Advances and Gaps in the Knowledge of Thermodynamics and Crystallography of Acid Mine Drainage Sulfate Minerals

Home , Bieberite, Melanterite, Meridianiite, Rozenite

MINERALOGY CHIMIA 2010, 64, No. 10 699 doi:10.2533/chimia.2010.699 Chimia 64 (2010) 699–704 © Schweizerische Chemische Gesellschaft Advances and Gaps in the Knowledge of Thermodynamics and Crystallography of Acid Mine Drainage Sulfate Minerals

Juraj Majzlan*

Abstract: Acidic and metal-rich waters produced by sulfide decomposition at mining sites are termed acid mine drainage (AMD). They precipitate a number of minerals, very often sulfates. The recent advances in thermodynamic properties and crystallography of these sulfates are reviewed here. There is a reasonable amount of data for the divalent (Mg, Ni, Co, Fe2+, Cu, Zn) sulfates and these data may be combined with and optimized by temperature-relative humidity brackets available in the literature. For the sulfates with Fe3+, most data exist for jarosite; for other minerals and phases in this system, a few calorimetric studies were reported. No data whatsoever are available for the Fe2+-Fe3+ sulfates. A significant advance is the development of the Pitzer model for Fe3+-sulfate solutions and its confrontation with the available thermodynamic and solubility data. In summary, our knowledge about the thermodynamic properties of the AMD sulfates is unsatisfactory and fragmented. Keywords: Acid mine drainage

Introduction population matches the quantities of the relatively recently.[12,13] A sizeable body of overall natural mass flow.[4] The mining of data accumulated during the last ten years, The ever-increasing demand for metals ores and salts does not dominate the an- partially because of interest in AMD, par- dictates the steady increase in the exploi- thropogenic mass flow but cannot be ne- tially because sulfate minerals have been tation of ore deposits, and, to a smaller glected. identified in Martian meteorites[14] or pre- extent, attempts to recycle the useful ele- The extraction of ores from the litho- sumed to be present on the surface of Mars ments. For example, 3,547 tonnes of gold sphere, fine grinding and disposal of un- (e.g. ref. [15]) and to have implications for were extracted in 2007.[1] This is a large wanted waste exposes many ore minerals, the understanding of the ancient evolution amount for a metal handled so cautiously in a fine particulate form, to the atmo- of this planet.[16,17] and carefully even in the smallest quanti- sphere and hydrosphere, thereby acceler- ties. Correspondingly larger amounts of ating their decomposition rates by many other metals are needed, some of them orders of magnitude.[5] The released acid- Mineral Parageneses reaching unimaginable magnitudes. The ity, metal cations, and anions can reach for- global population moves into cities which midable levels.[6] This phenomenon is well Acidic, metal-rich waters can be pro- [2] are becoming vast new ore deposits with known under the name acid mine drainage duced by weathering of pyrite (FeS2) or their buildings, traffic systems, and other (AMD) and is certainly not new, as can be pyrrhotite (Fe1-xS), rarely other sulfides, daily components of life. The human- documented by detailed descriptions in ref. in contact with atmosphere and surface induced fluxes of some elements have [7] among others. The mineralogy of AMD or underground water.[5] Such weathering already exceeded the natural fluxes (e.g. systems is rich, although many minerals most commonly occurs in waste products Cd[3]). Apart from creation and decompo- are only ephemeral, being washed away by of mining activities and the process and sition of the biomass, the today’s human rain or surface streams after a dry period. its products are called acid mine drainage Much of the mineralogical work on AMD (AMD). The composition of the two pri- has been reviewed.[5,8–10] Here I summarize mary sulfides (pyrite and pyrrhotite) dic- the recent advances in the thermodynamics tates the bulk ion load of the AMD waters, and crystallography of the sulfates found that is, their usually high concentration of in AMD systems. The thermodynamic iron, either ferrous or ferric, and sulfate. properties of iron oxides, halogenides, se- The acid waters attack and dissolve other lected sulfates, and many aqueous species minerals present and release their con- with iron have been critically summarized stituents into the aqueous fluids. These in an excellent review of Lemire et al.[11] constituents comprise major rock-forming and need not to be discussed here. Because elements such K, Na, Mg, Ca, Al but also the thermodynamic properties must be re- toxic and heavy metals and metalloids, for ferred to phases with known composition example Cu, Ni, Zn, Cd, As, and others. All and structure, this review provides also an these cations may combine with sulfate to *Correspondence: Prof. Dr. J. Majzlan update for each group of considered com- form AMD minerals. Beside sulfates, iron Friedrich-Schiller University Institute of Geosciences pounds in terms of their crystal structures. oxides, be they poorly or well crystalline, Burgweg 11 This update documents how variable is are abundant. The iron oxides are power- D-07749 Jena this ‘well known’ group of structures. No ful sorbents of many ions and thus in some Tel.: +49 3641 948700 Fax: +49 3641 948602 comprehensive overview of the sulfate cases prevent the formation of independent E-mail: [email protected] structures is necessary; this has been done iron-free or iron-poor phases. 700 CHIMIA 2010, 64, No. 10 MINERALOGY temperature [°C] temperature

