American Mineralogist, Volume 93, pages 7 21, Department of Geology, University of Maryland, College Park, Maryland, 20742, U.S.A.

Transcription

1 American Mineralogist, Volume 93, pages 7 21, 2008 Element mobility and scale of mass transport in the formation of quartz veins during regional metamorphism of the Waits River Formation, east-central Vermont Sarah C. Penniston-Dorland 1, * and John M. Ferry 2 1 Department of Geology, University of Maryland, College Park, Maryland, 20742, U.S.A. 2 Department of Earth and Planetary Sciences, Johns Hopkins University, Baltimore, Maryland 21218, U.S.A. Abstract Veins and adjacent alteration selvages in the Waits River Formation were investigated to determine whether associated mass transfer was due primarily to large-scale advection or small-scale diffusion. Samples of the vein, selvage, and adjacent wall rock were collected from the earliest and most numerous generation of veins in pelite and carbonate hosts from outcrops in the chlorite and the kyanite zones. Bulk compositions of selvages and unaltered wall rocks were compared using a reference frame defined by a combination of Zr, Ti, REEs, and U. Selvages from both outcrops typically exhibit losses of Si, K, Ba, and Rb relative to unaltered wall rock. Kyanite zone selvages show losses in Mg and Cs in addition. Differences in K, Ba, Rb, Cs, and Mg between the vein-selvage system as a whole and adjacent unaltered wall rock are possibly accounted for by development of micas in veins. The addition of Si and Ca to veins is not balanced by removal of Si and Ca from the selvages. Vein-selvage systems contain more Si than wall rock, with an overall addition of 40 mg Si/cm 3 to the chlorite zone and 55 mg Si/cm 3 to the kyanite zone. Mass balance of Si at the outcrop scale requires that >90% of the Si in quartz veins was derived externally. Quartz veins studied formed primarily by fluid flow and large-scale advective mass transfer with a relatively minor component of local mass transport by diffusion. The estimated time-integrated fluid flux necessary to produce the observed amount of quartz in veins in an entire outcrop is cm 3 fluid/cm 2 rock. Mineral inclusions in garnet and fracturing of garnets adjacent to veins indicate that formation of selvages and veins initiated prior to formation of garnet and continued after the end of garnet growth. Keywords: Mass transfer, regional metamorphism, veins, fluid flow Introduction Quartz veins are a common feature in regionally metamorphosed rocks. Their abundance in outcrops of regionally metamorphosed rocks can be 20 30% by volume (Ague 1994b; Ferry 1994). The large volume of silica in veins raises the question of whether they represent fossilized paths of large-scale fluid flow and mass transfer through the metamorphic terrain (Walther and Orville 1982; Yardley 1986; Walther 1990; Ferry and Dipple 1991; Ferry 1992; Ague 1994b; Breeding and Ague 2002; Masters and Ague 2005), segregations produced by local Si transport (Yardley 1975; Yardley and Bottrell 1992), or some combination of both. The problem of vein formation has further implications for a more general understanding of fluid flow and element transport during regional metamorphism. Investigations of fluid inclusions, veins, volatile contents of rocks, phase equilibria, reaction progress, and stable isotopic data all indicate that the volatile species H 2 O and CO 2 are mobile during metamorphism, but there is still uncertainty about other elements. Several studies have examined regional-scale mass transfer using bulk rock compositions of * sarahpd@geol.umd.edu samples far removed from veins. Shaw (1954, 1956) found no evidence for change in the concentrations of oxides (aside from H 2 O and CO 2 ) in average pelitic rock of the Littleton Formation, New Hampshire, during metamorphism. Subsequent analysis of Shaw s data using new statistical methods (Ague 1991) concluded that SiO 2, FeO, MgO, K 2 O, CaO, and Na 2 O could all have been mobile during metamorphism. A study of the pelitic rocks of the Wepawaug Schist, Connecticut showed that a range of elements can be mobile during metamorphism, including Si, P, Na, Mn, Zn, K, and Ba (Ague 1994a). Other studies of pelitic rocks, however, have concluded that there is no positive evidence for significant mass transfer of components other than H 2 O and CO 2 during metamorphism (e.g., Ferry 1982; Moss et al. 1995, 1996; Symmes and Ferry 1995). Ague (1997b) pointed out that there may be problems due to lithologic heterogeneity in the studies of Moss et al. (1995, 1996). The question of mass transfer in regionally metamorphosed impure carbonate rocks is clearer but still incomplete. Calc-silicate and argillaceous carbonate rocks can be depleted in K and Na during metamorphism (Tanner and Miller 1980; Ferry 1983). A significant, and sometimes insurmountable, problem in determining whether mass transfer has occurred during metamorphism is the lithologic heterogeneity in rocks prior to X/08/ $05.00/DOI: /am

2 8 metamorphism. At the outcrop scale, differences in composition due to pre-metamorphic heterogeneity are often much larger than changes in composition that develop from metamorphic processes. Study of metamorphic veins and their associated alteration selvages is an alternative approach to investigating element mobility during metamorphism in which the problem of pre-metamorphic rock composition is much better constrained by analysis of unaltered wall rock taken from the same lithologic layer as the selvage (e.g., Ague 1994b). Compositions of veins and associated selvages from the Wepawaug Schist, for example, indicate regional-scale mobility of Si, Na, and K that promoted the preferential growth of index minerals like garnet, staurolite, and kyanite in the altered selvages (Ague 1994b) and veins and selvages from the Dalradian metasediments in Stonehaven, Scotland, indicate regional-scale mobility of Si, Na, Ca, Sr, K, Rb, and Ba (Masters and Ague 2005). Analysis of the latest generation of veins and associated selvages from pelites hosting the Kanmantoo Copper Deposit in the Adelaide Fold Belt of southern Australia indicates large-scale mobility of Na, Ca, and Sr and local mobility of Si, Rb, K, and Ba (Oliver et al. 1998). In contrast, the compositions of selvages adjacent to veins hosted by pelitic schists from West Beach, Australia, indicate little or no mass transfer during metamorphism (Oliver and Bons 2001). In spite of the great promise of examining vein-selvage systems to evaluate the nature, amount, and spatial scale of element mobility during metamorphism, there have been too few studies to draw firm, general conclusions. The purpose of this study is to contribute to a better understanding of vein formation and element mobility during regional metamorphism through a study of veins and associated selvages in pelites and micaceous carbonate rocks from the Waits River Formation, east-central Vermont (Fig. 1). Specific questions addressed include: (1) is quartz in the veins mostly locally or externally derived; (2) was there significant mass transfer of elements other than Si in the formation of veins and selvages; (3) if so, is the mass transfer of these elements local or larger-scale; and (4) when did mass transfer occur in the metamorphic history of the rocks? Geologic background The Waits River Formation (Fig. 2) typically consists of micaceous carbonates and pelites interlayered on a scale of 1 10 m and lies stratigraphically below the Gile Mountain Formation (Fisher and Karabinos 1980). Hueber et al. (1990) concluded that the stratigraphic age of the two formations ranges from Silurian through Devonian based on the presence of plant fossils that date part of the Gile Mountain Formation as late Early Devonian and a Silurian U-Pb zircon age of 423 ± 4 Ma from a dike that cross-cuts the Standing Pond Volcanics member of the Waits River Formation. The Waits River Formation was deformed and regionally metamorphosed during the Acadian orogeny (Thompson et al. 1968; Thompson and Norton 1968; Osberg et al. 1989). Deformation involved the formation of west-verging nappes and subsequent doming (Woodland 1977). Isograds have been mapped in the pelites corresponding to the appearance of biotite, garnet, and kyanite (Fig. 2) and in the carbonates corresponding to the appearance of oligoclase, biotite, and amphibole (Lyons 1955; Doll et al. 1961; Ferry 1992, 1994). Final prograde mineral reaction during metamorphism was synchronous with the dome stage of deformation (Barnett and Chamberlain 1991; Menard and Spear 1994). The formation of monazite neoblasts at pressure and temperature conditions recorded by mineral equilibria in the kyanite zone of pelitic schists has been radiometrically dated at 353 ± 8 Ma (Wing et al. 2003). Mineral equilibria record P 8 kbar for the area and T from 475 C in the biotite zone to 550 C in the kyanite zone (Ferry 1994). Metamorphic fluids were composed primarily of H 2 O and CO 2 with minor H 2 S, CH 4, CO, H 2, and dissolved chlorides. Mineral-fluid equilibria record X CO2 from <0.03 to 0.2 (Ferry 1992, 1994; Penniston-Dorland and Ferry 2006). Figure 1. Representative V1 vein and adjacent selvage in kyanite zone pelitic schist. Selvage is ~1 cm thick. Waits River Formation, location S35-1.

3 9 Two outcrops were selected for this study: one in the chlorite zone and one in the kyanite zone (Figs. 2 and 3). Three distinct generations of originally planar veins are recognized. Criteria for identification of the different generations include vein structures and cross-cutting relationships with other veins and structural features (Table 1). The oldest veins (V1) are the most numerous and usually comprise the largest volume of veins within a given outcrop. The V1 veins are generally highly deformed (dismembered and folded), and they appear as pods and stringers parallel to layering and schistosity. The second generation (V2) usually is composed of relatively large, branching, and commonly discontinuous veins that cross-cut lithologic layers, folding, and schistosity. They are not numerous in a given outcrop, but because of their large size, their volume is close to or exceeds that of the V1 veins in some outcrops. The third generation of veins (V3) is the least deformed and hence youngest, and is generally composed of small and volumetrically minor veins (<0.1% of a given outcrop). Methods of investigation Field aspects of veins, associated selvages, and wall rock were recorded and measured in kyanite zone outcrop S35-1 and chlorite zone outcrop S40-1 (outcrops 21-2 and 27-11, respectively, of Ferry 1994). Vein selvages were identified in outcrop and hand sample by a higher concentration of aluminous minerals, such as garnets or micas, immediately adjacent to the vein and/or by a change in the color of the rock adjacent to the vein. Specifically, along traverses perpendicular to layering: (1) the generation of each vein (V1, V2, or V3) and its mineralogy were recorded; (2) the width of each vein and its associated selvage (if present) was measured with a steel measuring tape; and (3) the kind of wall rock between adjacent veins (pelite or carbonate) was noted and its thickness measured. Three samples of V1 veins were selected from outcrop S40-1 in the chlorite zone for detailed investigation [two hosted in carbonate rock (A, R) and one in pelite (D)], and three samples of V1 veins were selected from outcrop S35-1 in Figure 2. Geologic sketch map of the study area in east-central Vermont (after Lyons 1955; Doll et al. 1961; Ferry 1994). SDwr = Siluro-Devonian Waits River Formation; SDgm = Siluro-Devonian Gile Mountain Formation; pre-s = pre-silurian units, undifferentiated. Figure 3. Waits River Formation outcrops. (a) Chlorite-zone outcrop (S40-1) with single boudinaged vein visible. (b) Kyanite-zone outcrop (S35-1) with multiple folded and boudinaged veins visible.

