- Email: [email protected]
Diffusive reequilibration of quartz-hosted silicate melt and fluid inclusions: Are all metal concentrations unmodified?
Available online at www.sciencedirect.com
Geochimica et Cosmochimica Acta 73 (2009) 3013–3027 www.elsevier.com/locate/gca
Diffusive reequilibration of quartz-hosted silicate melt and fluid inclusions: Are all metal concentrations unmodified? Zoltan Zajacz a,c,*, Jacob J. Hanley b, Christoph A. Heinrich c, Werner E. Halter c, Marcel Guillong c a
Laboratory for Mineral Deposits Research, Department of Geology, University of Maryland, College Park, MD 20742, USA b Department of Geology, Saint Mary’s University, Halifax, Canada c Institute of Isotope Geochemistry and Mineral Resources, ETH Zu¨rich, Switzerland Received 8 September 2008; accepted in revised form 17 February 2009; available online 4 March 2009
Abstract Experiments were conducted to determine the extent and mechanism by which the composition of quartz-hosted silicate melt inclusions (SMI) and aqueous fluid inclusions (FI) can undergo post-entrapment modification via diffusion. Quartz slabs containing assemblages of SMI and FI were reacted with synthetic HCl bearing and metalliferous aqueous fluids at T = 500– 720 °C and P = 150–200 MPa. SMI from the single inclusion assemblages were analyzed by laser ablation inductively coupled plasma mass spectrometry (LA-ICPMS) and electron probe microanalysis (EPMA) before and after the experiments. Analyses revealed that rapid diffusion of the univalent cations Na+, Li+, Ag+, Cu+ and H+ occurred through the quartz from the surroundings, resulting in significant changes in the concentrations of these elements in the inclusions. Concentrations of other elements with an effective ionic radius larger than that of Ag+, or multiple valence states were not modified in the inclusions during the experiments. Our results warn inclusion‘‘ researchers that the interpretation of Na, Li, Cu and Ag concentrations from quartz-hosted SMI and FI should be treated critically. Ó 2009 Elsevier Ltd. All rights reserved.
1. INTRODUCTION Quartz-hosted fluid inclusions (FI) and silicate melt inclusions (SMI) are commonly considered to preserve the composition of hydrothermal fluids and evolved silicate melts, allowing natural processes that lead to the generation of ore deposits and volcanic eruptions to be traced and characterized chemically. However, observations of anomalously Na-deficient SMI compositions in natural samples (Aude´tat et al., 2000; Aude´tat and Pettke, 2003; Student and Bodnar, 2004; Zajacz et al., 2008) and modification of the Na, Cu and Ag content of natural SMI during experimental reheating (Kamenetsky and Danyushevsky, 2005; Zajacz et al., 2008) raise the question as to whether inclusions in quartz *
Corresponding author. Address: Laboratory for Mineral Deposits Research, Department of Geology, University of Maryland, College Park, MD 20742, USA. E-mail address: [email protected] (Z. Zajacz). 0016-7037/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.gca.2009.02.023
reliably represent the composition of fluids and silicate melts at the time of their entrapment. No systematic experimental study has been carried out to determine if such fluid and melt inclusions can be compositionally modified after entrapment by diffusion-controlled cation exchange through the quartz host between the inclusions and the surroundings of their host crystals. Previous studies focused only on the diffusion of molecular H2 and H2O in and out of the inclusions (Qin et al., 1992; Mavrogenes and Bodnar, 1994; Vityk et al., 2000; Severs et al., 2007). Aude´tat and Gu¨nther (1999) studied the compositional changes of fluid inclusions during their migration related to the deformation of quartz in natural systems and found that besides water loss from the inclusions, the ratios of the analyzed cations remained constant with the exception of Li. The present study, describes a series of experiments conducted on natural samples to determine which element concentrations can be significantly modified in quartz-hosted FI and SMI after their entrapment at P–T conditions typi-
3014
Z. Zajacz et al. / Geochimica et Cosmochimica Acta 73 (2009) 3013–3027
cal of sub-volcanic magma reservoirs and associated high temperature hydrothermal systems. 2. STARTING MATERIALS AND EXPERIMENTAL PROCEDURES To determine which cations are mobile in quartz, slabs of natural quartz containing FI and/or SMI were immersed in high temperature aqueous fluids of various compositions using conventional hydrothermal experimental techniques. Free-standing quartz crystals from miarolitic cavities sampled from plutons at Rito del Medio, New Mexico, USA (Sample #Rito2nd), Canada Pinabete, New Mexico, USA (Sample #Cana11a) and Erongo, Namibia (Sample #Erg03) containing FI and recrystallzed SMI, and quartz phenocrysts from a rhyolite sampled from Banska Stiavnica, Slovakia, (Sample #BnSt) containing glassy SMI were selected for experimental investigation. FI and SMI from Rito2nd and Cana11a occur in well-characterized inclusion assemblages along primary growth zones or in pseudosecondary trails (Aude´tat and Pettke, 2003; Zajacz et al., 2008). Inclusions in Erg03 occur in pseudosecondary trails. Both SMI and FI are free of any sign of decrepitation, necking or leakage. SMI are rhyolitic in composition, but show anomalously low Na2O content (0.10 ± 0.03 wt%, 1r, 8 analyses) compared to what is expected for a typical haplogranitic liquid. All SMI in miarolitic quartz were strongly recrystallized. Melt inclusions from Banska Stiavnica are glassy and occur isolated or in clusters of up to 15 inclusions that do not follow any obvious textural or growth feature. The inclusions vary between <5 and 220 lm in diameter. They are rhyolitic in composition with an Na2O content of 2.8 ± 0.1 wt% (1r, 10 analyses). Slabs of quartz were cut from petrographic thick sections (200–500 lm thickness) using a small diamond wheel, re-polished with clean diamond paste (3 and 1 lm) and cleaned through multiple washings with acetone in an ultrasonic cleaner before placing them into the metal capsules. Slabs from Cana 11a and Erg03 were cut parallel to the crystallographic c-axis, while slabs from Rito2nd were cut parallel to the growth surfaces. The size of the analyzed
inclusions ranged from 15 to 50 lm in diameter and they were 10–60 lm below the sample surface. To prepare the solutions added to the capsules, we used analytical grade aqueous HCl solution (16 m) and metal chloride reagents or AAS standard chloride solutions. High purity SiO2 powder (SiO2 > 99.9%) was added to all experimental charges to minimize the dissolution of the studied quartz chips. The components of each experimental charge as well as run conditions and duration are shown in Table 1. Capsules were weighed before and after welding, and after the experiments to check for potential fluid loss. In experiment #01 (Table 1) SMI were reheated in open (crimped but not sealed) Ag70Pd30 capsules in an atmosphere of Ar in pressure vessels composed of Nimonic105TM alloy. Samples were held at T = 710 °C, P = 150 MPa for 72 h, then at T = 740 °C, P = 180 MPa for 70 h. These conditions approached the predicted melt homogenization temperature, estimated from the water-saturated haplogranite solidus (Johannes and Holtz, 1996). Other open-capsule reheating experiments were described by Zajacz et al. (2008), thus here we describe only the one from which the inclusion bearing quartz chip has been used for further experiments (experiment #01). Experiments #02 to #07 were run in Rene´ 41 pressure vessels using an Ar pressure medium. Experiments #01 to #07 were initially pressurized so that the approximate target pressure was reached during heating to run temperature. Once at final run temperature, the desired final pressure was set. Experiment #08 was carried out in water pressure medium, and was near isobarically heated to run temperature. Experiments were air quenched after removal from the furnaces and cooled below 400 °C within 3 min. After removal from the capsules, the quartz chips commonly displayed a thin overgrowth of new quartz (few tens of lm thick) which was removed by polishing. The chips were cleaned through multiple washings in acetone in an ultrasonic cleaner prior to LA-ICPMS analysis. In order to constrain an internal standard for the LAICPMS analyses of single FI (Heinrich et al., 2003), microthermometric measurements were conducted before and after the experiments using a Linkam THMSG 600 heat-
Table 1 Summary of conducted experiments. Experiments T (°C) P (MPa) Time (h) sample #01 #02 #03 #04 #05
740 720 710 710 710
180 200 150 200 200
142 72 24 72 168
#06 #07 #08
650 500 650
200 150 150
72 168 72
Added solutions and solids Capsule
Rito2nd SiO2 powder BnSt–4 chips 4 m HCl–80 ll Rito2nd, BnSt–2 chips 1 m HCl–150 ll BnSt–2 chips H2O–120 ll Rito2nd, BnSt–2 chips Doped sol. #01–120 ll Haplogranite glass–12 mg PbO–3 mg, Cu–2 mg Ag70Pd30–1 mg Prefracured quartz core Erg03–3 chips 4 m HCl–130 ll Erg03–2 chips 4 m HCl–150 ll Cana11a doped sol. #02–100 ll AgCl–1 mg, Cu–2.5 mg
Location/vessel material
Ag70Pd30–crimped Au–sealed Au- sealed Au- sealed Au-sealed Au-sealed
Univ. of Bern/Nimonic 105 ETH Zurich/Rene 41 ETH Zurich/Rene 41 ETH Zurich/Rene 41 ETH Zurich/Rene 41
Au-sealed Au-sealed Pt-sealed
ETH Zurich/Rene 41 ETH Zurich/Rene 41 Univ. of Maryland/Rene 41
Diffusive reequilibration of quartz-hosted inclusions
ing-freezing stage mounted on a Nikon petrographic microscope. 3. ANALYTICAL TECHNIQUES Both fluid and melt inclusions were analyzed by using LA-ICPMS at ETH Zu¨rich. The system consists of a 193 nm ArF excimer laser to produce an energy homogenized beam profile (Gu¨nther et al., 1997) coupled with an ELAN 6100 DRC ICP quadrupole mass spectrometer. The laterally homogenous energy density of the laser beam allows controlled ablation with a flat bottomed ablation crater. The system permits online observation of the ablation signal which is a prerequisite for controlled inclusion analysis. Analyses of a maximum of 16 unknowns were bracketed by measurement of an external standard (NIST SRM 610) to allow linear drift correction. The linear dynamic range of 9 orders of magnitude of the detector allows simultaneous measurement of major and trace elements down to sub-lg/g level. This allows quantification of the measurements by normalizing the concentration of the major elements to 100 wt% minus the total volatile concentration. More detailed description of the quantification approach for homogeneous solids and fluid inclusions is given in (Longerich et al., 1996; Gu¨nther et al., 1998; Heinrich et al., 2003). Output energy of the laser source was adjusted to the minimum value required for controlled ablation of the host quartz for each sample (generally between 20 and 32 J/cm2 energy density on the sample surface). The volume of the oval shaped ablation cell was 1 or 8 cm3. Analyses of the SMI and the quantification of their composition were carried out using the method of Halter et al. (2002) and Zajacz and Halter (2007). The entire inclusions were ablated along with the host quartz and the contribution of the host quartz was subtracted from the mixed signal to obtain the composition of the SMI. To determine the amount of host to be subtracted, we used the average Al2O3 content of the melt inclusions as an internal standard, determined by EPMA on other inclusions from the same assemblages, that were rehomogenized previously (except for sample BnSt, in which the SMI were quenched to glass in nature). The method provides the bulk composition of recrystallized SMI, including the host mineral that generally crystallizes on the walls of SMI. The composition of fluid inclusions was quantified using NaCl wt% equivalent concentrations based on microthermometric observation of freezing point depression and salt dissolution tempreatures. Due to the presence of cations other than Na in significant concentrations, a salt correction has been carried out with all major cations (generally K, Fe, Mn) to obtain true Na concentrations following the method of (Gu¨nther et al., 1998). These Na concentrations were used as an internal standard to quantify the LAICPMS data for fluid inclusions. Homogenized silicate melt inclusions were analyzed for their major element composition by a JEOL SUPERPROBE JXA-8200 electron microprobe at ETH Zu¨rich. A 15 kV acceleration voltage was used. The homogenized inclusions are highly unstable under the electron beam, thus we used a beam current of 5 nA and the largest beam diam-
3015
eter possible, depending on the size of the melt inclusions (generally P8 lm). The Na signal was collected first, and was monitored online during each analysis. Only those measurements where no significant change could be observed in the Na count rate during analyses were further considered. Counting times of 20 s on the peak and 10 s on the background on both sides were used for all major elements except for Na, for which we used only 10 s peak and 5 s background counting. 4. RESULTS Since the investigated inclusions were kept intact throughout the experimental duration and did not crack or leak, there was no direct physical contact between the content of the inclusion (FI or SMI) and the outside environment and the only possible mechanism of mass transfer through the host quartz was diffusion. The Na2O content of the SMI increased significantly in all inclusions measured (from 0.05–0.7 to 1.5–3.5 wt% Na2O) during the re-homogenization experiment (#01) similarly to other open-capsule reheating experiments reported by Zajacz et al. (2008). Besides Na, concentrations of Ag and Cu increased in all inclusions analyzed by two and one order of magnitude, respectively (Fig. 1). Concentrations of all other analyzed elements remained unchanged. As no Ag, Cu or Na was present in the host quartz before the experiments in significant concentration (generally Na<1 lg/g, Cu < LOD of 0.1 lg/g, Ag < LOD of 0.03 lg/g) the observations suggest that these element enrichments in the inclusions, at least in part, were derived from an external source (surrounding the quartz host). The most likely source of Ag is the capsule material (Ag70Pd30). The Cu may originate from the capsule (36 lg/g Cu analyzed in the capsule material) or more likely the vessel material itself (up to 0.2 wt% Cu in the Nimonic 105 alloy). The presence of alloyed metallic Ag and Cu at run temperatures may have yielded large enough chemical potential gradients between the SMI and the exterior of the host quartz to allow metal transport through the Ar atmosphere simultaneously with rapid diffusion through the quartz. Transfer of metals through the Ar pressure media may have been possible despite the rather low vapor pressure of the metals at run temperature due to very rapid diffusion in the gas phase. Transfer through direct contact between the vessel, the capsule wall and the quartz chip may have been also possible. The source of Na enriched in the SMI is likely to be the large volume of quartz surrounding the inclusions. To raise the Na2O content of an SMI with a typical value of 2 wt%, a spherical volume of quartz with approximately 24 times the radius of the SMI is necessary (assuming 1 lg/g Na in the quartz) if all the Na leaves the quartz. Though such volume of quartz is available in some of the reheated chips, not in all of them. Therefore additional sources of Na have to be taken into consideration. In experiment #01 the quartz chip were placed into the capsule together with supporting SiO2 powder (about 3 times the weight of the quartz chip itself), which may have contained Na well above 1 lg/g (up to 20 lg/g, Alfa Aesar) and contributed to the Na enrichment in the SMI. Quartz chips from Cuasso al Monte, Baveno
3016
Z. Zajacz et al. / Geochimica et Cosmochimica Acta 73 (2009) 3013–3027
Fig. 1. Comparison of the compositions of SMI before and after experiment #01 (740 °C, 180 MPa, 142 h) in sample Rito2nd. Concentrations of Na, Ag and Cu strongly increased in the melt inclusions during the heating experiment, while other element concentrations remained unaltered. Cu and Ag were below the limits of detection before the experiment, thus we plotted the limits of detections. Major elements are in wt%, trace elements are in lg/g.
