Access the full text.
Sign up today, get DeepDyve free for 14 days.
Laboratory for Waste Management (LES), Paul Scherrer Institut, 5232 Geological field observations evidence that active and fossil Icelandic hydrothermal Villigen, Switzerland systems are typically embedded into an intercalation of almost completely altered and nearly unaltered volcanic rock layers. We investigated the reasons for this finding with help of geochemical reaction path calculations, by studying the mineralogical evolution of contrasting lithofacies–basalt flows and hyaloclastites at various temperatures and pressures, different recharge water composition, and gas content. From this study, we conclude that the initial porosity of protoliths and volume changes due to their transformation into secondary minerals are sufficient to explain the different extents of alteration as observed in field studies. In addition, we present a generalized kinetic model to estimate the alteration time of glassy fragments in hyaloclastite as a function of grain size, surface roughness, and temperature. Thistimewas foundtoberathershort,ranging from a few hours to a few years. Keywords: Hydrothermal alteration; Iceland; Lava flow; Hyaloclastite; Basalt; Basaltic glass; Secondary mineral; Porosity closure; Geochemical modeling Background Motivation Icelandic hydrothermal systems have been extensively studied in the context of high- enthalpy fluid circulation that allows efficient and cheap electrical power generation (e.g., Fridleifsson and Elders 2005). The aqueous alteration of basaltic rocks consumes CO , therefore playing a role in the global carbon cycle (Dessert et al. 2003; Shibuya et al. 2013). This feature makes basaltic rocks as a good candidate for CO disposal (Matter et al. 2009; Gislason et al. 2010; Aradottir et al. 2012). During the aqueous fluid circulation, the primary magmatic minerals or glass in the volcanic rocks (called “protolith” below) are generally in disequilibrium with the fluid because temperatures, pressures, and H O fugacity are quite different from those dur- ing the volcanic rock formation. This leads to rock alteration that consists of the irre- versible transformation of the protolith into a more stable phase assemblage of the (hydrated) secondary minerals, with possibly significant volume and porosity change. Over a given period of time, an increase of the porosity enhances the water circulation (if the pore space is connected, for example, via fractures), and therefore the extent of rock alteration. On the contrary, a decrease of the porosity would reduce the degree of alteration. A possible complete closure of porosity has been mentioned for systems © 2015 Thien et al. This is an open access article distributed under the terms of the Creative Commons Attribution License (http:// creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. Thien et al. Geothermal Energy (2015) 3:11 Page 2 of 32 prone to carbonate formation (Steefel and Lichtner 1998; Shao et al. 2013) and in nat- ural and technical analogues on clay/cement interaction (Gaboreau et al. 2011; Pitty and Alexander 2014). In such a case, the fluid circulation may stop, preventing further rock alteration. Rock formations The subsurface stratigraphy of Iceland mainly consists in a succession of hyaloclastite formations and lava flows (Weisenberger and Selbekk 2009; Banik et al. 2014) some- times intersected by basalt intrusions (Lonker et al. 1993; Mortensen et al. 2014). The thickness of individual layers is variable and irregular (Watton et al. 2013), varying between a few meters and a few hundred meters (Alfredsson et al. 2013). Hyaloclastites consist of glassy fragments (at least 90 % (Watton et al. 2013)) result- ing from the fast cooling of lava due to the contact with water or ice (e.g., Watton et al. 2013). They are present in many volcanic systems all around the world (Zierenberg et al. 1995; Altaner et al. 2003; Honnorez 2003; Claridge and Campbell 2008; Moore et al. 2012; Barfod and Fitton 2014; Kanayama et al. 2014; Motelib et al. 2014) and also in Mars (Hovius et al. 2008; Scanlon et al. 2014). Hyaloclastites exhibit high porosity up to 60 % (Franzson et al. 2010; Alfredsson et al. 2013); relatively high initial perme- −13 −12 2 ability from 10 to 10 m (Jarosch et al. 2008); and a wide range of granulometry: fine hyaloclastite (average grain diameter <1/16 mm), hyaloclastite sandstone (1/16– 2 mm), granular hyaloclastite (2–4 mm), and hyaloclastite breccia (4–64 mm) (Watton et al. 2013). Lava flows and basaltic intrusions mainly consist of dense crystalline basalts (Kristmannsdottir 1979) with primary porosity ranging from 5 to 40 % (Franzson et al. −15 −13 2 2008). Measured permeability generally varies between 1°10 and 1.5°10 m (Arnorsson 1995). Greatest permeabilities are vertical, due to faults and fractures. Hyaloclastites usually show more signs of hydrothermal alteration than basalt flows or intrusives (Wolff-Boenisch et al. 2006). Kristmannsdottir (1979) classified Icelandic geothermal zones into two areas: (1) High-temperature geothermal areas, within the zones of rifting and volcanism. Rocks are young (Quaternary), and are assumed to have not been altered previously at higher temperatures. Nevertheless, some authors evidenced temporary higher temperature conditions when magma was introduced into the geothermal reservoir, as in Krafla (Gudmundsson and Arnorsson 2005) and Reykjanes (Marks et al. 2010). A progressive hydrothermal mineral alteration sequence with increasing temperature and depth is generally observed (e.g., Marks et al. 2010). Above 200 °C, the bulk of primary minerals or glass is generally completely altered (Franzson et al. 2008). The degree of alteration also depends on the structure of the rock. The more permeable and the less crystalline the protolith is, the more is the extent of its alteration. When compared at a similar temperature, hyaloclastites are usually completely altered, whereas intrusive and compact basalts appear rather fresh (Kristmannsdottir 1979;Lonkeretal. 1993; Gudmundsson and Arnorsson 2002;Gudmundsson andArnorsson 2005; Marks et al. 2010). For instance, in Hvalfjördur area, secondary minerals mainly occur in the top and bottom parts of basaltic flow units (Weisenberger and Selbekk 2009). (2) Low-temperature geothermal areas, in the older Quaternary and Tertiary rock formations, in which rocks may have experienced previous low-grade metamorphic Thien et al. Geothermal Energy (2015) 3:11 Page 3 of 32 events. In low-temperature areas, the extent of protolith alteration ranges from 0 to 60 % (Schopka et al. 2006;Alfredsson et al. 2013). The massive observation of zeolites in Tertiary areas is in contradiction with the mineralogy observed in high temperature areas. This had been explained by a progressive low-grade metamorphism due to the burial of lava succession and higher heat flow caused by the influence of central volcano (Neuhoff et al. 1999; Weisenberger and Selbekk 2009). In those areas, altered −17 −11 2 hyaloclastites exhibit a wide range of permeability, from 1°10 and 1.5°10 m , positively correlated to the porosity (Frolova et al. 2005). The lowest permeabilities correspond to oldest hyaloclastites (2–2.5 m.y.) which experienced low-grade metamorphism. The secondary mineral–fluid equilibrium is closely approached at temperatures above 150 °C (Gudmundsson and Arnorsson 2005) or even sometimes already at 100 °C (Scott et al. 2014). Nevertheless, Lonker et al. (1993) identified some departures from equilibrium above 200 °C, which he attributed to some kinetic effects. Second- ary minerals were found in vesicles and fractures, as well as in the rock matrix as re- placement of protolith (Lonker et al. 1993). The presence of secondary minerals that do not match with the actual reservoir temperature can be explained by prior heat- ing episodes (Lonker et al. 1993; Arnorsson 1995; Marks et al. 2010). These earlier events could have been caused not only by magma intrusions but also by hydrostatic pressure increase due to the progressive burial of lava succession or the ice load dur- ing Pleistocene. The occurrence of wollastonite in Krafla (Gudmundsson and Arnorsson 2005), garnet in Krafla and Namafjall (Gudmundsson and Arnorsson 2005), or some zeolites in Hvalfjördur (Weisenberger and Selbekk 2009) are assumed to be the result of contact metamorphism rather than of purely hydrothermal alteration. Secondary minerals precipitated in Icelandic basaltic hydrothermal systems had been described in many studies (Table 1). In this table, only minerals are listed that were positively identified. For instance, smectites, when only identified as “smectites”, are not reported in this table because smectite is not a single mineral, but a mineral family which includes montmorillonites, nontronites, beidellites, saponites, and hectorites. Thermodynamic properties of them can be very different from one category to an- other one (Vieillard 2000). There is a lack of accurate smectite identification. Actu- ally, very accurate XRD patterns are necessary to differentiate a trioctahedral smectite from a dioctahedral smectite (Thien et al. 2010); this task is non-trivial if the sample contains some amorphous material (e.g., glass, amorphous silica). Some authors actually represented observed mineralogical assemblages as a function of depth (pressure) and/or temperature (Arnorsson 1995; Gudmundsson and Arnorsson 2005; Lonker et al. 1993; Marks et al. 2010). The consistency between those different profiles is not perfect, which can be explained by different recharge water compos- ition, geological heterogeneities, or previous thermal events. Overall, synthesis of the available information suggests a classification of Icelandic geothermal systems into three zones: – First zone is up to 100 °C (from the surface to 300-m depth) which contains smectites, calcite, low-temperature zeolites, and numerous sulfides, oxides, and hydroxides. Quartz is generally not present; chalcedony, a polymorph of SiO , is typically observed; 2 Thien et al. Geothermal Energy (2015) 3:11 Page 4 of 32 Table 1 Secondary minerals identified in basaltic systems with hydrothermal alteration Carbonates i t a,e,g,i,m,n,t i,j i Ankerite , Aragonite , Calcite , Ca-Mg-Fe carbonates , Dolomite . Hydroxides and oxides m t h,o,q,u q j,o n Anatase , Fe-oxide , Goethite , Gibbsite , Hematite , Limonite . Zeolites t,x,y a,n,t,y t a,l,n,t,y g,l,t,y y a,y t Analcime , Chabazite , Clinoptololite , Heulandite , Laumontite , Levyne , Mesolite , Mordenite , a,n,t f f,n,t,y a,t,y g,l,n,t y Scolecite , Stellerite , Stilbite , Thomsonite , Wairakite , Yugawairalite . Smectites i d i l,r l,o Ca-Fe-Mg smectites , Fe-Mg smectites , Mg-rich smectites , Interstratified smectite-chlorite , Montmorillonite , l a,v,w Fe-rich saponites , Saponites . Other silicates e,t g,k, l,n,t,v e,g,m,n,t p i,o,t n Adularia , Actinolite , Albite , low-Albite , Amorphous silica , Antigorite (poorly crystalline) , b,g,n,t c,e,g,i,l,m,n,t,v,w e,g,k,l,m,n,t,v e,m,n,t g t e,k Chalcedony , Chlorite , Epidote , Garnet , Grossular , Hedenbergite , Hornblende , o,q g,m,p m e,g,k,l,m,n,t b,e,g,m,n,t n g,n,t Kaolinite , K-feldspar , Oligoclase , Prehnite , Quartz , Talc , Wollastonite . Others n d,f ,l,m,n,v n l n n l n r Anatase , Anhydrite , Antlerite , Apatite , Brochantite , Bonatitte , Chalcopyrite , Covellite , Fluorite , n l g d,g,h,l,m,n,s,t g,t m l,m l,m,s Gypsum , Halite , Marcasite , Pyrite , Pyrrhotite , Rutile , Sphalerite , Titanite . a b c d e Alfredsson et al. (2013); Arnorsson (1975); Arnorsson et al. (1983); Ehlmann et al. (2012); Freedman et al. (2009); f g h i j Fridriksson et al. (2001); Gudmundsson and Arnorsson (2005); Gunnlaugsson (1982); Gysi and Stefansson (2012a); Gysi k l m n and Stefansson (2012b); Hreggvisdottir (1978); Kristmannsdottir (1979); Lonker et al. (1993); Marks et al. (2010); o p q r Markusson and Stefansson (2011); Miyashiro (1975); Navarre-Sitchler et al. (2011); Schiffman and Fridleifsson (1991); s t u v Sigurdsson (1970); Steingrimsson et al. (1986a, 1986b); Steinthorsson and Sveinbjörnsdottir (1981); Sveinbjörnsdottir w x y (1992); Tomasson and Kristman (1972); Walker (1960); Weisenberger and Selbekk (2009) – second zone is from 100 to 200 °C (from 300 m to 500–1000 m) containing mixed-layer smectites, other zeolites (wairakite and laumontite), calcite, chalcedony, various sulfides like pyrite; – third zone is above 200 °C which contains chlorite, quartz, albite, wollastonite, prehnite, and epidote. At highest temperature/largest depths, amphiboles (actinolite and hornblende) and garnet were observed. It is worth to mention that the first zone can be subject to atmospheric oxidation (Marks et al. 2010), whereas the deep zones are typically anoxic. Tectonic spreading movements pull rock formations away on either side of magmatic activity areas. It results in fossil geothermal systems, commonly found in Quaternary and Tertiary formations in Iceland (Arnorsson 1975). This implies that active high- enthalpy geothermal systems are relatively young. This study is mainly focused on high-enthalpy geothermal areas. Fluid circulation Recharge water in central-Icelandic hydrothermal systems is mainly meteoric water because it can easily penetrate highly permeable volcanic rocks (Arnorsson 1995; Marks et al. 2010). In Reykjanes and Svartsengi areas, recharge water mainly consists of seawater (Kristmannsdottir 