The Lovisa Stratiform Zn-Pb Deposit, Bergslagen, Sweden: Structure, Stratigraphy, and Ore Genesis

Mediumto high-grade metamorphosed, 1.9 Ga, stratiform, syngenetic Zn-Pb ± Ag sulfide deposits constitute an economically important type of ore deposit in the Bergslagen lithotectonic unit of the Fennoscandian Shield. The Lovisa Zn-Pb deposit occurs in a metamorphosed succession of rhyolitic ash-siltstone, rhyolitic mass flow deposits, limestone, and Fe formation, deposited at a stage of waning volcanism in Bergslagen. Accessory graphite, absence of Ce anomalies in shale-normalized rare earth element (REE) data, and absence of hematite in Mn-rich Fe formations stratigraphically below the Lovisa Zn-Pb deposit indicate a suboxic-anoxic depositional environment. The uppermost Mn-rich Fe formation contains disseminated, inferred syngenetic Pb-Ag mineralization with mainly negative δ34S values in sphalerite and galena (–6.1 to –1.9‰). Deposition of this Fe formation terminated during a pulse of explosive felsic volcanism. The Lovisa Zn-Pb deposit is interpreted to have formed in an alkali-rich brine pool developed immediately after this volcanic event, based on lithogeochemical and stratigraphic evidence. The first stage of mineralization deposited stratiform sphalerite mineralization with mainly positive δ34S values (–0.9 to 4.7‰). This was succeeded by deposition of more sphalerite-galena stratiform mineralization with δ34S values close to 0‰ (–2.1 to 1.5‰). The more galena-rich mineralization partitioned strain and was partly remobilized during later ductile deformation. The stratigraphic context, sulfide mineralogy, S isotopes, and alteration geochemistry suggest that the metalliferous fluids and the depositional environment were H2S deficient (S poor or SO4 dominant). The source of S is interpreted to have been a mixture of H2S derived from bacterial and thermochemical seawater sulfate reduction and S derived from leaching of volcanic rocks, with the latter becoming more important over time. Lovisa formed in a setting where basin subsidence was periodically punctuated by the deposition of thick, syneruptive felsic volcaniclastic mass flow deposits. Coeval volcanism was likely important for driving hydrothermal activity and supplying a reservoir of metals and S. However, the high rate of deposition of volcaniclastic sediment in Bergslagen also precluded the establishment of long-lived, deep, and anoxic environments favorable for accumulation of organic matter and H2S. This stratigraphic pattern is common in Bergslagen and may explain why large stratiform Zn-Pb deposits are uncommon in the region and restricted to the uppermost part of the metavolcanic succession, directly stratigraphically beneath postvolcanic pelitic rocks.


Introduction
The Lovisa mine exploits one of the most important styles of base metal sulfide mineralization in the Bergslagen mining district of southern Sweden, consisting of stratiform Zn-Pb ± Ag sulfides interbedded with fine-grained, metamorphosed, volcaniclastic rocks. With a strike extent of at least 1,100 m and a minimum depth of 425 m (Carlon, 1987), the known tonnage of Lovisa (past production and current reserves) is 1.15 million metric tonnes (Mt) grading 9.4 wt % Zn, 5.3 wt % Pb, and 10 to 20 ppm Ag. No description of the deposit exists in the scientific literature, with only unpublished exploration reports produced prior to the onset of mining (e.g., Carlon, 1987).
Lovisa has features in common with the 65-Mt Zinkgruvan deposit, which is the largest stratiform Zn-Pb-Ag deposit in the Bergslagen district. Similarities include the stratiform nature of the ore, a high sphalerite + galena to Fe sulfide ratio, the occurrence of stratiform Zn-Pb ore in a fine-grained succession above a thick metalimestone unit, and the local occurrence of stratabound, dolomite-hosted Cu-Co mineralization nearby. Differences include the smaller size, lower Ag content, and the occurrence of the ore in a succession that also hosts stratiform Fe oxide deposits.