and Mt. Malosa described by Zajacz et al. (2008) contained alkali-feldspar intergrown with the quartz, because it could not be entirely separated before the heating experiments. Therefore, in those chips the alkali-feldspar may serve as an additional source of Na to allow for the observed increase in Na concentrations (Zajacz et al., 2008). To verify the hypothesis derived from open-capsule reheating experiments, that elevated metal concentrations surrounding the quartz slabs induced metal enrichment in the SMI by diffusion through the quartz, additional experiments were conducted to reproduce the same and reversal processes reacting the quartz with various fluids of known composition in sealed capsules. Hydrogen chloride-rich fluids are known to be in equilibrium with silicate melts with an aluminum saturation index (‘‘ASI” = Al/[Na + K + 0.5Ca + Li + Rb + Cs]) at or higher than unity (Williams et al., 1997). Trace elements complexed by chloride should partition strongly into such a fluid, being incompatible in the coexisting melt. Thus, in experiments #02 and #03 (Table 1), quartz chips containing SMI with an ASI1.0 were run in capsules loaded with concentrated (1 or 4 m) and metal-free HCl solution at 710 °C and 150 or 200 MPa in order to impose a strong driving force (i.e., chemical potential gradient) for all elements that tend to partition/ dissolve into aqueous chloride solutions (e.g. Ca, Sr, Ba, Mn, Fe, Cu, Zn, Ag, Pb) to diffuse out the inclusions (Candela and Piccoli, 1995; Zajacz et al., 2008). As the mass of the HCl bearing fluid phase is several orders of magnitudes higher than that of all SMI together in the capsule, even elements that are only weakly soluble as chlorides (or have a low fluid/melt partition coefficient) should be depleted from the SMI if they can diffuse through the quartz. These experiments showed a significant decrease in wt% Na, Li, Ag and
Cu concentrations in the SMI compared to the initial values prior to exposure to the HCl solution, while concentrations of all the other analytes remained unchanged (Figs. 2–4; Tables 2 and 3). For comparison with HCl, interaction with pure H2O (experiment #04) resulted in a more moderate drop in Na, Li, Ag and Cu concentrations (Fig. 5; Table 3). To check if this process is reversible, a quartz chip (Rito5-2nd; experiment #05) containing SMI was reacted with a salt solution that contained NaCl, KCl, and FeCl2 in a proportion predicted to be in equilibrium with a normal granitic melt with an ASI1.00. Dissolved or solid sources of trace elements with variable ionic charges and cationic radii were added to this solution (Table 4). Lithium, Rb, Cs, Sr, Ba, Sb, La, Mo and Nb were added as dissolved chlorides or hydroxides, while elemental Cu, Ag and PbO were added as solid phases to the capsules. The starting solution had a pH of 3. A small amount of haplogranitic glass (200 MPa water saturated minimum composition; Johannes and Holtz, 1996) was added to the charge in order to monitor the final composition of the melt that equilibrated with the fluid phase. A pre-fractured quartz cylinder was added to trap fluid inclusions, allowing determination of the actual fluid composition at the experimental run conditions. After the experiment, SMI in the quartz chips, synthetic FI in the quartz cylinder, and the monitoring glass” were analyzed by LA-ICPMS. Concentrations of Li and Na significantly increased in the SMI during the experiment, approaching concentrations measured in the monitoring glass (Fig. 6). Even though they were included in the capsule load, concentrations of Ag and Cu remained very low in the monitoring glass, likely due to preferential alloying of these two metals with the Au capsule. This was con-
Diffusive reequilibration of quartz-hosted inclusions
3017
Fig. 2. Results of experiment #02 (720 °C, 200 MPa, 172 h, 4 m HCl solution) on SMI in magmatic quartz phenocrysts (sample BnSt). A significant drop in Na, Li and Cu concentrations is clearly observable. Major elements are in wt%, trace elements are in lg/g.
Fig. 3. Comparison of the compositions of SMI before and after experiment #03 (710 °C, 150 MPa, 24 h, 1 m HCl solution). Concentrations of Na, Ag, Cu and Li decreased one to two orders of magnitude during the experiment, while other element concentrations remained unaltered. Major elements are in wt%, trace elements are in lg/g.
firmed by the relatively low Cu and Ag concentrations measured in the synthetic FI (402 ± 88 and 3 ± 3 lg/g, respectively, Table 5). Thus, the experiment is inconclusive with respect to Cu as the final Cu concentration in the monitoring glass is identical within uncertainty to that in the SMI before and after the experiment. The concentration of Ag
displayed a moderate drop in the SMI, approaching the concentration detected in the monitoring glass. Concentrations of all other elements remained unchanged in the SMI, though in most cases they were present at much higher or lower concentrations in the monitoring glass at the end of the experiment than in the SMI. The change in the physical
3018
Z. Zajacz et al. / Geochimica et Cosmochimica Acta 73 (2009) 3013–3027
Fig. 4. Results of experiment #03 (710 °C, 150 MPa, 24 h, 1 m HCl solution) on SMI in magmatic quartz phenocrysts (sample BnSt). A significant drop in Na and Li concentrations is clearly observable. Cu and Ag concentrations were below the limits of detection after the experiment (Cu < 16.5 lg/g; Ag < 1.6 lg/g). Major elements are in wt%, trace elements are in lg/g.
appearance of a set of SMI in the sample Rito2nd, which was a subject to all kinds of experiments in this study is shown on Fig. 7. To confirm that the very rapid modification of Li, Na, Cu and Ag concentrations in SMI through the host quartz observed in the previous experiments may also occur in FI, further experiments (#06, 07 and 08) were conducted at lower temperatures. In experiment #06, quartz chips from sample Erg03 containing brine inclusions were put into con-
tact with 4 m aqueous HCl solution at 650 °C and 200 MPa for 72 h. Sodium concentrations dropped by more than 60% relative, while Li concentrations dropped an order of magnitude (Fig. 8; Table 6). All other element concentrations remained constant. Modification of the FI composition corresponded to a significant change in the NaCl dissolution temperatures (from >530 °C to 290–340 °C) and the range in temperature of vapor bubble disappearance (from 430–460 °C to 345–370 °C). Another experiment
Fig. 5. Comparison of the compositions of SMI before and after experiment #04 (710 °C, 200 MPa, 72 h, pure H2O). The drop in Na, Li, Ag and Cu concentrations is more moderate due to the use of H2O instead of HCl. Major elements are in wt%, trace elements are in lg/g.
Diffusive reequilibration of quartz-hosted inclusions
3019
Table 2 Average composition of silicate melt inclusions and host quartz in chip Rito2nd after various experiments. The composition of the monitoring glass from experiment #05 is shown in addition. Silicate melt inclusions
SiO2 TiO2 M2O3 FeO MnO MgO CaO Na2O K2O P Sc V Cu Zn As Rb Zr Nb Mo Ag Sn Sb Ba Cs La Ce Yb Hf Ta W Pb Bi Th U Li B ASI
Rito2nd original
Rito2nd after exp. #01
Rito2nd after exp. #03
Rito2nd after exp. #05
Monitoring glass” (exp.#05)
average (8)
1r
average (8)
1r
average (10)
1r
average (6)
1r
average (7)
1r
73.32 0.08 12.05 0.65 0.13 0.03 0.22 0.10 3.94 52.6 4.5 0.55 <0.33 51.0 18.1 321.9 253.0 134.5 7.3 <0.11 5.6 0.92 <0.38 9.3 33.9 40.3 5.72 16.7 8.15 6.1 31.5 0.81 71.5 35.3 379.2 21.9 2.360
0.15 0.01 0.00 0.03 0.01 0 00 0.05 0.03 0.23 21.2 0.7
72.15 0.04 12.05 0.61 0.15 0.02 0.35 2.25 4.04 46.5 5.9 0.21 25.86 44.2 6.0 365.3 186.1 126.1 8.7 158.68 5.9 1.26 0.11 12.5 51.4 50.8 6.50 14.7 7.38 7.8 37.7 1.50 55.1 44.2 369.0 51.3 1.331
0.45 0.01 0.00 0.03 0.00 0.00 0.05 0.63 0.31 19.6 0.8 0.00 13.45 2.8 0.8 10.1 47.3 8.4 3.0 71.96 2.4 0.25 0.00 1.1 11.6 4.8 3.99 2.4 0.92 1.4 1.6 0.12 5.9 13.1 34.3 8.8 0.133
73.33 0.04 12.05 0.68 0.16 0.03 0.24 0.03 4.06 68.4 6.8 0.17 1.99 63.9 4.0 418.8 211.0 147.1 16.3 1.02 5.0 3.12 1.16 14.6 50.1 53.7 5.59 16.6 8.59 10.8 39.5 2.04 62.9 40.0 2.4 46.6 2.486
0.25 0.01 0.00 0.07 0.02 0.01 0.08 0.01 0.13 30.4 2.6
74.75 0.05 12.02 0.64 0.16 0.02 0.29 2.16 4.01 95.0 6.6 <0.51 <1.60 65.4 4.5 410.5 191.0 153.8 9.2 0.31 8.0 2.93 <0.49 16.1 47.6 55.6 9.17 15.9 9.19 9.8 43.5 2.01 56.9 49.2 2800.4 46.2 1.16
0.32 0.01 0.00 0.03 0.01 0.00 0.04 0.38 0.15 26.5 1.2
74.77 0.01 12.65 0.27 0.00 0.04 0.14 3.35 2.75 53.8 0.4 0.55 0.48 1.1 0.4 9879.7 26.5 39.7 4.0 0.05 5.4 9.53 316.97 2912.2 4.8 8.1 0.22 0.8 0.05 <0.05 666.4 <0.04 0.3 0.2 4144.3 8.4 1.01
0.62 0.00 0.38 0.02 0.00 0.00 0.01 0.13 0.10 8.5 0.1 0.15 0.13 0.5 0.2 386.7 0.9 1.8 0.2 0.02 0.6 0.82 13.84 100.1 0.8 1.7 0.08 0.1 0.02
7.8 8.7 25 0 51.1 15.9 1.9 2.3 0.24 0.5 9.0 9.5 2.01 4.4 0.80 1.6 4.6 0.28 20.3 3.4 84.0 6.6 0.143
0.41 13.1 1.7 20.9 33.4 17.2 10.6 0.33 2.0 2.53 0.66 0.9 7.2 3.6 2.34 3.5 1.36 2.1 3.8 0.50 6.8 2.4 0.9 25.2 0.218
13.5 0.8 14.8 39.5 14.2 4.5 0.34 2.7 1.39 1.1 2.5 3.0 4.28 3.2 1.43 2.6 2.7 0.41 6.3 10.8 701.3 11.7 0.15
19.5 0.0 0.0 153.5 1.5 0.02
Host quartz
Ti Al Fe Na K Li P Sc Cu Ag Sn
average (3)
1r
average (4)
1r
average (6)
1r
average (5)
35.1 103.4 <5.67 <0.45 <1.99 15.2 26.1 0.49 <0.11 <0.05 2.99
23.0 21.4
33.5 68.6 <5.13 0.79 <0.83 6.6 30.1 0.71 <0.10 <0.03 4.01
2.9 8.5
33.0 81.4 <4.82 0.57 <2.24 <0.05 0.0 0.61 <0.24 <0.06 1.57
5.4 21.7
34.3 83.8 <6.92 0.85 1.35 21.6 16.5 0.54 <0.08 <0.02 1.19
4.2 5.0 0.05
0.60
0.01 3.5 9.2 0.27
1.40
0.13
0.0 0.10
0.24
1r 7.8 32.7 0.30 8.9 4.3 0.06 0.17
For silicate melt inclusions, major elements are in wt%, trace elements are in lg/g. For the host quartz, all elements are in lg/g and only clear, non-contaminated, inclusion free ablations were used. The lowest detection limits are shown.