4 10 Table 1. Vein descriptions V1 V2 V3 Thickness up to ~20 cm up to ~150 cm up to ~1.5 cm Mineralogy* Qtz ± Cal ± Ms ± Bt ± Po Qtz ± Cal ± Fsp ± Ky Cal or Qtz Selvages Up to ~40% of veins in an Up to 75% of veins in an No selvages observed outcrop have selvages outcrop have selvages Cross-cutting Usually conformable, at low Cross-cut V1 veins, schistosity, Cross-cut schistosity, lithologic layers relationships grades may cross-cut lithologic layers schistosity or layering Structure Boudinaged, folded May display slight boudinage, Planar, no deformation generally irregular in shape * Abbreviations follow Kretz (1983): Qtz = quartz, Cal = calcite, Ms = muscovite, Bt = biotite, Po = pyrrhotite, Fsp = feldspar, Ky = kyanite. Table 2. Vein and selvage percentages of outcrop Traverse Metamorphic zone (pelites) Total length of traverse V1 veins V2 veins V1 selvage V2 selvage S40-1 chlorite 3119 cm 4.5% 0% 0.4% 0% S35-1 kyanite 3782 cm 6.2% 6.0% 0.3% 0.1% the kyanite zone [one in carbonate rock (X) and two in pelite (G, S)]. As was the case for all quartz veins hosted by pelite in the chlorite zone outcrop, the vein in sample S40-1D did not have a selvage visible in hand sample. This sample was used to test whether there were any significant mineralogical or chemical changes adjacent to veins where no selvage is visible and to characterize the degree of chemical heterogeneity in rocks unaffected by formation of veins at the scale of hand specimens. The V2 and V3 veins were not analyzed in detail because either their selvages compose a much smaller volume of outcrop than do V1 selvages or they have no visible selvages at all (Table 2). Thin sections were made that included part of the vein, the entire selvage, and part of adjacent unaltered wall rock for each sample (Fig. 1). Wall rock and selvage for each sample were taken from the same lithologic layer. Layers appear internally homogenous in hand specimen. Mineral assemblages were determined in thin section with optical petrography and back-scattered electron (BSE) imaging using the JEOL JXA-8600 electron microprobe at Johns Hopkins University. Compositions of minerals were determined by electron microprobe analysis using wavelength-dispersive spectrometry, natural and synthetic mineral standards, and a ZAF correction scheme (Armstrong 1988). Mineral modes for selvages and wall rocks were measured by counting >2000 points in thin section using BSE imaging. Any uncertainty in the identification of a particular point was resolved by obtaining an energy-dispersive X-ray spectrum (EDS). Modal abundances of minerals were converted to molar abundances using mineral compositions and molar volumes of mineral components from Holland and Powell (1998). Graphs of the percentage of quartz with distance from the selvage-wall rock contact were created for the five selvage-bearing samples by point counting at intervals of <0.5 mm away from the vein across the selvage-wall rock contact with an average of 100 points for each interval. Maps that visually illustrate the mineralogical differences between selvage and wall rock were made from point counting results across the selvagewall rock contact. Modes of mineral inclusions in garnet from sample S35-1G were determined in a similar fashion by counting ~300 inclusions in garnet from both selvage (two garnets examined) and wall rock (one garnet examined) in thin section using BSE imaging and qualitative EDS analysis. A portion of each sample was separated into selvage and adjacent wall rock (or portions proximal to and distal from the vein in sample S40-1D) using a rock saw and rotating grinding lap. The mass and volume of portions of selvage and wall rock were measured and rock density was calculated from these measurements. The pairs for each sample were then crushed into a fine powder in a tungsten carbide container. The size of most samples of selvage was limited by the relatively narrow width of the selvage and the size of the sample collected. The amount of sample pulverized for selvages ranged from 7.8 g (S35-1S) to 67.6 g (S40-1R). The amount of sample pulverized for wall-rock analyses ranged from 24.4 g (S40-1D) to 50.2 g (S35-1G). Bulk-chemical analyses of major and trace elements using X-ray fluorescence (XRF) and inductively coupled plasma-mass spectrometry (ICP- MS) were obtained commercially from SGS Minerals Services. The accuracy and long-term reproducibility of analyses from this laboratory (formerly X-ray Assay Laboratories) have been assessed by Ferry (1988) and Ague (1994a). Estimates of whole-rock major-element composition were also made for one sample (S40-1D) from mineral modes and mineral compositions determined with the electron microprobe combined with published molar volumes of minerals. Results agree within error of measurement for most elements with those obtained by XRF analysis (see below). Uncertainties in the microprobe data, however, are larger than for XRF analyses because of the accumulated errors from both point counting and mineral analysis. For this reason and because XRF analysis samples a larger volume of rock, estimates of whole-rock composition were not made from modal and microprobe data for the other samples, and only XRF data were used for all calculations of mass transfer of the major elements. Results Abundances of veins and selvages in outcrop Vein and selvage abundances are listed in Table 2. In the chlorite-zone outcrop, all veins are V1 veins, and they comprise 4.5% of the traverse. No V2 or V3 veins were observed. Vein selvages comprise 0.4% of the length of the traverse. In the kyanite-zone outcrop, both V1 and V2 veins occur in roughly equal volumes along the traverse (6.2% and 6.0%, respectively). The V1 vein selvages account for 0.3% of the traverse, about the same as in the chlorite-zone outcrop. The total volume of selvages adjacent to V2 veins along the kyanite-zone traverse (0.1%) is less than the total volume of selvages adjacent to V1 veins. Mineral assemblages and modes of V1 veins and vein selvages The V1 veins are composed of quartz ± calcite ± minor muscovite, biotite, and/or