Zinkgruvan has variably been classified as a Broken Hilltype deposit (e.g., Beeson, 1990;Walters, 1996), sedimentary-exhalative (SEDEX) deposit (Hellingwerf, 1996), volcanogenic massive sulfide (VMS) deposit (Billström, 1991), and a VMS-SEDEX hybrid (Hedström et al., 1989). Recently, Jansson et al. (2017) argued that the deposit formed upon exhalation of oxidized saline brines at a near-neutral pH into a reduced (H2S > SO 4 2-) water column above a thick succession of dominantly felsic volcanic rocks. The inferred fluid chemistry differs from typical VMS models, Selwyn-type SEDEX models, and most fluid models for Broken Hill but has similarities to MacArthur-type SEDEX models (cf. Cooke et al., 2000). A continental back-arc setting with a high proportion of felsic metavolcanic rocks deposited in a shallow marine setting in the footwall and the siting of the deposit at just below the transition from rift-phase to sag-phase sediments are, however, features more similar to Broken Hill (Allen et al., 1996;Walters 1996).
In this study, we address the controls on ore formation at Lovisa based on new data from underground workings and recent exploration drilling, remapping of surface outcrops, whole-rock lithogeochemistry, and mineral chemical and isotope analysis (Pb and S). The focus is on constraining the depositional environment, timing and controls on ore formation, and the postore modifications resulting from the tectonic and metamorphic overprint. Due to the high proportion of stratiform Fe oxide deposits in the succession, their characteristics and possible genetic relationships to the stratiform Zn-Pb mineralization are evaluated.

Regional Geology
The Bergslagen mining district is located in the Bergslagen lithotectonic unit of the Fennoscandian Shield (Stephens and Andersson, 2015). The Bergslagen lithotectonic unit is dominated by plutonic rocks emplaced at various stages during the ca. 1.9 to 1.8 Ga Svecokarelian orogeny (Stephens et al., 2009). These intrusive rocks record variable degrees of tectonic and metamorphic overprint during two phases of ductile deformation and mainly amphibolite facies metamorphism at ca. 1.87 to 1.85 (D1 and M1) and ca. 1.84 to 1.79 Ga (D2 and M2; Fig. 1; Stephens and Andersson, 2015;Johansson and Stephens, 2017). The early plutonic rocks intruded a thick succession of supracrustal rocks, deposited at different stages during the evolution of a back-arc basin developed on continental crust (Allen et al., 1996). The supracrustal rocks commonly form inliers bound by plutonic rocks and shear zones (Fig. 1).
The lowermost preserved part of the stratigraphy consists of metasedimentary rocks that record a shallowing upward transition from turbidites to locally cross-bedded quartzite. This is overlain by a thick succession of dominantly felsic metavolcanic rocks that, based on original depositional facies, can be subdivided into a lower succession dominated by pyroclastic flow deposits, coherent lavas, and subvolcanic intrusions and an upper, more stratified succession in which originally silt-to sand-grain-size volcanic ash deposits (ash-siltstone and ash-sandstone of Allen et al., 1996) are interbedded with volcaniclastic mass flow deposits, limestone (now marble), banded iron formations (BIFs), and rare stratiform sulfide mineralization (Allen et al., 1996). Most Fe oxide and base metal sulfide deposits occur in this upper part of the succession, which, in addition to stratiform syngenetic deposits, also hosts base metals sulfide and Fe oxide deposits that replaced limestone (Allen et al., 1996 Jansson, 2011). The metavolcanic succession is overlain by fine-grained metasedimentary rocks. The Lovisa deposit is located in the northernmost part of the NE-trending Guldsmedshyttan syncline (Fig. 2). This syncline hosts many mineral deposits over a strike extent of ca. 45 km including BIF, Fe oxide deposits in marble and skarn 1 , and base metal sulfide deposits (Geijer and Magnusson, 1944;Carlon and Bleeker, 1988;. The syncline was interpreted as an F1 fold by Lundström (1983), who showed that it is isoclinal and doubly plunging with a steeply E dipping axial surface and an overturned eastern limb. Carlon and Bleeker (1988) demonstrated that the syncline was refolded around W-to NW-trending axial surfaces, resulting in a hook-like fold interference pattern northeast of Lovisa (Fig. 3;Bleeker, 1984;Carlon and Bleeker, 1988). The hinge of the F1 syncline has been traced southward to the area east of Striberg, whereas farther south the fold is truncated by NE-trending shears and fault zones (Stephens, 1998). Beunk and Kuipers (2012) suggested that the area is transgressed by a large system of approximately NE trending, sinistral shear zones that run parallel to the marble unit along its entire length (Fig. 2) and extends farther northeast toward Garpenberg (Fig. 1). This shear system was termed the "west Bergslagen boundary zone." No detailed pressure-temperature estimates exist for metamorphism in the Guldsmedshyttan syncline, but regional metamorphism was estimated to have reached amphibolite facies based on the presence of sillimanite, garnet, and cordierite (Lundström, 1983;Stephens et al., 2009).