3020
Z. Zajacz et al. / Geochimica et Cosmochimica Acta 73 (2009) 3013–3027
Table 3 Average composition of silicate melt inclusions and host quartz in chip BnSt after various experiment. Silicate melt inclusions BnSt original
SiO2 TiO2 AI2O3 FeO MnO MgO CaO Na2O K2O P Sc V Cu Zn As Rb Sr Zr Nb Mo Ag Sn Sb Ba Cs La Ce Yb Hf Ta W Pb Bi Th U Li B ASI
BnSt-experiment #02
BnSt-experiment #03
BnSt-experiment #04
average (10)
1r
average (7)
1r
average (3)
1r
average (7)
1r
71.36 0.04 13.60 0.98 0.05 0.05 0.72 2.81 5.82 52.9 3.11 0.76 10.00 33.9 7.2 229.4 43.4 75.5 19.6 2.31 0.17 9.5 0.85 188.7 9.74 26.7 49.2 2.48 2.54 1.98 3.81 26.7 0.33 28.2 8.5 269.2 60.3 1.11
0.27 0.01
74.18 0.05 13.60 0.93 0.05 0.05 0.89 0.06 5.46 57.7 6.39 0.62 1.08 29.2 14.0 239.9 54.6 94.3 20.0 1.89 0.19 4.7 0.78 235.8 9.31 31.4 56.7 3.30 3.74 2.78 1.79 29.4 0.44 32.9 10.4 0.8 61.1 1.78
0.74 0.01
70.23 0.07 13.60 0.98 0.05 0.05 0.50 0.02 4.93 <238.7 5.31 <9.11 <16.46 22.0 0.0 197.8 45.8 98.1 20.6 3.79 <1.58 <14.83 <3.41 277.0 7.04 35.6 62.6 <2.25 3.06 1.79 6.93 27.5 <1.51 29.9 6.9 6.8 46.9 2.17
0.38 0.03
72.93 0.04 13.60 0.90 0.05 0.06 0.77 0.92 5.92 48.0 4.26 0.73 0.33 135.3 7.8 290.7 42.8 78.8 18.8 3.73 0.06 5.7 0.57 327.3 8.90 35.5 67.9 2.82 3.72 1.99 3.01 30.8 0.27 26.8 7.3 6.6 54.9 1.46
1.22 0.01
0.08 0.01 0.01 0.13 0.09 0 23 23 9 0 27 0 28 7 90 79 1.5 15.9 13.5 13 3 1.5 1.14 0.02 4.7 0.55 145.7 1.82 5.1 9.1 1.99 0.80 0.32 0.75 2.1 0.14 3.1 1.8 120.6 10.2 0.03
0.09 0.00 0.01 0.10 0.08 0.39 41.1 6.94 0.24 0.00 8.1 8.5 24.5 16.1 10.8 2.9 0.96 0.03 1.2 0.12 144.2 2.05 6.0 13.2 0.76 0.77 0.89 0.50 5.5 0.11 4.3 1.5 0.0 25.2 0.32
0.09 0.01 0.01 0.13 0.00 0.30
27.1 0.0 14.5 5.0 8.7 2.5
99.9 0.51 4.2 5.3 0.00 0.67 5.6 4.1 0.5 7.3 0.15
0.08 0.02 0.03 0.57 0.34 0.71 36.0 1.08 0.89 165.0 2.5 115.7 15.4 13.9 3.5 2.99 0.00 1.1 0.31 252.1 3.23 10.8 26.4 1.53 1.91 0.60 1.08 9.1 0.01 5.7 2.2 5.0 7.0 0.23
Host quartz BnSt original
Ti Al Fe Na K Li P Sc Cu Ag
BnSt -experiment #02
BnSt -experiment #03
BnSt -experiment #04
average (12)
1r
average (7)
1r
average (4)
1r
average (5)
1r
33.0 96.3 7.0 0.71 <0.66 21.3 27.1 0.75 <0.09 <0.02
8.0 11.0 1.2 0.43
33.3 150.4 <10.03 0.77 <1.74 0.6 30.1 0.92 <0.15 <0.04
6.8 90.8 10.3 0.23 83.6
50.3 130.5 10.3 0.36 5.3 <0.05 19.3 0.67 <0.42 <0.03
6.8 35.1 3.0
26.9 95.0 3.0 1.14 <0.58 0.4 15.1 1.06 <0.43 <0.01
8.6 8.0
2.0 6.0 0.16
4.9 0.15
3.7 0.8 0.13
0.06 0.4 8.0 0.16
For the silicate melt inclusions, major elements are in wt%, trace elements are in lg/g. For the host quartz, all elements are in lg/g and only clear, non-contaminated, inclusion free ablations were used. The lowest detection limits are shown.
Diffusive reequilibration of quartz-hosted inclusions
3021
Fig. 6. Comparison of the compositions of SMI before and after experiment #05 (710 °C, 200 MPa, 168 h, doped salt solution). The composition of the haplogranite glass that has also been equilibrated with the doped fluid is shown as well. Li, Na and Ag shows obvious reequilibration as SMI compositions nearly adjust to that of the equilibrated glass. All other elements are unmodified. For some elements the experiment is non-conclusive as all three compositions are identical within uncertainty (clear fields). Major elements are in wt%, trace elements are in lg/g.
with the same constituents as #06 was carried out below the a/b transition temperature of quartz (experiment #07, 500 °C, 150 MPa, 168 h) to determine if diffusion of Na and Li is still rapid in this temperature range. Sodium content in the FI dropped about 40 relative%, and Li content dropped by over an order of magnitude (Fig. 9; Table 6). The NaCl dissolution temperatures dropped from >530 °C to 360–400 °C and the range in temperature of vapor bubble disappearance dropped from 430–460 °C to 350–400 °C. Concentrations of all analytes other than Na and Li remained constant during experiments #06 and #07. Table 4 Composition of doped solution #01 in experiment #05. In solution
(wt%)
As solids added to 150 ll solution
NaCl KCl FeCl2
10.15 3.71 3.29 (lg/g) 7901 9382 2733 201 208 206 1329 210 212 204
Cu Ag PbO
Li Rb Cs Sr Ba Sb La Sn Nb Mo
2 mg 1 mg 3 mg
The concentrations of Ag and Cu in FI did not show a distinct change in these two experiments. Therefore, an additional experiment (#08, 650 °C, 150 MPa, 72 h, intermediate density FI) was designed that imposed a stronger driving force for Cu and Ag diffusion by using a solution
Table 5 The composition of synthetic fluid inclusions trapped in quartz in experiment #05.
Li Na K Mn Fe Cu Zn Rb Sr Nb Mo Ag Sn Sb Cs Ba La Pb
Average (3)
1r
7258 39900 20272 24 12054 401 60 9648 204 1.1 111 2.8 58 8.3 3125 225 127 9706
424 1137 1936 31 2356 87 61 884 10 1.1 10 3.1 21 8.1 535 12 152 1124
Concentrations are shown in lg/g.
3022
Z. Zajacz et al. / Geochimica et Cosmochimica Acta 73 (2009) 3013–3027
Fig. 7. Change in the physical appearance of SMI in sample Rito2nd after various experiments. (a) The original appearance of the SMI, inclusions were strongly recrystallized due to slow cooling in nature; (b) The same inclusions after the reheating experiment (exp #01). The silicate melt quenches to a homogeneous glass. Note that the shown inclusions are relatively large, thus incompletely homogenized; (c) The same inclusions after depletion of Na in experiment (exp #03). Note that the inclusions rapidly recrystallized during the 3 min quench time likely due to the low Na2O content; (d) The same inclusions after experiment #05. Note that the elevated Na2O content causes the silicate melt to quench into a homogeneous glass. Two of the inclusions were ablated before the experiment.
Fig. 8. Results of experiment #06 (650 °C, 200 MPa, 72 h, 4 m HCl) on brine inclusions in miarolitic quartz crystals (sample Erg03). A significant drop in Na and Li concentrations is clearly observable, while other element concentrations remained unchanged. All element concentrations are shown in lg/g.
Diffusive reequilibration of quartz-hosted inclusions
3023
Fig. 9. Comparison of the compositions of FI before and after experiment #07 (500 °C, 150 MPa, 168 h, 4 m HCl solution). All element concentrations are shown in lg/g. A significant drop in Na and Li concentrations is clearly observable.
doped with Cu (5 wt% CuCl2 dissolved, +2.5 wt% solid Cu metal) and Ag (1 wt% solid AgCl). Additionally, the doped solution contained 6 wt% LiCl, 2 wt% NaCl and 1 wt% HCl. To reduce Cu and Ag loss to the capsule by alloying, a Pt capsule was used in this experiment. During this experiment the Na/Li molar ratio in the FI almost completely equalized (changed from 0.25 to 3.25) to that of the doped solution (Na/Li = 4.14). Copper concentration in the FI increased more than two orders of magnitude (from 7 lg/g to
2000 lg/g), while Ag concentrations increased more moderately from <2 lg/g to >7 lg/g (Fig. 10). Simultaneously to the modification of SMI and FI compositions, a change in the Li content of the host quartz was also observed (K, Cu and Ag contents were always below detection limits of 2, 0.1, 0.03 lg/g in the quartz, respectively). In the samples from miarolitic cavities, about 58 relative% of the Al content of the quartz (55–220 lg/g) was charge balanced by Li. In the magmatic phenocrysts,
Fig. 10. Comparison of the compositions of SMI before and after experiment #08 (650 °C, 150 MPa, 72 h, doped salt solution). All element concentrations are shown in lg/g.
3024
Z. Zajacz et al. / Geochimica et Cosmochimica Acta 73 (2009) 3013–3027
Table 6 Compositional data of analyzed fluids inclusions. Erg03 –original
Li B Na Al K Ca Mn Fe Cu Zn As Rb Sr Ag Sn Sb Cs Ba W Pb Bi
Erg03- experiment. #06
Erg03- experiment. #07
Cana11a–original
Cana11a–experiment #08
average(11)
1r
average(9)
1r
1606.0
269.4
6049.7
268.8
20609 3471 9065
615 568 1521
5968 2692 9600
901 728 1112
1639
124
1614
233
6.6 427.0
4.1 41.5
1971.6 480.8
440.3 157.9
390 0.7 1.5 14.7 15.2 118
40 0.5 1.2 6.7 2.8 20
463 0.9 7.8 15.2 13.0 136
41 0.6 4.0 5.3 3.5 15
266
17
292
26
average(8)
1r
average(3)
1r
average(5)
1r
88.9 255 153259 228 69682 2106 20226 106479 40.8 4734.1 1465 5889 369 4.1 14.0 8.1 1842 43.3 16.6 1277 62.5
44.2 91 4879 96 4285 1315 1324 6744 13.8 276.1 134 507 37 1.1 2.9 1.6 188 11.8 4.5 74 9.2
9.0 224 58733 343 72451 1117 21965 122034 69.5 4281.0 812 6671 377 4.1 14.5 6.6 2084 45.0 19.5 1184 54.5
1.5 115 6670 489 4963 323 1473 10167 7.2 848.3 358 455 50 0.3 10.8 1.3 177 0.8 4.0 170 6.0
3.3 300 92763 183 65534 767 19961 107393 31.0 4232.8 1355 5849 347 5.6 19.8 7.4 1869 44.4 13.8 1136 60.8
123 635 97 1427 843 386 1286 6.5 401.2 363 99 10 2.4 16.2 2.4 54 15.7 4.8 87 11.6
Host quartz Erg03 –original
Al Fe Na K Li Cu Ag
Erg03-experiment. #06
Erg03-experiment. #07
Cana11a–original
Cana11a– experiment #08
average(8)
1r
average(3)
1r
average(5)
1r
average(11)
1r
average(9)
1r
78 <7.4 9.3 5.6 12.7 <0.3 <0.02
19
79 <12.6 <2.11 <13.59 <0.28 <0.2 <0.06
6
73 <14.1 <0.5 <3.6 <0.29 <0.4 <0.06
5
83 <8.2 2.82 <1.4 14.2 <0.2 0.035
16
80 <12 0.9 <3.9 13 <0.6 <0.1
12
6.3 2.3 3
1.34 2.5 0.012
0.5 2.5
Concentrations are shown in lg/g. Only clear, non-contaminated, inclusion free ablations used to obtain the composition of the host quartz. The lowest detection limits are shown.
the same value is about 85 relative% (Fig. 11; Tables 2 and 3). The rest of Al must have originally been charge balanced by hydrogen. During interaction with HCl solution or pure water, Li leaves the quartz and 99 relative% of the Al content becomes charge balanced by H. During interaction with the doped salt solution in experiment #05, Li+ completely replaced H+ again. Sodium content of all investigated quartz slabs were originally below 1 lg/g, near or below limits of detection, thus changes could not be monitored. However, based on its maximum concentration, Na could play only a very subordinate role in charge balancing Al. 5. DISCUSSION Consistent modification of the composition of the SMI and FI and their host quartz in equilibrium with various fluids points to bulk diffusion of H, Li, Na, Cu and Ag or
their cations in the quartz. Considering the large number of studied elements with similar atomic and effective ionic radii (Shannon, 1976), concentration modification of only those elements that form cations in the univalent state suggests that these elements diffuse as charged species through the quartz structure, and that the tendency for rapid diffusion is valence-specific. If the effective ionic radii of known univalent cations are compared, it is apparent that the elements which rapidly diffused through quartz are the small˚ < Cu+ – 0.60 A ˚ < Na+– est (diffusing cations: Li+ – 0.59 A + ˚ ˚ 0.99 A < Ag – 1.00 A << non-diffusing cations: K+ – ˚ < Rb+ – 1.52 A ˚ < Cs+ – 1.67 A ˚ ). As Cl (the main 1.37 A anion present in the FI) is very unlikely to diffuse through ˚ ), quartz due to its very large effective ionic radius (1.81 A the diffusion of cations in and out of SMI and FI must have to be governed by exchange reactions involving cations that maintain charge balance. This is clearly observed in experiment #06, #07 and #08 whereby the concentration of Cl
Diffusive reequilibration of quartz-hosted inclusions
3025
Fig. 11. Compositional variability of studied quartz samples before and after the conducted experiments. The role of alkali element ions and H+ in charge balancing Al in the quartz is shown together with the aluminum saturation indices (ASI) of SMI. Data for Mt. Malosa, Cuasso al Monte and Ehrenfriedersdorf are taken from Zajacz et al. (2008).
did not change in the FI, but the cation ratios readjusted to approach the values in the external solutions. In experiments with pure aqueous HCl solutions (i.e., containing no trace metal or alkali cations), the most likely candidate cation that replaced Na+ and Li+ in the SMI and FI is H+, the smallest and most rapidly diffusing univalent cation. Pure H2O was less effective in causing the removal of Na, Li, Cu and Ag from the SMI than the aqueous HCl solutions, which can be explained by the elevated H+ activities and the presence of Cl that forms complexes with these cations in the HCl solution. The replacement of NaCl by HCl in the fluid inclusions will result in a decreasing density, which intuitively would suggest that the bubble disappearance temperatures should increase as opposed to our observations. However, neither the boiling curve nor the isochors are known in the HCl–H2O system at the studied P–T conditions. Therefore it cannot be excluded that the partial molar volume of the HCl enriched fluid is significantly larger at a specific P–T than it is for a fluid with the same Cl, but higher Na concentration, resulting in lower homogenization temperature. In the case of SMI, the loss of Na, Li, Cu and Ag must be accompanied by an increase in the water content since the metal-oxides are replaced by H2O via cation exchange through the quartz. This increase will, however, not be very significant when expressed in wt%, because of the large difference in the molar mass of the Na2O and H2O components. For example, a loss of 3 wt% Na2O will result in only a 0.87 wt% increase in the H2O content.