Fig. 1. Eh-pH diagram that shows two possible pathways for relative humidity [%] the evolution of an acid mine drainage system. The minerals not mentioned in the text but shown here are: goethite – α-FeOOH, green Fig. 2. Stability of sulfate minerals as a function of relative air rust – Fe2+-Fe3+ hydroxide with additional anions, schwertmannite humidity and temperature. The curves calculated from optimized [26] 2+ – ~FeO(OH)3/4(SO4)1/8, and ferrihydrite – ~Fe(OH)3. Schwertmannite thermodynamic data are: Mg sulfates, Cu, Ni, Zn, Fe , Co sulfates and ferrihydrite are poorly crystalline and metastable with respect to n (Grevel and Majzlan, in preparation). The curves for the Fe2(SO4)3· H2O goethite. were calculated from measured and estimated thermodynamic [30] data. The curve for rhomboclase and (H3O)Fe2(SO4)2 is the best ﬁt to the experimental data in ref. [31]. The curves in the range of very high relative humidity are dashed because the sulfate minerals In general, an AMD system proceeds deliquesce in this region. Minerals shown in the ﬁgure: bianchite ZnSO ·6H O, bieberite CoSO ·7H O, bonattite CuSO ·3H O, through several stages of development 4 2 4 2 4 2 chalcanthite CuSO ·5H O, epsomite MgSO ·7H O, goslarite before the final, stable or metastable but 4 2 4 2 ZnSO ·7H O, hexahydrite MgSO ·6H O, kieserite MgSO ·H O, kornelite kinetically inert state is achieved.[5] Fig. 1 4 2 4 2 4 2 Fe (SO ) ·~7H O, melanterite FeSO ·7H O, moorhouseite CoSO ·6H O, shows possible evolution pathways of an 2 4 3 2 4 2 4 2 morenosite NiSO4·7H2O, paracoquimbite Fe2(SO4)3·9H2O, retgersite AMD system. Crystallization of the solu- NiSO ·6H O, rhomboclase (H O )Fe(SO ) ·2H O, rozenite FeSO ·4H O. ble divalent metal sulfates depends on the 4 2 5 2 4 2 2 4 2 availability of the divalent cations and the degree of evaporation. As the system ma- tures, iron is being oxidized and is stored in iron sulfates or oxides, depending on the to polymerize (as opposed to octahedra ious MgSO4·nH2O phases by varying rela- pH value of the fluids. Note that the system that house Fe3+), the chemical composition tive humidity and temperature. Because must achieve relatively low pH values (<2) of these minerals is simply MSO4·nH2O. they were able to reverse the reactions, that to form the crystalline Fe3+ (or Fe2+-Fe3+) Even with this simple composition, the is, to observe hydration of the lower hy- sulfates described below. It may seem, variability in this group is remarkable. The drate and dehydration of the correspond- therefore, that these minerals are rare and structures of most of these phases have ing higher hydrate, these experiments are restricted to the large AMD systems which been solved earlier, reviewed in ref. [12] brackets in the sense of the brackets com- achieve such low pH values routinely. To- and further discussed in ref. [13]. A recent monly determined for high-temperature sca et al.[18] have shown that evaporation thorough study of the Mg sulfates identi- and high-pressure silicate equilibria. The of pools of waters with relatively high pH fied several new phases.[19] The structure univariant curves calculated from avail- [24,25] (~3) can lead to a final pH of almost ~0 and of the MgSO4 dihydrate (mineral sand- able thermodynamic values signifi- precipitation of many minerals mentioned erite[20]), 2.5 hydrate,[21] and the dodeca- cantly deviate from the experimentally in this paper. A common example from hydrate (mineral meridianiite[22]) were observed reversals.[27,28] To alleviate this many AMD sites are the powdery aggre- solved. The systems with the other diva- problem, Grevel and Majzlan[26] applied gates of copiapite which probably crystal- lent cations await similar attention and it is the mathematic programming analysis (cf. lize from small isolated evaporating pools very likely that new phases exist in those ref. [29]), a method that varies the avail- of acidic waters and disappear with the first systems as well. able thermodynamic values within their rain. As the acid-generating capacity of an The thermodynamic data for these uncertainties. The result of the calculations AMD system vanishes, the pH rises and phases have been summarized in an ex- is a phase diagram constructed from opti- the final product is goethite (Fig. 1). tensive compilation,[23,24] some other data mized thermodynamic data and satisfying may be found in ref. [25]. The former work the experimental observations (Fig. 2). We offers references to the older literature, the are currently in process of optimizing the Divalent Metal Sulfates latter has, unfortunately, no references scarcer data for the Fe2+, Co, Ni, and Zn whatsoever. The reliability of the thermo- sulfate phases and some preliminary re- Divalent metal sulfates are a diverse dynamic data varies from one mineral to sults are also depicted in Fig. 2. and large group of minerals. The divalent the other and can be tested by construct- The mixing properties in the continu- metals here include Fe2+, Mg, Zn, Ni, Co, ing phase diagrams. Most data exist for ous solid solutions between Ni, Mg, and Zn 2+ Mn . The minerals progress from frame- the system MgSO4-H2O and this system sulfates heptahydrates were investigated in work structures for the anhydrous forms to has been reviewed.[26] The sources of data a series of studies.[32–35] They showed that structures with isolated polyhedra or clus- are calorimetric data (summarized in ref. the thermodynamics of mixing in these ters thereof in heavily hydrated forms. The [24,25], new data in ref. [26]) and the tem- solid solutions is not ideal although the metals are coordinated octahedrally but perature-relative humidity reversals.[27,28] Gibbs free energies of mixing are rather because these octahedra have no tendency They determined the stability fields of var- small, up to 0.2 kJ/mol. MINERALOGY CHIMIA 2010, 64, No. 10 701

