Geochimica et Cosmochimica Acta, Vol. 63, No. 21, pp. 3635–3651, 1999 Copyright © 1999 Elsevier Science Ltd Printed in the USA. All rights reserved 0016-7037/99 $20.00 ⫹ .00
Pergamon
PII S0016-7037(99)00162-3
Oxy-substitution and dehydrogenation in mantle-derived amphibole megacrysts P. L. KING,1,* R. L. HERVIG,1,2 J. R. HOLLOWAY,1,3 T. W. VENNEMANN,4 and K. RIGHTER5 1
Geology Department, Arizona State University, Tempe, AZ 85287-1404 USA Center for Solid State Science, Arizona State University, Tempe, AZ 85287-1704 USA 3 Chemistry Department, Arizona State University, Tempe, AZ 85287-1604 USA 4 Institut Geochemie, Wilhelmstr. 56, D-72076, Tu¨bingen, Germany 5 Lunar & Planetary Laboratory, University of Arizona, Tuscon, AZ 85721 USA
2
(Received September 1, 1998; accepted in revised form April 19, 1999)
Abstract—Results from major element and hydrogen micro-analyses of titanium-rich mantle-derived amphiboles from the SW USA are combined with previous experimental studies. We show that the distinctive chemistry of mantle-derived amphiboles, especially relatively high Ti, variable ferric/ferrous iron, and hydrogen contents, result from both initial crystallization conditions and dehydrogenation. On the basis of previous experimental work, it is concluded that the Ti-rich nature of mantle-derived amphibole megacrysts is a result of crystallization from mafic-ultramafic melts at low to moderate pressure (ⱕ1.0 GPa), high temperature (⬎950°C) and low to moderate oxygen fugacity (ƒO2). We propose that those conditions change TiO2 and Al2O3 activity in the melt. Iron oxidation state in amphiboles is affected by ƒO2 or hydrogen fugacity (ƒH2) in the melt. In contrast to previous suggestions, it is not necessary to have low water activity (aH2O) to crystallize Ti-rich amphiboles. Mantle-derived amphiboles typically have homogeneous H contents. Megacrysts from maars and dikes have high H contents (OH ⬎ 1.1 atomic formula units) and individual crystals from a single locality have similar H contents. Amphiboles from lava flows and scoria cones have low to variable H contents (OH ⬍ 1.4 atomic formula units) and individual megacrysts from a single locality commonly have different H contents. Amphibole H contents and Fe3⫹/Fe2⫹ are a function of both initial crystallization conditions and dehydrogenation, with variations occurring due to different pressure-temperature-ƒH2-time paths. Amphibole dehydrogenation likely occurs at the surface or en route to the surface where ƒH2 is low, cooling is slow, or grain attributes tend to favor rapid H diffusion. We propose a model for calculating Fe3⫹ in Ti-rich kaersutites where Fe3⫹ ⫽ 2.47– 0.93(OH)–(Ti ⫹ Alvi). This equation takes into account crystallographic constraints within an amphibole structure. Our findings have implications for determining the primary oxygen fugacity of the mantle on Earth and Mars (using SNC meteorites). Amphiboles from rapidly cooled volcanic rocks have most likely retained their ‘primary’ OH and Fe3⫹/Fe2⫹ contents and are the best targets for calculating mantle oxygen fugacities and for stable isotopic analyses. Copyright © 1999 Elsevier Science Ltd noted the role that amphibole could have in mantle processes. Subsequently, experiments have been designed to determine amphibole stability as a function of pressure, temperature, composition, and volatile content (e.g., Nishikawa et al., 1970; Helz, 1973; Allen et al., 1975; Merrill and Wyllie, 1975; Allen and Boettcher, 1978; Jokobsson and Holloway, 1986; Oba et al., 1986; Clowe et al., 1988; Johnson et al., 1991; LaTourette et al., 1995; Popp et al., 1995b; Oba 1997). The volatile content of amphiboles has been used to estimate water activity (aH2O) of the melt from which the amphibole crystallized (Treiman, 1985; Harvey and McSween, 1992), and the hydrogen fugacity (ƒH2) of that melt. We examine the oxy-amphibole component in mantle-derived amphiboles using micro-analyses of megacrysts hosted in alkalic volcanic rocks from the southwestern USA (Fig. 1, details of the area are found in Wilshire et al., 1988). Samples are from lava flows, scoria cones, dikes and maars (Fig. 1). The micro-analytical data is used to calculate Fe3⫹ content and to investigate the chemical and physical parameters that control the oxy-component in amphiboles. We then present a model for the formation of H-deficient oxy-amphibole and discuss the implications of the findings for the origin of Ti-rich amphiboles and mantle ƒO2 and ƒH2.
1. INTRODUCTION
The role of volatile elements (principally H-C-O species) in melting and crystallization processes in the mantle has received a great deal of attention. Volatiles affect mantle melting, phase relationships, oxygen fugacity (ƒO2), and physical properties of melts such as density, viscosity and extrusive behavior. One of the difficulties in determining initial volatile content is that late-stage processes and magma emplacement processes (e.g., degassing, explosive eruption and alteration) often overprint the initial volatile contents. To determine the H content of the mantle, hydrous phases such as amphiboles and micas from alkali mafic rocks, and mantle-derived xenoliths and veins have been studied (e.g., Aoki, 1963; Best, 1970; Boettcher and O’Neil, 1980; McGuire et al., 1989; Deloule et al., 1991; Righter and Carmichael, 1993). Mantle amphiboles are often H-deficient oxy-amphiboles, commonly titanian pargasite or kaersutite (nomenclature of Leake et al., 1997), although they are also referred to as ‘basaltic hornblende’ or ‘oxy-hornblende’. Oxburgh (1964)
*Address reprint requests to P. King, Dept. Earth Sciences, University of Western Ontario, London ON N6A 5B7, Canada. 3635
3636
P. L King et al.
found similar results between Fe3⫹ (Mo¨ssbauer) and H content (manometry). The different conclusions could be due to inaccuracies in older H2O analyses performed using weight loss techniques. Saxena and Ekstro¨m (1970) identified the relation: Alvi ⫹ OH⫺ ⫽ Ti4⫹ ⫹ O2⫺
Fig. 1. Map of the southwestern USA showing the localities sampled for Ti-rich amphiboles. Localities are: 1) Lunar Craters; 2) Black Canyon (Hoover or Boulder Dam); 3) Toroweap (Grand Canyon); 4) Coliseum Maar; 5) Hopi Buttes dike (Joseph City); 6) Cleator; 7) Peridot Mesa (lava flow); 8) Soda Springs (cinder cone); 9) Kilbourne Hole; 10) Potrillo Maar (cinder cone). Crustal thickness estimates are taken from Wilshire et al. (1988).
This study differs from previous work on oxy-substitution in amphiboles (e.g. Leake, 1968; Saxena and Ekstro¨m, 1970; Popp et al., 1990; Popp and Bryndzia, 1992; Dyar et al., 1993) because it compares compositional changes within suites of Ca-amphiboles from individual localities and no alkali-rich amphiboles are included. In the past, oxy-substitution mechanisms have been investigated by combining analyses of amphiboles from multiple localities and we show that this results in false trends. Also, the micro-analytical data that we present have the advantage that mineral inclusions are excluded from the analysis and zonation can be examined. 1.1. Oxy-substitution/Dehydrogenation Mechanisms Proposed for Amphiboles The oxy-component in amphiboles has been studied for over 100 years. Early workers concentrated on changes in the optical properties of amphibole with heating and proposed that these resulted from oxidation (Schneider, 1891) or H-loss (Graham, 1926). Barnes (1930) summarized studies where Fe-bearing amphiboles were heated in the presence of more reducing gases at atmospheric pressure and the optical changes were reversed. This behavior is also observed at higher pressures (100 MPa, 650 –700°C: Clowe et al., 1988). Barnes (1930) introduced the idea that oxy-substitution occurs by both H-loss and oxidation. He did not examine potential substitutions in detail; however, his results support the substitution: Fe2⫹ ⫹ OH⫺ ⫽ Fe3⫹ ⫹ O2⫺ ⫹ 1/2 H2
(1)
Leake (1968) and Saxena and Ekstro¨m (1970) examined amphiboles reported in the literature and found no correlation between Fe2O3 and H2O content. In contrast, Aoki (1963) and Boettcher and O’Neil (1980) found a good inverse correlation between Fe2O3 (wet chemistry) and H2O⫹ content (loss on ignition and manometry respectively) and Dyar et al. (1992)
(2)
which is similar to other results: Ti ⬀ F content and Ti ⬀ (OH ⫹ F ⫹ Cl) content (Matson et al., 1984). We consider that all Ti is Ti4⫹ (Otten and Buseck, 1987; Waychunas, 1987), and that Ti3⫹ is negligible (c.f. Winchell, 1924). Substitutions 1 and 2 were examined in natural titanian pargasites and kaersutites (Popp and Brynzia, 1992) and synthetically treated amphiboles (Clowe et al., 1988; Phillips et al., 1988). Their results showed that Fe3⫹ as well as other octahedral ions are involved in oxidation, consistent with substitutions 1 and 2. A-site cations (Na, K, vacancy) may also have a role in maintaining charge balance in oxy-amphiboles (Rowbotham and Farmer, 1973; Phillips et al., 1988). Such a relation could have the form: A
(K⫹, Na⫹) ⫹ Fe2⫹ ⫽ Fe3⫹ ⫹ vacancy
(3)
Popp and coauthors (1995a) used results from hydrothermal experiments to show: 2.0 ⫽ Fe3⫹ ⫹ OH⫺ ⫹ F⫺ ⫹ Cl⫺ ⫹ Ti4⫹
(4)
They suggested that this relation is a combination of two 2⫹ substitution vectors: Fe3⫹O2⫺Fe⫺1 (OH⫺)⫺1 (substitution 1) n⫹ and Ti4⫹O2⫺R⫺1 (OH⫺)⫺1 (c.f. substitution 2) where Rn⫹ are octahedral cations (Alvi, Ti, Fe3⫹, Mg, Fe2⫹, Mn) selected to satisfy charge balance and stoichiometry. Popp and coauthors (1995a) also used natural samples to test relation 4 and a general trend was confirmed. Young and coauthors (1997) examined amphibole compositions using principle components analysis and concluded that substitution 1 (dehydrogenation) is the most important relation in oxy-substituted amphiboles. Their approach eliminated closure effects by examining exchange components (vectors) and also constrained the compositional space by examining data relative to an additive end member amphibole composition. 2. METHODS Amphibole megacrysts were collected from southwest USA (Fig. 1) and supplied by Righter and Carmichael (1993). The term ‘megacryst’ describes large crystals (1 to 7 cm long xenocrysts and phenocrysts) with at least one crystal face. We did not analyze amphiboles from veins, aggregates of crystals or with interstitial textural relations because those amphiboles may originate from other processes (e.g., Wilshire and Trask, 1971; Zanetti et al., 1997). The Potrillo Maar sample was partially surrounded by scoria indicating that it was derived from the scoria cone, not the maar. The Hopi Buttes and Cleator dike samples were surrounded by host rock. All the megacrysts from dikes were sampled at localities where the dike was less than 1.6 m wide. Most megacrysts were mounted as grains from broken fragments of a saw cut about 2 mm wide and approximately perpendicular to the c-axis of the crystal. At least 4 grain fragments were analyzed from each locality for major elements and H. One crystal from Cleator (CLRM) was sectioned approximately perpendicular to the c-axis to study variations between the core and rim. Our analyses were intended to characterize the oxy-component of the amphiboles, not specifically the compositional variation because the majority of mantle-derived
Oxy-substitution and dehydrogenation in amphibole megacrysts
3637
Table 1. Representative analyses of amphibole megacrysts. Locality
Crystal No.