Rapid diffusion of H+, Li+ and Na+ along the open channels parallel to the c-axis in the quartz structure was previously observed by applying an electric field to quartz (Verhoogen, 1952; White, 1970; Kronenberg et al., 1986; Kronenberg and Kirby, 1987; Girardet et al., 1988; Plata et al., 1988; Bahadur, 1993). Frischat (1970) and Rybach and Laves (1967) determined high diffusion coefficients for Na between 400 and 1000 °C using only a concentration gradient as the driving force. Our results show that at near-magmatic temperatures, concentration gradients impose a sufficient driving force to induce rapid diffusion of H+, Li+ and Na+ through quartz, and no electric field is needed. Besides these elements, rapid diffusion of Cu+ and Ag+ was also observable which is consistent with the very similar effective ionic radii and identical valence of these cations to those of Li+ and Na+, respectively. The diameter of the channels parallel to the c-axis in the quartz crystal structure is the limiting factor determining which univalent cations can diffuse (Verhoogen, 1952; White, 1970; Plata et al., 1988). The observation that the same process occurred in magmatic quartz phenocrysts and a variety of hydrothermal-type miarolitic quartz samples suggests that diffusion does not require any kind of special defect structures in the quartz. Though our experiments were not aimed at precise determination of diffusion coefficients, we can provide a first approximation for Na and Li. Considering the degree of reequilibration in the Na content of the SMI in quartz during heating experiments we can conclude that the SMI were modified to a very similar extent in experiments with a run time between 24 and 168 h, between 710 and 740 °C. This suggests
3026
Z. Zajacz et al. / Geochimica et Cosmochimica Acta 73 (2009) 3013–3027
Fig. 12. Degree of reequilibration of the Na concentration in SMI in quartz as a function of time at various diffusivities (shown as numbers in m2/s next to the curves). The modeling indicates that D is likely larger than 10 10 m2/s at 740 °C to allow achievement of nearly complete reequilibration within 24 h.
that the reequilibration of the SMI is nearly complete already after 24 h in this temperature range. We used Fick’s first law to model the degree of reequilbation of the SMI as a function of time. As the diffusion of Na+ is orders of magnitude slower perpendicular to the c-axis of the quartz than parallel to it, we accounted for mass transfer only parallel to the c-axis. We used the following boundary conditions based on our typical experimental conditions: (1) We assumed a spheric SMI with a diameter of 20 lm sitting 50 lm below the external surface of the quartz slab, which is perpendicular to the crystallographic c-axis. The distance from the other side of the quartz slab was considered to be infinite. The difference in the Na content of the quartz in equilibrium with a Na-poor (0.1 wt% Na2O) and a Na-rich (2 wt% Na2O) SMI were considered to be 1 lg/g. We assumed 1 lg/g Na content in equilibrium with the external environment of the quartz chip as well. Model calculations indicated that the diffusion coefficient of Na is likely larger than 10 10 m2/s at temperatures of 710– 740 °C (Fig. 12). This result is consistent with the data of Frischat (1970; 2.08 10 9 m2/s at 720 °C) but slightly higher than the value determined Rybach and Leaves (1967; 1.53 10 11 m2/s at 720 °C). Verhoogen (1952) also determined lower Na diffusion coefficients using electric field as a driving force (1.88 10 12 m2/s at 720 °C). We calculated a minimum required diffusion coefficient for Li as well based on experiment #05, in which the largest change in Li concentration was observed. We used the same boundary conditions, except for the initial concentration gradient, which was 21 lg/g for Li. Our estimated minimum value for the diffusion coefficient of Li is about 10 11 m2/s at 710 °C which is consistent with those determined by Verhoogen (1952) calculated for the same temperature (1.9 10 11 m2/s at 720 °C). 6. CONCLUSIONS AND IMPLICATIONS The experiments showed that concentrations of Na, Li, Cu, Ag and H can be rapidly modified in quartz-hosted sil-
icate melt and fluid inclusions after entrapment, via diffusion through the quartz at P–T conditions typical of upper crustal felsic magmatism and related high temperature ore-forming hydrothermal activity. As these elements diffuse as univalent cations, their mobility is limited by the necessity to maintain charge balance. The concentration of Cl (sensu lato salinity) cannot be modified in fluid inclusions, but the relative proportions of H, Na, Li, Cu and Ag can easily reequilibrate with subsequent generations of fluids, silicate melts, sulfide liquids, or precipitated sulfide minerals that come into contact with the host quartz. Since Na is usually the most abundant of these elements, and H/ Na ratios as high as those imposed in this study are unlikely to occur in natural hydrothermal fluids, very significant modification of the NaCl contents and volumetric properties of fluid inclusions in nature is unlikely. However, Li/ Na, Cu/Na and Ag/Na ratios can readily readjust to equilibrate with external fluids or melts via cation exchange. This may induce very significant post-entrapment modifications of the Li, Na, Cu and Ag concentrations in both silicate melt and fluid inclusions in quartz, and should be kept in mind during interpretation of this kind of compositional data. ACKNOWLEDGEMENTS We thank Andreas Aude´tat for providing us with samples from Rito del Medio and Canada Pinabete. We are grateful to Peter Ulmer for providing access to the cold seal pressure vessel facility at ETH Zu¨rich, and to Larryn Diamond and Monika Painsi for allowing and helping us to use the cold seal pressure vessel facility at the University of Bern. We thank the members of the Laboratory for Mineral Deposits Research for allowing us to use their facility to conduct the last experiment of this study at the University of Maryland. We are grateful to Adam C. Simon and an anonymous reviewer for their constructive comments and to Edward M. Ripley for his editorial work. This research was supported by the Swiss National Science Foundation (SNF) under project numbers 20020-107955 and PBEZ2-118876.
REFERENCES Aude´tat A. and Gu¨nther D. (1999) Mobility and H2O loss from fluid inclusions in natural quartz crystals. Contrib. Mineral. Petrol. 137, 1–14. Aude´tat A., Gu¨nther D. and Heinrich C. A. (2000) Magmatichydrothermal evolution in a fractionating granite: a microchemical study of the Sn-W-F-mineralized Mole Granite (Australia). Geochim. Cosmochim. Acta 64, 3373–3393. Aude´tat A. and Pettke T. (2003) The magmatic-hydrothermal evolution of two barren granites: a melt and fluid inclusion study of the Rito del Medio and Canada Pinabete plutons in northern New Mexico (USA). Geochim. Cosmochim. Acta 67, 97–121. Bahadur H. (1993) Hydroxyl defects and electrodiffusion (sweeping) in natural quartz crystals. J. Appl. Phys. 73, 7790–7797. Candela P. A. and Piccoli P. M. (1995) Model ore-metal from melts into vapor and vapor/brine mixtures. In Magmas, fluids and ore deposits (eds. Fluid Magmas, Or Deposits and J. F. H. Thompson). University of British Columbia, pp. 101–127. Frischat G. H. (1970) Sodium diffusion in natural quartz crystals. J. Am. Ceram. Soc. 53, 357–359.
Diffusive reequilibration of quartz-hosted inclusions Girardet C., Plata J., Breton J. and Hardisson A. (1988) Electrodiffusion of Alkali-Metal Ions and Protons in Quartz – a Diffusion-Compensation Model. Phys. Rev. B 38, 5648–5653. Gu¨nther D., Audetat A., Frischknecht R. and Heinrich C. A. (1998) Quantitative analysis of major, minor and trace elements in fluid inclusions using laser ablation inductively coupled plasma mass spectrometry. J. Anal. Atom. Spectrom. 13, 263– 270. Gu¨nther D., Frischknecht R., Heinrich C. A. and Kahlert H. J. (1997) Capabilities of an Argon Fluoride 193 nm excimer laser for laser ablation inductively coupled plasma mass spectrometry microanalysis of geological materials. J. Anal. Atom. Spectrom. 12, 939–944. Halter W. E., Pettke T., Heinrich C. A. and Rothen-Rutishauser B. (2002) Major to trace element analysis of melt inclusions by laser ablation ICPMS: methods of quantification. Chem. Geol. 183, 63–86. Heinrich C. A., Pettke T., Halter W. E., Aigner-Torres M., Audetat A., Gunther D., Hattendorf B., Bleiner D., Guillong M. and Horn I. (2003) Quantitative multi-element analysis of minerals, fluid and melt inclusions by laser ablation inductively coupled plasma mass spectrometry. Geochim. Cosmochim. Acta 67, 3473–3497. Johannes W. and Holtz F. (1996) Petrogenesis and experimental petrology of granitic rocks. Springer-Verlag, Berlin. Kamenetsky V. S. and Danyushevsky L. V. (2005) Metals in quartz-hosted melt inclusions: natural facts and experimental artifacts. Am. Mineral. 90, 1674–1678. Kronenberg A. K. and Kirby S. H. (1987) Ionic-conductivity of quartz – Dc time-dependence and transition in charge-carriers. Am. Mineral. 72, 739–747. Kronenberg A. K., Kirby S. H., Aines R. D. and Rossman G. R. (1986) Solubility and diffusional uptake of hydrogen in quartz at high water pressures – implications for hydrolytic weakening. J. Geophys. Res. Sol. Earth and Planets 91, 2723– 2744. Longerich H. P., Jackson S. E. and Gu¨nther D. (1996) Laser ablation inductively coupled plasma mass spectrometric transient signal data acquisition and analyte concentration calculation. J. Anal. Atom. Spectrom. 11, 899–904. Mavrogenes J. A. and Bodnar R. J. (1994) Hydrogen movement into and out of fluid inclusions in quartz – experimental-
3027
evidence and geologic implications. Geochim. Cosmochim. Acta 58, 141–148. Plata J., Breton J. and Girardet C. (1988) Theoretical-model for the electrodiffusion of M+ (M = Li, Na, K) ions in a quartz crystal. Phys. Rev. B 38, 3482–3493. Qin Z. W., Lu F. Q. and Anderson A. T. (1992) Diffusive reequilibration of melt and fluid inclusions. Am. Mineral. 77, 565–576. Rybach L. and Laves F. (1967) Sodium diffusion experiments in quartz crystals. Geochim. Cosmochim. Acta 31, 539–546. Severs M. J., Azbej T., Thomas J. B., Mandeville C. W. and Bodnar R. J. (2007) Experimental determination of H2O loss from melt inclusions during laboratory heating: evidence from Raman spectroscopy. Chem. Geol. 237, 358–371. Shannon R. D. (1976) Revised effective ionic-radii and systematic studies of interatomic distances in halides and chalcogenides. Acta Crystallogr. A 32, 751–767. Student J. J. and Bodnar R. J. (2004) Silicate melt inclusions in porphyry copper deposits: identification and homogenization behavior. Can. Mineral. 42, 1583–1599. Verhoogen J. (1952) Ionic diffusion and electrical conductivity in quartz. Am. Mineral. 37, 637–655. Vityk M. O., Bodnar R. J. and Doukhan J. C. (2000) Synthetic fluid inclusions. XV. TEM investigation of plastic flow associated with reequilibration of fluid inclusions in natural quartz. Contrib. Mineral. Petrol. 139, 285–297. White S. (1970) Ionic diffusion in quartz. Nature 225, 375–376. Williams T. J., Candela P. A. and Piccoli P. M. (1997) Hydrogenalkali exchange between silicate melts and two-phase aqueous mixtures: an experimental investigation. Contrib. Mineral. Petrol. 128, 114–126. Zajacz Z. and Halter W. (2007) LA-ICPMS analyses of silicate melt inclusions in co-precipitated minerals: quantification, data analysis and mineral/melt partitioning. Geochim. Cosmochim. Acta 71, 1021–1040. Zajacz Z., Halter W. E., Pettke T. and Guillong M. (2008) Determination of fluid/melt partition coefficients by LAICPMS analysis of coexisting fluid and silicate melt inclusions: controls on element partitioning. Geochim. Cosmochim. Acta 72, 2169–2197. Associate editor: Edward M. Ripley
Geochimica et Cosmochimica Acta 73 (2009) 3013–3027 www.elsevier.com/locate/gca
Diffusive reequilibration of quartz-hosted silicate melt and fluid inclusions: Are all metal concentrations unmodified? Zoltan Zajacz a,c,*, Jacob J. Hanley b, Christoph A. Heinrich c, Werner E. Halter c, Marcel Guillong c a
Laboratory for Mineral Deposits Research, Department of Geology, University of Maryland, College Park, MD 20742, USA b Department of Geology, Saint Mary’s University, Halifax, Canada c Institute of Isotope Geochemistry and Mineral Resources, ETH Zu¨rich, Switzerland Received 8 September 2008; accepted in revised form 17 February 2009; available online 4 March 2009
Abstract Experiments were conducted to determine the extent and mechanism by which the composition of quartz-hosted silicate melt inclusions (SMI) and aqueous fluid inclusions (FI) can undergo post-entrapment modification via diffusion. Quartz slabs containing assemblages of SMI and FI were reacted with synthetic HCl bearing and metalliferous aqueous fluids at T = 500– 720 °C and P = 150–200 MPa. SMI from the single inclusion assemblages were analyzed by laser ablation inductively coupled plasma mass spectrometry (LA-ICPMS) and electron probe microanalysis (EPMA) before and after the experiments. Analyses revealed that rapid diffusion of the univalent cations Na+, Li+, Ag+, Cu+ and H+ occurred through the quartz from the surroundings, resulting in significant changes in the concentrations of these elements in the inclusions. Concentrations of other elements with an effective ionic radius larger than that of Ag+, or multiple valence states were not modified in the inclusions during the experiments. Our results warn inclusion‘‘ researchers that the interpretation of Na, Li, Cu and Ag concentrations from quartz-hosted SMI and FI should be treated critically. Ó 2009 Elsevier Ltd. All rights reserved.