1979). During the Pleistocene, the recharge water in Reykjanes was meteoric (Marks et al. 2010). Arnorsson (1995) estimated with the available permeability data that the age of geothermal waters emerging in wells was no more than a few tens of years, and a few hundreds of years in areas having the lowest permeability. Thien et al. Geothermal Energy (2015) 3:11 Page 5 of 32 Fluid analyses from boreholes indicate the presence of gases (e.g., Stefansson and Arnorsson 2002; Kaasalainen and Stefansson 2011). The source of gas is not very well established but is commonly associated with young magma intrusions (Arnorsson 1995; Gudmundsson and Arnorsson 2002). When a depressurization boiling occurs, the vapor phase is produced, and volatiles like CO ,H S, and H preferentially enter 2 2 2 this phase. Vapor fraction, when exists, does not exceed 0.2 % by mass of the total fluid (Gudmundsson and Arnorsson 2002; Scott et al. 2014). Cross-checking of fluid analysis from different borehole analyses (Arnorsson 1995; Gudmundsson and Arnorsson 2002; Markusson and Stefansson 2011; Scott et al. 2014) allows establishing a range of possible aqueous CO and H S concentrations in Icelandic geothermal systems: 2 2 0.0003–0.35 M for CO ; 0.0001–0.05 M for H S. It is difficult to estimate those values 2 2 and to compare them with each other, because concentrations in both liquid and gas depends on temperature separation upon sampling (Gudmundsson and Arnorsson 2002). Because the most important part of the system consists in non-boiling zone (i.e., no steam), we recalculated the H S and CO concentrations considering liquid analysis, 2 2 gas analysis, and average proportion of gas in the fluid. In those studies, the measured pH of water is weakly alkaline (i.e., 8–9) and drops to sub-neutral (i.e., 5–6) when corrected according to the temperature increase with depth. Close to the surface, the gas phase due to depressurization boiling mixes with surface water which contains atmospheric oxygen (Markusson and Stefansson 2011), 2− oxidizing H StoSO . This results in a strong drop of pH (i.e., 2–3), which triggers 2 4 the acidic leaching of near-surface zones. Relation between porosity and permeability Porosity and permeability are two essential factors that drive the evolution of the hydrothermal system. The porosity and its connectivity define the permeability and the amount of mineral surface accessible to the reaction with water. In Icelandic reservoir rocks, the permeability generally increases with the porosity (e.g., Franzson et al. 2001). Such a correlation was measured for hyaloclastites (Frolova et al. 2005; Franzson et al. 2011), as well as for basalts (Sigurdsson and Stefansson 2002; Saar and Manga 1999). There are some exceptions. For instance, Franzson et al. (2001) re- ported a specific case of a young fresh olivine tholeiite lava flow, in which the permeability decreases with the porosity. In hyaloclastites, macropores can be filled out with secondary minerals when secondary microporosity is formed (Franzson et al. 2010). The correlation between porosity and per- meability is being influenced by the evolution of the pore space. It was reported that the connectivity of the pore space (which strongly influenced the permeability) is changing during the rock alteration (Navarre-Sitchler et al. 2011; Navarre-Sitchler et al. 2013). In this study, we calculated volume changes of mineral phases, which were trans- formed into a porosity value. The models give no information on connectivity or change in pore size/pore shape. It implies a correlation between porosity and perme- ability for a specific rock type. We can postulate that permeability changes result from changes in porosity. The need for modeling Most mineralogical field observations in basaltic hydrothermal systems, given in the literature and summarized in this paper, remain qualitative. There is a real lack of the Thien et al. Geothermal Energy (2015) 3:11 Page 6 of 32 quantitative mineralogical data. Consequently, the modeling approaches of rock alter- ation in those systems cannot be fully validated. To a wider extent, numerous modeling studies of the mineralogical evolution of basaltic systems have been done (Griffith and Shock 1997; Franzson et al. 2008; Navarre-Sitchler et al. 2011; Gysi and Stefansson 2011; Pham et al. 2011; Aradottir et al. 2012; Aradottir et al. 2013; Catalano 2013; Hellevang et al. 2013), but none of those reported the porosity evolution during reactive transport and water–rock inter- action. The evolution of porosity has potentially important implications for basalt alter- ation. Actually, in low-porosity rocks, changes in porosity noticeably modify effective diffusivity and permeability (Navarre‐Sitchler et al. 2009). Modeling the evolution of porosity is a difficult task that requires knowing the value of the initial connected por- osity and how it changes during the reaction. Some studies evidenced that the initial porosity decreases during the alteration of basaltic formations (Neuhoff et al. 1999; Gustavson 2006), but also the contrary (Navarre-Sitchler et al. 2013). In this study, the mineral volume evolution during basaltic rock hydrothermal alteration was simulated using a sequential reactors model. The sensitivity of different parameters was tested (initial protolith porosity, temperature and pressure, gas content, pore geometry) in order to assess what factors influence most the hydrothermal rock alter- ation in Icelandic hydrothermal systems. Methods Geochemical model setup Mineralogical assemblages were modeled using the GEM-Selektor v.3 geochemical modeling package (Kulik et al. 2013; Wagner et al. 2012). We considered the equilib- rium dissolution of a primary basaltic rock and a primary basaltic glass at various pressures and temperatures, and various initial water compositions (i.e., freshwater and seawater), with and without volcanic gas (H Sand CO ). 2 2 The mineralogical composition of basalt was calculated using a chemical composition given by Markusson and Stefansson (2011) (Table 2). The glass composition is given by Gysi and Stefansson (2011), see Table 3. In order to simplify the system, minor ele- ments K, Mn, Ti, P, and Zr were not considered. The mole amounts of basalt and glass were adjusted in order to reach the desired porosity per 1 dm of water. The compos- ition of the water is indicated in Table 4. For modeling the impact of volcanic gases, 0.3 M of CO and 0.2 M of H S were added to the system composition. Those gas 2 2 Table 2 Mineralogical composition of fresh basalt, calculated using the chemical composition given by Markusson and Stefansson (2011). Molar mass = 228.63 g/mol and density = 3.1 g/cm Mineral Formula M % mol Albite NaAlSi O 262.223 0.16 3 8 Anorthite CaAl Si O 277.41 0.24 2 2 8 Diopside CaMgSi O 216.55 0.25 2 6 Enstatite Mg Si O 200.7774 0.08 2 2 6 Forsterite Mg SiO 140.6931 0.09 2 4 Fayalite Fe SiO 203.773 0.16 2 4 Magnetite Fe O 231.54 0.02 3 4 Thien et al. Geothermal Energy (2015) 3:11 Page 7 of 32 Table 3 Chemical composition of fresh basaltic glass, per 1 mol Si, derived from Gysi and Stefansson (2011) Component Mol amount Al O 0.350 2 3 CaO 0.270 FeO 0.169 Fe O 0.012 2 3 MgO 0.260 Na O 0.080 SiO 1.000 Molar mass = 119.13 g/mol and density = 2.9 g/cm quantities allow to reproduce average H S and CO aqueous concentrations and pH 2 2 measured by Stefansson et al. (2011) in a similar system (250 °C, seawater-like fluid). A realistic model suitable for various hydrothermal systems should account for every mineral mentioned in Table 1. Nevertheless, in order to reduce the complexity of the chemical system, and in accordance with basalt and basaltic glass compositions in Tables 2 and 3, respectively, we did not consider accessory minerals containing the trace elements K, Mn, F, P, Cu, Ti, and Zr, assuming that their absence or presence is not expected to trigger relevant changes to the system. These elements can be incor- porated in major minerals (e.g., Curti 1999) or precipitate as accessory minerals (Kaasalainen and Stefansson 2012). In both cases, their amount is expected to be too small to noticeably affect the mineral volume of the system. In addition, the amount of trace elements varies as a function of the location of the field (Kaasalainen and Stefansson 2012). In this study, we wanted to present a general view of Icelandic hydrothermal system, rather than to consider a very specific case. Minerals which led to inconsistent phase assemblages were removed from the chemical system defin- ition; it is the case of goethite for instance, which is actually considered as a metasta- ble phase (Berner 2013). Selected minerals with their thermodynamic parameters are listed in Table 5. Table 4 Stoichiometry of sea salt used for setting up initial water compositions, taken from the tutorial of GEM-Selektor v.3 Element Mol amount C 0.00187109 Ca 0.010264347 Cl 0.54570368 H 0.002257133 K 0.010187321 Mg 0.053026052 Na 0.46738417 O 0.1193977 S 0.028173998 Si 0.000153078 To create the seawater-like solution, we mixed 1000 g of water with 35 g of this stoichiometry. To create the fresh water −5 −6 solution, we mixed 1000 g of water with 0.035 g of this stoichiometry, and 3°10 mol of CO . 1°10 mol of O was 2 2 added in order to insure stability within phase calculations Thien et al. Geothermal Energy (2015) 3:11 Page 8 of 32 Table 5 Selected minerals and their thermodynamic data Mineral Formula ΔG ° (J/mol) S° (J/mol/K) a (J/mol/K) b (J/mol/K) c (J/mol/K) V° (J/bar) Reference Albite Na(AlSi )O −3708313 207.15 342.59 0.0149 −2.10E + 07 10.025 S98 3 8 Low-albite Na(AlSi )O −3708313 207.15 258.15 0.0582 −6.28E + 06 10.007 S98 3 8 Analcime NaAlSi O (H O) −3078890 231.97 132.50 0.2555 −2.13E − 05 9.710 Neuhoff (2000) 2 6 2 Anhydrite CaSO −1321830 106.69 70.21 0.0987 0.00E + 00 4.594 S98 Anorthite Ca(Al Si )O −3991862 205.43 264.89 0.0619 −6.46E + 06 10.079 S98 2 2 8 Boehmite AlO(OH) −908974 48.45 60.40 0.0176 0.00E + 00 1.954 S98 Calcite CaCO −1128810 92.50 104.90 0.0050 −9.51E − 05 3.693 Holland and Powell (1998) Ca-chabazite Ca(Al Si )O (H O) −7156380 584.23 564.18 0.9033 −1.59E + 02 24.745 Neuhoff (2000) 2 4 12 2 6 Na-chabazite Na (Al Si )O (H O) −7115170 619.51 591.72 0.9192 −1.59E + 02 24.745 Neuhoff (2000) 2 2 4 12 2 6 Chalcedony SiO −854691 41.34 46.94 0.0343 −1.13E + 06 2.269 S98 Clinochlore-14A Mg Al(AlSi )O (OH) −8263350 410.50 1161.80 0.0101 −7.66E − 04 20.710 Holland and Powell (1998) 5 3 10 8 Daphnite-14A Fe Al(AlSi )O (OH) −6535560 545.00 1237.40 0.0014 −3.74E − 04 21.342 Holland and Powell (1998) 5 3 10 8 Diaspore AlO(OH) −913794 35.27 60.40 0.0176 0.00E + 00 1.776 S98 Diospside CaMg(SiO ) −3028296 143.09 221.21 0.0328 −6.59E + 06 6.609 S98 3 2 Dolomite CaMg(CO ) −2161510 156.00 305.89 −0.0049 0.00E + 00 6.439 Holland and Powell (1998) 3 2 Enstatite MgSiO −1459923 67.78 102.72 0.0198 −2.63E + 06 3.128 S98 Epidote Ca Fe|3|Al Si O (OH) −6070591 314.97 492.13 0.0536 −1.33E + 07 13.920 S98 2 2 3 12 Fayalite Fe SiO −1381695 148.32 152.76 0.0392 −2.80E + 06 4.639 S98 2 4 Ferrous oxide FeO −251446 60.75 50.72 0.0087 −3.14E + 05 1.200 S98 Forsterite Mg SiO −2056704 95.19 149.83 0.0274 −3.56E + 06 4.379 S98 2 4 Gibbsite Al(OH) −1155487 70.08 36.19 0.1908 0.00E + 00 3.196 S98 Grossular Ca Al Si O −6260549 254.68 435.21 0.0712 −1.14E + 07 12.530 S98 3 2 3 12 Gypsum CaSO (H O) −1797763 193.80 91.38 0.3180 0.00E + 00 7.469 S98 4 2 2 Halite NaCl −384120 72.13 45.94 0.0163 0.00E + 00 2.702 S98 Hedenbergite CaFe(SiO ) −2673568 170.29 229.33 0.0342 −6.28E + 06 6.827 S98 3 2 Thien et al. Geothermal Energy (2015) 3:11 Page 9 of 32 Table 5 Selected minerals and their thermodynamic data (Continued) Ca-heulandite Ca(Al Si )O (H O) −9,726,320 732.67 742.54 0.0632 −1.43E − 03 31.927 Neuhoff (2000) 2 7 18 2 6 Na-heulandite Na (Al Si )O (H O) −9371886 736.39 722.37 0.0631 −1.40E − 03 31.943 Neuhoff (2000) 2 2 7 18 2 6 Hydromagnesite Mg (OH) (CO ) (H O) −5864658 541.33 591.87 0.2731 −9.07E + 06 20.880 S98 5 2 3 4 2 4 K-feldspar K(AlSi )O −3746245 213.93 320.57 0.0180 −1.25E + 07 10.887 S98 3 8 Kaolinite Al Si O (OH) −3801720 203.70 436.70 −0.0034 −4.06E − 05 9.952 Holland and Powell (1998) 2 2 5 4 Laumontite Ca(Al Si )O (H O) −6800520 483.91 582.22 0.1448 −8.99E − 04 20.755 Neuhoff (2000) 2 4 12 2 4.5 Magnesite MgCO −1027740 65.10 186.40 0.0000 0.00E + 00 2.802 Holland and Powell (1998) Magnetite FeFe|3| O −1014930 145.73 91.55 0.2017 0.00E + 00 4.452 S98 2 4 Marcassite FeS|0|S|-2| −153335 36.82 74.81 0.0055 −1.28E + 06 2.394 Gronvold (1976) Pargasite Na(Ca Mg Al) (Al Si )O (OH) −11910710 669.44 861.07 0.1743 −2.10E + 07 27.350 S98 2 4 2 6 22 2 Fe-pargasite Na(Ca Fe Al) (Al Si )O (OH) −10569566 776.13 893.58 0.1799 −1.98E + 07 27.990 S98 2 4 2 6 22 2 Prehnite Ca Al Si O (OH) −5816166 271.96 383.25 0.1582 −8.20E + 06 14.033 S98 2 2 3 10 2 Pyrite FeS|0|S|-2| −160218 52.93 74.81 0.0055 −1.28E + 06 2.394 S98 Pyrrhotite FeS|-2| −100767 60.29 72.80 0.0000 0.00E + 00 1.820 S98 Quartz SiO −856239 41.34 46.94 0.0343 −1.13E + 06 2.269 S98 Ca-Fe_saponite Ca Fe|2| Al Si Al O (OH) −4777570 429.68 355.41 0.2305 −7.00E + 06 17.925 This study 0.19 2.62 0.38 3.24 0.76 10 2 Ca-Mg_saponite Ca Mg Al Si Al O10(OH) −5676610 335.99 352.40 0.2199 −8.00E + 06 17.925 This study 0.19 2.62 0.38 3.24 0.76 2 Na-Fe_saponite Na Fe|2| Al Si Al O (OH) −4773700 435.46 358.06 0.2346 −7.00E + 06 17.925 This study 0.38 2.62 0.38 3.24 0.76 10 2 Na-Mg_saponite Na Mg Al Si Al O10(OH) −5674020 346.04 355.05 0.2240 −8.00E + 06 17.925 This study 0.38 2.62 0.38 3.24 0.76 2 SiO2-amorphous SiO −848900 60.00 24.81 0.1975 −9.53E + 06 2.900 S98 Siderite FeCO −679540 105.02 48.66 0.1121 0.00E + 00 2.938 S98 Sulfur S|0| 0 31.80 14.98 0.0261 0.00E + 00 1.561 S98 Wairakite Ca(Al Si )O (H O) −6181576 439.74 420.07 0.1861 −6.87E + 06 18.687 S98 2 4 12 2 2 Wollastonite CaSiO −1544837 82.01 111.46 0.0151 −2.73E + 06 3.993 S98 Thermodynamic parameters of saponites were calculated using the method of Vieillard (2007) ΔG ° is the standard Gibbs energy of formation, S° is the standard absolute entropy, V° is the molar volume, all taken per mole at 1 bar 25 C; a, b, and c are the empirical coefficients of the heat capacity function Cp (T) = a + bT + c/T. S98 refers to the SUPCRT98 database (Shock et al. 1997) Thien et al. Geothermal Energy (2015) 3:11 Page 10 of 32 The chemical thermodynamic system was set up for Al, C, Ca, Cl, Fe, H, K, Mg, Na, O, S, Si, charge, assuming the overall charge neutrality. The properties of aqueous spe- cies and gases with their standard thermodynamic properties (and HKF EoS parameters for aqueous species) were taken from the SUPCRT98 database (Shock et al. 1997). Aqueous activity coefficients were calculated by GEM-Selektor v.3 by using the ex- tended Debye-Hückel equation consistent with SUPCRT98 data and HKF EoS, with the common ion size parameter equal to 0.372 nm and the third parameter b equal to 0.064 (as for NaCl electrolyte). The stability and mixing in the non-ideal gaseous fluid were calculated using the Peng-Robinson-Stryjek-Vera (PRSV) multicomponent fluid model (Stryjek and Vera 1986; Zezin et al. 2011). The presence of smectites had been widely evidenced in Icelandic hydrothermal sys- tems, but there is a lack of their accurate identification. Chemical analysis indicates a high content of Mg and Fe (Sveinbjörnsdottir 1992; Ehlmann et al. 2012; Gysi and Stefansson 2012a; Alfredsson et al. 2013), which is compatible with a saponite struc- ture. Saponites are very common corrosion products of basaltic glass alteration (Thien et al. 2010 and references therein) under anoxic or oxic weathering conditions (Catalano 2013). Montmorillonites have been sometimes identified, mainly in the near-surface acid leaching zone (Kristmannsdottir 1979; Markusson and Stefansson 2011). We therefore consider that secondary smectites are, in fact, saponites. Noting that basalt has the same Si/Al ratio as basaltic glass, we assumed that the Si/Al ratio of smectites is the same as the Si/Al ratio of primary basaltic glass (Thien 2014). The repartition of Al between tetrahedral sheet and octahedral sheets was assumed to be the same as in smectites of similar stoichiometry (Thien et al. 2010). The rest of the octahedral sheet can be filled with Fe and Mg. We used ideal solid solutions with both Fe and Mg end-members, allowing Mg/Fe content in the octahedral sheet to vary. In Icelandic hydrothermal systems, the smectite interlayer space is mainly occupied with Ca, and partially with Na in the presence of salt water (Sveinbjörnsdottir 1992). So, we considered two solid solutions, one containing Ca as interlayer cation, and the other one containing Na as interlayer cation. The knowledge of molar volumes is fundamental to calculate the porosity evolution in the system, especially considering that smectites are important secondary minerals involved. For thermodynamic calculations, the molar volume was obtained considering the basal spacing (001) of a similar phase (Thien et al. 2010). Saponite showing such a basal spacing value contains in fact one layer of water (Rinnert et al. 2005), although two or even layers of water can exist in the interlayer (Ferrage et al. 2010), triggering a swelling, and therefore increasing the molar volume. The degree of hydration increases with temperature and, to a less extent, with pressure (Schleicher et al. 2013). Vidal and Dubacq (2009) claim the contrary, but they considered in their modeling a decrease of water activity with temperature increase, whereas Schleicher et al. (2013) performed their experiments in humidity chamber. Also, an increase in the saponite layer charge enhances the interlayer hydration (Michot et al. 2005). Karmous et al. (2009) conducted hydration experiments at standard pressure and temperature by forcing the relative humidity (90 %) on saponites similar to those con- sidered (i.e., the same layer charge). His results indicate the incorporation of two layers of water, leading to a potential maximal expansion of the basal spacing at 16.32 Å when Ca is in the interlayer sheet, and 14.99 Å when Na is in the interlayer sheet. For Thien et al. Geothermal Energy (2015) 3:11 Page 11 of 32 saponites used in this study, this hydration corresponds to molar volumes of 238 cm / mol and 218 cm /mol, respectively. There is no clear evidence in the literature whether such samples can incorporate a third layer of water at temperatures and pressures con- sidered in this study. Considering that the interlayer charge of our smectites is relatively low compared to montmorillonites, and the fact that smectites found in the San Andreas Fault (2.7-km depth) contain two layers of water in spite of the pressure of 800 bar (Schleicher et al. 2013), we can reasonably assume (at pressures and temperatures consid- ered in this study) a maximum of two layers of water in saponites. For porosity calcula- tions, we therefore considered molar volumes established considering two layers of water. An amorphous gel phase inherent to glass alteration (e.g., Gin et al. 2001) was not taken into the model, since its composition and thermodynamic parameters are not known. The gel phase can be accounted for by amorphous silica, diaspore, boehmite, gibbsite, and probably secondary smectites because the composition of the gel is close to the composition of crystallized secondary minerals (Thien et al. 2012). Ideal solid solutions were also considered for other mineral phases: chabazite which includes theCaend-memberand theNaend-member; heulandite whichincludesthe Ca end-member and the Na end-member; chlorite which includes the daphnite end- member and the clinochlore end-member; and amphibole which includes the parga- site end-member and the Fe-pargasite end-member. Details about solid solutions and their implementation in GEM-Selektor v.3 are given in Kulik et al. (2010) and Wagner et al. (2012). Thermodynamics only indicates if considered minerals can precipitate or not but does not provide any information about the precipitation rates. A slowly precipitating phase may never be observed if there is a rapidly precipitating phase of similar compos- ition, even though thermodynamics predicts that one phase is more stable and the faster precipitating one is more soluble. Even if sufficient time is given to the reaction to proceed, the fast precipitation of a more soluble mineral may decrease the supersat- uration of the solution relative to the less soluble one, thus impeding the nucleation and growth of the less soluble mineral with a sluggish kinetics. For instance, chlorites systematically appear in the modeling results at any temperature, whereas in reality they are only observed at temperatures above 200 °C (e.g., Kristmannsdottir 1979; Gysi and Stefansson 2012a, 2012b). Smectites are normally present at low temperature instead of chlorites. Hence, we had to block the precipita- tion of chlorites at temperatures below 200 °C, allowing the precipitation of smectites and therefore making the model more realistic. The minerals suppressed in the model at different temperatures are indicated in Table 6. This is a very primitive account for mineral–water reaction kinetics to circumvent the absence of detailed kinetic rate equations in the model; one of the further improvements of the model would be the implementation of realistic dissolution and precipitation kinetic rates. We assume that not accounting for accurate kinetic rates does not qualitatively change the results re- garding the final mineral volumes. Sequential reactor model setup In order to simulate rock alteration upon fluid circulation, a batch reactor calculation was set up with the process script in GEM-Selektor v.3 code. This simplified reactive transport model consists of a single flow-through box containing the fluid and solid Thien et al. Geothermal Energy (2015) 3:11 Page 12 of 32 Table 6 Secondary minerals disabled to account for their very slow precipitation kinetics 50 °C; 5 MPa 100 °C; 10 MPa 150 °C; 15 MPa 200 °C; 20 MPa 250 °C; 25 MPa 300 °C; 30 MPa Quartz Quartz Quartz K-feldspath K-feldspath K-feldspath Albite_low Albite_low Albite_low Clinochlore Clinochlore Clinochlore Clinochlore Daphnite Daphnite Daphnite Daphnite Prehnite Prehnite Prehnite Prehnite Epidote Epidote Epidote Epidote Wollastonite Wollastonite Wollastonite Wollastonite Fe-pargasite Fe-pargasite Fe-pargasite Fe-pargasite Pargasite Pargasite Pargasite Pargasite Grossular Grossular Grossular Grossular Hedenbergite Hedenbergite Hedenbergite Hedenbergite Pyrite Pyrite Pyrite Pyrrhotite Laumontite Wairakite Magnesite Wairakite Hydromagnesite Dolomite Siderite Laumontite Wairakite The cut-offs for mineral precipitation were selected following Gudmundsson and Arnorsson (2005) and Marks et al. (2010) phases. The initial composition in terms of solid to fluid ratio is chosen such that fluid and solid phase volumes represent a porous rock with a desired porosity. The system is equilibrated, and after each equilibration step, the bulk of the equilibrated fluid phase (1 dm ) is replaced with a fluid with the initial composition, while the reacted solid phases are retained from the previous step. The initial fluid composition is either fresh- water or seawater (with or without volcanic gas), pre-equilibrated with the same mass of protolith as the initial one. For a given volumetric fluid flow rate per unit area, each of these so-called pore water exchange cycles can be associated with a certain equilibration time, which in addition allows to calculate kinetic constraints for mineral dissolution/precipitation (though this functionality was not used in the present study). The system is not constrained to a constant volume, because the volume of fluid phase is kept constant, while the volume of solid phase aggregate was allowed to vary. As the fluid volume is kept constant, the effective porosity would change with each flow and equilibration cycle. Therefore, the time associated with each cycle for a constant Darcy flux is also changing. Actually, for the same liquid flux, a porosity change implies a change of fluid velocity. Different fluid velocities imply different contact times. We made the assumption that the liquid flux is always the same. In reality, porosity changes might also imply changes of liquid flux, which can only be resolved in a fully coupled reactive transport model in which poros- ity changes imply permeability changes. In a primitive flow-through reactor model, it is even possible to accumulate precipitates in an unphysical way such that the overall solid volume exceeds the initial total volume. In a realistic volume-constrained system, Thien et al. Geothermal Energy (2015) 3:11 Page 13 of 32 this would cause the porosity clogging that would diminish the volume occupied by the fluid phase, thus stopping the system evolution. However, to make the comparison between the different cases easier, we have made the choice of not stopping the calcula- tions when the total volume of the system becomes unrealistic. In case of unrealistic volume since the first calculation step, it would be difficult to represent the contribu- tions of various secondary minerals. In addition, it would not be satisfactorily for the cases where the total mineral volume is just below or just above the maximum volume limit. This part is rather a sensitivity study than a predictive model of the reality. This approach implies that there is a correlation between porosity and permeability and that porosity changes lead to permeability changes. This simplification is necessary, but reasonable, as discussed in the introduction. The goal of this study was to investi- gate the alteration contrast between those two rock formations, rather than to build a model which predicts the detailed evolution of any possible intermediate case. For the initial systems, we considered three rock types: “low-porosity” basalt (10 % por- osity), “high-porosity” basalt (40 % porosity), and hyaloclastite (basaltic glass with 60 % porosity). We always used 1 dm of fluid, which was equilibrated with mineral volumes of 3 3 3 9dm , 1.5 dm , and 0.67 dm , respectively. Values of temperature and hydrostatic pres- sure vary as a function of the depth of the magma intrusion and the density of the rocks, respectively. In the present context, we approximated linear gradients of 10 MPa/1000 m and 100 °C/1000 m (Marks et al. 2010, Fig. 3). For each rock type, we considered tempera- tures and pressures of 50 °C and 5 MPa; 150 °C and 15 MPa; 250 °C and 25 MPa, to mimic a linear increase of temperature and pressure with depth. In addition to this single-box reactor setup, a multi-box-flux sequential reactor setup was used (Fig. 1). The principle is the same, but the water equilibrated in a given box is transferred to the next box at each time step by a zero-order (constant) flux. The in- put water source is constant, not pre-equilibrated with the protolith. The multi-box approach