Error
Coliseum
Cleator
MAAR
DIKE
Cleator rimmed crystal Ave. analyses
CM1
CM2
CM3
CL1
CL3
CL4
CL5
CLRM
CLRM
CM142
CM222
CM3.4
CL1.3
CL3.4
CL4.2
CL5.3
Rim
Core
40.69 40.65 4.45 4.82 14.34 13.51 9.69 10.39 na 0.01 0.08 0.15 14.11 12.50 11.56 11.14 2.51 2.41 1.90 2.20 0.00 0.11 0.00 0.02 0.08 0.13 1.53 1.47 0.01 0.01 100.93 99.51 charge corrected 5.94 6.05 2.03 1.93 0.43 0.44 1.18 1.29 0.49 0.54 0.01 0.02 3.07 2.77 1.81 1.78 0.71 0.70 0.35 0.42 0.00 0.01 16.03 15.94 0.00 0.01 0.03 0.06 1.49 1.46 22.48 22.48 0.17 0.14 1.02 1.16
37.65 3.29 16.74 18.10 0.01 0.44 7.38 10.94 1.97 2.42 0.07 0.01 0.12 1.38 0.01 100.51
38.32 3.24 16.68 15.57 0.00 0.17 8.82 11.21 1.93 2.53 0.05 0.01 1.12 1.47 0.01 100.12
39.43 2.27 15.89 15.90 0.10 0.35 9.58 10.94 2.44 2.12 0.00 0.04 0.14 1.59 0.01 100.79
38.10 3.60 16.74 15.45 0.00 0.12 9.04 11.22 1.98 2.57 0.02 0.02 0.09 1.37 0.01 100.31
38.66 5.75 13.90 11.63
38.21 3.32 15.61 17.77
0.13 12.04 12.49 2.08 1.75 0.09 0.02 0.17 1.19
0.23 8.16 11.17 2.07 2.42 0.04 0.01 0.14 1.14
99.87
5.78 2.22 0.80 2.32 0.38 0.06 1.69 1.80 0.59 0.47 0.01 16.11 0.00 0.06 1.41 22.53 ⫺0.02 2.32
5.82 2.18 0.81 1.98 0.37 0.02 2.00 1.82 0.57 0.49 0.01 16.06 0.00 0.06 1.49 22.45 ⫺0.10 1.98
5.94 2.06 0.77 2.00 0.26 0.04 2.15 1.77 0.71 0.41 0.00 16.11 0.01 0.07 1.59 22.33 ⫺0.04 2.00
5.79 2.21 0.78 1.96 0.41 0.02 2.05 1.83 0.58 0.50 0.00 16.12 0.01 0.04 1.39 22.56 ⫺0.02 1.96
5.78 2.16 0.29 1.45 0.65 0.02 2.68 2.00 0.60 0.33 0.01 15.97 0.00 0.08 1.19 22.73 0.43 1.02
wt% ⴞ 41.44 SiO2 TiO2 4.09 AI2O3 13.38 FeO tot 9.37 Cr2O3 na MnO 0.08 MgO 14.23 CaO 11.39 Na2O 2.66 K2O 1.69 P2O5 0.05 CI 0.00 F 0.08 H2O 1.64 H2O ⫾ 0.01 Total 100.09 Ions per formula unit, based on 24 oxygens, Si 0.03 6.07 Aliv 0.04 1.91 AIvi 0.04 0.40 Fe total 0.04 1.15 Ti 0.01 0.45 Mn 0.01 0.01 Mg 0.04 3.11 Ca 0.03 1.79 Na 0.02 0.76 K 0.01 0.32 P 0.01 0.01 Total 15.97 Cl 0.00 F (see text) 0.03 OH 1.60 O remain 22.36 Fe3⫹ calc 0.12 Fe2⫹ calc 1.03
100.3 5.85 2.10 0.71 2.27 0.38 0.03 1.86 1.83 0.61 0.47 0.01 16.14 0.00 0.07 1.16 22.77 0.29 1.98
*Where Fe3⫹ calc is less than zero, Fe2⫹ calc is taken to equal Fe total (Continued)
megacrysts are homogeneous (Binns et al., 1970; Irving, 1974; Righter and Carmichael, 1993). Electron microprobe data were obtained at Arizona State University (ASU) using a JEOL JXA 8600 electron microprobe with a 15 kV accelerating voltage, a 10 nA, 15 m beam and wavelength dispersive spectroscopy. For the Lunar Crater and Peridot Mesa amphiboles electron probe analyses were made at both ASU and at the University of Arizona (U of A) using a Cameca SX-50 with 20 kV accelerating voltage, and a 10 nA, 15 m beam. At each location, ZAF correction procedures were used and Na and K were the first elements analyzed. Analyses at both locations were calibrated against the Kakanui hornblende (Smithsonian 143965, Jarosewich, 1975) and a kaersutite from Harrat Hutaymah (Harvard Mineralogical Museum H336-92, McGuire et al., 1992). The U of A analyses are within a 2 sigma error of the analyses made at ASU which are reported in this paper. However, the ASU and U of A values for Al, F and Na are higher and for Mn are lower than those reported by Righter and Carmichael (1993). Errors on the ASU analyses were evaluated based on 1 standard deviation of the atomic formula unit (afu, based on a 24 oxygen formula) values for Kakanui hornblende which was typically analyzed every tenth analysis throughout an analytical session (Table 1).
Fluorine analysis was problematic because it is barely detected using a TAP spectrometer (Righter and Carmichael, 1993) and using a LDEI detector the F K-␣ peak (0.677 keV) overlaps with the Fe L-␣1 peak (0.705 keV). We applied a correction based on repeated analyses of the Kakanui hornblende (FeOtotal ⫽ 10.9 wt.% nominal value for F is 0.15 wt.%; Mason, 1966). This correction over-estimates F in amphiboles with FeOtotal greater than 10.9 wt.% by at most 0.07 afu and underestimates F in amphiboles with less FeOtotal by a maximum of 0.03 afu. However, those maximum values are probably excessive because there is no correlation between corrected F content and FeOtotal. Another problem with F analysis is that F may diffuse in apatite during calibration (Stormer et al., 1993), although this was not observed by Peng et al. (1997). For the Peridot Mesa and Lunar Craters samples, we used published Fe3⫹/Fe2⫹ (Righter and Carmichael, 1993; wet chemistry) to calculate Fe2O3 and FeO from total FeO determined on the electron probe. We chose to use Fe3⫹/Fe2⫹, rather than the FeO bulk analyses, since the ratio is less sensitive to small scale variations in FeOtotal content that are encountered with micro-analysis. Hydrogen was determined with a Cameca IMS-3F ion microprobe at ASU. The primary mass-filtered 16O⫺ ion beam was accelerated to 12.5
3638
P. L King et al. Table 1. Continued
Locality
Kilbourne Hole MAAR
Hopi Buttes
Black Canyon Dam
DIKE
DIKE
Crystal
KH
HP2
No.
KH.2
HP2.2
HP3
HP4
HP5
BD1
BD2
BD3
BD4
BD5
HP3.2
HP4.3
HP5.3
BD1.1
BD232
BD3.3
BD412
BD5.1
41.35 4.67 13.70 9.94 0.03 0.00 13.71 10.58 2.49 1.94 0.00 0.01 0.13 1.30 0.01 99.84
40.00 5.64 14.70 9.80 na 0.14 13.00 11.00 2.29 1.86 0.00 0.00 0.17 1.13 0.04 99.72
40.52 5.80 14.76 9.04 na 0.08 13.88 11.03 2.41 1.62 0.00 0.02 0.08 1.12 0.04 100.36
40.39 5.64 14.90 8.34 na 0.10 14.20 11.00 2.39 1.90 0.00 0.00 0.15 1.12 0.04 100.13
40.71 5.69 15.02 8.77 na 0.05 14.33 11.05 2.37 1.73 0.00 0.00 0.08 1.11 0.04 100.90
39.90 5.59 14.40 9.46 na 0.08 13.60 10.90 2.39 1.86 0.00 0.00 0.14 1.12 0.04 99.44
6.09 1.87 0.50 1.22 0.52 0.00 3.01 1.67 0.71 0.36 0.00 15.95 0.00 0.06 1.27 22.67 0.26 0.96
5.92 2.04 0.52 1.21 0.63 0.02 2.87 1.74 0.66 0.35 0.00 15.95 0.00 0.08 1.11 22.81 0.29 0.92
5.92 2.03 0.51 1.10 0.64 0.01 3.02 1.73 0.68 0.30 0.00 15.95 0.00 0.03 1.10 22.86 0.30 0.80
5.91 2.04 0.52 1.02 0.62 0.01 3.10 1.72 0.68 0.35 0.00 15.98 0.00 0.07 1.09 22.84 0.31 0.71
5.91 2.04 0.53 1.06 0.62 0.01 3.10 1.72 0.67 0.32 0.00 15.97 0.00 0.03 1.08 22.89 0.32 0.74
5.90 2.04 0.47 1.17 0.62 0.01 3.00 1.73 0.69 0.35 0.00 15.98 0.00 0.07 1.11 22.83 0.35 0.83
wt% 39.40 40.93 40.78 41.16 SiO2 TiO2 4.95 4.49 4.8 4.37 Al2O3 15.00 14.34 14.17 13.15 FeO total 10.20 9.20 9.74 11.33 Cr2O3 na 0.04 0.04 0.00 MnO 0.13 0.00 0 0.09 MgO 12.80 14.49 13.68 12.92 CaO 10.80 10.90 10.61 10.59 Na2O 2.40 2.44 2.48 2.45 K2O 1.80 2.04 2.003 2.07 P2O5 0.09 0.05 0 0.00 Cl 0.03 0.04 0.05 0.02 F 0.17 0.12 0.13 0.15 H2O 1.39 1.50 1.446 1.48 H2O ⫾ 0.02 0.01 0.01 0.01 Total 99.15 100.57 99.93 99.77 Ions per formula unit, based on 24 oxygens, charge corrected Si 5.87 5.97 6.01 6.11 Aliv 2.11 2.00 1.97 1.86 Alvi 0.53 0.47 0.49 0.44 Fe total 1.27 1.12 1.20 1.41 Ti 0.56 0.49 0.53 0.49 Mn 0.02 0.00 0.00 0.01 Mg 2.84 3.15 3.00 2.86 Ca 1.73 1.70 1.67 1.68 Na 0.69 0.69 0.71 0.70 K 0.34 0.38 0.38 0.39 P 0.01 0.01 0.00 0.00 Total 15.97 15.99 15.96 15.95 Cl 0.01 0.01 0.01 0.01 F 0.08 0.06 0.06 0.07 OH 1.38 1.46 1.42 1.46 O remain 22.53 22.48 22.51 22.47 Fe3⫹ calc 0.10 0.15 0.13 0.19 Fe2⫹ calc 1.17 0.97 1.07 1.22
(Continued)
kV at a current of 1 nA. Positive secondary ions were accelerated to ⬃4.5 kV and only those with 75 ⫾ 20 eV excess kinetic energy were allowed into the mass spectrometer (Ihinger et al., 1994). The entrance and exit slits were fully open. Analyses did not commence until the base pressure in the sample chamber was less than 5 ⫻ 10⫺8 torr which was achieved by baking the machine prior to an analytical session. Following Deloule et al. (1995), we used a liquid nitrogen cold trap for most analyses to help minimize the H background. For H analysis, the sample disc was polished slightly to remove the C coat used for electron probe analysis and then coated with Au. Four non-consecutive analyses were made on fragments of each amphibole, slightly deeper in the sample mount than the electron probe analysis. Background measurements were made every four to five analyses (about once an hour) using a number of nominally volatile-free samples: alkali basalt and andesite glasses prepared in a one atmosphere furnace, and Lake County plagioclase (SM115900). The beam was initially rastered over a 25 ⫻ 25 m2 area for 7 min prior to analysis with a stationary beam (10 to 15 m diameter). H⫹ intensities were normalized to 30Si⫹ and then converted to H2O using a calibration curve based on kaersutites to titanian pargasites for which H content was determined manometrically (Fig. 2). Four kaersutites were analyzed for H and C at University Tu¨bingen, Germany,
following techniques slightly modified from Vennemann and O’Neil (1993). We also analyzed other amphibole samples (McGuire et al., 1992; Dyar et al., 1993; Feldstein et al., 1996). The kaersutite to titanian pargasite calibration slope is essentially constant for three analytical sessions (Fig. 2). Matrix effects due to changing composition were considered minimal because of the limited compositional range. Thus, the major error source for the H⫹ ion microprobe analyses is the variation in the background analyses, while error from counting statistics is minor (⬃3% relative). The background error was determined as the average value for an entire analytical session. If the background measurements before and after each calibration measurements are used, then the slopes in Figure 2 are reduced by 0.01. We have chosen to use the average background values (c.f. King et al., 1998). Assuming that there is no error in the calibration slope, errors for OH content based on background reproducibility were from ⫾0.02 afu to ⫾0.05 afu (Table 1). 3. RESULTS
Representative micro-analyses of the amphiboles are presented in Table 1. In total, 130 H and major element microanalyses were obtained on 38 individual megacrysts. More than
Oxy-substitution and dehydrogenation in amphibole megacrysts
3639
Table 1. (Continued) Locality
Potrillo Maar SCORIA
Soda Springs SCORIA
Crystal
PM1
SS1
SS2
SS3
SS4
SS6
SS6
SS6
SS6
No.