1. INTRODUCTION Quartz-hosted fluid inclusions (FI) and silicate melt inclusions (SMI) are commonly considered to preserve the composition of hydrothermal fluids and evolved silicate melts, allowing natural processes that lead to the generation of ore deposits and volcanic eruptions to be traced and characterized chemically. However, observations of anomalously Na-deficient SMI compositions in natural samples (Aude´tat et al., 2000; Aude´tat and Pettke, 2003; Student and Bodnar, 2004; Zajacz et al., 2008) and modification of the Na, Cu and Ag content of natural SMI during experimental reheating (Kamenetsky and Danyushevsky, 2005; Zajacz et al., 2008) raise the question as to whether inclusions in quartz *
Corresponding author. Address: Laboratory for Mineral Deposits Research, Department of Geology, University of Maryland, College Park, MD 20742, USA. E-mail address: [email protected] (Z. Zajacz). 0016-7037/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.gca.2009.02.023
reliably represent the composition of fluids and silicate melts at the time of their entrapment. No systematic experimental study has been carried out to determine if such fluid and melt inclusions can be compositionally modified after entrapment by diffusion-controlled cation exchange through the quartz host between the inclusions and the surroundings of their host crystals. Previous studies focused only on the diffusion of molecular H2 and H2O in and out of the inclusions (Qin et al., 1992; Mavrogenes and Bodnar, 1994; Vityk et al., 2000; Severs et al., 2007). Aude´tat and Gu¨nther (1999) studied the compositional changes of fluid inclusions during their migration related to the deformation of quartz in natural systems and found that besides water loss from the inclusions, the ratios of the analyzed cations remained constant with the exception of Li. The present study, describes a series of experiments conducted on natural samples to determine which element concentrations can be significantly modified in quartz-hosted FI and SMI after their entrapment at P–T conditions typi-
3014
Z. Zajacz et al. / Geochimica et Cosmochimica Acta 73 (2009) 3013–3027
cal of sub-volcanic magma reservoirs and associated high temperature hydrothermal systems. 2. STARTING MATERIALS AND EXPERIMENTAL PROCEDURES To determine which cations are mobile in quartz, slabs of natural quartz containing FI and/or SMI were immersed in high temperature aqueous fluids of various compositions using conventional hydrothermal experimental techniques. Free-standing quartz crystals from miarolitic cavities sampled from plutons at Rito del Medio, New Mexico, USA (Sample #Rito2nd), Canada Pinabete, New Mexico, USA (Sample #Cana11a) and Erongo, Namibia (Sample #Erg03) containing FI and recrystallzed SMI, and quartz phenocrysts from a rhyolite sampled from Banska Stiavnica, Slovakia, (Sample #BnSt) containing glassy SMI were selected for experimental investigation. FI and SMI from Rito2nd and Cana11a occur in well-characterized inclusion assemblages along primary growth zones or in pseudosecondary trails (Aude´tat and Pettke, 2003; Zajacz et al., 2008). Inclusions in Erg03 occur in pseudosecondary trails. Both SMI and FI are free of any sign of decrepitation, necking or leakage. SMI are rhyolitic in composition, but show anomalously low Na2O content (0.10 ± 0.03 wt%, 1r, 8 analyses) compared to what is expected for a typical haplogranitic liquid. All SMI in miarolitic quartz were strongly recrystallized. Melt inclusions from Banska Stiavnica are glassy and occur isolated or in clusters of up to 15 inclusions that do not follow any obvious textural or growth feature. The inclusions vary between <5 and 220 lm in diameter. They are rhyolitic in composition with an Na2O content of 2.8 ± 0.1 wt% (1r, 10 analyses). Slabs of quartz were cut from petrographic thick sections (200–500 lm thickness) using a small diamond wheel, re-polished with clean diamond paste (3 and 1 lm) and cleaned through multiple washings with acetone in an ultrasonic cleaner before placing them into the metal capsules. Slabs from Cana 11a and Erg03 were cut parallel to the crystallographic c-axis, while slabs from Rito2nd were cut parallel to the growth surfaces. The size of the analyzed
inclusions ranged from 15 to 50 lm in diameter and they were 10–60 lm below the sample surface. To prepare the solutions added to the capsules, we used analytical grade aqueous HCl solution (16 m) and metal chloride reagents or AAS standard chloride solutions. High purity SiO2 powder (SiO2 > 99.9%) was added to all experimental charges to minimize the dissolution of the studied quartz chips. The components of each experimental charge as well as run conditions and duration are shown in Table 1. Capsules were weighed before and after welding, and after the experiments to check for potential fluid loss. In experiment #01 (Table 1) SMI were reheated in open (crimped but not sealed) Ag70Pd30 capsules in an atmosphere of Ar in pressure vessels composed of Nimonic105TM alloy. Samples were held at T = 710 °C, P = 150 MPa for 72 h, then at T = 740 °C, P = 180 MPa for 70 h. These conditions approached the predicted melt homogenization temperature, estimated from the water-saturated haplogranite solidus (Johannes and Holtz, 1996). Other open-capsule reheating experiments were described by Zajacz et al. (2008), thus here we describe only the one from which the inclusion bearing quartz chip has been used for further experiments (experiment #01). Experiments #02 to #07 were run in Rene´ 41 pressure vessels using an Ar pressure medium. Experiments #01 to #07 were initially pressurized so that the approximate target pressure was reached during heating to run temperature. Once at final run temperature, the desired final pressure was set. Experiment #08 was carried out in water pressure medium, and was near isobarically heated to run temperature. Experiments were air quenched after removal from the furnaces and cooled below 400 °C within 3 min. After removal from the capsules, the quartz chips commonly displayed a thin overgrowth of new quartz (few tens of lm thick) which was removed by polishing. The chips were cleaned through multiple washings in acetone in an ultrasonic cleaner prior to LA-ICPMS analysis. In order to constrain an internal standard for the LAICPMS analyses of single FI (Heinrich et al., 2003), microthermometric measurements were conducted before and after the experiments using a Linkam THMSG 600 heat-
Table 1 Summary of conducted experiments. Experiments T (°C) P (MPa) Time (h) sample #01 #02 #03 #04 #05
740 720 710 710 710
180 200 150 200 200
142 72 24 72 168
#06 #07 #08
650 500 650
200 150 150
72 168 72
Added solutions and solids Capsule
Rito2nd SiO2 powder BnSt–4 chips 4 m HCl–80 ll Rito2nd, BnSt–2 chips 1 m HCl–150 ll BnSt–2 chips H2O–120 ll Rito2nd, BnSt–2 chips Doped sol. #01–120 ll Haplogranite glass–12 mg PbO–3 mg, Cu–2 mg Ag70Pd30–1 mg Prefracured quartz core Erg03–3 chips 4 m HCl–130 ll Erg03–2 chips 4 m HCl–150 ll Cana11a doped sol. #02–100 ll AgCl–1 mg, Cu–2.5 mg
Location/vessel material
Ag70Pd30–crimped Au–sealed Au- sealed Au- sealed Au-sealed Au-sealed
Univ. of Bern/Nimonic 105 ETH Zurich/Rene 41 ETH Zurich/Rene 41 ETH Zurich/Rene 41 ETH Zurich/Rene 41
Au-sealed Au-sealed Pt-sealed
ETH Zurich/Rene 41 ETH Zurich/Rene 41 Univ. of Maryland/Rene 41
Diffusive reequilibration of quartz-hosted inclusions
ing-freezing stage mounted on a Nikon petrographic microscope. 3. ANALYTICAL TECHNIQUES Both fluid and melt inclusions were analyzed by using LA-ICPMS at ETH Zu¨rich. The system consists of a 193 nm ArF excimer laser to produce an energy homogenized beam profile (Gu¨nther et al., 1997) coupled with an ELAN 6100 DRC ICP quadrupole mass spectrometer. The laterally homogenous energy density of the laser beam allows controlled ablation with a flat bottomed ablation crater. The system permits online observation of the ablation signal which is a prerequisite for controlled inclusion analysis. Analyses of a maximum of 16 unknowns were bracketed by measurement of an external standard (NIST SRM 610) to allow linear drift correction. The linear dynamic range of 9 orders of magnitude of the detector allows simultaneous measurement of major and trace elements down to sub-lg/g level. This allows quantification of the measurements by normalizing the concentration of the major elements to 100 wt% minus the total volatile concentration. More detailed description of the quantification approach for homogeneous solids and fluid inclusions is given in (Longerich et al., 1996; Gu¨nther et al., 1998; Heinrich et al., 2003). Output energy of the laser source was adjusted to the minimum value required for controlled ablation of the host quartz for each sample (generally between 20 and 32 J/cm2 energy density on the sample surface). The volume of the oval shaped ablation cell was 1 or 8 cm3. Analyses of the SMI and the quantification of their composition were carried out using the method of Halter et al. (2002) and Zajacz and Halter (2007). The entire inclusions were ablated along with the host quartz and the contribution of the host quartz was subtracted from the mixed signal to obtain the composition of the SMI. To determine the amount of host to be subtracted, we used the average Al2O3 content of the melt inclusions as an internal standard, determined by EPMA on other inclusions from the same assemblages, that were rehomogenized previously (except for sample BnSt, in which the SMI were quenched to glass in nature). The method provides the bulk composition of recrystallized SMI, including the host mineral that generally crystallizes on the walls of SMI. The composition of fluid inclusions was quantified using NaCl wt% equivalent concentrations based on microthermometric observation of freezing point depression and salt dissolution tempreatures. Due to the presence of cations other than Na in significant concentrations, a salt correction has been carried out with all major cations (generally K, Fe, Mn) to obtain true Na concentrations following the method of (Gu¨nther et al., 1998). These Na concentrations were used as an internal standard to quantify the LAICPMS data for fluid inclusions. Homogenized silicate melt inclusions were analyzed for their major element composition by a JEOL SUPERPROBE JXA-8200 electron microprobe at ETH Zu¨rich. A 15 kV acceleration voltage was used. The homogenized inclusions are highly unstable under the electron beam, thus we used a beam current of 5 nA and the largest beam diam-
3015
eter possible, depending on the size of the melt inclusions (generally P8 lm). The Na signal was collected first, and was monitored online during each analysis. Only those measurements where no significant change could be observed in the Na count rate during analyses were further considered. Counting times of 20 s on the peak and 10 s on the background on both sides were used for all major elements except for Na, for which we used only 10 s peak and 5 s background counting. 4. RESULTS Since the investigated inclusions were kept intact throughout the experimental duration and did not crack or leak, there was no direct physical contact between the content of the inclusion (FI or SMI) and the outside environment and the only possible mechanism of mass transfer through the host quartz was diffusion. The Na2O content of the SMI increased significantly in all inclusions measured (from 0.05–0.7 to 1.5–3.5 wt% Na2O) during the re-homogenization experiment (#01) similarly to other open-capsule reheating experiments reported by Zajacz et al. (2008). Besides Na, concentrations of Ag and Cu increased in all inclusions analyzed by two and one order of magnitude, respectively (Fig. 1). Concentrations of all other analyzed elements remained unchanged. As no Ag, Cu or Na was present in the host quartz before the experiments in significant concentration (generally Na<1 lg/g, Cu < LOD of 0.1 lg/g, Ag < LOD of 0.03 lg/g) the observations suggest that these element enrichments in the inclusions, at least in part, were derived from an external source (surrounding the quartz host). The most likely source of Ag is the capsule material (Ag70Pd30). The Cu may originate from the capsule (36 lg/g Cu analyzed in the capsule material) or more likely the vessel material itself (up to 0.2 wt% Cu in the Nimonic 105 alloy). The presence of alloyed metallic Ag and Cu at run temperatures may have yielded large enough chemical potential gradients between the SMI and the exterior of the host quartz to allow metal transport through the Ar atmosphere simultaneously with rapid diffusion through the quartz. Transfer of metals through the Ar pressure media may have been possible despite the rather low vapor pressure of the metals at run temperature due to very rapid diffusion in the gas phase. Transfer through direct contact between the vessel, the capsule wall and the quartz chip may have been also possible. The source of Na enriched in the SMI is likely to be the large volume of quartz surrounding the inclusions. To raise the Na2O content of an SMI with a typical value of 2 wt%, a spherical volume of quartz with approximately 24 times the radius of the SMI is necessary (assuming 1 lg/g Na in the quartz) if all the Na leaves the quartz. Though such volume of quartz is available in some of the reheated chips, not in all of them. Therefore additional sources of Na have to be taken into consideration. In experiment #01 the quartz chip were placed into the capsule together with supporting SiO2 powder (about 3 times the weight of the quartz chip itself), which may have contained Na well above 1 lg/g (up to 20 lg/g, Alfa Aesar) and contributed to the Na enrichment in the SMI. Quartz chips from Cuasso al Monte, Baveno
3016
Z. Zajacz et al. / Geochimica et Cosmochimica Acta 73 (2009) 3013–3027
Fig. 1. Comparison of the compositions of SMI before and after experiment #01 (740 °C, 180 MPa, 142 h) in sample Rito2nd. Concentrations of Na, Ag and Cu strongly increased in the melt inclusions during the heating experiment, while other element concentrations remained unaltered. Cu and Ag were below the limits of detection before the experiment, thus we plotted the limits of detections. Major elements are in wt%, trace elements are in lg/g.