reveals possible alteration fronts. Actually, because water composition is buffered by minerals, the mineralogy (and therefore mineral volumes) may change along the distance traveled by water. Kinetic model for basaltic glass dissolution In fresh water, the dissolution rate of silicate glass is at its maximum. Upon this process, the solution is progressively loaded with dissolved silica. Simultaneously, an amorphous gel layer precipitates and the dissolution rate drops to a low value, the re- sidual rate. The residual rate is several orders of magnitude lower than the initial rate (Techer et al. 2001). The gel acts as a diffusion barrier by limiting the transport of water from the solution to the fresh glass (e.g., Rebiscoul et al. 2004). It makes the glass Fig. 1 The reactive transport model setup. After each time step, the water that has reacted in a given box is moved to the next box or lost in the output. The water is replaced by that from the previous box or the input water (constant initial composition) Thien et al. Geothermal Energy (2015) 3:11 Page 14 of 32 very durable; for instance, basaltic glasses may survive for millions of years under low- temperature subsurface conditions (Crovisier et al. 2003). The gel alteration layer of basaltic glass is commonly called “palagonite”. The secondary phase precipitation con- stitutes a sink of silica, and can therefore decrease the gel thickness, and therefore increase the dissolution rate (e.g., Thien et al. 2012). At equilibrium, minerals are constantly subject to a gross forward precipitation rate and a gross backward dissolution rate (e.g., Thien et al. 2014). Considering glass, there is no gross forward precipitation rate (i.e., glass cannot re-precipitate) because the glass can only be formed by fast cooling of a melt, which is not possible in the context of hydrothermal alter- ation. This implies that the equilibrium between glass and aqueous solution will never be reached, even in the case of strong silica saturation in the solution. Glass can only dissolve, even though very slowly. This specific feature makes determination of the glass “solubility” senseless; conversely, the glass alteration kinetics cannot rigorously be modeled using a clas- sical kinetic equation involving solubility constants or saturation indices. Therefore, many authors replaced the glass solubility by the gel solubility (Techer et al. 2001; Gysi and Stefansson 2011). But the stoichiometry and the solubility of the gel are difficult to estimate, and the problem can be simplified by considering amorphous silica or aluminum hydroxide (e.g., Bourcier et al. 1990). Those models are able to describe experimental data in far-from- equilibrium cases and can also describe the drop of the alteration rate, but cannot be used at close-to-equilibrium conditions because a residual alteration rate is not accounted for. A way to circumvent this problem is to add a constant term related to the residual rate (Strachan and Neeway 2014). A complete model of glass dissolution had been developed, the GRAAL model (Frugier et al. 2008). This model considers an instantan- eous hydration (i.e., alteration) of the glass. As soon as a part of the glass is altered, the increase of silica aqueous concentration triggers the formation of the gel. The gel limits the transport of water from the solution to the glass (diffusion barrier); the alteration rate consequently decreases. The precipitation of secondary phases (which influence the thickness of the gel via silicium mass balance) is accounted for. But GRAAL is a reactive transport model working at the scale of the glass grain. It is currently not pos- sible to embed this reactive transport model into another reactive transport model working at the scale of the hydrothermal system. The simplest approach would be to consider a constant dissolution rate. Actually, a compilation of measured basaltic glass dissolution rates (Grambow et al. 1986) indi- cates at Earth surface conditions an average value between 3 and 20 μm/1000 years in fresh water, and 0.1 μm/1000 years in silica-saturated water. The difference between those two conditions is about two orders of magnitude. The water located inside the hydrothermal system has already interacted with rocks. Hence, realistic glass dissol- ution rate in hydrothermal systems may be closer to that in the silica-saturated water rate than this of the fresh water. We therefore consider the average dissolution rate of abasalticglass of 0.1 μm/1000 years. The relation between a linear dissolution rate V (m/s)and agrowthratenormalized to thesurface R (mol/m /s) is: −6 V ¼ 10 R ð1Þ where M is the molar mass of the glass (g/mol) and d is the glass density (g/cm ). By −14 2 applying this equation, the rate of 0.1 μm/1000 years corresponds to 7.7°10 mol/m /s. By considering an average surface temperature of 5 °C corresponding to those data, Thien et al. Geothermal Energy (2015) 3:11 Page 15 of 32 alteration rates can be recalculated at each temperature using the activation energy. The ac- tivation energy of basaltic glass dissolution is of 72 kJ/mol (Crovisier et al. 2003), nearly the same as that for nuclear waste glass dissolution (e.g., Adocat et al. 2001). This value is able to accommodate the temperature variations of basaltic glass dissolution rates measured by Gislason and Oelkers (2003), at a given pH, for temperature ranges between 0 and 300 °C. By considering the datasets obtained at pH between 4 and 10, the average uncertainty is of one order of magnitude. The considered equation giving the amount of dissolved glass is: −E 1 1 ðÞ R T 278 m ¼ S e R Δt M ð2Þ aðÞ Δt ðÞ t p where m is the mass of glass altered (g) during the considered time interval Δt (s), a(Δt) S is the surface of the glass (m ), E is the activation energy of the glass (J/mol), R is (t) the gas constant (8.314 J/mol/K), and T is the temperature (K). The surface of the glass is recalculated at each time step: S ¼ S m ð3Þ ðÞ t SðÞ t ðÞ t where S is the specific surface area (m /g) and m the mass of glass (g), at the time t. s(t) (t) The specific surface area is assumed to vary during the dissolution process: ðÞ 0 S ¼ S ð4Þ SðÞ t SðÞ 0 ðÞ t where S is the initial specific surface area (m /g), and m the initial mass of glass S(0) (0) (g). This spherical model is the simplest way to account for the specific surface area increase during dissolution. Nevertheless, possible morphological changes make the reality more complex. When not measured by BET, the specific surface area is estimated as follows: S ¼ SRF ð5Þ SðÞ 0 d r where SRF is the surface roughness factor (unitless), and r the average grain radius (cm). The surface roughness factor accounts for the fact that grains are nor perfectly spherical. It was defined by Wolff-Boenisch et al. (2004) as the ratio between specific surface area measured by BET and geometric specific surface area. In case of perfectly smooth and calibrated grains, the surface roughness factor is equal to 1; otherwise it is greater than 1. A collection of Icelandic grain glasses indicates a surface roughness fac- tor between 3 and 206 (Wolff-Boenisch et al. 2004). The calculations were performed with an Excel® spreadsheet. The time step length was manually optimized, in order to limit the amount of calculations. It varies from 0.1 to 100,000 days, depending on the total time. Results and discussion Mineralogical evolution and volume changes The volume evolution of different minerals is presented in Figs. 2, 3, 4, 5 and 6. The re- placement of water performed at each time step corresponds to one pore water ex- 3 3 3 change cycle. The initial volumes of protoliths were 9 dm , 1.5 dm , and 0.67 dm , respectively, for low-porosity basalt, high-porosity basalt, and hyaloclastite. The initial volume of water was always 1 dm . Thien et al. Geothermal Energy (2015) 3:11 Page 16 of 32 Fig. 2 Single-box batch reactor modeling results with freshwater at 50 °C and 5 MPa Model of leaching in freshwater At 50 °C and 5 MPa, the primary basalt is already severely altered after the first pore water exchange cycle; in each case, fayalite, forsterite, and enstatite are fully dissolved (Fig. 2). For hyaloclastite, the basaltic glass does evidently no longer exist at the first pore water exchange cycle, because thermodynamic equilibrium is considered. In the presence of vol- canic gas, the whole primary basalt is nearly completely altered at the beginning of the simulation (at least 70 % in volume, after the first pore water exchange cycle). The more the initial porosity the more the primary rock is altered. Without volcanic gas, the pre- dominant minerals are saponite and chabazite (low-temperature zeolite). With volcanic gas, saponites and zeolites are less abundant; they are replaced by chalcedony, kaolinite, and carbonates. In such a case, marcasite and sulfur are always present. The mineral vol- ume of the low-porosity basalt systems is always greater than the initial value of 10 dm : in those systems, the whole initial porous volume is filled by secondary minerals, whereas it is not the case considering high-porosity basalt systems (altered mineral volume less than the initial value of 2.5 dm ). For hyaloclastites, the final mineral volume is largely less than the limit of 1.67 dm (i.e., the initial volume or the total volume available). At 150 °C and 15 MPa (Fig. 3), the results are similar to those at 50 °C and 5 MPa. Nevertheless, there is a difference for the hyaloclastite system: in the presence of volcanic gas, the mineral volume decreases with time (from the first pore water exchange cycle). Fig. 3 Single-box batch reactor modeling results with freshwater at 150 °C and 15 MPa Thien et al. Geothermal Energy (2015) 3:11 Page 17 of 32 Fig. 4 Single-box batch reactor modeling results with freshwater at 250 °C and 25 MPa At 250 °C and 25 MPa, chlorites are the most abundant secondary minerals (Fig. 4). They are less abundant in the presence of volcanic gas. The presence of volcanic gas fa- vors the precipitation of kaolinite, carbonates, and quartz and disfavors the precipita- tion of prehnite and wollastonite. Epidote never appears in the presence of volcanic gas; whereas pyrite is always present. The primary basaltic minerals diopside and albite are less stable in the presence of volcanic gases. In low-porosity basalt, the volume of secondary minerals exceeds the initial volume of the solid. Considering high-porosity sys- tems, the pore space is not completely filled by secondary minerals; the mineral volume de- creases with the time in hyaloclastite systems (from the first pore water exchange cycle). Model of leaching in seawater At 50 °C and 5 MPa, the evolution is similar as in the freshwater case (Fig. 5), but is character- ized by a more abundant precipitation of saponites and a less important precipitation of zeo- lites. Basalt minerals are less stable in seawater than in freshwater. With seawater and volcanic gas, the precipitation of sulfur is important and occupies a considerable part of the volume for low-porosity basalt and hyaloclastite. In those systems, the pore space can be com- pletely filled by secondary minerals after sufficient number of pore water exchange cycles. At 250 °C and 25 MPa, the evolution is also similar to that in the freshwater case (Fig. 6), but is characterized by larger amount of chlorites. The mineral volume for Fig. 5 Single-box batch reactor modeling results with seawater at 50 °C and 5 MPa Thien et al. Geothermal Energy (2015) 3:11 Page 18 of 32 Fig. 6 Single-box batch reactor modeling results with seawater at 250 °C and 25 MPa high-porosity basalt and hyaloclastite increases on the contrary to freshwater case, but the pore space is not completely filled as observed at 50 °C and 5 MPa. Discussion The calculations show that the presence of volcanic gas and the higher fluid/solid ratio (higher porosity) enhance the alteration of primary basalt. The complete dissolution of pri- mary basalt at high temperatures, reported in the literature (e.g., Franzson et al. 2008), seems to be supported by this modeling, at least when volcanic gases are involved. For albite being usually stable at high temperatures, it is difficult to evaluate which part comes from the protolith, and which part is secondary. Actually, at those conditions, feldspars can be primary minerals as well as secondary minerals (Larsson et al. 2002). Basalt primary min- erals appear to be more stable in freshwater than in seawater. Main secondary minerals are saponites below 250 °C, and chlorites above this temperature. The nature of the protolith (i.e., basalt or basaltic glass) has a limited influence on the secondary mineralogy. The effect of dissolved gases (assumed to originate from young magma intrusions) on the alteration is significant. The presence of volcanic gas disfavors precipitation of zeolites and saponites (or chlorites above 250 °C, in a less extent) and favors the precipitation of carbonates, quartz (or polymorphs like chalcedony), and kaolinite. Epidote never forms in the presence of volcanic gas, whereas pyrite and chalcedony are always observed. The sec- ondary mineralogy appears therefore as an indicator of the conditions of the alteration. The modeling indicates that the precipitation of zeolites requires a low amount of volcanic gas and preferably freshwater than seawater. It is consistent with the observations of Krist- mannsdottir (1979). Actually, she reported