The lowermost part of the stratigraphy on the northwest limb of the regional syncline consists of a package of feldspar-and quartzphyric, massive, felsic metavolcanic rocks overlain by rhyolitic, quartzphyric, medium-grained metavolcanic rocks (Carlon and Bleeker, 1988), termed the Storsjön Formationby Lundström (1983) (Fig. 4). Lundström (1983) and Stephens et al. (2009) interpreted most of these rocks as subvolcanic intrusions. This metavolcanic succession is overlain by the Usken Formation, beginning with a regionally extensive unit of hematite + quartz-rich BIFs from which Fe ore has been mined at many locations including Blanka, Stripa, Stråssa, and Striberg (Fig. 3). Mining mainly occurred in structurally thickened zones, wherein the Fe content has been upgraded by alteration involving conversion of hematite to magnetite and dissolution of quartz (Geijer and Magnusson, 1944). The footwall metavolcanic rocks to the Stråssa deposit contain porphyroblasts of magnetite, biotite, cordierite, and sillimanite (Carlon and Bleeker, 1988).
In the upper part of the stratigraphy, the syncline hosts one of the most laterally extensive marble units in Bergslagen, with a strike extent of ca. 45 km and a thickness of 50 to 250 m (Fig. 2). The marble is mainly dolomitic in the Lovisa area ( Fig. 3) but regionally ranges from calcitic to dolomitic (Carlon and Bleeker, 1988). Allen et al. (2003) demonstrated that at least the stratigraphic lower few hundred meters of the marble in the Älvlången area has a stromatolitic origin ( Fig. 2). For clarity, herein we refer to the marble unit as the "Håkansboda limestone." The succession between the BIF and the Håkansboda limestone is made of massive, aphyric to quartzphyric, rhyolitic metavolcanic rocks (Carlon and Bleeker, 1988). The transition upward to the Håkansboda limestone is gradational, marked by the appearance of rhyolitic ash-siltstone with interbedded calc-silicate rocks of calcic clinopyroxene and garnet-biotitehornblende with local magnetite and pyrrhotite, followed by the appearance of thin beds of dolomitic and calcitic marble (Carlon and Bleeker, 1988). Such stratiform, bedded calc-silicate rocks ("skarn beds") have elsewhere in Bergslagen been shown to include reaction skarns, formed by regional metamorphism of hybrid rock types containing a variable proportion of interbedded carbonate, variably altered volcaniclastic material, and locally Fe oxides (e.g., Hedström et al., 1989;Jansson and Allen, 2011;Jansson et al., 2013).
The Håkansboda limestone is massive in the lower part but grades upward (Fig. 3) into a succession dominated by rhyolitic ash-siltstone with lesser interbeds of marble, calc-silicate rock (clinopyroxene-amphibole), magnetite-bearing Fe formation, and coarse-grained volcaniclastic rock (Lundström, 1983;Carlon and Bleeker, 1988). The Lovisa Zn-Pb deposit is situated in this part of the stratigraphy in an area hosting several magnetite Fe formations that contain minor galena and Ag minerals. A similar type of Pb-Ag deposit was mined at Guldsmedshyttan, southwest of Lovisa ( Fig. 3; Tegengren, 1924).
The Usken Formation is overlain by the Mårdshyttan Formation, which consists of metamorphosed, originally mainly fine grained, sedimentary rocks that are correlated regionally in Bergslagen with the postvolcanic metasedimentary rocks that overlie the volcanic succession. Due to the depth of erosion, these rocks are not present in the Lovisa area but occur in a basinal culmination of the Guldsmedshyttan syncline south of Lovisa (Fig. 2).
Detailed correlation between individual beds of marble, bedded calc-silicate rock, and Fe formation of the Usken Formation across fold hinges and faults in the Lovisa area has not been successful to date, and various models for the local structure were proposed during mineral exploration in the area (Fig 3;Bleeker, 1984;Engvall and Persson, 1985;Carlon, 1987). A revised model for the structure and stratigraphic correlations is presented in this paper.

Methods
Only assay data and geologic logs remain from earlier exploration campaigns in the Lovisa area. This, along with scarcity of outcrops south of the mine, hindered evaluation of the southern part of the deposit. Our study is focused on underground workings in the currently mined northern part of the deposit. Underground mapping of three mine levels, surface mapping, and logging of new exploration drill holes formed the basis for producing new geologic maps, geologic cross sections, and a 3-D model of the deposit generated in the Leapfrog Geo computer software. Legacy reports provide a detailed lithological subdivision based on core logging of premining exploration drill holes, the most detailed account being a synthesis provided by Carlon (1987).