PM1.3
SS1.1
SS2.2
SS3.1
SS4.1
SS6.1
SS6.2
SS6.3
SS6.4
wt% SiO2 TiO2 Al2O3 FeO total Cr2O3 MnO MgO CaO Na2O K2O P2OS Cl F H2O H2O ⫾ Total Ions per formula Si Aliv Alvi Fe total Ti Mn Mg Ca Na K P Total Cl F OH O remain Fe3⫹ calc Fe3⫹ calc
38.61 40.29 39.80 39.40 4.16 4.74 4.49 4.94 16.00 13.60 13.30 14.10 13.00 16.50 17.50 16.30 na na na na 0.10 0.30 0.36 0.19 11.00 9.83 9.42 9.68 11.00 10.10 9.89 10.10 2.40 2.90 2.86 2.93 1.59 1.13 1.13 1.11 0.10 0.00 0.11 0.02 0.10 0.01 0.02 0.03 0.20 0.15 0.17 0.16 1.42 0.31 1.22 1.14 0.05 0.05 0.05 0.05 99.68 99.86 100.27 100.10 unit, based on 24 oxygens, charge corrected 5.80 6.07 6.02 5.95 2.19 1.77 1.92 2.00 0.64 0.65 0.45 0.51 1.63 2.08 2.21 2.06 0.47 0.54 0.51 0.56 0.01 0.04 0.05 0.02 2.46 2.21 2.12 2.18 1.77 1.63 1.60 1.63 0.70 0.85 0.84 0.86 0.30 0.22 0.22 0.21 0.01 0.00 0.01 0.00 16.01 16.05 15.96 15.99 0.03 0.00 0.01 0.01 0.10 0.07 0.08 0.07 1.42 0.31 1.23 1.15 22.46 23.61 22.68 22.77 0.04 0.99 0.36 0.32 1.60 1.09 1.85 1.73
39.50 4.95 13.80 16.10 na 0.26 9.83 10.10 2.99 1.13 0.07 0.02 0.16 0.81 0.05 99.71
39.50 5.15 13.60 15.10 na 0.28 10.10 10.30 2.97 1.11 0.00 0.00 0.15 0.58 0.05 98.84
40.10 4.90 13.60 16.20 na 0.27 9.93 10.40 2.97 1.10 0.07 0.02 0.14 0.36 0.05 100.06
39.50 5.30 13.80 15.80 na 0.23 10.10 10.50 2.88 1.12 0.02 0.02 0.14 0.92 0.05 100.33
39.50 4.67 13.30 17.20 na 0.30 9.43 10.10 2.95 1.11 0.05 0.05 0.16 0.24 0.05 99.06
5.98 1.93 0.53 2.04 0.56 0.03 2.22 1.64 0.88 0.22 0.01 16.03 0.01 0.07 0.82 23.10 0.61 1.42
6.00 1.87 0.56 1.92 0.59 0.04 2.29 1.68 0.88 0.22 0.00 16.04 0.00 0.07 0.59 23.34 0.77 1.15
6.03 1.81 0.60 2.04 0.55 0.03 2.23 1.68 0.87 0.21 0.01 16.05 0.01 0.07 0.36 23.56 0.98 1.05
5.94 1.98 0.47 1.99 0.60 0.03 2.26 1.69 0.84 0.21 0.00 16.01 0.01 0.06 0.92 23.01 0.54 1.44
6.03 1.80 0.60 2.19 0.54 0.04 2.14 1.65 0.87 0.22 0.01 16.08 0.01 0.08 0.24 23.67 1.11 1.08 (Continued)
75% of the analyses have total oxide (including H2O) and anion sums between 99.0 and 101.0 wt.%. Amphiboles include titanian pargasites and kaersutites (nomenclature of Leake et al., 1997) with Ti contents greater than 0.40 afu for most samples. The Cleator and some Toroweap samples are titanian pargasites (Ti contents between 0.26 and 0.40 afu). Magnesium numbers (100 ⫻ atomic Mg/(Mg ⫹ Fetotal)) range from 41 to 78. Amphiboles were recalculated on a 24 oxygen mole basis (following Jackson et al., 1967) and cation site calculations were made following Schumacher (1997). Overall, the major substitution vectors observed within iv amphiboles from a single locality are M4NaSiCa⫺1 Al⫺1 (plaiv gioclase), TiAl2 Mg⫺1Si⫺2 (Ti-tschermak), FeMg⫺1 and vi TiAlivAl⫺1 Si⫺1 (Fig. 3 a– d). Note that the MgFe⫺1 substitution (Fig. 3c) causes parallel trends on diagrams that include Mg (Figs. 3b and 3e). In most cases, the major element compositions do not differ between fragments of a megacryst. The Black Canyon Dam megacrysts have the most homogeneous major element com-
position (also reported by Nielson and Nakata, 1994). The Toroweap amphiboles show wide compositional variation, as noted by Best and Brimhall (1974). The Peridot Mesa samples also show a large compositional range, possibly because some crystals are fragments of cumulates. Amphiboles from host rocks that were emplaced by different processes (e.g., dike, lava flow, scoria cone, or maar) have similar major element compositions (Fig. 3 a– e). That is, there is no relationship between composition and emplacement process. However, the H contents of amphiboles from host rocks emplaced under different conditions are bimodal (Fig. 4). Amphiboles from rapidly cooled rocks, dikes and maars have relatively high H contents (OH ⬎ 1.1 afu, median OH ⫽ 1.4 afu). Amphiboles from scoria cones and lava flows have a wide range of H contents (OH ⫽ 0 to 1.4 afu, median OH ⫽ 0.9 afu). The OH content measured by SIMS is generally lower than the OH content predicted if the O(3) site is fully occupied by OH, F and Cl (Fig. 4, as noted by previous workers, e.g., Dyar et al., 1993).
3640
P. L King et al. Table 1. (Continued)
Locality
Crystal No.
Peridot Mesa LAVA
Toroweap LAVA
Lunar Craters LAVA
SC1
SC3
SC5
TR1
TR2
TR4
TR5
LC1
LC2
LC2
SC1.1
SC3.3
SC5.3
TR1.1
TR2.2
TR4.4
TR5.2
LC1.1
LC2.1
LC2.2
40.29 3.91 15.40 8.25 na 0.05 14.80 10.40 2.93 1.64 0.05 0.01 0.11 1.24 0.04 99.07
41.09 3.85 14.90 7.96 na 0.06 14.90 10.60 2.79 1.80 0.00 0.030 0.014 1.29 0.04 99.40
41.09 3.81 15.50 8.18 na 0.12 14.60 10.30 2.98 1.63 0.09 0.00 0.11 1.25 0.04 99.66
39.69 6.21 15.43 10.81 na 0.03 11.90 11.14 2.58 1.12 0.00 0.00 0.15 0.33 0.01 99.39
40.17 5.82 15.51 10.88 na 0.08 12.09 11.54 2.60 0.98 0.00 0.02 0.12 0.81 0.01 100.63
39.82 5.73 15.48 11.01 na 0.15 11.76 11.47 2.72 1.08 0.00 0.00 0.10 0.80 0.01 100.11
5.93 2.03 0.65 1.02 0.43 0.01 3.25 1.64 0.84 0.31 0.01 16.10 0.00 0.05 1.22 22.73 0.26 0.76
6.02 1.94 0.63 0.98 0.42 0.01 3.25 1.66 0.79 0.34 0.00 16.05 0.01 0.06 1.26 22.67 0.24 0.73
6.01 1.96 0.71 1.00 0.42 0.01 3.18 1.61 0.84 0.30 0.01 16.06 0.00 0.05 1.22 22.73 0.21 0.79
5.89 1.99 0.71 1.34 0.69 0.00 2.63 1.77 0.74 0.21 0.00 15.98 0.00 0.07 0.32 23.61 0.77 0.57
5.89 2.03 0.65 1.34 0.64 0.01 2.64 1.81 0.74 0.18 0.00 15.95 0.00 0.06 0.79 23.15 0.45 0.89
5.89 2.04 0.65 1.36 0.64 0.02 2.59 1.82 0.78 0.20 0.00 15.99 0.00 0.04 0.78 23.17 0.45 0.91
Wt% 39.78 40.10 39.54 40.89 SiO2 TiO2 4.87 4.79 5.48 3.60 Al2O3 13.98 14.01 14.76 15.70 FeO total 17.01 16.98 13.38 10.20 Cr2O3 na na na na MnO 0.26 0.18 0.14 0.14 MgO 8.94 8.69 10.79 13.20 CaO 10.55 10.30 10.24 10.20 Na2O 2.46 2.47 2.88 2.79 K2O 1.97 1.96 1.25 1.27 P2O5 0.02 0.00 0.00 0.00 Cl 0.03 0.04 0.03 0.01 F 0.10 0.11 0.00 0.11 H2O 0.50 0.99 1.01 1.39 H2O ⫾ 0.02 0.01 0.01 0.04 Total 100.47 100.61 99.49 99.50 Ions per formula unit, based on 24 oxygens, charge corrected Si 6.01 6.06 5.49 6.03 Aliv 1.86 1.88 2.01 1.96 Alvi 0.63 0.61 0.60 0.77 Fe total 2.15 2.15 1.68 1.26 Ti 0.55 0.54 0.62 0.40 Mn 0.03 0.02 0.02 0.02 Mg 2.01 1.96 2.42 2.90 Ca 1.71 1.67 1.65 1.61 Na 0.72 0.72 0.84 0.80 K 0.38 0.38 0.24 0.24 P 0.00 0.00 0.00 0.00 Total 16.07 15.99 16.02 15.99 Cl 0.01 0.01 0.01 0.00 F 0.05 0.05 0.00 0.05 CH 0.50 1.00 1.01 1.37 O remain 23.44 22.94 22.98 22.58 Fe3⫹ calc 0.82 0.38 0.31 0.03 Fe2⫹ calc 1.33 1.76 1.37 1.23
*Where Fe3⫹ calc is less than zero, Fe2⫹ calc is taken to equal Fe total
Our reconnaissance data (typically four analyses per megacryst) show that H contents within single megacrysts are homogeneous (e.g., LC2; Table 1) or less commonly heterogeneous (e.g., SS6; Table 1). Typically, the range of OH contents within a megacryst is less than ⫾0.2 afu. Mantle amphiboles from other localities also have no significant variation in H content (measured as counts per second/primary beam intensity; Deloule et al., 1991). Amphiboles from scoria cones and lava flows show the most variability between individual crystals from a single locality (e.g., Lunar Craters).