pyrrhotite. Approximate visual estimates of mineral abundances in veins from both chlorite- and kyanitezone outcrops were determined using a petrographic microscope for 19 vein samples. Quartz typically comprises % of the vein and calcite comprises <5 40% of the vein. Muscovite is present in small amounts (<1%) in most veins. Biotite and pyrrhotite are not common and, where present, also occur in small amounts (<1%). Selvages typically range in thickness from <5 mm to 2.5 cm. The boundary between the selvage and the wall rock is usually <5 mm wide. The mineral assemblages and modes for the distal and proximal portions of sample S40-1D and for each of the five selvage-wall rock pairs are listed in Table 3. The designation of epidote/allanite is used for epidote-group minerals that have cores that appear bright in BSE imaging. These cores are enriched in one or more elements of high atomic number found in allanite, such as Ce, La, Nd, Y, Th, and U. Pelite sample S40-1D contains ~1% small garnets that were likely stabilized at the conditions of the chlorite zone by their high Mn content (Table 4). In sample S40-1D, which has no visible selvage, there is no statistically significant difference in mineral modes between rock

5 11 Table 3. Mineral assemblages and modes (volume percent) Sample S40-1A S40-1R S40-1D S35-1X S35-1G S35-1S Rock type* Carbonate Carbonate Pelite Carbonate Pelite Pelite Wr Sel Wr Sel Distal Proximal Wr Sel Wr Sel Wr Sel Quartz Plagioclase Muscovite Chlorite Biotite Calcite Ankerite Garnet Ilmenite Rutile Apatite Epidote/Allanite Tourmaline Paragonite Pyrrhotite Zircon Amphibole * Wr = wall rock, Sel = Selvage. immediately adjacent to the vein and rock further away, confirming that there is no selvage on the basis of either macroscopic or petrographic examination. On the other hand, there are significant differences in modes between selvage and wall rock in the other five samples having selvages visible in hand specimen. The mineralogical differences between selvage and wall rock are primarily in mineral abundance rather than in mineral assemblage. Modal quartz is less in the selvages than in adjacent wall rock, and in most cases, near zero through much of the selvage. This depletion in quartz is visible in the point-counting maps (Fig. 4) and point-counting traverses performed across selvages into wall rock (Fig. 5). In contrast, compared to wall rock, selvages contain more modal apatite and tourmaline, more calcite (in carbonate hosts), and more ilmenite + rutile (in kyanite-zone samples). Qualitatively, the differences in modes between selvage and adjacent wall rock require a bulk-chemical difference as well. Timing of vein selvage formation Garnets in sample S35-1G contain evidence that vein and selvage formation for a single V1 vein spanned a wide range in conditions of metamorphism and hence time. The assemblage and modes of mineral inclusions in garnets from the selvage are very different from those in garnet from adjacent wall rock (Fig. 6). The total volume of inclusions is 16.0% of the wall-rock garnet whereas the total volume of inclusions is 3.4% of the selvage garnets. Mineral inclusions in garnets from unaltered wall rock are primarily quartz with minor ilmenite, epidote/allanite, zircon, pyrrhotite, and calcite (Fig. 6a; Table 5). Inclusions in garnets from the adjacent alteration selvage are mostly ilmenite and epidote/allanite with minor zircon, quartz, apatite, pyrrhotite, tourmaline, plagioclase, calcite, and rutile. (Fig. 6b; Table 5). The paucity of quartz inclusions and abundance of plagioclase inclusions in garnet from the selvage compared to garnet from the wall rock mirrors the difference in the relative abundance of quartz and plagioclase between the matrices of the selvage and wall rock themselves (Tables 3 and 5). This correlation indicates that some and probably most of the alteration involved in formation of the selvages occurred at a relatively low grade of metamorphism before the garnets grew. Garnets in selvages from the same sample, S35-1G, have been fractured and altered adjacent to the vein, with one fracture running through the middle of a garnet (Fig. 7). This observation suggests that at least some V1 veins or parts of V1 veins formed when garnets were already present, at a higher grade of metamorphism, and that the fracturing and alteration may have been caused by the vein-forming event. The evidence for selvage and vein formation within a single vein at two distinctly different grades of metamorphism suggests that alteration and concomitant fluid flow occurred over a long duration during metamorphism. Mineral compositions Representative mineral compositions are reported in Table 4. Mineral compositions in the selvage and wall rock for a given sample are similar, as are those in the distal and proximal portions of sample S40-1D. Plagioclase from the kyanite zone is irregularly zoned. Approximately 20 grains of plagioclase were selected and analyzed in each of the high-grade samples. The selection method was designed to minimize preferential analysis of either cores or rims. A grid of regularly spaced points was set up, and the grain or part of a grain nearest each point on the grid was analyzed. The average plagioclase composition as well as the range of plagioclase compositions is reported for each sample. Whole-rock geochemistry Bulk-rock XRF and ICP-MS data are reported in Table 6. Detection limits and uncertainties of analysis are listed in Appendix 1 1. Results for several elements represent probable contamination from the tungsten carbide container used for crushing samples. 1 Deposit item AM , Appendix 1 (detection limits and uncertainties of analysis). Deposit items are available two ways: For a paper copy contact the Business Office of the Mineralogical Society of America (see inside front cover of recent issue) for price information. For an electronic copy visit the MSA web site at go to the American Mineralogist Contents, find the table of contents for the specific volume/issue wanted, and then click on the deposit link there.