Jarosite products. The thermodynamic properties of jarosite were then derived by extrapo- The solid solution between jarosite s.s., lation of the data to 298.15 K, using the natrojarosite, and hydronium jarosite is prob- activity-molality relation of the aqueous ably the most common crystalline sulfate in ions valid only for T = 298.15 K. In ad- acid mine drainage. Jarosite is isostructural dition, Stoffregen[58] reported that jarosite with alunite whose structure was solved decomposes either to hematite or Fe(OH) [36] [37] by Hendricks. Recently, Sato et al. SO4, depending on the activity of sulfate in refined the structure of a suite of jarosite- the aqueous solution. We have found (Ma- group minerals. The numerous substitutions jzlan, unpublished) that jarosite decom- within jarosite can place constraints on the poses invariably to a mixture of hematite composition of natural fluids which precipi- and yavapaiite (KFe(SO4)2) at 0.2 GPa and tated this mineral.[38,39] The structure is built temperatures of 200–600 ºC. To compound by sheets of octahedra which house Fe3+. the problem further, in most of the stud- The sheets are decorated by sulfate tetrahe- ies cited in this paragraph, jarosite was dra and the monovalent ions reside between implicitly assumed to be stoichiometric, the sheets in a large site with coordination although, as discussed above, the prepa- number of 12. These principal features are Fig. 3. A segment from the structure of ration of a stoichiometric phase requires maintained throughout the jarosite family of jarosite showing a vacancy at the Fe3+ site. significant effort. minerals; deviations from the trigonal sym- An alternative approach is to mea- Note that the H2O molecule shown is located metry or localization of cations at otherwise approximately (1/6)c above the plane deﬁned sure the formation enthalpy and standard vacant sites have been reported.[40–42] The by the Fe3+ cations. entropy of these phases separately and to Δ o structure is further complicated by the abun- combine them into G f values. Drouet and dance of defects and species that are difficult Navrotsky[59] measured the formation en- to detect by diffraction techniques or most metric samples by NMR.[43,44] The entire thalpy of a series of jarosite phases by high- types of spectroscopy. picture is further complicated by the fact temperature calorimetry and proposed the The elusive hydronium ion is usually that the A site may not be fully occupied. best thermodynamic values. Their best val- called upon when the monovalent cations The occupancy of this site was only 91% ues, however, do not satisfy the relation- [45] Δ o Δ o Δ o (mostly Na and K in nature) do not occupy in pure synthetic jarosite, showing that ship G f = H f – T S f, and therefore the 12-fold coordinated site completely. the common assumption that the difference should be used with caution, if at all. They The existence of the hydronium ion in the between 100 % and (Na+K) content on the also measured the formation enthalpy of jarosite structure is difficult to prove by dif- A site is the hydronium ion may be in er- hydronium jarosite, later re-measured (us- fraction techniques. Nielsen et al.[43] have ror. Gale et al.[46] used atomistic simulation ing the same sample) by acid-solution cal- applied nuclear magnetic resonance spec- of a hydronium alunite structure and found orimetry.[45] The discrepancy between the 2 + Δ o troscopy (NMR) on the H nuclei in a set that the H3O ion assumes 12 symmetry- two H f values is astounding, almost 30 of synthetic, deuterated jarosite samples. related positions in its cavity. In none of kJ/mol. The reason for such large discrep- They detected three different local deuter- these positions, however, coincides the ancy is not clear, but the data (measured on environments, Fe2OD, FeOD2, and D2O/ penetrative three-fold symmetry axis with enthalpy and entropy) in ref. [45] show in- + + D3O in jarosite (Fig. 3). The paramagnetic the local three-fold axis of the H3O ion. ternal consistency with the available phase nature of the jarosite at room temperature There was a significant effort expended diagrams and are therefore considered to was helpful, and not obstructive, as it sepa- to synthesize stoichiometric, defect-free be superior. The formation enthalpy of Pb- rated the signal into three easily resolvable jarosite samples. One synthesis route uses jarosite was recently reported.[60] Fermi contact shifts of δ ≈ 237, 70, and metallic iron as a starting product and the The only so far published low-tempera- 3+ 0 ppm, respectively. The Fe2OD groups consequent slow supply of Fe into the so- ture heat capacity and entropy data are those are rigid in terms of the motion visible lution as the means for the assembly of a for hydronium jarosite.[45] Majzlan et al.[61] [47] o in the NMR spectra. The FeOD2 groups, stoichiometric structure. This procedure measured Cp and S for a series of stoichio- on the other hand, undergo rapid flips on also allows syntheses of jarosite phases metric jarosite phases and a non-stoichio- the NMR time scale. They arise from the with V3+ and Cr3+,[48] unknown from nature, metric potassium-rich jarosite. Striking is charge compensation of the Fe3+ vacancies and was driven mostly by the interest in the the difference in the estimate of So for the and were used in these experiments to de- unusual magnetic properties of jarosite.[49] K-jarosite of 388.9 J·mol–1·K–1[58] and the termine the concentration of the vacancies. The other route is based on lowering the experimental value of 427.4 J·mol–1·K–1.[61] + The sites usually assumed to be occupied activity of H2O and H3O in the aqueous Such differences further underscore the by monovalent cations are populated by solution by massive addition of Li+ which need for experimental measurements, es- + [50,51] D2O or D3O groups. The mechanism of does not enter the jarosite structure. pecially in systems where vacancies, de- vacancy introduction can be described by The earlier studies on thermodynamic fects, deviations from nominal stoichiom- an exchange reaction properties of jarosite minerals have been etry, and unusual magnetic properties are summarized.[10] They were mostly based the rule, not an exception. + + → + M(OH)4O2 + 3H +H3O (OH2)4O2 on evaluation of the solubility of jarosite 3+ [52–56] + H2O + M in aqueous media. The results of those studies where jarosite was allowed Fe(III) Sulfates except Jarosite where M is either Al3+[44] or Fe3+. There- to dissolve in water (as opposed to crys- fore, the hydronium ions are consumed tallizing jarosite from a solution) could These minerals are chemically and by the vacancies and can only exist in the be questioned because the dissolution of structurally variable (see Table 1 for a list near-stoichiometric regions in jarosite (or jarosite could be incongruent.[57] Stoffre- with the chemical formulae). In general, alunite, the Al analogue of jarosite). This gen[58] adopted a different approach, that the structures progressively depolymerize conclusion was confirmed by the study of a is, evaluation of high-temperature equilib- with the increasing hydration state, passing series of stoichiometric and non-stoichio- ria between jarosite and its decomposition from the anhydrous framework structures 702 CHIMIA 2010, 64, No. 10 MINERALOGY