that zeolites are less common in areas with sa- line fluids. The high amount of zeolites sometimes reported in the literature is not supported by this study. Nevertheless, it had been assumed that zeolites are the conse- quence of progressive low-temperature metamorphism (Weisenberger and Selbekk 2009), rather than hydrothermal alteration. The porosity evolution of the systems follows the pattern of the initial porosity of the protolith. For the low-porosity basalt, a complete filling of initial porosity by secondary minerals is systematically observed, regardless of the conditions. It could explain why cer- tain layers of compact intrusive basalt still appear fresh: if there is no water circulating, no significant alteration can occur. The field observation reported by Weisenberger and Thien et al. Geothermal Energy (2015) 3:11 Page 19 of 32 Selbekk (2009), indicating that secondary minerals mainly occur in the top and bottom parts of basaltic flow units, is consistent with this concept. Actually, the high-porosity and permeability of hyaloclastites implies high liquid fluxes; this water is necessarily in contact with the edges of low-porosity basaltic flow units (but can also penetrate to them along the fractures). The water in contact with the edges of the low-porosity basalt units alters them, mineralogical changes trigger a complete closure of the porosity, and secondary minerals are therefore visible in the top and bottom parts of the basalt units, as well as in the fractures. For the high-porosity basalt, the initial porosity is decreased, but the volume of precipitates is not enough to block the transport in the system. The interpretation of those results is not certain. In fact, below 250 °C saponites are the main secondary min- erals, and there are uncertainties about their molar volumes. Nevertheless, a tendency can be considered: the more the initial porosity of the basalt the less the system has the ten- dency to be clogged. For hyaloclastites, the porosity decrease upon alteration is negligible, except that at low temperature, in seawater systems with volcanic gas, the possibility of porosity closure after many pore water exchange cycles cannot be excluded. By using a single-box batch reactor model, the mineralogical evolution can be followed as a function of time, but not as a function of space. In fact, the water that reacted with the rocks will react again with “less reacted rocks” because of the fluid circulation, and secondary minerals along the reaction pathway are not necessarily the same. To clarify this, the multi-box batch reactor model was used. The calculations were only done for hyaloclastites; for basalts, they do not make much sense because unrealistic volumes are obtained from the beginning of the simulation. The results at 250 °C and 25 MPa (Fig. 7a) show that secondary minerals and total mineral volumes are the same as when consider- ing the single-box model (Figs. 4, 6). In addition, a mineral zonation can be seen. Note that this simplified modeling approach is based on assumptions that are not neces- sarily realistic for the whole range of conditions. Firstly, we calculated water/rock ratios on the basis of the total porosity. It implies that the water can react with the total amount of protolith. This hypothesis seems to be reasonable for hyaloclastite formations, but less rea- sonable for lava flows in which the low permeability suggests that part of the pores are not connected. Possible modifications of the pore connectivity during the alteration (Navarre- Sitchler et al. 2011) were also not accounted for. Secondly, the dissolution kinetics was not accounted for: the total amount of protolith reacts instantaneously with the water. Such an approach is only true if the dissolution kinetics is fast. It might be the case for hyaloclastites because they are completely altered, but not for lava flows that are fresh. This simple mod- eling is not necessarily realistic, but our model is not supposed to represent the whole real- ity, a complex reactive transport model accounting for kinetics is necessary to approach this goal. Nevertheless, it shows the potential effect of the initial porosity (expressed as water/solid volume ratio) on the evolution of the system. Let us see for which conditions the model is realistic and propose complementary explanations for other conditions. The complete alteration of hyaloclastites It was previously shown by using a single-box batch reactor model that hyaloclastites could be potentially completely altered, because the transformation of glass into second- ary minerals does not trigger any noticeable volume change. In this section, we firstly investigate if this complete alteration is kinetically possible. Thereafter, we investigate for which conditions the approach used in the last section is realistic. Thien et al. Geothermal Energy (2015) 3:11 Page 20 of 32 Fig. 7 a Multi-box batch reactor modeling results for hyaloclastite at 250 °C and 25 MPa. The initial volumes 3 3 of protolith and water are of 0.67 dm and 1 dm , respectively. b Multi-box batch reactor modeling results 3 3 for basalt at 250 °C and 25 MPa. The initial fluid/rock ratio (0.04 dm of basalt for 1 dm of water) is such as in a fracture from a 14- to 40-μm diameter Alteration times The model described in the “Methods” section was used to calculate the time necessary to reach 99 % of the alteration as a function of surface roughness factor (SRF) (Fig. 8a) and temperature (Fig. 8b). The time necessary to the complete alteration is propor- tional to the grain size and also proportional to the SRF. In other words, the alteration is proportional to the specific surface area. The more the specific surface area, the faster Thien et al. Geothermal Energy (2015) 3:11 Page 21 of 32 1/32 mm 1/16 mm 1 mm 4 mm 64 mm 1.E-01 1.E+01 1.E+03 1.E+05 Time (days) 1/32 mm 1/16 mm 1 mm 4 mm 50 64 mm 1.E-01 1.E+01 1.E+03 1.E+05 1.E+07 1.E+09 Time (days) Fig. 8 a Time necessary to reach 99 % of the total alteration of basaltic glass grains of different diameters, as a function of surface roughness factor (SRF), at 200 °C. b Time necessary to reach 99 % of the total alteration of basaltic glass grains of different diameters, as a function of temperature. SRF = 6 the alteration occurs. In the considered examples, the alteration time varies up to six orders of magnitude as function of specific surface area. The effect of temperature can be more im- portant (Fig. 8b) and does not depend on the specific surface area. Because the residual alter- ation rate (i.e., minimum dissolution rate) of basaltic glass was considered, the calculated alteration durations represent “upper limits”. In certain cases (e.g., diluted water), alteration durations can be two orders of magnitude faster than those represented in this study. Upon hydrothermal conditions, the alteration of glassy fragments of hyaloclastites is very fast: from a few hours to a few years. It represents a huge contrast compared to the several millions years of surface basalt life duration (Crovisier et al. 2003). In Hellisheidi, the high temperature hydrothermal circulation did not occur; hence, hyaloclastites are only partially altered (Alfredsson et al. 2008). The measurements performed in this site give an oppor- tunity to validate the model used in this study, but the task is not trivial because of the huge geological heterogeneities and the limited amount of available data. We focus on a hyaloclastite layer located between 200- and 400-m depth. The published data indicate an average temperature of 16 °C, an average alteration of 30 %, and a grain size of 7 mm in diameter (Alfredsson et al. 2013). By using these temperature and grain size data, and sup- posing a SRF of 6 (an average value from considering the results of Wolff-Boenisch et al. (2004)), the model presented in this study indicates that 200,000 years are necessary to alter 30 % of the protolith. It is fully in accord with the age range from 125,000 to 275,000 years given by Alfredsson et al. (2013). Nevertheless, in the considered system, the water supply is not unlimited if a given time is considered. The rock alteration consumes water, and the alteration stops if T (°C) SRF Thien et al. Geothermal Energy (2015) 3:11 Page 22 of 32 water is no longer available. The water is supposed to be initially present in the pore space of system and is renewed with a certain fluid velocity. If the availability of water limits the alteration, the results presented in this study do not make any sense. The amount of water necessary to completely alter 1 m of basaltic glass with a porosity of 60 % was calculated using the GEM-Selektor v.3 package. Depending on temperature and gas content, the amount of water necessary to completely alter the basaltic glass 3 3 ranges between 52 and 76 dm . It is largely inferior to 600 dm , the amount of water contained in 1 m of this rock. The water potentially present in the pore spaces of hya- loclastites is largely sufficient to completely achieve their alteration. It implies that fluid velocity is not necessarily a limiting parameter for the alteration of hyaloclastites. Alteration time versus fluid velocity In a sequential reactor simulation, the water of each box is completely replaced after equilibration with the solid at each time step. Thus, the duration of the time step corre- sponds to the time necessary to react with the whole amount of protolith in the node, i.e., this duration is proportional to the solid/water mass ratio in the box. Conversely, for an assumed flux of water through it, there is an optimal box size, defined by the time step duration and the characteristic diffusion length. −9 2 We considered an average value of water self-diffusion coefficient of 2.1°10 m /s at 25 °C (Harris and Woolf 1980). It is close to values established for solutes in water which −9 2 −9 2 vary (according to the ion species) between 0.6°10 m /s and 9.3°10 m /s at 25 °C −9 2 (Flury and Gimmi 2002). Those authors also reported a value of 1.5°10 m /s for NaCl electrolyte. As composition of the solution is likely to change, it is easier to consider the water self-diffusion coefficient, which in this study represents an average diffusion coeffi- cient of solutes in water. Because the diffusion length depends on the square root of the diffusion coefficient, the uncertainties on the diffusion coefficient do not noticeably influ- ence the results. This value has to be corrected to account for porosity. We used a simpli- fied Archie law (the effective diffusion coefficient is equal to the porosity multiplied by the pore diffusion coefficient). The initial porosity in hyaloclastites is 0.6, assumed to remain constant during the alteration. This implies the effective diffusion coefficient of water of −9 2 1.26°10 m /s at 25 °C. Unfortunately, Harris and Woolf (1980) did not report the diffu- sion coefficient for temperatures higher than 100 °C. But we used their data to establish the activation energy of 17 kJ/mol; thus, diffusion coefficients can be recalculated for the desired temperature. The effect of pressure is negligible in the range of pressures of inter- est. The distance crossed by a molecule affected by diffusion can be approximated as: pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ e ¼ D t ð6Þ where e is the distance (m), D the diffusion coefficient (m /s), and t the time (s). For each temperature, there is a length for which the diffusion velocity equals the transport velocity, and a corresponding time. If the glass can be completely altered dur- ing this time, the approach used in the “Mineralogical evolution and volume changes” section is realistic. By considering the residence time of water (Arnorsson 1995) along a hypothetical fluid pathway (going down to 3-km depth and going back to the surface), −7 −5 one can estimate a fluid velocity range between 1°10 m/s and 1°10 m/s. It is consist- −7 ent with the value estimated in Hellisheidi by Aradottir et al. (2012): 8°10 m/s. The time for which the transport equals the diffusion increases with the temperature be- cause the diffusion coefficient increases with the temperature whereas the fluid velocity does not depend on temperature. This time is represented in Fig. 9. Thien et al. Geothermal Energy (2015) 3:11 Page 23 of 32 1.00E-07 1.00E-06 1.00E-05 0.001 0.01 0.1 1 10 100 Time (days) Fig. 9 Time for which the flow rate equals the diffusion, for different fluid velocities (m/s). For a given fluid velocity, when this calculated time is greater than or equal to the total dissolution time of the protolith, equilibrium dissolution can be considered One can see that this time is very sensitive to the value of the fluid velocity. A vari- ation of one order of magnitude of the fluid velocity influences the time by two or- ders of magnitude. When comparing with the alteration times given on Fig. 8, by −6 considering an average fluid velocity of 1°10 m/s and a SRF of 6, it appears that the concept of complete dissolution and alteration is only valid for finest grain sizes at temperatures of 300 °C and above. By considering a SRF of 200 and the same fluid velocity, the concept is also valid at temperatures of 200 °C. By considering the lowest values of fluid velocity, the concept is valid for a wider range of conditions, for in- stance for the grains of a 1/16-mm diameter with a low SRF at 200 °C. By considering the lowest values of fluid velocity, this concept