4. EVALUATION OF PROPOSED SUBSTITUTION MECHANISMS
Figure 5 illustrates the relations between Fe3⫹ and OH⫺ content where Fe3⫹/Fe2⫹ was determined using wet chemistry (Righter and Carmichael, 1993) or Mo¨ssbauer spectroscopy (Dyar et al., 1993). We chose Dyar’s data set because they
determined H content using modern analytical techniques (manometry), rather than weight loss techniques. Also, Popp et al. (1995a) comment that analyses from other data sets are of dubious quality (e.g., Popp and Bryndzia, 1992). The data in Figure 5 support previous workers’ findings that Fe3⫹ and H2O⫹ or H content are inversely related in mantle kaersutites (Aoki, 1963; Boettcher and O’Neil, 1980; Dyar et al., 1993). The relationship does not rule out substitution 1. Furthermore, most samples have Fe3⫹ contents below the maximum possible Fe3⫹ content based on the stoichiometry of substitution 1, measured OH content and assumptions that (O ⫹ OH) equals 24 afu and initial Fe2⫹ is greater than 2 afu. That relation means that there is sufficient Fetotal for all Fe3⫹ to be produced by H-loss from the amphibole megacryst via substitution 1. Figure 5 also illustrates the relation proposed by Popp et al. (1995a). They used similar amphibole data to argue that this relation can be used to predict Fe3⫹ if H contents are known.
Oxy-substitution and dehydrogenation in amphibole megacrysts
3641
Table 1. Continued Locality Crystal No.
Lunar Craters LAVA LC2
LC2
LC3
LC5
LC6
LC2.3
LC2.4
LC3.1
LC5.3
LC6.2
Wt% SiO2 40.45 40.11 TiO2 5.56 5.56 Al2O3 15.48 15.39 FeO total 10.82 10.99 Cr2O3 na na MnO 0.08 0.00 MgO 12.27 12.25 CaO 11.67 11.18 Na2O 2.65 2.62 K2O 1.05 0.98 P2O5 0.02 0.05 Cl 0.00 0.02 F 0.08 0.12 H2O 0.94 0.83 H2O ⫾ 0.01 0.01 Total 101.07 100.09 Ions per formula unit, based on 24 Si 5.91 5.91 Aliv 2.03 2.02 vi Al 0.63 0.65 Fe total 1.32 1.35 Ti 0.61 0.62 Mn 0.01 0.00 Mg 2.67 2.69 Ca 1.83 1.76 Na 0.75 0.75 K 0.20 0.18 P 0.00 0.01 Total 15.96 15.95 Cl 0.00 0.00 F 0.04 0.06 OH 0.92 0.81 O remain 23.05 23.13 Fe3⫹ calc 0.37 0.45 Fe2⫹ calc 0.95 0.91
39.71 40.26 39.88 5.82 5.58 5.87 15.16 15.51 15.37 11.14 10.50 10.94 na na na 0.06 0.10 0.08 11.91 12.43 11.60 11.39 11.54 11.57 2.53 2.70 2.92 1.16 1.11 1.14 0.02 0.00 0.00 0.00 0.02 0.00 0.15 0.12 0.14 0.22 1.04 1.10 0.02 0.01 0.01 99.27 100.90 100.61 oxygens, charge corrected 5.90 5.89 5.88 1.95 2.06 2.09 0.70 0.61 0.58 1.39 1.28 1.35 0.65 0.61 0.65 0.01 0.01 0.01 2.64 2.71 2.55 1.81 1.81 1.83 0.73 0.77 0.83 0.22 0.21 0.22 0.00 0.00 0.00 16.01 15.97 15.98 0.00 0.00 0.00 0.07 0.05 0.06 0.22 1.01 1.08 23.71 22.93 22.85 0.91 0.30 0.23 0.48 0.98 1.12
That relation also shows that Ti may have a role in chargebalancing H. Some authors have suggested that A-site cations (Na, K, vacancy) are also important in oxy-substitution relations, for example, substitution 3. Popp and Bryndzia (1992) suggested that Fe3⫹ varies with Aliv ⫹ M4Na content. We did not observe any correlation in the mantle amphibole data set that suggests a role for A-site or alkali element substitutions (Fig. 6), probably because no alkali-rich amphiboles were included. Because Fe3⫹ is correlated with OH there is similarly no bivariate relation observed between A-site cations and OH content. 5. CALCULATING THE FE3ⴙ CONTENT OF TI-RICH AMPHIBOLES
A statistical analysis of Ti-rich amphiboles with known Fe3⫹/Fe2⫹ (from Fig. 5) shows that overall OH content correlates well with (Alvi ⫹ Ti ⫹ Fe3⫹) content (r2 ⫽ 0.93; Fig. 7a). The linear regression in Figure 7a leads to the following equation:
Fig. 2. Calibration lines for ion microprobe measurement of H2O for the SW USA kaersutite to titanian pargasite megacrysts and those analyzed for H using manometry by Dyar et al. (1993). Secondary ions with excess kinetic energies of 75 ⫾ 20 eV were detected. Errors in H⫹/30Si⫹ ⫻ SiO2 (weight fraction) values result from the background, not from counting statistics. The slope for the calibration is essentially constant over different analytical sessions. Note that sample “Feldstein” is a hornblende, not a kaersutite.
Fe3⫹ ⫽ 2.47 ⫺ 0.93 (OH) ⫺ (Ti ⫹ Alvi)
(5)
This correlation allows us to calculate Fe3⫹/Fe2⫹ for the samples with H micro-analyses and better evaluate oxy-substitution mechanisms. The methods used to account for charge balance are discussed in detail in Appendix 1. For Ti-rich amphiboles, most models for predicting Fe3⫹/ 2⫹ Fe directly from electron microprobe analyses are not applicable because 4⫹ cations are not included in charge balance and site occupancy assumptions. The model for predicting Fe3⫹ content (Eqn. 5, Fig. 7a, r2 ⫽ 0.93) compares favorably with the technique proposed by Popp et al. (1995a). Error propagation on several elements, using two different analytical techniques, results in error bars of ⫾0.2 afu. The error is likely smaller (⬍⫾0.1 afu) because calculated Fe3⫹ contents are within error of synchrotron micro-X-ray absorption near edge structure measurements of 11 amphiboles with a range of Fe3⫹ contents (King et al., in prep.). The H-Alvi-rich, Ti-poor pargasites from the Cleator dike have negative calculated Fe3⫹ contents (Table 1). They represent an extrapolation of Eqn. 5 beyond the compositional range of the kaersutites shown in Figure 7a. We do not recommend that the equation be applied to samples with relatively high Alvi (0.8 afu). Furthermore, the octahedral cations Mn3⫹, Sc3⫹ and V3⫹ are not included in Eqn. 5 and if they are in sufficient concentrations they may have a role in charge-balance of the H-rich samples. The Cleator samples have up to 0.12 wt.% Cr2O3 (Table 1) although most have Cr2O3 contents below the electron microprobe detection limit and so it is unlikely to affect Fe calculations. Another complication is that exsolution of Fe-Ti oxides changes amphibole compositions and invalidates Eqn. 5. It is important to consider these factors when applying any Fe3⫹ calculation model for amphiboles. The amphiboles from dikes and maar deposits all have low calculated Fe3⫹ content (less than 0.42 afu; Table 1), reflecting
3642
P. L King et al.
iv Fig. 3. Major substitutions observed in the mantle-derived amphiboles from the southwest USA. a) M4NaSiCa⫺1Al⫺1 iv vi (plagioclase) substitution, b) TiAliv 2 Mg⫺1Si⫺2 (Ti-tschermak) substitution, c) FeMg⫺1 substitution; d) TiAl Al⫺1 Si⫺1 substitution, e) M4NaAlviCa⫺1 Mg⫺1 (glaucophane) substitution. Cations are calculated using a 24 O formula. Errors are calculated using the errors shown in Table 1 and following the technique described in the text. Note that the Ti-tschermak substitution would converge to one line if Fe2⫹ was added to Mg.
Oxy-substitution and dehydrogenation in amphibole megacrysts
3643
eruption of host rocks (Schneider, 1891; Aoki, 1963; Dyar et al., 1992). Initial crystallization conditions influence amphibole compositions overall. Dehydrogenation changes H and Fe3⫹/ Fe2⫹ contents, with no major structural changes to the amphibole.
6.1. Initial Crystallization Conditions Variations in amphibole chemistry as a function of bulk composition and crystallization conditions were documented by Gilbert et al. (1981), but more recent work has expanded the available data set. To evaluate oxy-substitution and dehydrogenation, we summarize work on the octahedral cation contents (Ti, Alvi and Fe) and H contents of amphiboles.
6.1.1. Composition
Fig. 4. Histograms of OH content (afu) in amphiboles from SW USA alkali basalts. A total of 130 analyses are included from 10 localities and 38 individual megacrysts. Note that the dikes and maar deposits have relatively restricted OH contents greater than 1.1 afu with a median value of 1.4 afu. These samples include Black Canyon (formerly Hoover Dam or Boulder Dam) dike, a dike in the Hopi Buttes volcanic field, a dike in the Cleator district, Arizona; Kilbourne Hole (maar), New Mexico and Coliseum Maar (Hopi Buttes), Arizona. Samples from lava flows and/or scoria cones show a wide range of OH contents. These sample localities include Easy Chair Crater, Lunar Craters, Nevada; Peridot Mesa and Toroweap flows, Arizona; Soda Springs scoria cone, Arizona; and a scoria cone within Potrillo Maar, New Mexico.
their high H content. In contrast, the amphiboles from lava flows and scoria cones have variable calculated Fe3⫹ content (0.03 to 1.17 afu). Because most the amphiboles have homogeneous H contents, the calculated Fe3⫹ value is homogeneous within a sample. For the samples with variable H contents we predict that Fe3⫹ content varies on a micro-scale (Table 1), consistent with Fe3⫹/Fe2⫹ micro-analyses of other amphiboles (Delaney et al., 1996; Dyar et al., 1998). 6. WHAT AFFECTS MANTLE-DERIVED AMPHIBOLE COMPOSITIONS?
Mantle-derived amphibole megacrysts have high Ti and variable Fe3⫹/Fe2⫹ and H contents that have been ascribed to two major processes: First, the chemical features may result from oxy-substitution in amphiboles during crystal growth in low hydrogen fugacity (low ƒH2) conditions (Dyar et al., 1993; Popp et al., 1995b). Second, the chemical features may result from dehydrogenation of amphiboles during the ascent and
Amphibole megacrysts in alkali basalts grow from maficultramafic bulk compositions. Sources could include metasomatically-derived melt or fluid (review in Ionov et al., 1997), the basaltic host melt (Binns, 1969; Green and Hibberson, 1970), and mantle cumulates, dikes and pegmatites (Best, 1970; Wilshire and Trask, 1971; Irving, 1974; Righter and Carmichael, 1993). In order to distinguish between different sources it is necessary to examine other mineral phases, magma compositions and amphibole trace element compositions (e.g., Ionov and Hofmann, 1995; Zanetti et al., 1995). Such a study is beyond the scope of this paper; however, the cleavage in the megacrysts suggests they were transported free floating in a melt.