Table 1. Sulfates of Fe3+ which may be found in the AMD systems trigonal polymorph is the mineral mika- AFe (SO ) (OH) jarosite group (A = Na+, K+, H O+) saite, the monoclinic polymorph is so far 3 4 2 6 3 unknown from nature. The formation en- AFe (SO ) (OH) ·20H O copiapite group (A = 2/3Fe3+, 2/3Al3+, Mg, Zn, Fe2+) 4 4 6 2 2 thalpy was measured or estimated in refs Fe(OH)SO ·nH O butlerite, parabutlerite, ﬁbroferrite 4 2 [71–73] and the values roughly agree. The [73,74] (Fe,Al)2(SO4)3·nH2O mikasaite, lausenite, kornelite, coquimbite, entropy values on the other hand, dis- paracoquimbite, quenstedtite agree significantly, mostly because Pan- [74] Fe O(SO ) ·nH O hohmannite, metahohmannite, amarantite kratz and Weller measured data down to 2 4 2 2 50 K and then extrapolated to 0 K. Doing (H O )Fe(SO ) ·2H O rhomboclase 5 2 4 2 2 so, they missed most of the magnetic en- n KFe(SO4)2· H2O yavapaiite, krausite, goldichite tropy and underestimated the total entropy Na Fe(OH)(SO ) ·3H O sideronatrite at T = 298.15 K by about 10%. This prob- 2 4 2 2 lem was avoided by measuring heat capac- Na Fe(SO ) ·3H O ferrinatrite 3 4 3 2 ity down to 1 K, well below the magnetic MgFe(SO ) (OH)·7H O botryogen [73] 4 2 2 transition.

NaMg2Fe5(SO4)7(OH)6·33H2O slavikite Enthalpies of formation were measured by acid-solution calorimetry for