and Mt. Malosa described by Zajacz et al. (2008) contained alkali-feldspar intergrown with the quartz, because it could not be entirely separated before the heating experiments. Therefore, in those chips the alkali-feldspar may serve as an additional source of Na to allow for the observed increase in Na concentrations (Zajacz et al., 2008). To verify the hypothesis derived from open-capsule reheating experiments, that elevated metal concentrations surrounding the quartz slabs induced metal enrichment in the SMI by diffusion through the quartz, additional experiments were conducted to reproduce the same and reversal processes reacting the quartz with various fluids of known composition in sealed capsules. Hydrogen chloride-rich fluids are known to be in equilibrium with silicate melts with an aluminum saturation index (‘‘ASI” = Al/[Na + K + 0.5Ca + Li + Rb + Cs]) at or higher than unity (Williams et al., 1997). Trace elements complexed by chloride should partition strongly into such a fluid, being incompatible in the coexisting melt. Thus, in experiments #02 and #03 (Table 1), quartz chips containing SMI with an ASI1.0 were run in capsules loaded with concentrated (1 or 4 m) and metal-free HCl solution at 710 °C and 150 or 200 MPa in order to impose a strong driving force (i.e., chemical potential gradient) for all elements that tend to partition/ dissolve into aqueous chloride solutions (e.g. Ca, Sr, Ba, Mn, Fe, Cu, Zn, Ag, Pb) to diffuse out the inclusions (Candela and Piccoli, 1995; Zajacz et al., 2008). As the mass of the HCl bearing fluid phase is several orders of magnitudes higher than that of all SMI together in the capsule, even elements that are only weakly soluble as chlorides (or have a low fluid/melt partition coefficient) should be depleted from the SMI if they can diffuse through the quartz. These experiments showed a significant decrease in wt% Na, Li, Ag and
Cu concentrations in the SMI compared to the initial values prior to exposure to the HCl solution, while concentrations of all the other analytes remained unchanged (Figs. 2–4; Tables 2 and 3). For comparison with HCl, interaction with pure H2O (experiment #04) resulted in a more moderate drop in Na, Li, Ag and Cu concentrations (Fig. 5; Table 3). To check if this process is reversible, a quartz chip (Rito5-2nd; experiment #05) containing SMI was reacted with a salt solution that contained NaCl, KCl, and FeCl2 in a proportion predicted to be in equilibrium with a normal granitic melt with an ASI1.00. Dissolved or solid sources of trace elements with variable ionic charges and cationic radii were added to this solution (Table 4). Lithium, Rb, Cs, Sr, Ba, Sb, La, Mo and Nb were added as dissolved chlorides or hydroxides, while elemental Cu, Ag and PbO were added as solid phases to the capsules. The starting solution had a pH of 3. A small amount of haplogranitic glass (200 MPa water saturated minimum composition; Johannes and Holtz, 1996) was added to the charge in order to monitor the final composition of the melt that equilibrated with the fluid phase. A pre-fractured quartz cylinder was added to trap fluid inclusions, allowing determination of the actual fluid composition at the experimental run conditions. After the experiment, SMI in the quartz chips, synthetic FI in the quartz cylinder, and the monitoring glass” were analyzed by LA-ICPMS. Concentrations of Li and Na significantly increased in the SMI during the experiment, approaching concentrations measured in the monitoring glass (Fig. 6). Even though they were included in the capsule load, concentrations of Ag and Cu remained very low in the monitoring glass, likely due to preferential alloying of these two metals with the Au capsule. This was con-
Diffusive reequilibration of quartz-hosted inclusions
3017
Fig. 2. Results of experiment #02 (720 °C, 200 MPa, 172 h, 4 m HCl solution) on SMI in magmatic quartz phenocrysts (sample BnSt). A significant drop in Na, Li and Cu concentrations is clearly observable. Major elements are in wt%, trace elements are in lg/g.
Fig. 3. Comparison of the compositions of SMI before and after experiment #03 (710 °C, 150 MPa, 24 h, 1 m HCl solution). Concentrations of Na, Ag, Cu and Li decreased one to two orders of magnitude during the experiment, while other element concentrations remained unaltered. Major elements are in wt%, trace elements are in lg/g.
firmed by the relatively low Cu and Ag concentrations measured in the synthetic FI (402 ± 88 and 3 ± 3 lg/g, respectively, Table 5). Thus, the experiment is inconclusive with respect to Cu as the final Cu concentration in the monitoring glass is identical within uncertainty to that in the SMI before and after the experiment. The concentration of Ag
displayed a moderate drop in the SMI, approaching the concentration detected in the monitoring glass. Concentrations of all other elements remained unchanged in the SMI, though in most cases they were present at much higher or lower concentrations in the monitoring glass at the end of the experiment than in the SMI. The change in the physical
3018
Z. Zajacz et al. / Geochimica et Cosmochimica Acta 73 (2009) 3013–3027
Fig. 4. Results of experiment #03 (710 °C, 150 MPa, 24 h, 1 m HCl solution) on SMI in magmatic quartz phenocrysts (sample BnSt). A significant drop in Na and Li concentrations is clearly observable. Cu and Ag concentrations were below the limits of detection after the experiment (Cu < 16.5 lg/g; Ag < 1.6 lg/g). Major elements are in wt%, trace elements are in lg/g.
appearance of a set of SMI in the sample Rito2nd, which was a subject to all kinds of experiments in this study is shown on Fig. 7. To confirm that the very rapid modification of Li, Na, Cu and Ag concentrations in SMI through the host quartz observed in the previous experiments may also occur in FI, further experiments (#06, 07 and 08) were conducted at lower temperatures. In experiment #06, quartz chips from sample Erg03 containing brine inclusions were put into con-
tact with 4 m aqueous HCl solution at 650 °C and 200 MPa for 72 h. Sodium concentrations dropped by more than 60% relative, while Li concentrations dropped an order of magnitude (Fig. 8; Table 6). All other element concentrations remained constant. Modification of the FI composition corresponded to a significant change in the NaCl dissolution temperatures (from >530 °C to 290–340 °C) and the range in temperature of vapor bubble disappearance (from 430–460 °C to 345–370 °C). Another experiment
Fig. 5. Comparison of the compositions of SMI before and after experiment #04 (710 °C, 200 MPa, 72 h, pure H2O). The drop in Na, Li, Ag and Cu concentrations is more moderate due to the use of H2O instead of HCl. Major elements are in wt%, trace elements are in lg/g.
Diffusive reequilibration of quartz-hosted inclusions
3019
Table 2 Average composition of silicate melt inclusions and host quartz in chip Rito2nd after various experiments. The composition of the monitoring glass from experiment #05 is shown in addition. Silicate melt inclusions
SiO2 TiO2 M2O3 FeO MnO MgO CaO Na2O K2O P Sc V Cu Zn As Rb Zr Nb Mo Ag Sn Sb Ba Cs La Ce Yb Hf Ta W Pb Bi Th U Li B ASI
Rito2nd original
Rito2nd after exp. #01
Rito2nd after exp. #03
Rito2nd after exp. #05
Monitoring glass” (exp.#05)
average (8)
1r
average (8)
1r
average (10)
1r
average (6)
1r
average (7)
1r
73.32 0.08 12.05 0.65 0.13 0.03 0.22 0.10 3.94 52.6 4.5 0.55 <0.33 51.0 18.1 321.9 253.0 134.5 7.3 <0.11 5.6 0.92 <0.38 9.3 33.9 40.3 5.72 16.7 8.15 6.1 31.5 0.81 71.5 35.3 379.2 21.9 2.360
0.15 0.01 0.00 0.03 0.01 0 00 0.05 0.03 0.23 21.2 0.7
72.15 0.04 12.05 0.61 0.15 0.02 0.35 2.25 4.04 46.5 5.9 0.21 25.86 44.2 6.0 365.3 186.1 126.1 8.7 158.68 5.9 1.26 0.11 12.5 51.4 50.8 6.50 14.7 7.38 7.8 37.7 1.50 55.1 44.2 369.0 51.3 1.331
0.45 0.01 0.00 0.03 0.00 0.00 0.05 0.63 0.31 19.6 0.8 0.00 13.45 2.8 0.8 10.1 47.3 8.4 3.0 71.96 2.4 0.25 0.00 1.1 11.6 4.8 3.99 2.4 0.92 1.4 1.6 0.12 5.9 13.1 34.3 8.8 0.133
73.33 0.04 12.05 0.68 0.16 0.03 0.24 0.03 4.06 68.4 6.8 0.17 1.99 63.9 4.0 418.8 211.0 147.1 16.3 1.02 5.0 3.12 1.16 14.6 50.1 53.7 5.59 16.6 8.59 10.8 39.5 2.04 62.9 40.0 2.4 46.6 2.486
0.25 0.01 0.00 0.07 0.02 0.01 0.08 0.01 0.13 30.4 2.6
74.75 0.05 12.02 0.64 0.16 0.02 0.29 2.16 4.01 95.0 6.6 <0.51 <1.60 65.4 4.5 410.5 191.0 153.8 9.2 0.31 8.0 2.93 <0.49 16.1 47.6 55.6 9.17 15.9 9.19 9.8 43.5 2.01 56.9 49.2 2800.4 46.2 1.16
0.32 0.01 0.00 0.03 0.01 0.00 0.04 0.38 0.15 26.5 1.2
74.77 0.01 12.65 0.27 0.00 0.04 0.14 3.35 2.75 53.8 0.4 0.55 0.48 1.1 0.4 9879.7 26.5 39.7 4.0 0.05 5.4 9.53 316.97 2912.2 4.8 8.1 0.22 0.8 0.05 <0.05 666.4 <0.04 0.3 0.2 4144.3 8.4 1.01
0.62 0.00 0.38 0.02 0.00 0.00 0.01 0.13 0.10 8.5 0.1 0.15 0.13 0.5 0.2 386.7 0.9 1.8 0.2 0.02 0.6 0.82 13.84 100.1 0.8 1.7 0.08 0.1 0.02
7.8 8.7 25 0 51.1 15.9 1.9 2.3 0.24 0.5 9.0 9.5 2.01 4.4 0.80 1.6 4.6 0.28 20.3 3.4 84.0 6.6 0.143
0.41 13.1 1.7 20.9 33.4 17.2 10.6 0.33 2.0 2.53 0.66 0.9 7.2 3.6 2.34 3.5 1.36 2.1 3.8 0.50 6.8 2.4 0.9 25.2 0.218
13.5 0.8 14.8 39.5 14.2 4.5 0.34 2.7 1.39 1.1 2.5 3.0 4.28 3.2 1.43 2.6 2.7 0.41 6.3 10.8 701.3 11.7 0.15
19.5 0.0 0.0 153.5 1.5 0.02
Host quartz
Ti Al Fe Na K Li P Sc Cu Ag Sn
average (3)
1r
average (4)
1r
average (6)
1r
average (5)
35.1 103.4 <5.67 <0.45 <1.99 15.2 26.1 0.49 <0.11 <0.05 2.99
23.0 21.4
33.5 68.6 <5.13 0.79 <0.83 6.6 30.1 0.71 <0.10 <0.03 4.01
2.9 8.5
33.0 81.4 <4.82 0.57 <2.24 <0.05 0.0 0.61 <0.24 <0.06 1.57
5.4 21.7
34.3 83.8 <6.92 0.85 1.35 21.6 16.5 0.54 <0.08 <0.02 1.19
4.2 5.0 0.05
0.60
0.01 3.5 9.2 0.27
1.40
0.13
0.0 0.10
0.24
1r 7.8 32.7 0.30 8.9 4.3 0.06 0.17
For silicate melt inclusions, major elements are in wt%, trace elements are in lg/g. For the host quartz, all elements are in lg/g and only clear, non-contaminated, inclusion free ablations were used. The lowest detection limits are shown.