6.1.2. Pressure and temperature Pressure and temperature control TiO2 and Al2O3 activities during amphibole crystallization. Increasing pressure results in higher Alvi contents and lower Ti contents in amphiboles from sub- and super-solidus experiments (Holloway and Burnham, 1972; Helz, 1973; Oba et al., 1986; Adam and Green, 1994; Fujinawa and Green, 1997; Ernst and Liu, 1998 and references therein). High temperatures (⬎950°C) favor Alvi-Ti-rich Caamphibole compositions (Holloway, 1973; Allen and Boettcher, 1978; Wallace and Green, 1991). Adam and Green (1994) studied alkali basalt compositions and proposed some qualitative pressure and temperature-dependent substitutions. Substitutions that may be favored by low-moderate pressures (ⱕ1.0 vi Gpa) and high temperatures (ⱖ950°C) include TiAlivAl⫺1 Si⫺1, iv iv M4 TiAl2 Mg⫺1Si⫺2 and NaSiCa⫺1Al⫺1. Note that Ernst and Liu (1998) showed that the substituvi tion M4NaAlviCa⫺1Mg⫺1 (glaucophane) is the major pressure-dependent relation affecting Alvi content in amphiboles synthesized from metamorphosed basalt up to 950°C. This substitution does not explain the Alvi contents of the SW USA mantle-derived amphiboles (Fig. 3e), probably because they crystallized at higher temperatures from different bulk compositions. Future experimental work on H content and Fe3⫹/Fe2⫹ of synthesized amphiboles is needed to evaluate the role of pressure and temperature on oxy-substitution.
3644
P. L King et al.
Fig. 5. Measured Fe3⫹ versus OH content (afu based on a 24 oxygen amphibole formula) for the kaersutite to titanian pargasite samples. Note that Fe3⫹ and OH contents have a correlation of r2 ⫽ 0.76. Also shown on the diagram is Fe3⫹ versus (OH ⫹ F ⫹ Cl ⫹ Ti) content. That relation was used by Popp et al. (1995a) to determine a model for Fe3⫹ calculation, however, we propose that the relation shown in Figure 7a is a superior method. Error bars for Fe3⫹ content are not included because they were not given in the original data sources.
6.1.3. Volatile activities and ƒO2 Volatile activities and fugacities are often inter-related. For instance, in the presence of a fluid phase, water fugacity (ƒH2O) is a function of total pressure and aH2O: fH2O ⫽ f oH2O ⫻ aH2O, where ƒ oH2O is the fugacity of pure H2O at pressure and temperature. Oxygen fugacity and ƒH2 may also be related to H2O because in an equilibrium melt: H2O ⫽ H2 ⫹
⁄ O2. Thus, amphibole H contents may be related to H2O, ƒO2 and/or ƒH2. Several authors have proposed that Ti-rich amphiboles crystallize from H-poor melts (e.g., Dyar et al., 1993; Watson et al., 1994). However, based on phase equilibria, anhydrous minerals such as olivine and pyroxene are more likely to crystallize from a H-poor basaltic melt than amphibole (e.g., Holloway, 1973). 12
Fig. 6. Measured Fe3⫹ versus M4Na ⫹ Aliv and ANa contents. Note that there is not a good correlation between those variables, in contrast to suggestions that oxy-substitution follows substitution 3. Error bars are not included because they were not given in the original sources.
Oxy-substitution and dehydrogenation in amphibole megacrysts
3645
Fig. 7. a (Ti ⫹ Alvi ⫹ Fe3⫹) versus OH content for samples with known Fe3⫹/Fe2⫹, where cations are calculated based on a 24 oxygen formula. Note the high correlation coefficient (r2 ⫽ 0.93). This relation is used to calculate Fe3⫹ for samples with known OH content (Eqn. 5). Error bars are not included because they were not given in the original data sources. b) Alvi versus OH content for mantle-derived amphiboles from the SW USA. Error bars are calculated as described in the text and given in Table 1. For OH content, the error is within the size of the symbol unless noted in the legend. c) Ti versus OH content for mantle-derived amphiboles from the SW USA.
As far as we are aware, there are no published H contents for experimentally produced Ti-rich amphiboles from extremely H-poor melts. SIMS analyses of amphiboles synthesized from melts with moderate H contents (H2O ⫽ 4.27 wt.%), show high to hyperstoichiometric H content (LaTourette et al., 1995).
Kuroda et al. (1975) showed that in Fe-free tremolite-richterites the H2O content increased regularly over 0.5–2.0 GPa in the presence of a pure H2O fluid. In general, Ti-rich synthetic amphiboles crystallize at moderate to low ƒO2 conditions. Several workers have found that
3646
P. L King et al.
the Ti content of amphiboles decreases at high oxygen fugacities in basaltic compositions (hematite-magnetite buffer and aH2O ⫽ 1, Helz, 1973; Adam and Green, 1994; King et al., 1999). In contrast, Ti-rich amphibole rims on kaersutite seed crystals were produced using a Fe-Ti-rich basalt composition at the hematite-magnetite buffer with aH2O ⱕ 1 (Johnson et al., 1991). Titanian pargasites have been produced experimentally at other ƒO2s and aH2Os by Holloway and Burnham (1972, nickel-bunsenite or NNO, aH2O ⫽ 0.6), Helz (1973, quartzfayalite-magnetite, QFM, buffer, aH2O ⫽ 1), Johnson et al. (1991, QFM, aH2O ⱕ 1), King et al. (1999, ferrosilite-fayalitemagnetite and ⬃1–2 log units below NNO buffer, aH2O ⱕ 1), LaTourette et al. (1995, ⬃1–2 log units below NNO buffer, aH2O ⬍ 1), and Jakobsson and Holloway (1986, graphite-ironwu¨stite buffer, aH2O ⫽ 0.4). Amphibole Fe3⫹/Fe2⫹ content increases and H decreases with increasing ƒO2 and decreasing ƒH2 (King et al., 1999). The role of ƒH2 in controlling amphibole compositions was examined using hydrothermal experiments where amphibole was equilibrated with a fluid, not a melt (Popp et al., 1995b). That technique is thermodynamically valid; however, melt composition, especially alkali content, controls Fe3⫹/Fe2⫹ of silicate melts at a particular ƒO2 (e.g., Sack et al., 1980; Kilinc et al., 1983; Kress and Carmichael, 1988). A consequence is that melt composition, in addition to ƒH2 (or ƒO2) control the Fe3⫹/Fe2⫹ in amphibole. Therefore, we concur with Popp and coauthors (1995b) that their experiments were only calibrated for one amphibole composition (Vulcan’s Throne).
Miyagi et al., 1998). Amphiboles from rapidly cooled rocks have generally high H contents. As mentioned above, the hydroxyl contents within single megacrysts are typically homogeneous and H contents vary between individual crystals from a single locality, especially at lava flows and scoria cones. This data must be reconciled with other information, including ascent rates and surface effects, and H diffusion data. 6.2.2. Ascent rate and surface effects
The correlation between octahedral cation (Alvi ⫹ Ti ⫹ Fe ) and H contents could be due to local charge balance of the octahedral cations (M1-3 sites) and the O2⫺ and OH⫺ groups in the O(3) anion site (Jenkins and Hawthorne, 1995; Oberti and Sardone, 1995). If H content is controlled by a simple bivariate mechanism then H may be expected to correlate with Ti and/or Alvi content (e.g., substitution 2), independent of Fe3⫹ content. That prediction is not borne out by the data for most localities (Fig. 7 b,c). No relation between Ti and H content is observed for a single locality, although there is a correlation overall. Most localities show no relation between Alvi and H content, excepting Soda Springs and Lunar Craters amphiboles which have a slight increase in OH content with decreasing Alvi content. Furthermore, the rimmed Cleator amphibole (CLRM) does not show any variation in H content, although Ti and Alvi contents vary (section 7). Crystallographic closure effects may also affect bivariate correlations (e.g., Young et al., 1997). A principle components analysis of the data indicates that the effects of the vectors n⫹ Ti4⫹O2⫺R⫺1 (OH⫺)⫺1 and Fe3⫹O2⫺(Fe2⫹OH⫺)⫺1 are difficult to separate (King, 1999; similar to Young et al., 1997). Misleading correlations may also result from dependent rela⫺ tions, such as the substitution TiO2⫺ 2 Mg⫺1OH⫺2 (Zanetti et al., 1995) which is linked to MgFe⫺1 and dehydrogenation (substitution 1).
Amphibole oxy-substitution has been related to ascent rate (Aoki, 1963; Dyar et al., 1992; Miyagi et al., 1998), although this work was superseded by Dyar et al. (1993). Magma ascent models for alkali basalts suggest transport from the base of the crust (30 – 40 km for the southwest USA; Fig. 1) in approximately 30 hours (Spera, 1984). Diffusion data has been used to show that amphiboles are unlikely to dehydrogenate during ascent (Dyar et al., 1993) and we evaluate this in more detail below (section 6.2.3). Furthermore, slower ascent, especially in a H-poor magma, will promote amphibole breakdown to anhydrous minerals (pyroxene, olivine, and nepheline or plagioclase). Since amphiboles from alkali basalts are typically unrimmed, magma ascent rates are too fast for breakdown to occur. Subaerial environments are characterized by low ƒH2 conditions that are conducive to H-loss from amphibole (Substitution 1), particularly if temperatures are favorable. Evidence that megacrysts can encounter high temperatures in an oxidizing environment includes the observation of pseudobrookite lamellae in an amphibole from Lunar Craters (LC5; OH content ⫽ 0.89 to 1.02 afu, Fe3⫹ content ⫽ 0.26 to 0.38 afu; Righter and Carmichael, 1993). Pseudobrookite is only stable at high temperatures (⬎585°C) and high oxygen fugacities (NNO ⫹ 4.8), that exceed those calculated for the host alkali basalt lava (NNO ⫺ 0.5). This difference in calculated ƒO2 is explained if the pseudobrookite exsolved from the amphibole at the air-rock interface when the cooling magma was dominated by anhydrous minerals that could not change Fe3⫹/Fe2⫹ and maintain charge balance. Volcanic environments that promote low ƒH2 include shallow, degassing and convecting magma chambers with episodic eruptions (e.g., Clague et al., 1995), or flows where lava drainback oxidizes the subsurface magma (e.g., Wallace and Anderson, 1998). The low ƒH2 environment may promote dehydrogenation, especially when amphiboles cool slowly relative to H diffusion rates. Cooling rate may also be lowered by slow or episodic effusion and thick lava piles. In contrast, amphiboles from rapidly cooled volcanic rocks (maars and dikes), are less likely to encounter prolonged high temperatures and low ƒH2 conditions. Thus, the rapidly cooled amphiboles are less likely to dehydrogenate at the surface or en route to the surface. Grain attributes (such as inclusion and fracture density) will also affect the opportunity for an amphibole to dehydrogenate. Also, if an amphibole is insulated from temperature and ƒH2 gradients, for example, hosted by a xenolith or erupted under water, it will have less opportunity to lose H (e.g., Aoki, 1963).