6 12 Table 4a. Representative mineral compositions: Calcite Sample S40-1R S35-1X S35-1G Wr Sel Wr Sel Wr Sel Ca Mg Fe Mn Oxide sum No. analyses Notes: Analyses of calcite, ankerite, muscovite, biotite, and chlorite are averages of 4 22 spot analyses of grains in thin section. Analyses of plagioclase feldspars are averages of random spot analyses of grains per thin section; range of % An is taken from these analyses. Mineral formulae for muscovite, paragonite, and biotite are cations per 11 O atoms (less H 2 O); for chlorite, cations per 14 O atoms (less H 2 O); for amphibole, cations per 23 O atoms (less H 2 O); for tourmaline, cations per 31 O atoms (less H 2 O); for calcite, cations per oxygen atom (less CO 2 ); for ankerite, cations per 2 O atoms (less CO 2 ). Oxide sum refers to the average sum of oxide wt%, excluding B 2 O 3, CO 2 and H 2 O, and with all Fe as FeO. Dist = distal, Prox = proximal, Wr = wall rock, Sel = selvage. Table 4b. Representative mineral compositions: Ankerite Sample S40-1R S40-1D S35-1X Wr Sel Dist Prox Wr Sel Ca Mg Fe Mn Oxide sum No. analyses Table 4c. Representative mineral compositions: Plagioclase Sample S40-1A S40-1D S35-1X S35-1S Wr Sel Dist Prox Wr Sel Wr Sel Ave X an Ave X ab Ave X or Oxide sum No. analyses Range %An Table 4d. Representative mineral compositions: Muscovite Sample S40-1A S40-1R S35-1G Wr Sel Wr Sel Wr Sel K Na Fe Mg Mn Ti Al VI Al IV Si Oxide sum No. analyses Table 4e. Representative mineral compositions: Paragonite Sample S40-1R Wr Sel K Na Fe Mg Mn Ti Al VI Al IV Si Oxide sum No. analyses 6 7 Table 4f. Representative mineral compositions: Biotite Sample S35-1X S35-1G Wr Sel Wr Sel K Na Fe Mg Mn Ti Al VI Al IV Si Oxide sum No. analyses Table 4g. Representative mineral compositions: Amphibole Sample S35-1X S35-1G Wr Sel Sel Na Ca Fe Mg Mn Ti Al Si Oxide sum No. analyses Table 4h. Representative mineral compositions: Tourmaline Sample S40-1R S35-1S Wr Sel Wr Sel Na Ca Fe Mg Mn Ti Al Si Oxide sum No. analyses Table 4i. Representative mineral compositions: Garnet Sample S40-1D S35-1X S35-1G Dist Prox Wr Sel Wr Sel Ca Fe Mg Mn Ti Al Si Oxide sum No. analyses Table 4j. Representative mineral compositions: Chlorite Sample S40-1D S35-1X S35-1G Dist Prox Wr Sel Wr Sel Fe Mg Mn Ti Al Si Oxide sum No. analyses

7 13 Table 5. Modes of mineral inclusions in garnet, sample S35-1G Wall rock (n = 378) Selvage (n = 299) Quartz 93% 5% Ilmenite 3% 41% Epidote/Allanite 1% 33% Zircon 1% 5% Apatite 0% 4% Pyrrhotite 1% 3% Tourmaline 0% 3% Plagioclase 0% 2% Calcite 1% 1% Rutile 0% 1% Note: Inclusions counted within one garnet for wall rock determination and within two garnets for selvage determination. n is the number of points counted. The container is composed primarily of W, C, and Co with lesser amounts of Ta, Ti, and Nb. Measured concentrations of Co, Ta, Ti, and Nb were plotted against that of W, and there are linear correlations of both Co and Ta with W. None of these elements would be expected to have large concentrations in typical pelitic and carbonate rocks, and thus the measured concentrations of Co, Ta, and W are likely primarily due to contamination. Cobalt, Ta, and W therefore are not included in Table 6. Because measured concentrations of Ti and Nb do not show any correlation with W, they are not considered to reflect significant contamination from the crusher, and are reported in Table 6. Detailed discussion of the whole-rock geochemistry follows. Mass transfer during selvage formation A geochemical reference frame Figure 4. Mineral maps of selvages and wall rocks created by point counting minerals at 0.25 mm intervals. (a) Mineral map of chlorite-zone metacarbonate rock (S40-1D) without visible selvage. Quartz (dark blue) at top of map defines the vein. There is no detectable depletion of quartz in wall rock adjacent to the vein. (b) Mineral map of chloritezone metacarbonate rock (S40-1A). The vein is off the top of the map. Region with almost no quartz (mostly pink calcite) at top of map is selvage, region below is wall rock. The transition from selvage with almost no quartz to wall rock with abundant quartz is sharp (<5 mm wide). (c) Mineral map of kyanite-zone pelitic rock (S35-1G). The vein is off the top of the map. Region with almost no quartz (mostly yellow plagioclase and red garnet) at top of map is selvage. The transition from selvage with almost no quartz to wall rock with abundant quartz is also sharp (<5 mm wide). To quantitatively assess gains and losses of elements during formation of the veins and selvages, an element or group of elements first must be chosen that is believed to have been immobile or almost so during alteration (e.g., Gresens 1966; Grant 1986) to avoid the closure problem (Ague and van Haren 1996). The ratio of the concentration of element i in the altered rock to the concentration of the element i in the unaltered wall rock, Ci'/C i (hereafter referred to as the concentration ratio), for an immobile element can be used to adjust the concentrations of the other elements in the altered rock to avoid the problem of closure. Selection of such an immobile reference frame also circumvents the effects of volume change during alteration. Elements that appear in vein minerals were eliminated from the list of possible immobile elements, including Si and Al (some V2 veins in the Waits River Formation contain kyanite). Elements for which there is evidence for mobility or for which analytical uncertainties are large were further eliminated (Ague and van Haren 1996). Specifically, mobility of several elements during metamorphism, including K, Na, Ca, Fe, Mg, Mn, Al, P, Rb, Ba, Sr, Zn, Ni, and/or Y, have been documented by Shaw (1954, 1956), Tanner and Miller (1980), Ferry (1983), and Ague (1994a, 2003). Elements with excessively large analytical uncertainties include Cr, Li, Sc, Ga, Ge, V, Cs, Eu, Nb, Tb, Hf, Lu, Tm, and Ti (when TiO2 or Ti <0.2 wt%). The method of Baumgartner and Olsen (1995) was then followed, in which several elements with overlapping concentration ratios were used to define a reference frame. Elements used to define the immobile reference frame include Ti (when TiO2 or Ti >0.2 wt%), Zr, La, Ce, Pr, Nd, Sm, Gd, Dy, Ho, Er, Yb, and U. There is documented mobility of Ti in some geologic settings (e.g., Ague 2003). However, in