Fe2(SO4)3·5H2O, kornelite, coquimbite, to heavily hydrated structures with isolated to infinite chains. At least two slightly dif- paracoquimbite, rhomboclase, and fer- polyhedra or polyhedral clusters. ferent structures can be recognized among ricopiapite.[30,75] Entropy estimates were In their thorough study, Posnjak and the copiapite-group minerals.[67] Voltaite, a presented for all these phases in the two Merwin[62] identified a number of phas- phase with cubic morphology but anisotro- publications. Ackermann et al.[30] have es whose structure is today known. The pic behavior in crossed nicols, was shown shown that, using reasonable estimates of last remaining structures were those of to have a slight tetragonal distortion.[68] In entropies, a sensible phase diagram for the [63] [64] n Fe2(SO4)3·5H2O, and (H3O)Fe(SO4)2. their study of hydration and dehydration of Fe2(SO4)3· H2O phases can be constructed The former phase may or may not corre- the Fe(iii) sulfates, Xu et al.[69] detected an (Fig. 2). Such phase diagram can be further spond to the mineral lausenite and the ques- unknown phase or phases. Obviously, these combined and optimized with the temper- tion can be probably solved only if this rare minerals still hide many secrets which await ature-relative humidity brackets,[69,76,77] al- mineral is found again.[63] The structure of to be discovered. though these experiments may be plagued metahohmannite was solved from powder Much less is known about the thermo- by sluggishness, disequilibrium, and XRD data,[65] a remarkable work given the dynamics of these phases. Hemingway et the presence of unknown phases. These triclinic symmetry and the presence of im- al.[70] estimated the properties of a number brackets are now not only available for the purities in the sample. It may seem from of these phases. Although it could be ar- Fe2(SO4)3·nH2O phases (l.c.) but also for [31] the previous few sentences that little re- gued that the absence of data at that time the pair (H3O)Fe(SO4)2-rhomboclase mains to be done with the crystal structures justified the estimates, the phase relation- (Fig. 2). An approximate Eh-pH phase of Fe(iii) sulfates. This impression is, of ships in systems with a delicate thermo- diagram (Fig. 4) agrees with the field ob- course, false. Interesting and unusual varia- dynamic balance cannot be constructed servations but a diagram that shows the tions were found among the phases close only with estimates. The Fe(iii) sulfates coexistence of the aqueous phase with the to coquimbite;[66] with an increasing Al are such a system. individual minerals[62,80] deviates, at least content, the coquimbite structure changes Several papers addressed the ther- visually, much from the experimental data from having isolated polyhedral clusters modynamics of anhydrous Fe2(SO4)3; its (Fig. 5). The discrepancy in the magnitude

pε

Fig. 4. A schematic pH-Eh diagram for pyrite, melanterite, and a suite Fig. 5. Contours of the ratio of molalities of HSO4 and SO4, as of Fe3+ sulfate and oxide minerals. The ﬁxed activities of aqueous predicted by the Pitzer model parameters from ref. [81]. The thick a 2– a 3+ a 2+ a species are: (SO4 ) = 3.50, (Fe ) = (Fe ) = 1.20, (H2O) = 0.75. dashed line represents the composition of the solution in equilibrium

Data for pyrite from ref. [78]. Data from melanterite from optimized with rhomboclase, (H5O2)Fe(SO4)2·2H2O, calculated from the Pitzer thermodynamic data (Grevel and Majzlan, in preparation). Data for the model[81] and thermodynamic data from ref. [75]. The circles show the Fe3+ phases: hydronium jarosite,[45] rhomboclase and ferricopiapite,[75] experimentally determined solution compositions in equilibrium with goethite,[79] copiapite and roemerite – adjusted estimates from ref. [70]. rhomboclase from refs [82–84]). MINERALOGY CHIMIA 2010, 64, No. 10 703