6.2. Dehydrogenation of Ti-rich Amphiboles
6.2.3. H diffusion data
The data from the SW USA amphibole megacrysts indicate that H content is related to emplacement process (Fig. 4;
Existing diffusion data show that a 10 mm amphibole at 1100 to 1250°C loses about 90% of its H over about 3 to 30
6.1.4. Crystallographic factors 3⫹
Oxy-substitution and dehydrogenation in amphibole megacrysts
years (Dyar et al., 1993; their Fig. 6). Such time scales exceed those expected for amphibole megacrysts cooling at the surface or during ascent (Hardee, 1980; Spera, 1984). To dehydrogenate, an amphibole requires H diffusivities about 3 to 4 orders of magnitude greater than the existing data at reasonable temperatures and times (1100 to 1250°C, 3 to 30 days). The corundum is that amphibole H contents are correlated with emplacement type. Possible explanations are that the diffusion data (pargasite and ferroan pargasite; Graham et al., 1984) are strongly dependent on composition and cannot be applied to natural samples. This seems unlikely because our samples have similar compositions (also see Dyar et al., 1993). Another possibility is that H diffusion in nature involves a different diffusing species than that measured in the D/H diffusion studies. Alternately, H diffusivity (DH⫹) may change as a function of ƒO2. No quantitative data exist for DH⫹ as a function of ƒO2, but several qualitative observations have been made: Amphiboles rapidly dehydrogenate when heated in air or under CO2. For example, amphiboles with grain sizes ranging from a few to 100 m lose H over periods of less than 10 min to several days at 700 to 850°C (Graham, 1926; Clowe et al., 1988; Miyagi et al., 1998). Unfortunately, quantitative DH⫹ data cannot be extracted from those experiments because of inadequate characterization of H content, grain size and shape, and/or time. Qualitatively, the data in oxidizing conditions suggest H diffusion several orders of magnitude higher than experiments between NNO and the hematite-magnetite buffer (Graham et al., 1984). In other words, dehydrogenation is possibly slow in Graham and coauthor’s (1984) experiments because of unfavorable ƒH2 gradients (substitution 1). This conclusion is supported by reequilibration experiments, showing that amphiboles may dehydrogenate or hydrogenate dependent on the ƒH2 (Popp et al., 1995b). Extrapolated H diffusion rates could be appropriate in a magma at moderate ƒO2s, but should be viewed as a minimum estimate in more oxidizing conditions (low ƒH2). If oxidizing conditions are sufficient to raise logDH⫹ by 2 log units, a 10 mm amphibole will dehydrogenate about 90% in less than one month. 6.3. Pressure-Temperature-Volatile Fugacity-Time Paths In Figure 8, we summarize some possible conditions that may affect the opportunities for amphibole oxy-substitution or dehydrogenation. Amphibole oxy-substitution could occur in a magma chamber and be controlled by the activities of H2O, TiO2, Al2O3, Fe2O3 and FeO in the melt. Experiments investigating H and Fe3⫹/Fe2⫹ partitioning between basanitic melt and amphibole indicate that both are a function of ƒH2 (King et al., 1999). The oxide activities are a function of pressure, temperature and bulk composition, and crystal chemical constraints also affect H and Fe3⫹/Fe2⫹ in amphiboles. An independent constraint on magma chamber H content could be obtained if amphiboles contained melt inclusions (somewhat unlikely based on their cleavage). Rimmed amphiboles potentially provide information on the pressure-temperature-ƒH2-time path of a megacryst. There are several reports of kaersutites with dark brown rims, inferred to have high-H kaersutite cores and low-H oxy-kaersutite rims (Aoki, 1963; M. Roden pers. comm. in Dyar et al., 1992).
3647
Fig. 8. Schematic pressure versus temperature diagram showing amphibole stability and some of the factors that affect the oxy-component of amphiboles. Initial crystallization conditions, such as pressure, temperature and ƒH2 may control amphibole H and octahedral cation contents. Slow ascent results in amphibole breakdown to anhydrous minerals (pyroxene, olivine, plagioclase or nepheline) or dehydrogenation. Amphibole megacrysts that are insulated from ƒH2 gradients and cool quickly have less opportunity to dehydrogenate. Crystal attributes, such as size, fracture and inclusion density also affect the likelihood of amphibole dehydrogenation.
However, core and rim compositions were not reported and the extent of oxy-substitution was not determined in any single crystal. We were unable to find H variation in an amphibole crystal provided by Aoki, although the crystal was not optically zoned. Dyar and coauthors (1998) found that H⫹ (SIMS count rate) varied with Fe3⫹/Fe2⫹ in mantle-derived amphiboles; however, H content was not quantified. One amphibole from Cleator (CLRM) has a titanian pargasite core and kaersutite rim (Table 1; King et al. 1998). The higher Ti and Ca with lower Alvi and Si are consistent with crystallization of the rim at lower pressure and temperature. These chemical changes are decoupled from H content which is constant within error throughout the crystal. The lack of H zonation collaborates the experimental evidence suggesting that the Ti content of amphiboles is not simply correlated with aH2O. The data also indicate that the pressure-temperatureƒH2-time path maintained constant H in the amphibole or that H diffusion rates were sufficient to homogenize the H content. This is consistent with the apparently homogeneous distribution of H within most amphibole megacrysts and also allows for different amphibole megacrysts to have different H contents if they follow different pressure-temperature-ƒH2-time paths. The SW USA data indicate a role for dehydrogenation since H content is related to Fe3⫹/Fe2⫹. Also, amphibole H contents correlate with emplacement mechanism, with the most variable H contents found at lava flows and scoria cones. At those localities, amphibole dehydrogenation likely occurs at the surface or en route to the surface, especially where ƒH2 is low and favors high DH⫹. Alternately, extrapolated DH⫹ values may not reflect H diffusion in nature.
3648
P. L King et al.
7. MANTLE ƒO2 AND ƒH2 CALCULATIONS AND STABLE ISOTOPES
The ƒO2 of the mantle in magma source regions is still a debated topic, despite extensive work (e.g., Arculus and Delano, 1981; Egler, 1983; Ulmer et al., 1987; Taylor and Green, 1988; Wood et al., 1990; Blundy et al., 1991; Carmichael, 1991; Ballhaus and Frost, 1994; Canil et al., 1994). The composition of mantle-derived amphiboles may help define ƒO2 in the source region. However, to unmask the effects of oxidation and dehydrogenation relations, it is best to examine amphiboles with high H content such as those from rapidly cooled dikes or maar deposits (Fig. 4) or xenoliths. The samples from dikes and maars contain close to zero calculated Fe3⫹ (based on Eqn. 5) which would suggest that the natural amphiboles crystallized under reducing conditions (King et al., 1999). On the whole, nominally anhydrous minerals are a better target for ƒO2 calculations than amphiboles because there is not the added complication of Fe3⫹/Fe2⫹ changes due to H loss. Similarly, in order to determine stable isotope contents of mantle phases it is best to examine amphiboles with high H contents. Dehydrogenation could explain the relatively large range of stable isotope values for mantle-derived amphiboles, compared with mantle-derived micas (e.g., Boettcher and O’Neil, 1980; Feldstein et al., 1996). 8. IMPLICATIONS FOR SNC METEORITES
Watson et al. (1994) proposed that the high Ti contents of amphiboles from melt inclusions in the shergottite-nakhlitechassignite (SNC) meteorites are consistent with low H content at the time of crystallization. That conclusion was based on terrestrial studies (Popp and Bryndzia, 1992). However if, as we contend, the H contents of terrestrial Ti-rich amphiboles are largely related to dehydrogenation and oxidation then their conclusions should be re-evaluated. We have calculated Fe3⫹ using measured H contents (0.1 to 0.2 afu) of different amphibole fragments of Chassigny and Zagami meteorites (Floran et al., 1978; Treiman, 1985; Johnson et al., 1991; Watson et al., 1993; Watson et al., 1994). Chassigny kaersutite has a calculated Fe3⫹ content of 0.93 to 1.01 afu and Zagami kaersutite has a calculated Fe3⫹ content of 0.62 to 0.70 afu. Relative to the terrestrial samples (Fig. 5), Chassigny has similar Fe3⫹ concentrations, while Zagami has lower Fe3⫹ contents. Using crystal chemical considerations, the calculated Fe3⫹ for the SNC meteorite kaersutites would tend to be lower than terrestrial samples because of their relatively higher Cr3⫹ (0.05 to 0.08 afu). If Cr3⫹ contents are added to Ti ⫹ Alvi in Eqn. 5 then the calculated Fe3⫹ for the SNC samples would be close to the field of terrestrial samples in Figure 5. The calculated Fe3⫹/Fetotal values are close to those calculated by Mysen and coauthors (1998) for Zagami (⬃0.3, their value ⬃0.35) and slightly higher for Chassigny (⬃0.7 to 0.8, their value ⬃0.6). If the relatively high calculated Fe3⫹ contents for the SNC amphiboles are accurate, it is not inconceivable that they have dehydrogenated (and oxidized) some time in their history. Such a process would explain their high calculated Fe3⫹/Fe2⫹ relative to other estimates of martian mantle ƒO2 that range from 3 to 4 log units below QFM (Ghosal et al., 1998) to approximately QFM (Smith and Hervig, 1979; Stolper and McSween,
1979). Another explanation is that the melt inclusions hosting amphibole have high ƒO2 as a result of crystallization in a closed system and do not represent the martian mantle. 9. CONCLUSIONS
Amphibole H contents and Fe3⫹/Fe2⫹ are controlled by pressure-temperature-ƒH2-time paths of the megacryst. Previous experimental work shows that Ti-rich amphibole crystallization is favored in mafic-ultramafic compositions at low pressure, high temperature and low-moderate ƒO2. Those conditions increase the activity of TiO2 and Al2O3 in the melt, which may be expressed in the amphibole composition by increased TiAliv 2 vi Mg⫺1Si⫺2 and/or TiAlivAl⫺1 Si⫺1 substitutions. Amphibole H content and Fe3⫹/Fe2⫹ is influenced by ƒH2 or ƒO2. Extrapolated DH⫹ values may inadequately describe H diffusion in nature because a different diffusing species may be involved, or ƒH2 changes DH⫹. In either case, the compositions of the mantle-derived amphiboles indicate that dehydrogenation, related to emplacement mechanism, controls the oxyamphibole component (substitution 1). Dehydrogenation occurs at localities where individual megacrysts encounter variable cooling and eruption rates, and ƒH2 gradients may exist either at the surface or en route to the surface (e.g., scoria cones and lavas). The homogeneous and relatively high H content of amphiboles from rapidly cooled host rocks (maars and dikes) indicate that dehydrogenation is not as extensive in those environments. Small changes are observed in Ti or Alvi contents with H content, indicating that substitution does not occur in a bivariate manner (substitution 2). Also, there is no evidence for a role of A-site cations in oxy-substitution, probably because previous studies included alkali amphiboles. Instead, oxy-substitution 2⫹ involves the entire octahedral strip following Fe3⫹O2⫺Fe⫺1 n⫹ (OH⫺)⫺1 (substitution 1) and Ti4⫹O2⫺R⫺1 (OH⫺)⫺1 (e.g., Young et al., 1997). In contrast to previous proposals, it is not necessary to have low water activity (aH2O) to crystallize Ti-rich amphiboles. The available data indicates that bulk composition, pressure, temperature and ƒH2 or ƒO2 influence Ti and Fe3⫹/Fe2⫹ in amphiboles (e.g., King et al., 1999). Thus, the SNC meteorites (Chassigny and Zagami) may not have crystallized from a H-poor melt. An empirical model for calculating Fe3⫹ content of Ti-rich amphiboles was proposed, based on Alvi, Ti, and H content. The model is best applied to kaersutites with low H and Alvi contents, although future work on other trivalent cations (e.g., Cr3⫹) may extend this model to other compositions. Acknowledgments—This work was supported by NSF grant EAR9614325 (Hervig and Holloway) and partially by NSF EAR-9506494 (Holloway). Helpful reviews were provided by M. D. Dyar, C. M. Graham and R. Popp. Thanks to those who supplied samples: E. Y. Anthony (Kilbourne Hole), M. Best (Toroweap), M. D. Dyar (collection from Dyar et al., 1993), S. Feldstein and R. Lange (SN-161), C. Francis (Harrat Hutaymah), and E. Jarosewich (Kakanui). E. Y. Anthony and D. Bell provided detailed directions to sampling sites. D. T. Lescinsky helped collect samples. Discussions with D. M. Burt, G. Moore and J. Tyburczy were beneficial. D. Desonie improved the writing style of early versions of the manuscript. J. Clark (ASU) and T. Teska (U of A) assisted with electron probe analysis and the ASU electron microprobe was purchased with NSF grant EAR-8408163. M.