8 14 Table 6. XRF and ICP-MS whole rock data Sample Method Units S40-1A S40-1R S40-1D S35-1X S35-1G S35-1S Wr Sel Wr Sel Dist Prox Wr Sel Wr Sel Wr Sel SiO 2 XRF wt% Al 2 O 3 XRF wt% CaO XRF wt% MgO XRF wt% Na 2 O XRF wt% K 2 O XRF wt% Fe 2 O 3 XRF wt% MnO XRF wt% TiO 2 XRF wt% P 2 O 5 XRF wt% Cr 2 O 3 XRF wt% <0.01 <0.01 <0.01 < LOI XRF wt% Sum XRF wt% Rb XRF ppm Sr XRF ppm Y XRF ppm Zr XRF ppm Nb XRF ppm 2 <2 2 < Ba XRF ppm Al ICP-MS wt% Ba ICP-MS ppm Ca ICP-MS wt% Cr ICP-MS ppm Fe ICP-MS wt% K ICP-MS wt% Li ICP-MS ppm <10 <10 <10 < Mg ICP-MS wt% Mn ICP-MS ppm Ni ICP-MS ppm P ICP-MS wt% Sc ICP-MS ppm <5 <5 <5 < Sr ICP-MS ppm Ti ICP-MS wt% V ICP-MS ppm Zn ICP-MS ppm Zr ICP-MS ppm Bi ICP-MS ppm <0.1 < <0.1 <0.1 < Cd ICP-MS ppm <0.2 <0.2 <0.2 <0.2 <0.2 <0.2 <0.2 < < Ce ICP-MS ppm Cs ICP-MS ppm Dy ICP-MS ppm Er ICP-MS ppm Eu ICP-MS ppm Ga ICP-MS ppm Gd ICP-MS ppm Ge ICP-MS ppm <1 <1 <1 < Hf ICP-MS ppm < Ho ICP-MS ppm La ICP-MS ppm Lu ICP-MS ppm Mo ICP-MS ppm <2 <2 <2 <2 <2 <2 < <2 <2 Nb ICP-MS ppm Nd ICP-MS ppm Pb ICP-MS ppm Pr ICP-MS ppm Rb ICP-MS ppm Sm ICP-MS ppm Sn ICP-MS ppm <1 <1 <1 < Tb ICP-MS ppm Th ICP-MS ppm Tl ICP-MS ppm <0.5 <0.5 <0.5 < <0.5 <0.5 <0.5 < Tm ICP-MS ppm U ICP-MS ppm Y ICP-MS ppm Yb ICP-MS ppm Notes: Wr = wall rock, Sel = selvage, Dist = distal, Prox = proximal. Concentrations for Be, Cu, ag, As, and In were below the detection limit. Concentrations for W, Co, and Ta not reported. Italics are measurements that are below 20 times the detection limit. this study the concentration ratio for Ti in each sample falls in line with the other immobile elements used. For each sample, a range of values was defined for the immobile reference frame with upper and lower bounds based on the smallest interval in C i '/C i that included all of the elements considered immobile. In addition, a minimum value for the range of C i '/C i = 0.6 was set by the range of the XRF and ICP-MS data for rock adjacent to and distal from the vein in chlorite-zone sample S40-1D (Fig.

9 15 Figure 5. Modal percentage of quartz along traverses perpendicular to the selvage-wall rock contact in five samples studied in detail. Quartz is depleted in selvages relative to wall rock in all samples, and in most samples large portions of the selvage do not have any quartz. Different scales in each panel. The outer boundary of the selvage was determined for most samples by the location at which the abundance of quartz increased above ~0%. The inner boundary of the wall rock was defined as the location beyond which the abundance of quartz remained relatively constant (keeping in mind the inherent heterogeneity of the rock). The transition zone was the region between the outer boundary of the selvage and the inner boundary of the wall rock. Sample S40-1R was an exception to this characterization, since the selvage exhibited a gradational increase in quartz over a greater distance than the other samples and there was only a very small (<1 mm) region in which quartz was ~0%. For S40-1R, the inner boundary of the wall rock was defined similarly to the other samples, and that boundary was also defined as the outer boundary of the selvage. 8a) that shows no visual, petrographic, or chemical evidence for development of an alteration selvage. The range in C i '/C i = 0.6 therefore is believed to be a representative minimum estimate of uncertainties introduced both by the heterogeneity in the premetamorphic bulk composition of individual hand specimens and by the analytical errors associated with measured concentrations of the collection of immobile elements. Plots of the concentration ratio for the distal and proximal portions of sample S40-1D and for the five wall rock-selvage pairs in the other samples are shown in Figure 8 along with lines showing the upper and lower bounds on the range of values for the immobile reference frame. Geochemical, volume, and mineralogical changes during selvage formation Geochemical changes. If the concentration ratio for a particular element or species falls above the immobile reference frame (black symbols, Fig. 8), then the element was added during alteration, and if it falls below the immobile reference frame (open symbols, Fig. 8), then it was lost during alteration. If the

10 16 concentration ratio of a given element falls within the immobile reference frame (gray symbols, Fig. 8) then there is no conclusive evidence for either loss or gain. The quantitative change in mass for element i (ΔCi/C i) can be calculated following Ague (1994a): Ci C Figure 6. Photomicrographs of garnets in sample S35-1G under crossed polars. (a) Garnets in unaltered wall rock have a large abundance of inclusions most of which are quartz. (b) Garnets in adjacent altered selvage have fewer inclusions most of which are ilmenite and epidote/ allanite. Figure 7. Photomicrograph of garnet adjacent to V1 quartz-calcite vein (right-hand side of image) in sample S35-1G under crossed polars. Garnet is fractured and altered to chlorite adjacent to vein. i = Cref C ' ref * Ci' Ci 1 (1) using the concentration ratios at the upper and lower bounds of the immobile reference frame for each sample as the limiting values for C'ref/C ref. Chlorite-zone samples. Measured XRF, ICP-MS, and electron microprobe data for portions of pelite sample S401D that are adjacent