrial abundances of the rare earths, in Origin and distribution of the elements, L.H. Ahrens, Ed., Pergamon Press, New York: 889-912. 70 Henderson, P., 1982. Inorganic Geochemistry. Pergamon Press, Oxford: 353p. Hildreth, E.W., 1977. The magma chamber of the Bishop Tuff: Gradients in temperature, pressure and composition [Ph.D. thesis]. Berkeley, University of California, 328 p. Hildreth, E.W., 1979. The Bishop Tuff: Evidence for the origin of compositional zonation in silicic magma chambers. Geol. Soc. Am. Special Paper 180, pp. 43-75. Hildreth, E.W., 1981. Gradients in silicic magma chambers: Implications for lithospheric magmatism. J. Geophys. Res., 86: 10153-10192. Hyndman, D.W., 1985. Petrology of igneous and metamorphic rocks, 2nd ed., McGraw Hill, New York: 786p. Kistler, R.W., 1968. Potassium-argon ages of volcanic rocks in Nye and Esmeralda Counties, Nevada, in Nevada Test Site: Geol. Soc. Am. Mem. 110, p. 251-262. Langmuir, C.H., Vocke, R.D., Hanson, G.N., and Hart, S.R., 1977. A general mixing equation: applied to the petro- genesis of basalts from Iceland and Reykjanes Ridge. Earth Planet. Sci. Lett., 37: 380-392. Lipman, P.W., 1965. Chemical comparison of glassy and crystalline volcanic rocks. Geol. Soc. Am. Special Paper 82: 260-261. Lipman, P.W., 1971. Iron-titanium oxide phenocrysts in compositionally zoned ash-flow sheets from southern Nevada. J. Geol., 79: 438-456. Lipman, P.W., Christiansen, R.L. and O’Conner, J.T., 1966. A compositionally zoned ash-flow sheet in southern Nevada. U.S. Geol. Surv. Prof. Paper 524-F: 47 pp. Mahood, G.A., 1981. A summary of the geology and petrology of the Sierra La Primavera, Jalisco, Mexico. Jour. Geophys. Res., 86, No. B11: 10137-10152. Mahood, G.A., and Hildreth, E.W., 1983. Large partition coefficients for trace elements in high-silica rhyolites. Geochem. Cosmochim. Acta, 47: 11-30. Marvin, R.F., Byers, F.M., Jr., Mehnert, H.H., Orkild, P.P. and Stern, T.W., 1970. Radiometric ages and stratigraphic sequence of volcanic and plutonic rocks, southern Nye and western Lincoln Counties, Nevada. Geol. Soc. Am. Bull., 81: 2657-2676. 71 Michael, P.J., 1983. Chemical differentiation of the Bishop Tuff and other high-silica magmas through crystallization processes. Geology, 11: 31-34. Miller, C.F. and Middlefehldt, D.W., 1984. Extreme fractionation in felsic magma chambers: a product of liquid-state diffusion or fractional crystallization? Earth Planet. Sci. Lett., 68: 151-158. Mills, J.G., Jr., and Rose, T.P., 1986. Geochemistry of glassy pumices from the Timber Mountain Tuff, Southwestern Nevada (abstract). EOS, Trans., AGU, 67: 1262. Nash, W.P., and Crecraft, H.R., 1985. Partition coefficients for trace elements in silicic magmas. Geochem. Cosmochim. Acta, 49: 2309-2322. Noble, D.C., 1965. Ground-water leaching of sodium from quickly cooled volcanic rocks (abstract). Am. Mineral., 50: 289. Noble, D.C., 1967. Sodium, potassium, and ferrous iron contents of some secondarily hydrated natural silicic glasses. Am. Mineral., 52: 280-286. Noble, D.C., Krushensky, R.D., McKay, E.J., and Ege, J.R., 1967. Geologic map of the Dead Horse Flat quadrangle, Nye County, Nevada: U.S. Geol. Survey Geol. Quad. Map GQ—614. Noble, D.C., Vogel, T.A., Weiss, S.I., Erwin, J.W., McKee, E.B., and Younker, L.W., 1984. Stratigraphic relations and source areas of ash-flow sheets of the Black Mountain and Stonewall Mountain Volcanic Centers, Nevada. J. Geophys. Res., 89: 8593-8602. Reid, J.B., Evans, O.C., and Fates, D.C., 1983. Magma mixing in granitic rocks of the central Sierra Nevada, California. Earth Planet. Sci. Lett., 66: 243-261. Schuraytz, B.C., Vogel, T.A. and Younker, L.W., 1983. Pumice heterogeneity in the Topopah Spring Member of the Paintbrush Tuff: Implications for ash-flow eruption dynamics (abstract). EOS, Trans., AGU, 64: 896. Schuraytz, B.C., Vogel, T.A. and Younker, L.W., 1986. Geochemical gradients in the Topopah Spring Member of the Paintbrush Tuff: Evidence for eruption across a magmatic interface. Lawrence Livermore National Laboratory report UCRL-53698: 59 pp. 72 Smith, R.L., 1960. Ash Flows. Geol. Soc. Am. Bull., 71: 795-842. Smith, R.L., 1979. Ash-flow magmatism. Geol. Soc. Am. Special Paper 180: 5-27. Smith, R.L., and Bailey, R.A., 1966. The Bandelier Tuff - A study of ash-flow eruption cycles from zonal magma chambers. Bull. Volcanol., 29: 83-103. Spera, E.J., 1984. Some numerical experiments on the withdrawal magma from crustal reservoirs. J. Geophys. Res., 89: 8222-8236. Vogel, T.A., Younker, L.W., Wilband, J.T., and Kampmueller, E., 1984. Magma mixing: the Marsco Suite, Isle of Skye, Scotland. Contrib. Mineral. Petrol., 79: 411-423. Vogel, T.A., Ryerson, E.J., Noble, D.C., and Younker, L.W., 1987. Limits to magma mixing based on chemistry and mineralogy of pumice fragments erupted from a chemically zoned magma body. J. Geol., 95: 659-670. Williams, H., 1941. Calderas and their origin. University of California, Publications in Geological Sciences, 25: 239-346. Wolff, J.A., and Storey, M., 1984. Zoning in highly alkaline magma bodies. Geological Magazine, 121: 563-575. Wright, T.L., 1974. Presentation and interpretation of chemical data for igneous rocks. Contrib. Mineral. nICHIan STATE UNIV. LIBRARIES IllWWW"I"WWIMIIIHIUW\IHIIIWINIHI 31293000592158