3020
Z. Zajacz et al. / Geochimica et Cosmochimica Acta 73 (2009) 3013–3027
Table 3 Average composition of silicate melt inclusions and host quartz in chip BnSt after various experiment. Silicate melt inclusions BnSt original
SiO2 TiO2 AI2O3 FeO MnO MgO CaO Na2O K2O P Sc V Cu Zn As Rb Sr Zr Nb Mo Ag Sn Sb Ba Cs La Ce Yb Hf Ta W Pb Bi Th U Li B ASI
BnSt-experiment #02
BnSt-experiment #03
BnSt-experiment #04
average (10)
1r
average (7)
1r
average (3)
1r
average (7)
1r
71.36 0.04 13.60 0.98 0.05 0.05 0.72 2.81 5.82 52.9 3.11 0.76 10.00 33.9 7.2 229.4 43.4 75.5 19.6 2.31 0.17 9.5 0.85 188.7 9.74 26.7 49.2 2.48 2.54 1.98 3.81 26.7 0.33 28.2 8.5 269.2 60.3 1.11
0.27 0.01
74.18 0.05 13.60 0.93 0.05 0.05 0.89 0.06 5.46 57.7 6.39 0.62 1.08 29.2 14.0 239.9 54.6 94.3 20.0 1.89 0.19 4.7 0.78 235.8 9.31 31.4 56.7 3.30 3.74 2.78 1.79 29.4 0.44 32.9 10.4 0.8 61.1 1.78
0.74 0.01
70.23 0.07 13.60 0.98 0.05 0.05 0.50 0.02 4.93 <238.7 5.31 <9.11 <16.46 22.0 0.0 197.8 45.8 98.1 20.6 3.79 <1.58 <14.83 <3.41 277.0 7.04 35.6 62.6 <2.25 3.06 1.79 6.93 27.5 <1.51 29.9 6.9 6.8 46.9 2.17
0.38 0.03
72.93 0.04 13.60 0.90 0.05 0.06 0.77 0.92 5.92 48.0 4.26 0.73 0.33 135.3 7.8 290.7 42.8 78.8 18.8 3.73 0.06 5.7 0.57 327.3 8.90 35.5 67.9 2.82 3.72 1.99 3.01 30.8 0.27 26.8 7.3 6.6 54.9 1.46
1.22 0.01
0.08 0.01 0.01 0.13 0.09 0 23 23 9 0 27 0 28 7 90 79 1.5 15.9 13.5 13 3 1.5 1.14 0.02 4.7 0.55 145.7 1.82 5.1 9.1 1.99 0.80 0.32 0.75 2.1 0.14 3.1 1.8 120.6 10.2 0.03
0.09 0.00 0.01 0.10 0.08 0.39 41.1 6.94 0.24 0.00 8.1 8.5 24.5 16.1 10.8 2.9 0.96 0.03 1.2 0.12 144.2 2.05 6.0 13.2 0.76 0.77 0.89 0.50 5.5 0.11 4.3 1.5 0.0 25.2 0.32
0.09 0.01 0.01 0.13 0.00 0.30
27.1 0.0 14.5 5.0 8.7 2.5
99.9 0.51 4.2 5.3 0.00 0.67 5.6 4.1 0.5 7.3 0.15
0.08 0.02 0.03 0.57 0.34 0.71 36.0 1.08 0.89 165.0 2.5 115.7 15.4 13.9 3.5 2.99 0.00 1.1 0.31 252.1 3.23 10.8 26.4 1.53 1.91 0.60 1.08 9.1 0.01 5.7 2.2 5.0 7.0 0.23
Host quartz BnSt original
Ti Al Fe Na K Li P Sc Cu Ag
BnSt -experiment #02
BnSt -experiment #03
BnSt -experiment #04
average (12)
1r
average (7)
1r
average (4)
1r
average (5)
1r
33.0 96.3 7.0 0.71 <0.66 21.3 27.1 0.75 <0.09 <0.02
8.0 11.0 1.2 0.43
33.3 150.4 <10.03 0.77 <1.74 0.6 30.1 0.92 <0.15 <0.04
6.8 90.8 10.3 0.23 83.6
50.3 130.5 10.3 0.36 5.3 <0.05 19.3 0.67 <0.42 <0.03
6.8 35.1 3.0
26.9 95.0 3.0 1.14 <0.58 0.4 15.1 1.06 <0.43 <0.01
8.6 8.0
2.0 6.0 0.16
4.9 0.15
3.7 0.8 0.13
0.06 0.4 8.0 0.16
For the silicate melt inclusions, major elements are in wt%, trace elements are in lg/g. For the host quartz, all elements are in lg/g and only clear, non-contaminated, inclusion free ablations were used. The lowest detection limits are shown.
Diffusive reequilibration of quartz-hosted inclusions
3021
Fig. 6. Comparison of the compositions of SMI before and after experiment #05 (710 °C, 200 MPa, 168 h, doped salt solution). The composition of the haplogranite glass that has also been equilibrated with the doped fluid is shown as well. Li, Na and Ag shows obvious reequilibration as SMI compositions nearly adjust to that of the equilibrated glass. All other elements are unmodified. For some elements the experiment is non-conclusive as all three compositions are identical within uncertainty (clear fields). Major elements are in wt%, trace elements are in lg/g.
with the same constituents as #06 was carried out below the a/b transition temperature of quartz (experiment #07, 500 °C, 150 MPa, 168 h) to determine if diffusion of Na and Li is still rapid in this temperature range. Sodium content in the FI dropped about 40 relative%, and Li content dropped by over an order of magnitude (Fig. 9; Table 6). The NaCl dissolution temperatures dropped from >530 °C to 360–400 °C and the range in temperature of vapor bubble disappearance dropped from 430–460 °C to 350–400 °C. Concentrations of all analytes other than Na and Li remained constant during experiments #06 and #07. Table 4 Composition of doped solution #01 in experiment #05. In solution
(wt%)
As solids added to 150 ll solution
NaCl KCl FeCl2
10.15 3.71 3.29 (lg/g) 7901 9382 2733 201 208 206 1329 210 212 204
Cu Ag PbO
Li Rb Cs Sr Ba Sb La Sn Nb Mo
2 mg 1 mg 3 mg
The concentrations of Ag and Cu in FI did not show a distinct change in these two experiments. Therefore, an additional experiment (#08, 650 °C, 150 MPa, 72 h, intermediate density FI) was designed that imposed a stronger driving force for Cu and Ag diffusion by using a solution
Table 5 The composition of synthetic fluid inclusions trapped in quartz in experiment #05.
Li Na K Mn Fe Cu Zn Rb Sr Nb Mo Ag Sn Sb Cs Ba La Pb
Average (3)
1r
7258 39900 20272 24 12054 401 60 9648 204 1.1 111 2.8 58 8.3 3125 225 127 9706
424 1137 1936 31 2356 87 61 884 10 1.1 10 3.1 21 8.1 535 12 152 1124
Concentrations are shown in lg/g.
3022
Z. Zajacz et al. / Geochimica et Cosmochimica Acta 73 (2009) 3013–3027
Fig. 7. Change in the physical appearance of SMI in sample Rito2nd after various experiments. (a) The original appearance of the SMI, inclusions were strongly recrystallized due to slow cooling in nature; (b) The same inclusions after the reheating experiment (exp #01). The silicate melt quenches to a homogeneous glass. Note that the shown inclusions are relatively large, thus incompletely homogenized; (c) The same inclusions after depletion of Na in experiment (exp #03). Note that the inclusions rapidly recrystallized during the 3 min quench time likely due to the low Na2O content; (d) The same inclusions after experiment #05. Note that the elevated Na2O content causes the silicate melt to quench into a homogeneous glass. Two of the inclusions were ablated before the experiment.
Fig. 8. Results of experiment #06 (650 °C, 200 MPa, 72 h, 4 m HCl) on brine inclusions in miarolitic quartz crystals (sample Erg03). A significant drop in Na and Li concentrations is clearly observable, while other element concentrations remained unchanged. All element concentrations are shown in lg/g.
Diffusive reequilibration of quartz-hosted inclusions
3023
Fig. 9. Comparison of the compositions of FI before and after experiment #07 (500 °C, 150 MPa, 168 h, 4 m HCl solution). All element concentrations are shown in lg/g. A significant drop in Na and Li concentrations is clearly observable.
doped with Cu (5 wt% CuCl2 dissolved, +2.5 wt% solid Cu metal) and Ag (1 wt% solid AgCl). Additionally, the doped solution contained 6 wt% LiCl, 2 wt% NaCl and 1 wt% HCl. To reduce Cu and Ag loss to the capsule by alloying, a Pt capsule was used in this experiment. During this experiment the Na/Li molar ratio in the FI almost completely equalized (changed from 0.25 to 3.25) to that of the doped solution (Na/Li = 4.14). Copper concentration in the FI increased more than two orders of magnitude (from 7 lg/g to
2000 lg/g), while Ag concentrations increased more moderately from <2 lg/g to >7 lg/g (Fig. 10). Simultaneously to the modification of SMI and FI compositions, a change in the Li content of the host quartz was also observed (K, Cu and Ag contents were always below detection limits of 2, 0.1, 0.03 lg/g in the quartz, respectively). In the samples from miarolitic cavities, about 58 relative% of the Al content of the quartz (55–220 lg/g) was charge balanced by Li. In the magmatic phenocrysts,
Fig. 10. Comparison of the compositions of SMI before and after experiment #08 (650 °C, 150 MPa, 72 h, doped salt solution). All element concentrations are shown in lg/g.
3024
Z. Zajacz et al. / Geochimica et Cosmochimica Acta 73 (2009) 3013–3027
Table 6 Compositional data of analyzed fluids inclusions. Erg03 –original
Li B Na Al K Ca Mn Fe Cu Zn As Rb Sr Ag Sn Sb Cs Ba W Pb Bi
Erg03- experiment. #06
Erg03- experiment. #07
Cana11a–original
Cana11a–experiment #08
average(11)
1r
average(9)
1r
1606.0
269.4
6049.7
268.8
20609 3471 9065
615 568 1521
5968 2692 9600
901 728 1112
1639
124
1614
233
6.6 427.0
4.1 41.5
1971.6 480.8
440.3 157.9
390 0.7 1.5 14.7 15.2 118
40 0.5 1.2 6.7 2.8 20
463 0.9 7.8 15.2 13.0 136
41 0.6 4.0 5.3 3.5 15
266
17
292
26
average(8)
1r
average(3)
1r
average(5)
1r
88.9 255 153259 228 69682 2106 20226 106479 40.8 4734.1 1465 5889 369 4.1 14.0 8.1 1842 43.3 16.6 1277 62.5
44.2 91 4879 96 4285 1315 1324 6744 13.8 276.1 134 507 37 1.1 2.9 1.6 188 11.8 4.5 74 9.2
9.0 224 58733 343 72451 1117 21965 122034 69.5 4281.0 812 6671 377 4.1 14.5 6.6 2084 45.0 19.5 1184 54.5
1.5 115 6670 489 4963 323 1473 10167 7.2 848.3 358 455 50 0.3 10.8 1.3 177 0.8 4.0 170 6.0
3.3 300 92763 183 65534 767 19961 107393 31.0 4232.8 1355 5849 347 5.6 19.8 7.4 1869 44.4 13.8 1136 60.8
123 635 97 1427 843 386 1286 6.5 401.2 363 99 10 2.4 16.2 2.4 54 15.7 4.8 87 11.6
Host quartz Erg03 –original
Al Fe Na K Li Cu Ag
Erg03-experiment. #06
Erg03-experiment. #07
Cana11a–original
Cana11a– experiment #08
average(8)
1r
average(3)
1r
average(5)
1r
average(11)
1r
average(9)
1r
78 <7.4 9.3 5.6 12.7 <0.3 <0.02
19
79 <12.6 <2.11 <13.59 <0.28 <0.2 <0.06
6
73 <14.1 <0.5 <3.6 <0.29 <0.4 <0.06
5
83 <8.2 2.82 <1.4 14.2 <0.2 0.035
16
80 <12 0.9 <3.9 13 <0.6 <0.1
12
6.3 2.3 3
1.34 2.5 0.012
0.5 2.5
Concentrations are shown in lg/g. Only clear, non-contaminated, inclusion free ablations used to obtain the composition of the host quartz. The lowest detection limits are shown.
the same value is about 85 relative% (Fig. 11; Tables 2 and 3). The rest of Al must have originally been charge balanced by hydrogen. During interaction with HCl solution or pure water, Li leaves the quartz and 99 relative% of the Al content becomes charge balanced by H. During interaction with the doped salt solution in experiment #05, Li+ completely replaced H+ again. Sodium content of all investigated quartz slabs were originally below 1 lg/g, near or below limits of detection, thus changes could not be monitored. However, based on its maximum concentration, Na could play only a very subordinate role in charge balancing Al. 5. DISCUSSION Consistent modification of the composition of the SMI and FI and their host quartz in equilibrium with various fluids points to bulk diffusion of H, Li, Na, Cu and Ag or
their cations in the quartz. Considering the large number of studied elements with similar atomic and effective ionic radii (Shannon, 1976), concentration modification of only those elements that form cations in the univalent state suggests that these elements diffuse as charged species through the quartz structure, and that the tendency for rapid diffusion is valence-specific. If the effective ionic radii of known univalent cations are compared, it is apparent that the elements which rapidly diffused through quartz are the small˚ < Cu+ – 0.60 A ˚ < Na+– est (diffusing cations: Li+ – 0.59 A + ˚ ˚ 0.99 A < Ag – 1.00 A << non-diffusing cations: K+ – ˚ < Rb+ – 1.52 A ˚ < Cs+ – 1.67 A ˚ ). As Cl (the main 1.37 A anion present in the FI) is very unlikely to diffuse through ˚ ), quartz due to its very large effective ionic radius (1.81 A the diffusion of cations in and out of SMI and FI must have to be governed by exchange reactions involving cations that maintain charge balance. This is clearly observed in experiment #06, #07 and #08 whereby the concentration of Cl
Diffusive reequilibration of quartz-hosted inclusions
3025
Fig. 11. Compositional variability of studied quartz samples before and after the conducted experiments. The role of alkali element ions and H+ in charge balancing Al in the quartz is shown together with the aluminum saturation indices (ASI) of SMI. Data for Mt. Malosa, Cuasso al Monte and Ehrenfriedersdorf are taken from Zajacz et al. (2008).
did not change in the FI, but the cation ratios readjusted to approach the values in the external solutions. In experiments with pure aqueous HCl solutions (i.e., containing no trace metal or alkali cations), the most likely candidate cation that replaced Na+ and Li+ in the SMI and FI is H+, the smallest and most rapidly diffusing univalent cation. Pure H2O was less effective in causing the removal of Na, Li, Cu and Ag from the SMI than the aqueous HCl solutions, which can be explained by the elevated H+ activities and the presence of Cl that forms complexes with these cations in the HCl solution. The replacement of NaCl by HCl in the fluid inclusions will result in a decreasing density, which intuitively would suggest that the bubble disappearance temperatures should increase as opposed to our observations. However, neither the boiling curve nor the isochors are known in the HCl–H2O system at the studied P–T conditions. Therefore it cannot be excluded that the partial molar volume of the HCl enriched fluid is significantly larger at a specific P–T than it is for a fluid with the same Cl, but higher Na concentration, resulting in lower homogenization temperature. In the case of SMI, the loss of Na, Li, Cu and Ag must be accompanied by an increase in the water content since the metal-oxides are replaced by H2O via cation exchange through the quartz. This increase will, however, not be very significant when expressed in wt%, because of the large difference in the molar mass of the Na2O and H2O components. For example, a loss of 3 wt% Na2O will result in only a 0.87 wt% increase in the H2O content.