to and distal from the quartz vein fall near a concentration ratio of 1, which is the value expected for no mass change [log(c'/c ) = 0, dashed horizontal line in Fig. 8a]. Accordingly, there is no conclusive evidence for gain or loss of mass of any analyzed element. Not only is there no visual or modal evidence for development of an alteration selvage adjacent to the vein in sample S40-1D, but also there is no conclusive geochemical evidence for the development of an alteration selvage. Concentration ratios for numerous elements are illustrated in Figures 8b and 8c for the two chlorite-zone selvage-wall rock pairs (carbonate rocks S40-1A and S40-1R). In both cases, errors on the concentration ratios for Na and Ti are large (Figs. 8b and 8c) because of large analytical errors in XRF analyses of elements in low concentration. The concentration ratio for Zr based on ICP-MS analysis for sample S40-1A is anomalously large compared to the ratio based on XRF analysis (Fig. 8b). The result is contrary to expectation because Zr is traditionally used as a reliable immobile reference frame (e.g., Breeding and Ague 2002; Ague 2003; Masters and Ague 2005). The high Zr concentration ratio from ICP-MS analysis is likely the result of a nugget effect, an accidental incorporation of slightly more zircon in the aliquot of selvage sample S40-1A used for ICP-MS analysis. This datum was therefore not included in the determination of the immobile reference frame. Even if the higher Zr analysis is accepted as part of the immobile reference frame, however, conclusions are not changed. For both samples S40-1A and S40-1R Si, K, and Ba fall below the immobile reference frame (Figs. 8b and 8c). The datum for Rb falls below the immobile reference frame in sample S40-1A but within error of the immobile reference frame in sample S40-1R. No element in either sample plots significantly above the immobile reference frame. During formation of alteration selvages around veins hosted by carbonate rocks in the chlorite zone, therefore, Si, K, and Ba were consistently lost from the protolith, Rb was lost from one but not both samples, and no elements were added. The relative magnitudes of Si, K, Rb, and Ba loss were 34 82, 13 61, 11 66, and 17 66%, respectively (Table 7). A range of values results both from uncertainty in the immobile reference frame and from a possible difference in the process of alteration between the samples. Kyanite-zone samples. Concentration ratios are illustrated in Figures 8d 8f for the three kyanite-zone selvage-wall rock pairs (pelite samples S35-1G and S35-1S and carbonate sample S35-1X). As for carbonate samples from the chlorite

11 17 Figure 8. Concentration ratios (log scale) for major, minor, and trace elements for analyzed material adjacent to and distal from the vein in sample S40-1D and for the five analyzed selvage-wall rock pairs. Triangles, data from modal and electron microprobe analyses; diamonds, XRF data; squares, ICP-MS data. Error bars are ±2σ. When error bars are not shown, they are smaller than the size of the symbol. Dashed horizontal line indicates concentration ratio for no mass transfer. Solid horizontal lines show range of the immobile reference frame. Gray symbols are elements that fall within the immobile reference frame (indicating no conclusive evidence for gain or loss during alteration), elements with black symbols fall above (indicating gain during alteration), and elements with unfilled symbols fall below (indicating loss during alteration). Black x indicates an element that was used to define the immobile reference frame. zone, the concentration ratio for Na is large for sample S35-1X (Fig. 8d) because of large analytical errors in XRF analyses of elements in low concentration. In sample S35-1X, the concentration ratios for Si, K, and Rb fall below the immobile reference frame, similar to selvages in carbonate rocks in the chlorite zone. In addition, concentration ratios for Mg and Cs fall below the reference frame (Fig. 8d). The concentration ratios for immobile elements in sample S35-1S span a larger range of values than for any other. This is likely the geochemical expression of a larger degree of lithologic heterogeneity visible in the hand specimen of the rock. Furthermore, the distinctive pattern of concentration ratios of REEs in Figure 8f suggests a second example of a nugget effect, in this case caused by an accidental incorporation of excess garnet into the aliquot of selvage sample S35-1S and/or of too little garnet in the aliquot of wall rock used for ICP-MS analysis. Even if the REE data for sample S35-1S in Figure 8f are ignored, however, conclusions are not changed. Concentration ratios for Mg, K, Rb, Cs, and Ba in both of the kyanite-zone pelites plot below the immobile reference frame (Figs. 8e and 8f). The concentration ratio for Si for sample S35-1G also plots significantly below the immobile reference frame while that for sample S35-1S falls at the lower end of the immobile reference frame. For sample S35-1S, concentration ratios for Al, Na, Sr, V, and Th additionally plot below the immobile reference frame, and the concentration ratio for P plots above the immobile reference frame. During formation of alteration selvages in the kyanite zone, therefore, Si, K, Rb, Cs, and Mg were consistently lost from both protoliths (with the possible exception of Si in sample S35-1S); Ba was lost from both pelite samples S35-1G and S35-1S; and Al, Na, Sr, Th, and V were lost from and P added to sample S35-1S. The magnitudes of Si, K, Rb, and Ba loss were 0 76, 24 63, 25 67, and 7 65%, respectively (Table 7). As in the chlorite zone, a range of values results both from uncertainty in the immobile reference frame and from a possible difference in the process of alteration between the samples.