Oxy-substitution and dehydrogenation in amphibole megacrysts Drake provided access to the U of A electron microprobe. A. Higgs helped with electronic aspects of the ion microprobe. REFERENCES Adam J. and Green T. H. (1994) The effects of pressure and temperature on the partitioning of Ti, Sr and REE between amphibole, clinopyroxene and basanitic melt. Chem. Geol. 117, 219 –233. Allen J. C. and Boettcher A. L. (1978) Amphiboles in andesite and basalt: II. Stability as a function of P-T-fH2O-fO2. Amer. Mineral. 63, 1074 –1087. Allen J. C., Boettcher A. L., and Marland G. (1975) Amphiboles in andesite and basalt: I. Stability as a function of P-T-fO2. Amer. Mineral. 60, 1069 –1085. Aoki K. (1963) The kaersutites and oxykaersutites from alkalic rocks of Japan and surrounding areas. J. Petrol. 4, 198 –210. Arculus R J. and Delano J. W. (1981) Intrinsic oxygen fugacity measurements: Techniques and results for spinels from upper mantle periodotites and megacryst assemblages. Geochim. Cosmochim. Acta 45, 899 –913. Ballhaus C. and Frost R. R. (1994) The generation of oxidized CO2bearing basaltic melts from reduced CH4-bearing upper mantle sources. Geochim. Cosmochim. Acta 58, 4931– 4940. Barnes V. E. (1930) Changes in hornblende at about 800°C. Amer. Mineral. 15, 393– 417. Best M. G. (1970) Kaersutite-peridotite inclusions and kindred megacrysts in basanitic lavas, Grand Canyon, Arizona. Contrib. Mineral. Petrol. 27, 25– 44. Best M. G. and Brimhall W. H. (1974) Late Cenozoic alkalic basaltic magmas in the western Colorado Plateaus and the Basin and Range Transition Zone, U.S.A., and their bearing on mantle dynamics. Geol. Soc. Amer. Bull. 85, 1677–1690. Binns R. A. (1969) High-pressure megacrysts in basanitic lavas near Armidale, New South Wales. Amer. J. Sci. 267A, 33– 49. Binns R. A., Duggan M. B., Wilkinson J. F. G., and Kalocsai G. I. Z. (1970) High pressure megacrysts in alkaline lavas from northeastern New South Wales. Amer. J. Sci. 269, 132–168. Blundy J. D., Brodholt J. P., and Wood B. J. (1991) Carbon-fluid equilibria and the oxidation state of the upper mantle. Nature 349, 321–324. Boettcher A. L. and O’Neil J. R. (1980) Stable isotope, chemical, and petrographic studies of high-pressure amphiboles and micas: Evidence for metasomatism in the mantle source regions of alkali basalts and kimberlites. Amer. J. Sci. 280-A, 594 – 621. Canil D., O’Heill H. S. C., Pearson D. G., Rudnick R. L., McDonough W. F., and Carswell D. A. (1995) Ferric iron in peridotites and mantle oxidation states. Earth Planet. Sci. Lett. 123, 205–220. Carmichael I. S. E. (1991) The redox state of basic and silicic magmas: A reflection of their source regions? Contrib. Mineral. Petrol. 106, 129 –141. Clague D., Moore J. G., Dixon J. E., and Friesen W. B. (1995) Petrology of submarine lavas from Kilauea’s Puna Ridge, Hawaii. J. Petrol. 36, 299 –349. Clowe C. A., Popp R. K., and Fritz S. J. (1988) Experimental investigation of the effect of oxygen fugacity on the ferric-ferrous ratios and unit cell parameters of four natural clinoamphiboles. Amer. Mineral. 73, 500 –506. Delaney J. S., Bajt S., Sutton S. R., and Dyar M. D. (1996) In situ microanalysis of Fe3⫹/⌺Fe in amphibole by x-ray absorption near edge structure (XANES) spectroscopy. In Mineral Spectroscopy: A Tribute to Roger G. Burns, (ed. M. D. Dyar, C. McCammon, and M. W. Schaefer), Vol. 5, pp. 165–171. Geochem. Soc. Spec. Pub. Deloule E., Albarede F., and Sheppard S. M. F. (1991) Hydrogen isotope heterogeneities in the mantle from ion probe analysis of amphiboles from ultramafic rocks. Earth Planet. Sci. Lett. 105, 543–553. Deloule E., Paillat O., Pichavant M., and Scaillat B. (1995) Ion microprobe determination of water in silicate glasses: Methods and applications. Chem. Geol. 125, 19 –28. Dyar M. D., Delaney J. S., Sutton S. R., Graham C. M., and Kinny P. (1998) Comparison of microanalysis and bulk analysis of ferric iron, water and D/H in mantle kaersutite. Geol. Soc. Amer. Abstr. 30, A186.
3649
Dyar M. D., Mackwell S. J., McGuire A. V., Cross L. R., and Robertson J. D. (1993) Crystal chemistry of Fe3⫹ and H⫹ in mantle kaersutite: Implications for mantle metasomatism. Amer. Mineral. 78, 968 –979. Dyar M. D., McGuire A. V., and Mackwell S. J. (1992) Fe3⫹/H⫹ and D/H in kaersutites: Misleading indicators of mantle source fugacities. Geology 20, 565–568. Eggler D. H. (1983) Upper mantle oxidation state: Evidence from olivine-orthopyroxene-ilmenite assemblages. Geophys. Res. Lett. 10, 365–368. Ernst W. G. and Liu J. (1998) Experimental phase-equilibrium study of Al- and Ti-contents of calcic amphibole in MORB: A semiquantitative thermobarometer. Amer. Mineral. 83, 952–969. Feldstein S. N., Lange R. A., Vennemann T., and O’Neill J. R. (1996) Ferric-ferrous ratios, H2O contents and D/H ratios of phlogopite and biotite from lavas of different tectonic regimes. Contrib. Mineral. Petrol. 126, 51– 66. Floran R. J., Prinz M., Hlava P. F., Keil K., Nehru C. E., and Hinthorne J. R. (1978) The Chassigny meteorite: A cumulate dunite with hydrous amphibole-bearing melt inclusions. Geochim. Cosmochim. Acta 42, 1213–1229. Fujinawa A. and Green T. H. (1997) Partitioning behavior of Hf and Zr between amphibole, clinopyroxene, garnet and silicate melts at high pressure. Eur. J. Mineral. 9, 379 –391. Garcia M. O., Muenow D. W., and Liu N. W. K. (1980) Volatiles in Ti-rich amphibole megacrysts, southwest USA. Amer. Mineral. 65, 306 –312. Ghosal S., Sack R. O., Ghiorso M. S., and Lipschutz M. E. (1998) Evidence for a reduced, Fe-depleted martian mantle source region of shergottites. Contrib. Mineral. Petrol. 130, 346 –357. Gilbert M. C., Helz R. T., Popp R. K., and Spear F. S. (1982) Experimental studies of amphibole stability. In Amphiboles: Petrology and Experimental Phase Relations. (ed. D. R. Veblen and P. H. Ribbe), Vol. 9B pp. 229 –353. Mineralogical Society America. Graham C. M., Harmon R. S., and Sheppard S. M. F. (1984) Experimental hydrogen isotope studies: Hydrogen isotope exchange between amphibole and water. Amer. Mineral. 69, 128 –138. Graham W. A. P. (1926) Variations in the chemical composition of hornblende from different types of igneous rocks. Amer. Mineral. 11, 118 –123. Green D. H. and Hibberson W. (1970) Experimental duplication of conditions of precipitation of high-pressure phenocrysts in a basaltic magma. Phys. Earth Planet. Int. 3, 247–254. Hardee H. C. (1980) Solidification in Kilauea Iki lava lake. J. Volcanol. Geotherm. Res. 7, 211–223. Harvey R. P. and McSween H. Y. (1992) The parent magma of the nakhlite meteorites: Clues from melt inclusions. Earth Planet. Sci. Lett. 111, 467– 482. Helz R. T. (1973) Phase relations of basalts in their melting range at PH2O ⫽ 5 kb as a function of oxygen fugacity. Part 1. Mafic phases. J. Petrol. 14, 249 –302. Holloway J. R. (1973) The system pargasite-H2O-CO2: A model for melting of a hydrous mineral with a mixed-volatile fluid; I. Experimental results to 8 kbar. Geochim. Cosmochim. Acta 37, 651– 666. Holloway J. R. and Burnham C. W. (1972) Melting relations of basalt with equilibrium water pressure less than total pressure. J. Petrol. 13, 1–29. Ionov D. A., Griffin W. L., and O’Reilly S. Y. (1997) Volatile-bearing minerals and lithophile trace elements in the upper mantle. Chem. Geol. 141, 153–184. Ionov D. A. and Hofmann A. W. (1995) Nb-Ta-rich mantle amphiboles and micas: Implications for subduction-related metasomatic trace element fractionations. Earth Planet. Sci. Lett. 131, 341–356. Irving A. J. (1974) Megacrysts from the Newer Basalts and other basaltic rocks of southeastern Australia. Geol. Sci. Amer. Bull. 85, 1503–1514. Jackson E. D., Stevens R. E., and Bowen R. W. (1967) A computerbased procedure for deriving mineral formulas from mineral analyses. U.S. Geol. Surv. Prof. Pap. 575-C C23–C31. Jakobsson S. and Holloway J. R. (1986) Crystal-liquid experiments in the presence of a C-O-H fluid buffered by graphite ⫹ iron ⫹ wu¨stite: Experimental method and near-liquidus relations in basanite. J. Volcanol. Geotherm. Res. 29, 265–291.