is also valid at 150 °C if the highest possible values of specific surface area are considered (not presented on Fig. 8, but can be deduced by extrapolation, considering that the alteration time is roughly pro- portional to the SRF). Our box reactor batch modeling approach is mostly valid for the highest temperature areas, from 200 to 300 °C as a function of the grain morphology. For the other cases, we consequently overestimate the amount of glass that reacts with the fluid. An alternative approach would be to increase the fluid/rock ratio, to let the amount of protolith that can be altered during a considered time react. But such an approach is not necessarily more realistic, because when the initial amount of proto- lith is completely altered, secondary products continue to be altered in the model, whereas in reality more protolith still dissolves. A realistic model suitable for the whole range of conditions is necessarily a complex reactive transport model account- ing for mineral–water kinetic rates. We have seen that the transformation of glass into secondary minerals does not change the total mineral volume at 60 % initial porosity and that the more the initial porosity the less is the mineral volume change. Letting less glass to react (i.e., equivalent to set up porosity above 60 %) cannot reduce the initial porosity, as suggested by the results presented in the “Mineralogical evolution and volume changes” section. The conclusions about hyaloclastites derived in the last section are still valid: the alteration of hyaloclastites does not affect the porosity significantly. Such alteration is typically fast because of high specific surface areas and hydrothermal temperatures. T(°C) Thien et al. Geothermal Energy (2015) 3:11 Page 24 of 32 The limited alteration of lava flows As seen in the introduction, in lava flows, water circulates mostly through fractures. The fractures could be formed by the cooling contraction, but also necessarily by the tectonic extension occurring in the rift zones. For hydraulic reasons, the superposition of hyaloclastite formations and lava flows necessarily implies that the water circulates faster through the lava flows than in the hyaloclastites. Consequently, the amount of water cannot be a limiting parameter for the alteration of the basalt fracture surfaces. The water slowly diffuses through the more or less connected pores, in a perpendicular direction to the fracture flux (e.g., MacQuarrie and Mayer 2005). Mineral volume changes in pores and fractures In our modeling approach, we assumed that porosity changes are only due to mineral volume changes and this leads to permeability changes that are positively correlated with porosity variations. The approach developed in the “Mineralogical evolution and volume changes” sec- tion indicates that the initial fluid/rock ratio controls the evolution of the system and cannot be rigorously applied to the case of lava flows, which are not a homogeneous porous medium. Nevertheless, there should be a threshold fluid/rock ratio at which the system is clogged (i.e., the volume of secondary minerals is equal to the initial volume of protolith with voids). Above this ratio, the water circulation in the system is not blocked and the alteration can proceed. With the results presented in the “Mineral- ogical evolution and volume changes” section, it can be stated that the threshold ratio ranges conjugate to porosity range from 10 to 40 %. We calculated with GEM-Selektor v.3 that the threshold initial volumetric water/rock ratio is approximately 0.4 and barely varies with the temperature or the initial water composition. It corresponds to a total porosity of 30 %. By considering a range of possible specific surface areas from 0.4 to 2.3 m /g for a fresh basalt (Simonyan et al. 2012) and a perfectly cylindrical shape for pores and fractures, it can be simply calculated that the water/rock ratio of 0.4 corre- sponds to a pore or fracture diameter range between 0.2 and 1.3 μm, completely filled by water. This is greater than the average pore diameter of 45 nm measured in a fresh basalt (Simonyan et al. 2012). Consequently, the pores in basalt can be potentially clogged, as well as the sub-micrometric aperture fractures. The knowledge of permeability allows estimating the average aperture of fractures by using the cubic law for a relation between the water debit, the measured pressure gradient, and the cube of the size of the fracture plane. Combining this law with the Darcy’s law yields (Committee on Fracture Characterization and Fluid Flow, National Research Council 1996): b ¼ 12 T ð7Þ where T is the transmissivity (m ), and b the thickness of the fracture plane (m). It is the simplest approximation considering smooth walls and regular fractures. Actually, real fractures have rough walls and variable apertures, but this reasoning still gives some insight into how much porosity/aperture reduction is possible. Arnorsson (1995) used permeability measurements to estimate the transmissivity. He −12 −12 3 found a transmissivity in the range between 2°10 and 8°10 m . The application of Equation 7 gives the average fracture aperture between 288 and 458 μm. It is larger than Thien et al. Geothermal Energy (2015) 3:11 Page 25 of 32 the threshold diameter of <1 μm as previously calculated. In other words, such fractures cannot be clogged during the alteration and contribute to the fluid circulation during the whole life of the hydrothermal system. It is important to note that calculated values are only average values. It means that smaller fractures can exist, as well as larger fractures. The multi-box sequential reactor modeling was performed by considering a high 3 3 arbitrary fluid/rock ratio. The initial volume of basalt is 40 cm for 1 dm of water. It corresponds to aporeora fractureof14to40 μm in diameter, depending of spe- cific surface area. We can note that such a fluid/rock ratio can also represent a pore or afractureofadifferent diameter,byconsidering limiting dissolution kinetics. Vol- canic gas or seawater have the tendency to increase the total mineral volume, but this increase is insignificant when compared to the total volume available (Fig. 7b). Secondary minerals are the same as those givenbyall thepreviousmodelings. Precipitation kinetics versus diffusion The existence of a threshold water/rock ratio indicates that the fractures of average aperture are never blocked, whereas the pores are potentially blocked. In this approach, the diffusion was not accounted for. Effective diffusion coefficients have been measured in fresh basalts, indicating that a non-negligible proportion of pores are necessarily connected, or were connected at the early stage of the life of the hydrothermal system. Higher effective diffu- sion coefficients result in more advanced diffusion fronts. But, if the precipitation kinetics of secondary minerals is fast enough to consider a water/rock ratio of 0.4 (i.e., the threshold water/rock ratio calculated in this section) or less, the alteration is blocked because of mineral volume changes or severely decreased if the diffusion can continue through the secondary minerals. An increase of effective diffusion coefficient of water molecules in fresh −11 −10 2 basalt from 5.05°10 to 1.19°10 m /s was measured as temperature increases from 5 to 50 °C, implying an activation energy of 12.3 kJ/mol (Simonyan et al. 2012). By using Equation 6, it can be calculated that in 10,000 years, the diffusion front would propagate 6 m at 50 °C, and 15 m at 250 °C. We have no clear idea about the thickness of the alteration zone around the fractures; neither if there is a sharp boundary or a progressive transition, but such a diffusion length is in contradiction with the fact that lava flows are rather fresh. In addition, it is worth to note that the age of the fresh mid-ocean ridge basalt mentioned above was estimated at 8.64 Ma (Simonyan et al. 2012). The authors did notprovide anyinformation aboutthe temperature, butitis likelythat somehydrothermal circulation occurred in such systems. There is a competition between diffusion and second- ary mineral precipitation kinetics. The effect of the precipitation kinetics seems to be greater than the effect of the initial diffusion. The diffusion coefficient in unaltered basalts decreases during the alteration, because the mineral volume increase reduces the diameter of the pores. In this way, the bulk basalt becomes resistant from further alteration. The case of non-connected pores The observation of pores that are partially or completely filled with secondary minerals in otherwise unaltered rock raises the question of pore connectivity and how these minerals were formed, i.e., by slow equilibration with the basaltic rock in a closed pore or by hydrothermal alteration due to mass transport in a connected pore space. The possible clogging of narrower pores and fractures (as discussed before) implies that a majority of pores and vesicles are isolated from the water circulation. The calculations Thien et al. Geothermal Energy (2015) 3:11 Page 26 of 32 in the “The complete alteration of hyaloclastites” section were based on the assumption that water can freely circulate in the pore space. The alteration in isolated, non- connected pores or vesicles (i.e., closed system) was simulated with an additional set of model calculations (Fig. 10). Unlike the calculations in the “The complete alteration of hyaloclastites” section, these ones started with a given amount of water (1 dm ) and a relatively small amount of basalt. After equilibration of the system, more basalt was added, thus the system was successively equilibrated with more and more basalt. Figure 10 presents the simulation results for pure water without gases. We conducted simulations for various initial water compositions and found that the evolution of the system shows no significant dependency on the initial water composition or gas content. Unlike for the cases with water circulation (Figs. 2 and 7), the initial relatively small amount of water is never exchanged, which limits the influence of its initial composition on large amounts of basalt. We found for the isolated pores that, independent from temperature, the volume of secondary minerals is nearly equal to the total volume available in the system. This finding is in agreement with the field observation that the pore spaces of the vesicles are partially or totally filled with secondary minerals (Weisenberger and Selbekk 2009). The calculated mineralogy for isolated pores does not substantially differ from the calculated mineralogy of the connected pore space. The most significant difference concerns the occurrence of the following sequence of zeolites in the model described in the literature (e.g., Gudmundsson and Arnorsson 2005): low-temperature zeolites up to 100 °C, laumontite from 100 to 200 °C, and wairakite above 200 °C. The calculations for the connected pore space did not predict the precipitation of laumontite and wairakite. Therefore, it might be possible that the presence of those two phases is related to the occurrence of isolated pores or vesicles. A simplified conceptual model The results obtained in this study allow us to propose a simplified conceptual model of an Icelandic hydrothermal system (Fig. 11). The model is partial because we did not consider the highest temperature zones. We assume that the water circulation is mainly due to convection. Rainwater or seawater penetrates into the system, and will be pro- gressively heated up, because the temperature increases with the depth. In lava flows, the water circulates along the fractures (vertical permeability) and, to small extent, can penetrate perpendicularly to the fracture flow into the basalt, via diffusive transport. But the low fluid/rock ratio in the pores adjacent to the fractures makes them com- pletely clogged during the transformation of primary minerals into secondary minerals. The water can no longer diffuse through the basalt pores, and the basalt becomes “pro- tected”. This potential pore clogging and the existence of a threshold pore/fracture aperture is in accord with the field observation that basalt flows are in most cases ra- ther fresh, and that secondary minerals are found in and around the fractures, and at the interface with the hyaloclastites. The high permeability and porosity of hyaloclastite formations makes the water circulation easier. The water circulation may have a hori- zontal component because hyaloclastite layers are surrounded by basalt flow layers that are less permeable. The transformation of glass into secondary minerals does not sig- nificantly change the porosity, which allows the alteration to progress. The complete Thien et al. Geothermal Energy (2015) 3:11 Page 27 of 32 Fig. 10 Closed reactor modeling results for basaltic glass in pure water. The initial amount of water is of 1 dm and the basaltic glass was sequentially added in order to simulate a constant alteration rate in a closed medium. The total volume available is equal to the sum of the added volume of protolith and the added volume of water Fig. 11 Conceptual picture of water flow in the hydrothermal system Thien et al. Geothermal Energy (2015) 3:11 Page 28 of 32 alteration of small glassy fragments is rather fast, from a few days to a few hours, due to a low grain size, high surface roughness, and high temperature. Conclusions Field observations show