12 18 Table 7. Calculated range of % mass change during formation of the selvages (ΔC i /C ) 100 i Sample Si K Rb (ICP-MS) Ba (ICP-MS) S40-1D 35.2% + 9.9% 11.0% +50.9% 11.5% +50.1% 14.0% +45.8% S40-1A 82.1% 71.1% 60.6% 36.4% 66.3% 45.7% 66.0% 45.2% S40-1R 65.2% 34.3% 53.9% 13.1% 53.0% 11.2% 56.1% 17.1% S35-1G 76.4% 69.5% 62.5% 51.4% 47.0% 31.3% 37.3% 18.8% S35-1S 37.2% 0.5% 63.2% 41.8% 67.5% 48.5% 64.6% 43.9% S35-1X 47.3% 11.4% 55.0% 24.3% 55.6% 25.4% 44.7% 7.0% Overview of geochemical changes. Regardless of grade or type of host rock, formation of selvages around metamorphic veins in the Waits River Formation typically involved significant depletions in K, Rb, Ba, and Si. There was significant loss of K from all five analyzed selvages and significant loss of Rb, Ba, and Si from all but one. In samples from the kyanite zone, there is additional systematic depletion of both pelite and carbonate wall rocks in Mg and Cs during formation of selvages. Regardless of the details of the immobile reference frame, the amount of element loss and gain was quite variable from sample to sample. The qualitative similarity among samples is further evidence that much of the chemical change in the selvages during formation of veins occurred at a very early stage of metamorphism (conditions of the chlorite zone or at lower grade). The behavior of Mg and Cs, however, indicate important differences in element mobility related to grade. Volume changes. The change in volume during alteration (referenced to the original volume of wall rock) can be determined from the concentration ratio for the immobile reference frame for the sample. If a sample experienced no volume change, the concentration ratio of the immobile reference frame is 1. If volume decreased during alteration, the concentration ratio of the immobile reference frame generally is >1 (the exceptions are isochemical volume removal, e.g., dissolution of calcite in a pure calcite limestone and polymorphic phase change accompanied by a density change, e.g., conversion of calcite in a pure calcite limestone to aragonite), and if volume increased during alteration, it is <1. The fractional volume change ( V/V ) can be calculated using values of the concentration ratio of immobile elements along with values of rock density for the selvage (ρ') and wall rock (ρ ): V ρ C = 1 (2) V ρ' C ' Results for each sample are presented in Table 8, with a range of volume loss/gain calculated using the upper and lower bounds of C'/C for the immobile reference frame. As expected, application of the analysis to portions of selvage-free sample S40-1D that are adjacent to and distal from the vein indicates no volume change within error of measurement. Two of the five selvages experienced a significant decrease in volume during alteration of 4 67%, including both kyanite-zone pelite samples. For the other three samples with selvages (chlorite- and kyanite-zone carbonate samples S40-1A, S40-1R, and S35-1X), the evidence for a change in volume is inconclusive because of uncertainty in the choice of the immobile reference frame. Mineralogical differences. The amount of quartz is systematically less in selvages than in adjacent unaltered wall rock (Table 3; Fig. 5). The depletion in quartz is the mineralogical expression of whole-rock depletion in Si during formation of the Table 8. Calculated volume change during formation of the selvages Sample Rock Upper Lower Bounds on % Volume type C /Cº C /Cº change ( V/Vº) 100 S40-1D pelite % +13.8% S40-1A carbonate % + 0.3% S40-1R carbonate % +47.1% S35-1G pelite % 57.9% S35-1S pelite % 3.9% S35-1X carbonate % +12.3% selvages (Table 7). Uncertainty in the immobile reference frame unfortunately prohibits formulation of mineral-fluid reactions that would relate in detail the changes in whole-rock chemistry during development of the alteration selvages to corresponding mineralogical changes. Discussion Mass transfer at the vein scale The question whether there was an overall gain or loss of elements in the formation of a given vein-selvage system as a whole is addressed by a comparison between the mass of elements removed from the selvage and mass added to the vein. For this calculation, the thickness of both selvages and veins must be known. In the kyanite-zone outcrop, most veins were highly deformed, and accurate determination of the original proportion of vein and selvage widths for a single vein is not possible (Fig. 3b). In the chlorite-zone outcrop, however, the veins are boudinaged but much less deformed (Fig. 3a). A meaningful determination of the average widths of vein and selvage prior to deformation was obtained by measuring both along 23 m of a single vein at 15 cm intervals (a distance much less than that between boudins). Measured widths for each were then averaged. If there was no mass transfer of element i into or out of the vein-selvage system as a whole, the mass of i per volume removed from the selvage times the average thickness of the selvage zone prior to selvage formation must equal the mass added per volume to the vein times the average thickness of the vein: C i,sel ρ sel x sel = C i,vein ρ vein x vein (3) where x sel and x vein are the average thickness of selvage prior to selvage formation and vein respectively (the thicknesses of selvage on both sides of the vein were added together). Values for C i,sel for each element in each sample were determined from the range of ΔC i '/C i calculated from equation 1 using the upper and lower bounds on the immobile reference frame along with measured values for C i (wall-rock concentrations of each element). Values for C i,vein were determined from estimated modes of minerals (quartz, calcite) assuming ideal mineral compositions. The density of the selvage material prior to selvage formation (ρ sel ) was determined from measurements of sample mass and volume of the wall rock.