of the thermodynamic values is, however, properties of aqueous species in a wide Potentiale, Nutzung, Risiken und Kosten’, not very large, because the position of the range of H SO and Fe (SO ) molalities. Vieweg, 1997. 2 4 2 4 3 [5] J. L. Jambor, in ‘Environmental aspects of curve plotted in Fig. 5 is a very sensitive Care must be taken, however, for solutions mine wastes’, Eds J. L. Jambor, D. W. Blowes, function of the thermodynamic data. Tosca where the Fe2(SO4)3 molality is high and A. I. M. Ritchie, Mineralogical Association of [81] Canada 2003, 31, 117. et al. concluded that the extent of the that of H2SO4 is low. In this case, ion pair- divergence is ‘encouraging’ and I would ing of Fe3+ and sulfate may render even this [6] D. K. Nordstrom, C. N. Alpers, Proc. Natl. only add that it is a good starting point for complicated model inaccurate. Acad. Sci. USA 1999, 96, 3455. [7] A. Breithaupt, Berg- und hüttenmännische further studies. Thermodynamic data for Zeitung 1852, 11, 65; J. D. Dana, ‘Manual of all other minerals listed in Table 1 but not Mineralogy’ Durrie & Peck, New Haven, 1848; explicitly mentioned in this paragraph are Outlook and Implications H. Rose, Ann. Phys. 1833, 27, 309. unknown. Thermodynamic properties of [8] J. L. Jambor, D. K. Nordstrom, C. N. Alpers, 2+ 3+ Rev. Mineral. Geochem. 2000, 40, 303. Fe -Fe sulfates are also unknown. As shown in this short review, the ther- [9] J. M. Bigham, D. K. Nordstrom, Rev. Mineral. modynamic database for the AMD sulfates Geochem. 2000, 40, 351. has expanded in the last decade so that at [10] R. E. Stoffregen, C. N. Alpers, J. L. Jambor, Aqueous Solutions least preliminary phase diagrams can be Rev. Mineral. Geochem. 2000, 40, 454. drawn. Although the progress with the Fe3+ [11] R. J. Lemire, U. Berner, C. Musikas, D. A. Palmer, P. Taylor, O. Tochiyama, ‘Chemical [81] The thermodynamic description of the sulfates is ‘encouraging’, it is far from thermodynamics of iron’, Nuclear Energy AMD systems remains incomplete if the satisfactory. Data for the Fe2+-Fe3+ sulfates Agency (Organisation for Economic Coopera- properties of the aqueous media are un- are missing and the data for the divalent tion and Development), 2010. known. The aqueous solutions typical for sulfates, much more abundant, are in a [12] F. C. Hawthorne, S. V. Krivovichev, P. C. Burns, Rev. Mineral. Geochem. 2000, 40, 1. these systems are usually concentrated, need of critical evaluation and optimiza- [13] M. Schindler, D. M. C. Huminicki, F. C. with the extremes up to eight molal H2SO4 tion. Hawthorne, Can. Mineral. 2006, 44, 1403. concentrations.[6] Simple molality-activity The last point that needs discussion [14] F. M. McCubbin, N. J. Tosca, A. Smirnov, H. (e.g. Debye-Hückel) models are not valid are the implications that the phase dia- Nekvasil, A. Steele, M. Fries, D. H. Lindsley, for such solutions. Instead, the more com- grams have for the prediction of the natu- Geochim. Cosmochim. Acta 2009, 73, 4907. [15] S. W. Squyres, J. P. Grotzinger, R. E. Arvidson, plicated model developed by K. Pitzer and ral assemblages. First, the AMD systems J. F. Bell, W. Calvin, P. R. Christensen, B. coworkers[85,86] can be used. abound with metastability and sluggish- C. Clark, J. A. Crisp, W. H. Farrand, K. E. Herkenhoff, J. R. Johnson, G. Klingelhöfer, A. Initially, the H2SO4 solutions were ness. Therefore, the thermodynamic calcu- treated with the assumption of complete lations only whisper about the way the sys- H. Knoll, S. M. McLennan, H. Y. McSween, [87] – R. V. Morris, J. W. Rice, R. Rieder, L. A. dissociation. The inclusion of HSO4 tem may go. Second, even if equilibrium Soderblom, Science 2004, 306, 1709. provided much more reliable predictions is assumed and thermodynamic data are [16] J. J. Papike, J. M. Karner, C. K. Shearer, of the properties.[88] Lassin et al.[89] have on hand, phase diagrams with no meaning Geochim. Cosmochim. Acta 2006, 70, 1309. [17] J. J. Papike, P. V. Burger, J. M. Karner, C. K. shown that at high H2SO4 molalities, the whatsoever can be calculated because of an Pitzer model must also include H SO 0 as a improper choice of phases. For example, Shearer, V. W. Lueth, Am. Mineral. 