that hyaloclastites are completely altered (i.e., all the protolith had been transformed into secondary minerals), and compact basalts exhibit a wide range of alteration stages (i.e., from not altered to completely altered). In this study, we explore possible reasons that can explain this observation. The initial water/rock ratio has been found to be more important for the evolution of the system than water com- position or gas content. The water/rock ratio depends on the porosity and its connect- ivity (or on the size of a considered pore or fracture), and on the amount of rocks that can react during a given period of time. By considering that the initial pore space is completely filled with water, the lower is the initial water/rock ratio, the more the porosity (or pore or fracture size) decreases during the alteration. In more extreme cases, no connected porosity remains; the water can therefore no longer circulate: the system is impermeable. The complete alteration of hyaloclastites is possible because of high initial water/ rock ratio which is due to high initial connected porosity and high reactivity of small glass grains. Even if all the initial amount of protolith reacts with water, the available porosity does not noticeably change; the alteration can consequently proceed. This process is fast (from a few hours to a few years) compared to the lifetime of a hydro- thermal system (hundreds to thousands years). It is due to the high specific surface area of glassy fragments, as well as to elevated temperatures. This study highlights the fact that in any water–rock interaction study, consideration of dissolution or precipita- tion kinetics requires a good knowledge of the morphology of the minerals. The role of the specific surface area appears to be much more important than that of the kinetic rate constant. In lava flows, the water mainly circulates in the fractures. The average size of fractures corresponds to a water/rock ratio that is high enough not to produce any significant porosity change during the alteration, whatever the advancement of the reaction. The fluid flow in the fractures is therefore not affected by precipitation. There is a diffusive exchange of solutes across the fracture walls between the water in the fracture and the pore water of the adjacent rock matrix. The water/rock ratio in the pores close to the fracture walls is so low that on the long term, they become completely filled with secondary minerals. The pore connectivity is thus obstructed, preserving the rest of the basalt from subsequent alteration. Such clogging is only possible under the condition that the precipitation kinetics of secondary minerals is fast enough. The extent of the alteration rim around the fracture appears to be deter- mined by the competition between the precipitation kinetics of secondary minerals and the diffusive transport of solutes in the pore network. Future field observations will provide more data about the extent of alteration rims and also about the fracture size distribution in lava flows. This information is important for the parameterization of a more detailed geochemical reactive transport model that will describe the system evolution by simultaneously accounting for precipitation and dissolution kinetics (i.e., reaction rates and reactive surface areas), diffusion in pores, flow in pores and fractures, as well as heat conductivity changes of the rocks with alteration. Thien et al. Geothermal Energy (2015) 3:11 Page 29 of 32 Competing interests The authors declare that they have no competing interests. Authors’ contributions All the authors contributed to the literature search, the interpretation of modeling results and the elaboration of the conceptual model. BT did the kinetic model for basaltic glass dissolution, the geochemical reaction path setup and the calculations. All the authors read, enhanced and approved the final manuscript. Acknowledgements The financial support was provided by the Swiss National Science Foundation under the grant number CRSII2_141843/ 1 (Sinergia COTHERM). We also acknowledge the editorial board of Geothermal Energy for giving us the opportunity to present our work as an open issue, and two anonymous reviewers for their suggestions which helped to improve the quality and the readability of this paper. Received: 11 March 2015 Accepted: 20 May 2015 References Advocat T, Jollivet P, Crovisier JL, del Nero M (2001) Long-term alteration mechanisms in water for SON68 radioactive borosilicate glass. J Nucl Mater 298:55–62 Alfredsson HA, Hardarsson BS, Franzson H, Gislason SR (2008) CO2 sequestration in basaltic rock at the Hellisheidi site in SW Iceland: stratigraphy and chemical composition of the rocks at the injection site. Mineral Mag 72:1–5 Alfredsson HA, Oelkers EH, Hardarsson BS, Franzson H, Gunnlaugsson E, Gislason SR (2013) The geology and water chemistry of the Hellisheidi, SW-Iceland carbon storage site. Int J Greenh Gas Con 12:399–418 Altaner SP, Ylagan RF, Savin SM, Aronson JL, Belkin HE, Pozzuoli A (2003) Geothermometry, geochronology, and mass transfer associated with hydrothermal alteration of a rhyolitic hyaloclastite from Ponza Island, Italy. Geochim Cosmochim Acta 67:275–288 Aradottir ESP, Sonnenthal EL, Björnsson G, Jonsson H (2012) Multidimensional reactive transport modeling of CO mineral sequestration in basalts at the Hellisheidi geothermal field, Iceland. Int J Greenh Gas Con 9:24–40 Aradottir ESP, Sigfússon B, Sonnenthal EL, Björnsson G, Jónsson H (2013) Dynamics of basaltic glass dissolution–capturing microscopic effects in continuum scale models. Geochim Cosmochim Acta 121:311–327 Arnorsson S (1975) Application of silica geothermomether in low-temperature hydrothermal areas in Iceland. Am J Sci 275(7):763–784 Arnorsson S (1995) Geothermal systems in Iceland: structure and conceptual models-I. High-temperature areas. Geothermics 24(5/6):561–602 Arnorsson S, Gunnlaugsson E, Svavarsson H (1983) The chemistry of geothermal waters in Iceland.2. Mineral equilibria and independent variables controlling water compositions. Geochim Cosmochim Acta 47(3):547–566 Banik TJ, Wallace PJ, Höskuldsson A, Miller CF, Bacon CR, Furbish DJ (2014) Magma-ice-sediment interactions and the origin of lava/hyaloclastite sequences in the Sida formation, South Iceland. Bull Volcanol 76(785):1–19 Barfod DN, Fitton JG (2014) Pleistocene volcanism on São Tomé, Gulf of Guinea, West Africa. Quat Geochronol 21:77–89 Berner RA (2013) From black mud to Earth System science: a scientific autobiography. Am J Sci 313:1–60 Bourcier WL, Peiffer DW, Knauss KG, McKeegan, KD, Smith DK (1990) A kinetic model for borosilicate glass dissolution based on the dissolution affinity of a surface alteration layer. Scientific Basis for Nuclear Waste Management XIII (76):209–2016 Catalano JG (2013) Thermodynamic and mass balance constraints on iron-bearing phyllosilicates formation and alteration pathways on early Mars. J Geophys Res: Planets 118:2124–2136 Claridge GC, Campbell IB (2008) Zeolites in Antarctic soils: examples from Coombs hills and marble point. Geoderma 144:66–72 Committee on Fracture Characterization and Fluid Flow, National Research Council (1996) Rock Fractures and Fluid Flow: Contemporary Understanding and Applications. The National Academies Press, National research Council, Washington DC, USA. Crovisier JL, Advocat T, Dussossoy JL (2003) Nature and role of natural alteration gels formed on the surface of ancient volcanic glasses (Natural analogs of waste containment glasses). J Nucl Mater 321:91–109 Curti E (1999) Coprecipitation of radionuclides with calcite: estimation of partition coefficients based on a review of laboratory investigations and geochemical data. Appl Geochem 14:433–445 Dessert C, Dupré B, Gaillardet J, François LM, Allègre CJ (2003) Basalt weathering laws and the impact of basalt weathering on the global carbon cycle. Chem Geol 202(3–4):257–273 Ehlmann BL, Bish DL, Ruff SW, Mustard JF (2012) Mineralogy and chemistry of altered Icelandic basalts: Application to clay mineral detection and understanding aqueous environments on Mars. J Geophys Res 117: 1-27 Ferrage E, Lanson B, Michot LJ, Robert JL (2010) Hydration properties and interlayer organization of water and ions in synthetic Na-smectite with tetrahedral layer charge. Part 1. Results from X-ray diffraction profile modeling. J Phys Chem C 114(10):4515–4526 Flury M, Gimmi T (2002) Solute diffusion. Methods and Soil Analysis, Part 4, Physical Methods: 1323–1351. Edited by Dane JH and Topp GC, Soil Science Society of America, Madison WI Franzson H, Gudlaugsson SP, Fridleifsson GO (2001) Petrophysical properties of Icelandic rocks. Proceedings of the 6th Nordic Symposium on Petrophysics, Tronheim, Norway. In: Knut Backe & Ton Loermans (ed), The Norwegian University of Science and Technology. Trondheim, Norway Franzson H, Zierenber R, Schiffman P (2008) Chemical transport in geothermal systems in Iceland. Evidence from hydrothermal alteration. J Volcanol Geotherm Res 173:217–229 Thien et al. Geothermal Energy (2015) 3:11 Page 30 of 32 Franzson H, Gudfinnsson GH, Helgadóttir HM, Frolova J (2010) Porosity, density and chemical composition relationships in altered Icelandic hyaloclastites. In: Birkle & Torres-Alvarado (ed), Water-Rock Interaction. London: Taylor & Francis Group; 199–202 Franzson H, Gudfinnsson GH, Frolova J, Helgadóttir HM, Pauly B, Mortensen AK, Jakobsson SP (2011) Icelandic Hyaloclastite Tuffs, Petrophysical Properties, Alteration and Geochemical Mobility. Report prepared for National Energy Authority and Reykjavík Energy. Iceland: ISOR, Reykjavik; ÍSOR-2011/064 Freedman AJE, Bird DK, Arnorsson S, Fridriksson T, Elders WA, Fridleifsson GO (2009) Hydrothermal minerals record CO(2) partial pressures in the Reykjanes geothermal system, Iceland. Am J Sci 309(9):788–833 Fridleifsson GO, Elders WA (2005) The Iceland deep drilling project: a search for deep unconventional geothermal resources. Geothermics 34(3):269–285 Fridriksson T, Neuhoff PS, Arnórsson S, Bird DK (2001) Geological constraints on the thermodynamic properties of the stilbite—stellerite solid solution in low-grade metabasalts. Geochim Cosmochim Acta 65(21):3993–4008 Frolova J, Ladygin V, Franzson H, Sigurdsson O, Stefánsson V, Shustrov V (2005) Petrophysical properties of fresh to mildly altered hyaloclastite tuffs. Proceedings of World Geothermal Congress 2005, Antalya, Turkey, pp 24–29 Frugier P, Gin S, Minet Y, Chave T, Bonin B, Godon N, Lartigue JE, Jollivet P, Ayral A, De Windt L, Santarini G (2008) SON68 nuclear glass dissolution kinetics: current state of knowledge and basis of the new GRAAL model. J Nucl Mater 380:8–21 Gaboreau S, Prêt D, Tinseau E, Claret F, Pellegrini D, Stammose D (2011) 15 years of in situ cement–argillite interaction from tournemire URL: characterisation of the multi-scale spatial heterogeneities of pore space evolution. Appl Geochem 26(12):2159–2171 Gin S, Ribet I, Couillard M (2001) Role and properties of the gel formed during nuclear glass alteration: importance of gel formation conditions. J Nucl Mater 298:1–10 Gislason SR, Oelkers EH (2003) Mechanism, rates, and consequences of basaltic glass dissolution: II. An experimental study of the dissolution rates of basaltic glass as a function of pH and temperature. Geochim Cosmochim Acta 20:3817–3832 Gislason SR, Wolff-Boenisch D, Stefansson A, Oelkers EH, Gunnlaugsson E, Sigurdardottir H, Sigfusson B, Broecker WS, Matter JM, Stute M, Axelsson G, Fridriksson T (2010) Mineral sequestration of carbon dioxide in basalt: a pre- injection overview of the CarbFix project. Int J Greenh Gas Con 4(3):537–545 Grambow B, Jercinovic MJ, Ewing RC, Byers CD (1986) Weathered basalt glass: a natural analogue for the effects of reaction progress on nuclear waste glass alteration. Scientific Basis for Nucl Waste Mgt IX 50:363–272 Griffith LL, Shock EL (1997) Hydrothermal hydration of Martian crust: Illustration via geochemical model calculations. J Geophys Res 102:9135–9143 Gronvold F (1976) Heat capacities in critical regions. Plenary Lectures Presented at the Fourth International Conference on Chemical Thermodynamics Université des Sciences et Techniques de Languedoc, Montpellier, France 26–30 August 1975. Chemical Thermodynamics, edited by J. Rouquerol and R. Sabbah, Elsevier. Gudmundsson BT, Arnorsson S (2002) Geochemical monitoring of the Krafla and Namafjall geothermal areas, N-Iceland. Geothermics 31:195–243 Gudmundsson BT, Arnorsson S (2005) Secondary mineral-fluid equilibria in the Krafla and Namafjall geothermal systems, Iceland. Appl Geochem 20(9):1607–1625 Gunnlaugsson AS (1982) The chemistry of iron in geothermal systems in Iceland. J Volcanol Geotherm Res 14(3–4):281–299 Gustavson JE (2006) Analysis of porosity evolution during low temperature metamorphism of basaltic lava and implications for fluid flow. University of Florida, Master thesis Gysi AP, Stefansson A (2011) CO -water-basalt interaction. Numerical simulation of low temperature CO sequestration 2 2 into basalts. Geochim Cosmochim Acta 75:4728–4751 Gysi AP, Stefansson A (2012a) Mineralogical aspects of CO2 sequestration during hydrothermal basalt alteration-an experimental study at 75 to 250° C and elevated pCO(2). Chem Geol 306:146–159 Gysi AP, Stefansson A (2012b) Experiments and geochemical modeling of CO2 sequestration during hydrothermal basalt alteration. Chem Geol 306:10–28 Harris KR, Woolf LA (1980) Pressure and temperature