Rapid diffusion of H+, Li+ and Na+ along the open channels parallel to the c-axis in the quartz structure was previously observed by applying an electric field to quartz (Verhoogen, 1952; White, 1970; Kronenberg et al., 1986; Kronenberg and Kirby, 1987; Girardet et al., 1988; Plata et al., 1988; Bahadur, 1993). Frischat (1970) and Rybach and Laves (1967) determined high diffusion coefficients for Na between 400 and 1000 °C using only a concentration gradient as the driving force. Our results show that at near-magmatic temperatures, concentration gradients impose a sufficient driving force to induce rapid diffusion of H+, Li+ and Na+ through quartz, and no electric field is needed. Besides these elements, rapid diffusion of Cu+ and Ag+ was also observable which is consistent with the very similar effective ionic radii and identical valence of these cations to those of Li+ and Na+, respectively. The diameter of the channels parallel to the c-axis in the quartz crystal structure is the limiting factor determining which univalent cations can diffuse (Verhoogen, 1952; White, 1970; Plata et al., 1988). The observation that the same process occurred in magmatic quartz phenocrysts and a variety of hydrothermal-type miarolitic quartz samples suggests that diffusion does not require any kind of special defect structures in the quartz. Though our experiments were not aimed at precise determination of diffusion coefficients, we can provide a first approximation for Na and Li. Considering the degree of reequilibration in the Na content of the SMI in quartz during heating experiments we can conclude that the SMI were modified to a very similar extent in experiments with a run time between 24 and 168 h, between 710 and 740 °C. This suggests
3026
Z. Zajacz et al. / Geochimica et Cosmochimica Acta 73 (2009) 3013–3027
Fig. 12. Degree of reequilibration of the Na concentration in SMI in quartz as a function of time at various diffusivities (shown as numbers in m2/s next to the curves). The modeling indicates that D is likely larger than 10 10 m2/s at 740 °C to allow achievement of nearly complete reequilibration within 24 h.
that the reequilibration of the SMI is nearly complete already after 24 h in this temperature range. We used Fick’s first law to model the degree of reequilbation of the SMI as a function of time. As the diffusion of Na+ is orders of magnitude slower perpendicular to the c-axis of the quartz than parallel to it, we accounted for mass transfer only parallel to the c-axis. We used the following boundary conditions based on our typical experimental conditions: (1) We assumed a spheric SMI with a diameter of 20 lm sitting 50 lm below the external surface of the quartz slab, which is perpendicular to the crystallographic c-axis. The distance from the other side of the quartz slab was considered to be infinite. The difference in the Na content of the quartz in equilibrium with a Na-poor (0.1 wt% Na2O) and a Na-rich (2 wt% Na2O) SMI were considered to be 1 lg/g. We assumed 1 lg/g Na content in equilibrium with the external environment of the quartz chip as well. Model calculations indicated that the diffusion coefficient of Na is likely larger than 10 10 m2/s at temperatures of 710– 740 °C (Fig. 12). This result is consistent with the data of Frischat (1970; 2.08 10 9 m2/s at 720 °C) but slightly higher than the value determined Rybach and Leaves (1967; 1.53 10 11 m2/s at 720 °C). Verhoogen (1952) also determined lower Na diffusion coefficients using electric field as a driving force (1.88 10 12 m2/s at 720 °C). We calculated a minimum required diffusion coefficient for Li as well based on experiment #05, in which the largest change in Li concentration was observed. We used the same boundary conditions, except for the initial concentration gradient, which was 21 lg/g for Li. Our estimated minimum value for the diffusion coefficient of Li is about 10 11 m2/s at 710 °C which is consistent with those determined by Verhoogen (1952) calculated for the same temperature (1.9 10 11 m2/s at 720 °C). 6. CONCLUSIONS AND IMPLICATIONS The experiments showed that concentrations of Na, Li, Cu, Ag and H can be rapidly modified in quartz-hosted sil-
icate melt and fluid inclusions after entrapment, via diffusion through the quartz at P–T conditions typical of upper crustal felsic magmatism and related high temperature ore-forming hydrothermal activity. As these elements diffuse as univalent cations, their mobility is limited by the necessity to maintain charge balance. The concentration of Cl (sensu lato salinity) cannot be modified in fluid inclusions, but the relative proportions of H, Na, Li, Cu and Ag can easily reequilibrate with subsequent generations of fluids, silicate melts, sulfide liquids, or precipitated sulfide minerals that come into contact with the host quartz. Since Na is usually the most abundant of these elements, and H/ Na ratios as high as those imposed in this study are unlikely to occur in natural hydrothermal fluids, very significant modification of the NaCl contents and volumetric properties of fluid inclusions in nature is unlikely. However, Li/ Na, Cu/Na and Ag/Na ratios can readily readjust to equilibrate with external fluids or melts via cation exchange. This may induce very significant post-entrapment modifications of the Li, Na, Cu and Ag concentrations in both silicate melt and fluid inclusions in quartz, and should be kept in mind during interpretation of this kind of compositional data. ACKNOWLEDGEMENTS We thank Andreas Aude´tat for providing us with samples from Rito del Medio and Canada Pinabete. We are grateful to Peter Ulmer for providing access to the cold seal pressure vessel facility at ETH Zu¨rich, and to Larryn Diamond and Monika Painsi for allowing and helping us to use the cold seal pressure vessel facility at the University of Bern. We thank the members of the Laboratory for Mineral Deposits Research for allowing us to use their facility to conduct the last experiment of this study at the University of Maryland. We are grateful to Adam C. Simon and an anonymous reviewer for their constructive comments and to Edward M. Ripley for his editorial work. This research was supported by the Swiss National Science Foundation (SNF) under project numbers 20020-107955 and PBEZ2-118876.
REFERENCES Aude´tat A. and Gu¨nther D. (1999) Mobility and H2O loss from fluid inclusions in natural quartz crystals. Contrib. Mineral. Petrol. 137, 1–14. Aude´tat A., Gu¨nther D. and Heinrich C. A. (2000) Magmatichydrothermal evolution in a fractionating granite: a microchemical study of the Sn-W-F-mineralized Mole Granite (Australia). Geochim. Cosmochim. Acta 64, 3373–3393. Aude´tat A. and Pettke T. (2003) The magmatic-hydrothermal evolution of two barren granites: a melt and fluid inclusion study of the Rito del Medio and Canada Pinabete plutons in northern New Mexico (USA). Geochim. Cosmochim. Acta 67, 97–121. Bahadur H. (1993) Hydroxyl defects and electrodiffusion (sweeping) in natural quartz crystals. J. Appl. Phys. 73, 7790–7797. Candela P. A. and Piccoli P. M. (1995) Model ore-metal from melts into vapor and vapor/brine mixtures. In Magmas, fluids and ore deposits (eds. Fluid Magmas, Or Deposits and J. F. H. Thompson). University of British Columbia, pp. 101–127. Frischat G. H. (1970) Sodium diffusion in natural quartz crystals. J. Am. Ceram. Soc. 53, 357–359.
Diffusive reequilibration of quartz-hosted inclusions Girardet C., Plata J., Breton J. and Hardisson A. (1988) Electrodiffusion of Alkali-Metal Ions and Protons in Quartz – a Diffusion-Compensation Model. Phys. Rev. B 38, 5648–5653. Gu¨nther D., Audetat A., Frischknecht R. and Heinrich C. A. (1998) Quantitative analysis of major, minor and trace elements in fluid inclusions using laser ablation inductively coupled plasma mass spectrometry. J. Anal. Atom. Spectrom. 13, 263– 270. Gu¨nther D., Frischknecht R., Heinrich C. A. and Kahlert H. J. (1997) Capabilities of an Argon Fluoride 193 nm excimer laser for laser ablation inductively coupled plasma mass spectrometry microanalysis of geological materials. J. Anal. Atom. Spectrom. 12, 939–944. Halter W. E., Pettke T., Heinrich C. A. and Rothen-Rutishauser B. (2002) Major to trace element analysis of melt inclusions by laser ablation ICPMS: methods of quantification. Chem. Geol. 183, 63–86. Heinrich C. A., Pettke T., Halter W. E., Aigner-Torres M., Audetat A., Gunther D., Hattendorf B., Bleiner D., Guillong M. and Horn I. (2003) Quantitative multi-element analysis of minerals, fluid and melt inclusions by laser ablation inductively coupled plasma mass spectrometry. Geochim. Cosmochim. Acta 67, 3473–3497. Johannes W. and Holtz F. (1996) Petrogenesis and experimental petrology of granitic rocks. Springer-Verlag, Berlin. Kamenetsky V. S. and Danyushevsky L. V. (2005) Metals in quartz-hosted melt inclusions: natural facts and experimental artifacts. Am. Mineral. 90, 1674–1678. Kronenberg A. K. and Kirby S. H. (1987) Ionic-conductivity of quartz – Dc time-dependence and transition in charge-carriers. Am. Mineral. 72, 739–747. Kronenberg A. K., Kirby S. H., Aines R. D. and Rossman G. R. (1986) Solubility and diffusional uptake of hydrogen in quartz at high water pressures – implications for hydrolytic weakening. J. Geophys. Res. Sol. Earth and Planets 91, 2723– 2744. Longerich H. P., Jackson S. E. and Gu¨nther D. (1996) Laser ablation inductively coupled plasma mass spectrometric transient signal data acquisition and analyte concentration calculation. J. Anal. Atom. Spectrom. 11, 899–904. Mavrogenes J. A. and Bodnar R. J. (1994) Hydrogen movement into and out of fluid inclusions in quartz – experimental-
3027
evidence and geologic implications. Geochim. Cosmochim. Acta 58, 141–148. Plata J., Breton J. and Girardet C. (1988) Theoretical-model for the electrodiffusion of M+ (M = Li, Na, K) ions in a quartz crystal. Phys. Rev. B 38, 3482–3493. Qin Z. W., Lu F. Q. and Anderson A. T. (1992) Diffusive reequilibration of melt and fluid inclusions. Am. Mineral. 77, 565–576. Rybach L. and Laves F. (1967) Sodium diffusion experiments in quartz crystals. Geochim. Cosmochim. Acta 31, 539–546. Severs M. J., Azbej T., Thomas J. B., Mandeville C. W. and Bodnar R. J. (2007) Experimental determination of H2O loss from melt inclusions during laboratory heating: evidence from Raman spectroscopy. Chem. Geol. 237, 358–371. Shannon R. D. (1976) Revised effective ionic-radii and systematic studies of interatomic distances in halides and chalcogenides. Acta Crystallogr. A 32, 751–767. Student J. J. and Bodnar R. J. (2004) Silicate melt inclusions in porphyry copper deposits: identification and homogenization behavior. Can. Mineral. 42, 1583–1599. Verhoogen J. (1952) Ionic diffusion and electrical conductivity in quartz. Am. Mineral. 37, 637–655. Vityk M. O., Bodnar R. J. and Doukhan J. C. (2000) Synthetic fluid inclusions. XV. TEM investigation of plastic flow associated with reequilibration of fluid inclusions in natural quartz. Contrib. Mineral. Petrol. 139, 285–297. White S. (1970) Ionic diffusion in quartz. Nature 225, 375–376. Williams T. J., Candela P. A. and Piccoli P. M. (1997) Hydrogenalkali exchange between silicate melts and two-phase aqueous mixtures: an experimental investigation. Contrib. Mineral. Petrol. 128, 114–126. Zajacz Z. and Halter W. (2007) LA-ICPMS analyses of silicate melt inclusions in co-precipitated minerals: quantification, data analysis and mineral/melt partitioning. Geochim. Cosmochim. Acta 71, 1021–1040. Zajacz Z., Halter W. E., Pettke T. and Guillong M. (2008) Determination of fluid/melt partition coefficients by LAICPMS analysis of coexisting fluid and silicate melt inclusions: controls on element partitioning. Geochim. Cosmochim. Acta 72, 2169–2197. Associate editor: Edward M. Ripley