2007, 92, 2 4 444. species. Pitzer model parameters have been one could choose anhydrous FeSO4 for the [18] N. J. Tosca, S. M. McLennan, B. C. Clark, J. derived for solutions containing Mg-SO4, diagram in Fig. 4 and argue that such phase P. Grotzinger, J. A. Hurowitz, A. H. Knoll, C. Schröder, S. W. Squyres, Earth Planet. Sc. Lett. Fe(II)-SO4, Ni-SO4, Al-SO4 , and other sys- should be found in nature. It is unlikely tems (see ref. [90] for a review) relevant for that we will see such a diagram. Yet, there 2005, 240, 122. [33] [19] S. J. Chipera, D. T. Vaniman, Geochim. the AMD waters. Proskurina et al. re- are analogous diagrams, for example, for Cosmochim. Acta 2007, 71, 241. ported Pitzer coefficients for the MgSO4-, systems (AMD or not) polluted with anti- [20] H. W. Ma, D. L. Bish, H. W. Wang, S. J. Chipera, Am. Mineral. 2009, 94, 622. NiSO4-, and ZnSO4-H2O systems. Pitzer- mony (see ref. [93] for more information) model coefficients for solutions with Fe(iii) that provide a misleading picture about the [21] H. W. Ma, D. L. Bish, H. W. Wang, S. J. and SO were reported only recently by solid and aqueous speciation of elements. Chipera, Am. Mineral. 2009, 94, 1071. 4 [22] R. C. Peterson, R. Wang, Geology 2006, 34, 957. [81] Tosca et al. who derived them from the Therefore, phases for construction of the [23] C. W. DeKock, U.S. Bur. Mines, Inf. Circ. 1982, isopiestic data.[91,92] As an example, Fig. 5 phase diagrams should be always chosen 8910. 2– – [24] C. W. DeKock, U.S. Bur. Mines, Inf. Circ. 1986, depicts the speciation of SO4 and HSO4 carefully and critically, even if their ther- in the Fe(iii)-sulfate solutions. The fraction modynamic data are sound. 9081. – [25] D. D. Wagman, W. H. Evans, V. B. Parker, R. H. of HSO4 steeply increases from infinite di- Schumm, I. Halow, S. M. Bailey, K. L. Churney, Acknowledgements lution as the H2SO4 molality is increased R. L. Nuttall, J. Phys. Chem. Ref. Data 1982, but reaches a broad minimum between I thank T. Armbruster for the invitation 11, Suppl. 2. to write this review article, R. C. Peterson [26] K.-D. Grevel, J. Majzlan, Geochim. Cosmochim. 3.5–4.5 m H2SO4; a similar trend is seen for pure H SO solutions.[88] Generally, the and J. J. Papike for the reviews, and K.- Acta 2009, 73, 6805. 2 4 [27] I.-M. Chou, R. R. Seal, Astrobiology 2003, 3, fraction of HSO – decreases as Fe (SO ) D. Grevel for the unpublished data for the 4 2 4 3 divalent metal sulfates. The ﬁnancial support 619. is added to the solution. An exception in of the Deutsche Forschungsgemeinschaft in [28] I.-M. Chou, R. R. Seal, J. Geophys. Res. – Planets 2007, 112, 10.1029/2007JE002898. this trend are solutions with high H2SO4 projects on mineralogy, crystallography, and [29] R. G. Berman, M. Engi, H. J. Greenwood, T. H. molality where the addition of Fe2(SO4)3 thermodynamics of the sulfate minerals is – Brown, J. Petrol. 1986, 27, 1331. causes an increase in the HSO4 concentra- greatly appreciated. [30] S. Ackermann, T. Armbruster, B. Lazic, S. tion. The effect of these changes upon the Doyle, K.-D. Grevel, J. Majzlan, Am. Mineral. structures of co-existing minerals will have Received: April 24, 2010 2009, 94, 1620. to be examined in further detail. The pre- [31] W. Xu, J. Parise, J. Hanson, Am. Mineral. dicted pH values vary between 2 and –2 for (submitted) 2010. [1] ‘Gold Demand Trends’, Full year and fourth [32] I. A. Kasatkin, E. B. Treivus, T. I. Ivanova, the molalities considered by the model in quarter 2007, issued by The World Gold Russ. J. Phys. Ch. 2000, 73, 1880. ref. [81]. It is interesting to note that solu- Council and GFMS Ltd. [33] O. V. Proskurina, E. S. Mal’tseva, A. V. tions with very high Fe (SO ) molality but [2] S. Norra, Environ. Sci. Pollut. Res. 2009, 16, Rumyantsev, L. V. Puchkov, Russ. J. Phys. Ch. 2 4 3 539. low H SO molality have relatively ‘high’ 2001, 75, 343. 2 4 [3] P. Baccini, P. H. Brunner, ‘Metabolism of the [34] O. V. Proskurina, L. V. Puchkov, A. V. pH values around 2. The Pitzer parameters anthroposphere’ Springer, Berlin, 1991. Rumyantsev, Russ. J. Phys. Ch. 2001, 75, 163. presented[81] describe the thermodynamic [4] K. Heinloth, ‘Die Energiefrage – Bedarf und [35] O. V. Proskurina, A. V. Rumyantsev, N. A. 704 CHIMIA 2010, 64, No. 10 MINERALOGY