3650
P. L King et al.
Jarosewich E. (1975) Chemical analysis of five minerals for microprobe standards. Smith. Contrib. Earth Sci. 9, 83– 84. Jenkins D. M. and Hawthorne F. C. (1995) Synthesis and Reitveld refinement of amphibole along the join Ca2Mg5Si8O22F2NaCA2Mg4Ga3Si6O22F2. Can. Mineral. 33, 13–24. Johnson M. C., Rutherford M. J., and Hess P. C. (1991) Chassigny petrogenesis: Melt compositions, intensive parameters, and water contents of Martian (?) magmas. Geochim. Cosmochim. Acta 55, 349 –366. Kilinc A., Carmichael I. S. E., Rivers M. L., and Sack R. O. (1983) The ferric-ferrous ratio of natural silicate liquids equilibrated in air. Contrib. Mineral. Petrol. 83, 136 –140. King P. L., Delaney J. S., and Dyar, M. D., in preparation. Determining micro-Fe3⫹ content of Ti-rich amphiboles by direct SmX analysis, calculations and vector analysis. King P. L., Hervig R. L., Delaney J. S., Vennemann T., Holloway J. R., and Righter K. (1998) Micro-H and Fe3⫹ contents of Ti-rich amphibole megacrysts from the SW USA. Geol. Soc. Amer. Abstr. 30, A24. King P. L., Hervig R. L., Holloway J. R., and Delaney J. S. (1999) Partitioning of H and Fe3⫹/(Fetotal) between amphibole and basanitic melt as a function of oxygen fugacity. EOS, Trans. Amer. Geophys. Union, 80, S358. Kress V. C. and Carmichael I. S. E. (1988) Stoichiometry of the iron oxidation reaction in silicate melts. Amer. Mineral. 73, 1267–1274. LaTourette T., Hervig R. L., and Holloway J. R. (1995) Trace element partitioning between amphibole, phlogopite, and basanite melt. Earth Planet. Sci. Lett. 135, 13–30. Leake B. E. (1968) A catalog of analyzed calciferous and subcalciferous amphiboles together with their nomenclature and associated minerals. Geol. Soc. Amer. Spec. Pap. 98, 210. Leake B. E. and 21 others (1997) Nomenclature of amphiboles: Report of the subcommittee on amphiboles of the International Mineralogical Association, commission on new minerals and mineral names. Can. Mineral. 35, 219 –246. Mason B. (1966) Pyrope, augite, and hornblende from Kakanui, New Zealand. N. Z. J. Geol. Geophys. 9, 474 – 480. Matson D. W., Muenow D. W., and Garcia M. O. (1984) Volatiles in amphiboles from xenoliths, Vulcan’s Throne, Grand Canyon, Arizona, USA. Geochim. Cosmochim. Acta 48, 1629 –1636. McGuire A. V., Dyar M. D., and Ward K. A. (1989) Neglected Fe3⫹/Fe2⫹ ratios: A study of Fe3⫹ content of megacrysts from alkali basalts. Geology 17, 687– 690. McGuire A. V., Francis C. A., and Dyar M. D. (1992) Mineral standards for electron microprobe analysis of oxygen. Amer. Mineral. 77, 1087–1091. Merrill R. B. and Wyllie P. J. (1975) Kaersutite and kaersutite eclogite from Kakanui, New Zealand: Water-excess and water-deficient melting to 30 kilobars. Geol. Soc. Amer. Bull. 86, 555–570. Miyagi I., Matsubaya O., and Nakashima S. (1998) Change in D/H ratio, water content and color during dehydration of hornblende. Geochem. J. 32, 33– 48. Mysen B. O., Virgo D., Popp R. K., and Bertka C. M. (1998) The role of H2O in Martian magmatic systems. Amer. Mineral. 83, 942–946. Nielson J. E. and Nakata J. K. (1994) Mantle origin and flow sorting megacryst-xenolith inclusions in mafic dikes of Black Canyon, Arizona. U.S. Geol. Surv. Prof. Pap. 1541, 41 p. Nishikawa M., Kushiro I., and Uyeda S. (1970) Stability of natural hornblende at high water pressures: Preliminary experiments. Japan. J. Geol. Geogr. XLI, 41–50. Oba T. (1997) The stability fields of kaersutite and its substitution of R2⫹ ⫹ 2Si ⫽ Ti ⫹ 2AlIV. In Synthetic and natural rock systems (ed. A. K. Gupta, K. Onuma, and M. Arima), pp. 126 –138. Allied Publishers Ltd. Oba T., Yagi K., and Hariya Y. (1986) Stability relation of kaersutite, reinvestigated on natural and synthetic samples. Morphology and Phase Equilibria of Minerals: IMA 1982, 353–363. Oberti R. and Sardone N. (1995) Synthesis and crystal-structure refinement of synthetic fluor-pargasite. Can. Mineral. 33, 25–31. Oberti R., Ungaretti L., Cannillo E., and Hawthorne F. C. (1992) The behaviour of Ti in amphiboles: I. Four- and six-coordinate Ti in richterite. Eur. J. Mineral. 4, 425– 439. Otten M. T. and Buseck P. R. (1987) The oxidation state of Ti in
hornblende and biotite determinated by electron energy loss spectroscopy with inferences regarding the Ti substitution. Phys. Chem. Min. 14, 45–51. Oxburgh E. R. (1964) Petrological evidence for the presence of amphibole in the upper mantle and its petrogenetic and geophysical implications. Geol. Mag. 101, 1–19. Peng G., Luhr J. F., and McGee J. J. (1997) Factors controlling sulfur concentrations in volcanic apatite. Amer. Mineral. 82, 1210 –1224. Phillips M. W., Popp R. K., and Clowe C. A. (1988) Structural adjustments accompanying oxidation-dehydrogenation in amphiboles. Amer. Mineral. 73, 500 –506. Popp R. K. and Bryndzia L. T. (1992) Statistical analysis of Fe3⫹, Ti and OH in kaersutite from alkalic igneous rocks and mafic mantle xenoliths. Amer. Mineral. 77, 1250 –1257. Popp R. K., Phillips M. W., and Harrell J. A. (1990) Accommodation of Fe3⫹ in natural, Fe3⫹-rich, calcic and subcalcic amphiboles: Evidence from published chemical analyses. Amer. Mineral. 75, 163–169. Popp R. K., Virgo D., and Phillips M. W. (1995a) H deficiency in kaersutitic amphiboles: Experimental verification. Amer. Mineral. 80, 1347–1350. Popp R. K., Virgo D., Yoder H. S., Jr., Hoering T. C., and Phillips M. W. (1995b) An experimental study of phase equilibria and Fe oxy-component in kaersutitic amphibole: Implications for the fH2 and aH2O in the upper mantle. Amer. Mineral. 80, 534 –548. Righter K. and Carmichael I. S. E. (1993) Mega-xenocrysts in alkali olivine basalts: Fragments of disrupted mantle assemblages. Amer. Mineral. 78, 1230 –1245. Rowbotham G. and Farmer V. C. (1973) The effect of ‘‘A’’ site occupancy upon the hydroxyl stretching frequency in clinoamphiboles. Contrib. Mineral. Petrol. 38, 147–149. Saxena S. K. and Ekstro¨m T K. (1970) Statistical chemistry of calcic amphiboles. Contrib. Mineral. Petrol. 26, 276 –284. Schneider C. (1891) Sur Kenntniss basaltischer hornblenden. Zeits. Krystallogr. Mineral. 18, 580 –584. Schumacher J. C. (1997) Appendix 2: The estimation of the proportion of ferric iron in the electron-microprobe analysis of amphiboles. Can. Mineral. 35, 238 –246. Smith J. V. and Hervig R. L. (1979) Shergotty meteorite: Mineralogy, petrography and minor elements. Meteoritics 14, 121–142. Spera F. J. (1984) Carbon dioxide in petrogenesis III: Role of volatiles in the ascent of alkaline magma with special reference to xenolithbearing mafic lavas. Contrib. Mineral. Petrol. 88, 217–232. Stolper E. and McSween Jr. H. Y. (1979) Petrology and origin of the shergottite meteorites. Geochim. Cosmochim. Acta 43, 1475–1498. Stormer Jr. J. C., Pierson M. L., and Tacker R. C. (1993) Variation of F and Cl X-ray intensity due to anisotropic diffusion in apatite during electron microprobe analysis. Amer. Mineral. 78, 641– 648. Taylor W. R. and Green D. H. (1988) Measurement of reduced peridotite-C-O-H solidus and implications for redox melting of the mantle. Nature 332, 349 –352. Treiman A. H. (1985) Amphibole and hercynite spinel in Shergotty and Zagami: Magmatic water, depth of crystallization, and metasomatism. Meteoritics 20, 229 –243. Ulmer G. C., Grandstaff D. E., Weiss D., Moats M. A., Buntin T. J., Gold D. P., Hatton C. J., Kadik A., Koseluk R. A., and Rosenhauer M. (1987) The mantle redox state: An unfinished story? Geol. Soc. Amer. Spec. Pap. 215, 5–23. Vennemann T. W. and O’Neil J. R. (1993) A simple and inexpensive method of hydrogen isotope and water analyses of minerals and rocks based on zinc reagent. Chem. Geol. 103, 227–234. Wallace M. E. and Green D. H. (1991) The effect of bulk rock composition on the stability of amphibole in the upper mantle: Implications for solidus positions and mantle metasomatism. Contrib. Mineral. Petrol. 44, 1–19. Watson L. L., Hutcheon I. D., Esstein S., and Stolper E. M. (1993) D/H ratios and water contents of amphiboles in magmatic inclusions in Chassigny and Shergotty. Meteoritics 28, 456 – 457. Watson L. L., Hutcheon I. D., Epstein S., and Stolper E. M. (1994) Water on Mars: Clues from deuterium/hydrogen and water contents of hydrous phases in SNC meteorites. Science 265, 86 –90. Waychunas G. A. (1987) Synchrotron radiation XANES spectroscopy of Ti in minerals: Effects of Ti bonding distances, Ti
Oxy-substitution and dehydrogenation in amphibole megacrysts valence and site geometry on absorption edge structure. Amer. Mineral. 72, 89 –101. Wilshire H. G., Meyer C. E., Nakata J. K., Calk L. C., Shervais J. W., Nielson J. E., and Schwarzman E. C. (1998) Mafic and ultramafic xenoliths from volcanic rocks of the western United States. U.S. Geol. Surv. Prof. Paper 1443, 179 p. Wilshire H. G. and Trask N. J. (1971) Structural and textural relationships of amphibole and phlogopite in peridotite inclusions, Dish Hill, Califormia. Amer. Mineral. 56, 240 –255. Winchell A. N. (1924) Studies in the amphibole group. Amer. J. Sci. 7, 287–310. Wood B. J., Bryndzia L. T., and Johnson K. E. (1990) Mantle oxidation state and its relationship to tectonic environment and fluid speciation. Science 248, 337–345. Young E. D., Virgo D., and Popp R. K. (1997) Eliminating closure in mineral formulae with specific application to amphiboles. Amer. Mineral. 82, 790 – 806. Zanetti A., Oberti R., Bottazzi P., and Vannucci R. (1997) Fe3⫹ contents in partially-dehydrogenated high-T amphiboles by combining SREF, EMP and SIMS analyses: An insight into intensive parameters of upper-mantle systems Terra Nova 9, 437. Zanetti A., Vannucci R., Oberti R., and Dobosi G. (1995) Traceelement composition and crystal-chemistry of mantle amphiboles from the Carpatho-Pannonian region. Acta Vulcanol. 7, 265–276. APPENDIX 3⫹
To calculate Fe in amphiboles with known H contents a 24 oxygen basis was used . First, the weight percent oxide in
3651
column 1 of the table is converted to the equivalent per hundred weight (column 2), by dividing the wt.% of the oxide by (molecular weight of the oxide) / (number of positive valences in the formula). For the anions (excluding O), the wt.% measured is divided by the molecular weight of the anion. The cations are converted to ions per formula unit, based on 24 oxygens (column 3) by multiplying column 2 by a normalizing factor (48/sum cations) divided by the valence of the cation. Anions are multiplied by the normalizing factor. Then Fe3⫹ is calculated using Fe3⫹ ⫽ 2.47⫺(0.93 ⫻ OH afu) ⫺ (Ti afu ⫹ Alvi afu). Fe2⫹ is calculated using Fe2⫹ ⫽ Fetotal ⫺ Fe3⫹ calculated. The excess charge is equal to the Fe3⫹ content (afu) because (initial charge) ⫽ (2 ⫻ Fetotal) and (charge of calculated 3⫹ Fe2⫹ ⫹ calculated Fe3⫹) ⫽ (2 ⫻ (Fe2⫹ )) ⫹ total ⫺ Fe (3 ⫻ Fe3⫹). A charge factor is calculated using 48/(48 ⫹ excess charge) and then the analysis recalculated (column 4). There is a slight excess charge for some samples with this technique due to rounding. 10.1. Example of the Anion-Based H-Equivalent Method of Calculating Ions per Formula, Calculation of Fe3ⴙ, and Calculation of Charge-Balanced Formula See Table 2.
Table 2. Example of the anion-based H-equivalent method of calculating ions per formula, calculation of Fe3⫹, and calculation of charge-balanced formula. Wt.% Column SiO2 TiO2 Al2O3 FeO total MnO MgO CaO Na2O K2O P2O5 Cl F H2O Total
Sample CM1.4.2 1
Equiv. per hundred wt.
Ions per formula based on 24 O
2
41.44 4.09 13.38
2.76 0.20 0.79
9.37 0.08 14.23 11.39 2.66 1.69 0.05 0.00 0.08 1.64 100.09
0.26 0.00 0.71 0.41 0.09 0.04 0.00 0.00 0.00 0.18 5.44
Normalizing factor ⫽ 48/(sum anions) ⫽ 48/5.44.
3 Si Ti AlIV AlVI Fe total Mn Mg Ca Na K P Cl F OH Total O3 site O remain Calc. Fe3⫹ Excess charge Charge factor
6.09 0.45 1.91 0.41 1.15 0.01 3.12 1.79 0.76 0.32 0.01 0.00 0.03 1.61 16.01 0.36 22.36 0.12 0.12 0.998
Recalculated 4 6.07 0.45 1.91 0.40 1.15 0.01 3.11 1.79 0.76 0.32 0.01 0.00 0.03 1.60 15.97 0.36 22.36 0.12 0.01