dependence of the self diffusion coefficient of water and oxygen-18 water. J.C.S. Faraday I 76:377–385 Hellevang H, Dypvik H, Kalleson E, Pittarello L, Koeberl C (2013) Can alteration experiments on impact melts from El’gygytgyn and volcanic glasses shed new light on the formation of the Martian surface? Meteorit Planet Sci 48:1287–1295 Holland TGB, Powell R (1998) An internally-consistent thermodynamic dataset for phases of petrological interest. J Metamorphic Geology 16:309–344 Honnorez J (2003) Hydrothermal alteration vs. ocean-floor metamorphism. A comparison between two case histories: the TAG hydrothermal mound (Mid-Atlantic Ridge) vs. DSDP/ODP Hole 504B (Equatorial East Pacific). Comptes Rendus Geoscience 335(10–11):781–824 Hovius N, Lea-Cox A, Turowski JM (2008) Recent volcano–ice interaction and outburst flooding in a Mars polar cap re-entrant. Icarus 197(1):24–38 Hreggvisdottir H (1978) The greenschist to amphibolite facies transition in the Nesjavellir hydrothermal system. Stanford University, CA, M.Sc. thesis Jarosch A, Gudmundsson MT, Högnadottir T, Axelsson G (2008) Progressive cooling of the hyaloclastite ridge at Gjalp, Iceland, 1996–2005. J Volcanol Geotherm Res 170(3–4):218–229 Kaasalainen H, Stefansson A (2011) Sulfur speciation in natural hydrothermal waters, Iceland. Geochim Cosmochim Acta 75:2777–2791 Kaasalainen H, Stefansson A (2012) The chemistry of trace elements in surface geothermal waters and steam, Iceland. Chem Geol 330(331):60–85 Kanayama K, Umino S, Ishizuka O (2014) Shallow submarine volcano group in the early stage of island arc development: geology and petrology of small islands south off Hahajima main island, the Ogasawara Islands. J Asian Earth Sci 85:1–25 Thien et al. Geothermal Energy (2015) 3:11 Page 31 of 32 Karmous MS, Ben Rhaiem H, Robert JL, Lanson B, Ben Haj Amara A (2009) Charge location effect on the hydration + 2+ properties of synthetic saponite and hectorite saturated by Na ,Ca cations: XRD investigation. App Clay Sci 46:43–50 Kristmannsdottir H (1979) Alteration of basaltic rock by hydrothermal activity at 100–300 C. In: International clay conference 1978 (edited by Mortland MM and farmer VC). Elsevier, Amsterdam, pp 359–367 Kulik DA, Vinograd VL, Paulsen N, Winkler B (2010) (Ca, Sr)CO3 aqueous–solid solution systems: from atomistic simulations to thermodynamic modelling. Physics and Chemistry of the Earth 35:217–232 Kulik DA, Wagner T, Dmytrieva SV, Kosakowski G, Hingerl FF, Chudnenko KV, Berner U (2013) GEM-Selektor geochemical modeling package: revised algorithm and GEMS3K numerical kernel for coupled simulation codes. Computat Geosci 17:1–24, http://gems.web.psi.ch Larsson D, Grönvold K, Oskarsson N, Gunnlaugsson E (2002) Hydrothermal alteration of plagioclase and growth of secondary feldspar in the Hengill Volcanic Centre, SW Iceland. J Volcanol Geotherm Res 114(3–4):275–290 Lonker SW, Franzson H, Kristmannsdottir H (1993) Mineral-fluid interactions in the Reykjanes and Svartsengi geothermal systems, Iceland. Am J Sci 293(7):605–670 MacQuarrie KTB, Mayer KU (2005) Reactive transport modeling in fractured rock. A state-of-the-science review. Earth-Sci Rev 72:189–227 Marks N, Schiffman P, Zierenberg RA, Hj F, Fridleifsson GO (2010) Hydrothermal alteration in the Reykjanes geothermal system: Insights from Iceland deep drilling program well RN-17. J Volcanol Geotherm Res 189(1–2):172–190 Markusson SH, Stefansson A (2011) Geothermal surface alteration of basalts, Krysuvik Iceland-alteration mineralogy, water chemistry and the effects of acid supply on the alteration process. J Volcanol Geotherm Res 206(1–2):46–59 Matter JM, Broecker WS, Stute M, Gislason SR, Oelkers EH, Stefansson A, Wolff-Boenisch D, Gunnlaugsson E, Axelsson G, Björnsson G (2009) Permanent carbon dioxide storage into basalt: the CarbFix pilot project, Iceland. Energy Procedia 1(1):3641–3646 Michot LJ, Bihannic I, Pelletier M, Rinnert E, Robert JL (2005) Hydration and swelling of synthetic Na-saponites: influence of layer charge. Am Mineral 90:166–172 Miyashiro A (1975) Metamorphism and Metamorphic belts. London: George Allen & Unwin Ltd.; 1975 Moore LN, Mueller WU, Daigneault R (2012) In situ hydroclastic fragmentation of subaqueous ponded lavas; New Senator caldera, Abitibi greenstone belt, Quebec, Canada. Precambrian Res 214–215:44–59 Mortensen AK, Egilson P, Gautason B, Arnadottir S, Gudmundsson A (2014) Stratigraphy, alteration mineralogy, permeability and temperature conditions of well IDDP-1, Krafla, NE-Iceland. Geothermics 49:31–41 Motelib AA, Khalaf EA, Al-Marzouki H (2014) Growth, destruction and facies architecture of effusive and explosive volcanics in the Miocene Shama basin, southwest of Saudi Arabia: subaqueous–subaerial volcanism in a lacustrine setting. J Volcanol Geotherm Res 277:51–76 Navarre-Sitchler A, Steefel CI, Sak PB, Brantley SL (2011) A reactive-transport model for weathering rind formation on basalt. Geochim Cosmochim Acta 75(23):7644–7667 Navarre-Sitchler AK, Cole DR, Rother G, Jin L, Buss HL, Brantley SL (2013) Porosity and surface area evolution during weathering of two igneous rocks. Geochim Cosmochim Acta 109:400–413 Navarre‐Sitchler A, Steefel CI, Yang L, Tomutsa L, Brantley SL (2009) Evolution of porosity and diffusivity associated with chemical weathering of a basalt clast. J Geophys Res: Earth Surf (2003–2012), 114(F2): 1-19 Neuhoff P (2000) Thermodynamic properties of parageneses of rock-forming zeolites. Ph.D. thesis, Stanford University, Stanford, Calif Neuhoff PS, Fridriksson T, Arnorsson S, Bird DK (1999) Porosity evolution and mineral paragenesis during low-grade metamorphism of basaltic lavas at Teigarhorn, eastern Iceland. Am J Sci 299:467–501 Pham VTH, Lub P, Aagaard P, Zhu C, Hellevang H (2011) On the potential of CO2–water–rock interactions for CO2 storage using a modified kinetic model. Int J Greenh Gas Con 5(4):1002–1015 Pitty AF, Alexander WR (2014) A natural analogue study of cement buffered , hyperalkaline groundwaters and their interaction with a repository host rock IV : an examination of the Khushaym Matruk (central Jordan ) and Maqarin (norhtern Jordan) sites. Technical Report BG-TR-11-02 Vol. 2. Switzerland: Bedrock Geosciences, Auenstein Rebiscoul D, Van der Lee A, Rieutord F, Né F, Spalla O, El-Mansouri A, Frugier P, Ayral A, Gin S (2004) Morphological evolution of alteration layers formed during nuclear glass alteration: new evidence of a gel as a diffusive barrier. J Nucl Mater 326:9–18 Rinnert E, Carteret C, Humbert B, Fragneto-Cusani G, Ramsay JDF, Delville A, Robert JL, Bihannic I, Pelletier M, Michot LJ (2005) Hydration of a synthetic clay with tetrahedral charges: a multidisciplinary experimental and numerical study. J Phys Chem B 109(49):23745–23759 Saar MO, Manga M (1999) Permeability-porosity relationship in vesicular basalts. Geophys Res Lett 26:111–114 Scanlon KE, Head JW, Wilson L, Marchant DR (2014) Volcano–ice interactions in the Arsia Mons tropical mountain glacier deposits. Icarus 237(15):315–339 Schiffman P, Fridleifsson GO (1991) The smectite chlorite transition in drillhole NJ-15, Nesjavellir geothermal field, Iceland – XRD, BSE and Electron-microscope investigations. Journal of Metamorphic Geology 9(6):679–696 Schleicher AM, Hofmann H, van der Pluijm BA (2013) Constraining clay hydration state and its role in active fault systems. Geochem Geophys Geosyst 14(4):1039–1052 Schopka HH, Gudmundsson MT, Tuffen H (2006) The formation of Helgafell, southwest Iceland, a monogenetic subglacial hyaloclastite ridge: Sedimentology, hydrology and volcano-ice interaction. J Volcanol Geotherm Res 152:359–377 Scott S, Gunnarsson I, Arnorsson S, Stefansson A (2014) Gas chemistry, boiling and phase segregation in a geothermal system, Hellisheidi, Iceland. Geochim Cosmochim Acta 124:170–189 Shao H, Kosakowski G, Berner U, Kulik DA, Mäder U, Kolditz O (2013) Reactive transport modeling of the clogging process at Maqarin natural analogue site. Phys Chem Earth, parts A/B/C 64:21–31 Shibuya T, Tahata M, Ueno Y, Komiya T, Takai K, Yoshida N, Maruyama S, Russell MJ (2013) Decrease of seawater CO2 concentration in the Late Archean: an implication from 2.6 Ga seafloor hydrothermal alteration. Precambrian Res 236:59–64 Thien et al. Geothermal Energy (2015) 3:11 Page 32 of 32 Shock EL, Sassani DC, Willis M, Sverjensky DA (1997) Inorganic species in geologic fluids: correlations among standard molal thermodynamic properties of aqueous ions and hydroxide complexes. Geochim Cosmochim Acta 61:907–950 Sigurdsson H (1970) The petrology and chemistry of the Setberg volcanic region and of the intermediate and acid rocks of Iceland. Durham, UK: Ph.D. thesis, Durham University Sigurdsson O, Stefansson V (2002) Porosity structure of Icelandic basalt. Proc Estonian Acad Sci Geol 51:33–46 Simonyan AV, Dultz S, Behrens H (2012) Diffusive transport of water in porous fresh to altered mid-ocean ridge basalts. Chem Geol 306–307:63–77 Steefel CI, Lichtner PC (1998) Multicomponent reactive transport in discrete fractures: II: Infiltration of hyperalkaline groundwater at Maqarin, Jordan, a natural analogue site. J Hydr 209: 200–224. Stefansson A, Arnorsson S (2002) Gas pressures and redox reactionsingeothermalfluidsinIceland. ChemGeol 190(1–4):251–271 Stefansson A, Arnorsson S, Gunnarsson I, Kaasalainen H, Gunnlaugsson E (2011) The geochemistry and sequestration of H2S into the geothermal system at Hellisheidi, Iceland. J Volcanol Geotherm Res 202:179–188 Steingrimsson B, Gudmundsson A, Sigurdsson O, Gunlaugsson E (1986a) Nesjavellir well NJ-11. National Energy Authority Report OS-86025/JHD-0.5, in Icelandic Steingrimsson B, Fridleifsson GO, Sverrisdottir G, Tulinius H, Sigurdsson O, Gunnlaugsson E (1986b) Nesjavellir well NJ-15, Drilling, investigation and production characteristics. National Energy Authority Report OS-96029/JHD-09, in Icelandic Steinthorsson S, Sveinbjörnsdottir AE (1981) Opaque minerals in geothermal well no.7, Krafla, Northern Iceland. J Volcanol Geotherm Res 10:245–261 Strachan DM, Neeway JJ (2014) Effects of alteration product precipitation on glass dissolution. Appl Geochem 45:144–157 Stryjek R, Vera JH (1986) PRSV2: a cubic equation of state for accurate vapor—liquid equilibria calculations. Can J Chem Eng 64:820–826 Sveinbjörnsdottir AE (1992) Composition of geothermal minerals from saline and dilute fluids – Krafla and Reykjanes, Iceland. Lithos 17:301–315 Techer I, Advocat T, Lancelot J, Liotard JM (2001) Dissolution kinetics of basaltic glasses: control by solution chemistry and protective effect of the alteration film. Chem Geol 176:235–263 Thien BMJ (2014) A simple way to constrain the stoichiometry of secondary smectites upon glass alteration. Appl Geochem 42:45–46 Thien B, Godon N, Hubert F, Angeli F, Gin S, Ayral A (2010) Structural identification of a trioctahedral smectite formed by the aqueous alteration of a nuclear glass. Appl Clay Sci 49(3):135–141 Thien BMJ, Godon N, Ballestero A, Gin S, Ayral A (2012) The dual effect of Mg on the long-term alteration rate of AVM nuclear waste glasses. J Nucl Mater 427:297–310 Thien BMJ, Kulik DA, Curti E (2014) A unified approach to model uptake kinetics of trace elements in complex aqueous - solid solution systems. Appl Geochem 41:135–150 Tomasson J, Kristman H (1972) High-temperature alteration minerals and thermal brines, Reykjanes, Iceland. Contrib Mineral Petr 36(2):123–137 Vidal O, Dubacq B (2009) Thermodynamic modelling of clay dehydration, stability and compositional evolution with temperature, pressure and H O activity. Geochim Cosmochim Acta 73(21):6544–6564 Vieillard P (2000) A new method for the prediction of Gibbs free energies of formation of hydrated clay minerals based on the electronegativity scale. Clays and Clay Minerals 48(4):459–473 Vieillard P (2007) THERMOCHIMIE- estimation des enthalpies de formation des phyllosilicates anhydres. Hydrasa-INSU – FRE3114 N°2007-03, 20.2007, Report, in French Wagner T, Kulik DA, Hingerl FF, Dmytrieva SV (2012) GEM-Selektor geochemical modeling package: TSolMod library and data interface for multicomponent phase models. Can Mineral 50:1173–1195 Walker GPL (1960) Zeolite zones and dike distribution in relation to the structure of the basalts of eastern Iceland. J Geol 68(5):515–528 Watton TJ, Jerram DA, Thordarson T, Davies RJ (2013) Theree-dimensional facies variations in hyaloclastite deposits. J Volcanol Geotherm Res 250:19–33 Weisenberger T, Selbekk RS (2009) Multi-stage zeolite facies mineralization in the Hvalfjördur area, Iceland. Int J Sci 98:985–999 Wolff-Boenisch D, Gislason SR, Oelkers EH, Putnis CV (2004) The dissolution rates of nuclear glasses as a function of their composition at pH 4 and 10.6, and temperatures from 25 to 74 °C. Geochim Cosmochim Acta 68(23):4843–4858 Wolff-Boenisch D, Gislason SR, Oelkers EH (2006) The effect of crystallinity on dissolution rates and CO2 consumption capacity of silicates. Geochim Cosmochim Acta 70(4):858–870 Zezin DY, Migdisov AA, Williams-Jones AE (2011) PVTx properties of H O-H S fluid mixtures at elevated temperature 2 2 and pressure based on new experimental data. Geochim Cosmochim Acta 75:5483–5495 Zierenberg RA, Schiffman P, Jonasson IR, Tosdal R, Pickthorn W, McClain J (1995) Alteration of basalt hyaloclastite at the off-axis Sea Cliff hydrothermal field, Gorda Ridge. Chem Geol 126:77–99
Geothermal Energy – Springer Journals
Published: Jun 23, 2015
Access the full text.
Sign up today, get DeepDyve free for 14 days.