Charykov, Russ. J. Phys. Ch. 2002, 76, 1399. Geol. Bull. 1989, 4, 281. Donaldson, B. F. Woodfield, J. Boerio-Goates, [36] S. B. Hendricks, Am. Mineral. 1937, 22, 773. [55] D. Baron, D. Palmer, Geochim. Cosmochim. J. Chem. Thermodyn. 2005, 37, 802. [37] E. Sato, I. Nakai, R. Miyawaki, S. Matsubara, Acta 1996, 60, 185. [74] L. B. Pankratz, W. W. Weller, U.S. Bur. Mines, Neues Jb. Miner. Abh. 2009, 185, 313. [56] D. Baron, D. Palmer, Geochim. Cosmochim. Rep. Invest. 1969, 7280. [38] G. A. Desborough, K. S. Smith, H. A. Lowers, Acta 2002, 66, 2841. [75] J. Majzlan, A. Navrotsky, B. McCleskey, C. N. G. A. Swayze, J. M. Hammarstrom, S. F. [57] A. M. L. Smith, K. A. Hudson-Edwards, W. E. Alpers, Eur. J. Mineral. 2006, 18, 175. Diehl, R. W. Leinz, R. L. Driscoll, Geochim. Dubbin, K. Wright, Geochim. Cosmochim. Acta [76] E. A. Hasenmueller, D. L. Bish, Lunar Planet. Cosmochim. Acta 2010, 74, 1041. 2006, 70, 608. Sci. Conf. 2005, #1164. [39] P. V. Burger, J. J. Papike, C. K. Shearer, J. M. [58] R. E. Stoffregen, Geochim. Cosmochim. Acta [77] S. J. Chipera, D. T. Vaniman, D. L. Bish, Lunar Karner, Geochim. Cosmochim. Acta 2009, 73, 1993, 57, 2417. Planet. Sci. Conf. 2007, #1409. 3248. [59] C. Drouet, A. Navrotsky, Geochim. Cosmochim. [78] R. A. Robie, B. S. Hemingway, ‘Thermodynamic [40] J. Göttlicher, B. Gasharova, H. Bernotat-Wulff, Acta 2003, 67, 2063. properties of minerals and related substances at Abstr. Programs-Geol. Sci. Am. 2000, 32, A180. [60] F. L. Forray, A. M. L. Smith, C. Drouet, A. 298.15 K and 1 bar (105 Pascals) and at higher [41] I. E. Grey, W. G. Mumme, P. Bordet, S. J. Mills, Navrotsky, K. Wright, K. A. Hudson-Edwards, temperatures’, U.S. Geol. Survey Bull. 1995, 2131. Can. Mineral. 2008, 46, 1355. W. E. Dubbin, Geochim. Cosmochim. Acta [79] J. Majzlan, K.-D. Grevel, A. Navrotsky, Am. [42] I. E. Grey, W. G. Mumme, S. J. Mills, W. D. 2010, 74, 215. Mineral. 2003, 88, 855. Birch, N. C. Wilson, Am. Mineral. 2009, 94, [61] J. Majzlan, P. Glasnák, R. A. Fisher, M.-A. [80] H. E. Merwin, E. Posnjak, Am. Mineral. 1937, 676. White, M. Johnson, B. Woodfield, J. Boerio- 22, 567. [43] U. G. Nielsen, J. Majzlan, C. P. Grey, Chem. Goates, Phys. Chem. Miner. 2010, 37, 635. [81] N. J. Tosca, A. Smirnov, S. M. McLennan, Mater. 2008, 20, 2234. [62] E. Posnjak, H. E. Merwin, J. Am. Chem. Soc. Geochim. Cosmochim. Acta 2007, 71, 2680. [44] U. G. Nielsen, J. Majzlan, B. Phillips, M. 1992, 44, 1965. [82] F. K. Cameron, C. Robinson, J. Phys. Chem. Ziliox, C. P. Grey, Am. Mineral. 2007, 92, 587. [63] J. Majzlan, C. Botez, P. W. Stephens, Am. 1907, 11, 641. [45] J. Majzlan, R. Stevens, J. Boerio-Goates, B. Mineral. 2005, 90, 411. [83] F. Wirth, B. Bakke, Z. anorg. Chem. 1914, 87, 13. F. Woodfield, A. Navrotsky, M. Crawford, P. [64] R. C. Peterson, E. Valyashko, R. Want, Can. [84] W. H. Baskerville, F. K. Cameron, J. Phys. Burns, T. G. Amos, Phys. Chem. Miner. 2004, Mineral. 2009, 47, 625. Chem. 1935, 39, 769. 31, 518. [65] F. Scordari, G. Ventruti, A.F. Gualtieri, Am. [85] K. S. Pitzer, J. Phys. Chem. 1973, 77, 2300. [46] J. D. Gale, K. Wright, K. A. Hudson-Edwards, Mineral. 2004, 89, 365. [86] K. S. Pitzer, ‘Activity coefficients in electrolyte Am. Mineral. 2010, 95, 1109. [66] F. Demartin, C. Castellano, C. Gramaccioli, I. solutions’, CRC Press Boca Raton 1991. [47] D. Grohol, D. G. Nocera, D. Papoutsakis, Phys. Campostrini, Can. Mineral. 2010, 48, 323. [87] K. S. Pitzer, R. N. Roy, L. F. Silvester, J. Am. Rev. B 2003, 67, 064401. [67] J. Majzlan, R. Michallik, Mineral. Mag. 2007, Chem. Soc. 1977, 99, 4930. [48] D. Grohol, D. G. Nocera, J. Am. Chem. Soc. 71, 557. [88] S. L. Clegg, J. A. Rard, K. S. Pitzer, J. Chem. 2002, 124, 2640. [68] A. Ertl, M. D. Dyar, J. M. Hughes, F. Soc. Faraday T. 1994, 90, 1875. [49] D. Grohol, D. G. Nocera, Chem. Mater. 2007, Brandstatter, M. E. Gunter, M. Prem, R. C. [89] A. Lassin, C. Kervévan, L. André, M. Azaroual, 19, 3061. Peterson, Can. Mineral. 2008, 46, 661. Goldschmidt conf. 2009, A724. [50] L. C. Basciano, R. C. Peterson, Am. Mineral. [69] W. Xu, N. J. Tosca, S. M. McLennan, J. B. [90] C. Ptacek, D. Blowes, Rev. Mineral. Geochem. 2007, 92, 1464. Parise, Am. Mineral. 2009, 94, 1629. 2000, 40, 513. [51] L. C. Basciano, R. C. Peterson, Am. Mineral. [70] B. S. Hemingway, R. R. Seal, I. M. Chou, U.S. [91] A. V. Rumyantsev, S. Hagemann, H. C. Moog, 2008, 93, 853. Geol. Survey, Open File Report 2002, 02-161. Z. phys. Chem. 2004, 218, 1089. [52] A. V. Zotov, G. D. Mironova, V. L. Rusinov, [71] R. Barany, L. H. Adami, U.S. Bur. Mines, Rep. [92] M. Velazquez-Rivera, D. A. Palmer, R. M. Geokhimiya 1973, 5, 739. Invest. 1965, 6687. Kettler, J. Solution Chem. 2006, 35, 1699. [53] C. H. M. Kashkay, Y. U. B. Borovskaya, M. A. [72] H. H. Kellogg, T. Metall. Soc. AIME 1964, 230, [93] M. Filella, P. A. Williams, N. Belzile, Environ. Babazad, Geochem. Int. 1975, 12, 115. 1622. Chem. 2009, 6, 95. [54] C. N. Alpers, D. K. Nordstrom, J. W. Ball, Sci. [73] J. Majzlan, A. Navrotsky, R. Stevens, M.