Whole-rock lithogeochemistry samples were collected from the underground workings, drill core, and surface outcrops.     Fig. 3. Geologic map of the northernmost part of the Guldsmedshyttan syncline, based on data from the current investigation and legacy data from Bleeker (1984), Carlon (1987), and Carlon and Bleeker (1988). Grid is Swedish national grid RT90. Note that all supracrustal rocks are metamorphosed but are here named according to the interpreted premetamorphic precursor. Abbreviations as in Figure 1. FW = footwall, HW = hanging wall.
Drill holes 1501, 1502, and 1507 were systematically sampled at 5-to 10-m intervals. Cleaned samples lacking veins or weathered surfaces were sent to ALS Chemex in Öjebyn, Sweden, for sample preparation. The pulps were sent to Acme Laboratories, Vancouver, Canada, for analysis using their LF202 routine. Major oxides and Ba were determined by inductively coupled plasma-emission spectroscopy (ICP-ES) and refractory trace elements by ICP-mass spectrometry (ICP-MS), both following lithium borate fusion and nitric acid digestion. Zinc and Pb contents >1 wt % were determined by single element ICP-ES following aqua regia digestion. Total S and carbon were determined by the Leco method. Boron was determined by ICP-ES following NaOH fusion at ALS Vancouver, Canada. Analytical quality was monitored with certified reference materials disguised as unknowns. Returned values for all elements differed by less than 5% from the corresponding certified reference values.
Continuous sections through the ore-bearing interval were sampled by Lovisagruvan AB, Stråssa, Sweden, and analyzed for major and trace elements by MS Analytical, Langley, Canada. Sample pulps were fused by borate flux in a hightemperature controlled muffle furnace, and the resulting beads were dissolved in 4% nitric acid. Major oxides were determined by ICP-ES, and refractory trace elements (Ba, La, Lu, Cr, Cs, Ga, Hf, Nb, Rb, Sn, Sr, Ta, Th, U, V, W, Y, and Zr) were determined by ICP-MS. Other trace elements were determined by ICP-ES following aqua regia digestion. Total S and carbon were determined by a Leco carbon and S analyzer. Sulfide-rich samples were quantified by ICP-ES following aqua regia (Ag) or four-acid digestion (Pb, Zn). Where Zn and Pb contents exceeded the upper detection limits of the acid digestions (>10 wt %), these elements were quantified using wet titration techniques. Analytical quality was monitored with method blanks, duplicates, and certified and/or verified in-house reference materials. The uncertainties for the different methods (in % of the reported values) are (1) 5% for titration, (2) ±7.5% for borate fusion and sulfide-rich four-acid, (3) ±10% for total C and S, and (4) ±15% for aqua regia.
Electronic microprobe analyses on polished thin sections were performed using a Jeol JXA-8530F HyperProbe electron probe microanalyser at Uppsala University, Sweden. Standard operating used conditions were 15-kV accelerating voltage and 15-nA beam current. International reference materials were used for calibration and standardization. Analytical precision based on measurements on Smithsonian Institute mineral standards have shown that uncertainties are ≤1.5% standard devisation (s.d.) for oxides at concentrations of >10 wt %, ≤2.2% s.d. for oxides at concentrations of 5 to 10 wt %, and ≤10% s.d. for oxides at lower concentrations (Barker et al., 2015).

Lovisa Zn+Pb+Ag
Rhyolitic ash-siltstone with thin interbeds of limestone and rare calcite-banded iron formation   Lundström (1983). The stratigraphy in the Usken Formation is revised according to results from the current study, whereas the Storsjön and Mårdshyttan Formations are drawn schematically based on Lundström (1983) and Carlon and Bleeker (1988). The color legend is consistent with Figures 3, 5, and 6. Abbreviations: Fsp = feldspar, Hem = hematite, Mag = magnetite, Qz = quartz.

Storsjön fm
The S isotope composition of handpicked mineral separates was determined at the University of Salamanca, Spain, using a conventional method described by Robinson and Kusakabe (1975). Calibration to the Cañon Diablo Troilite (CDT) standard was performed by analysis of a suite of international reference materials (IAEA-S-1, IAEA-S-4, NBS-123, and NBS-127) and internal laboratory standards in conjunction with analysis of the unknowns. The δ 34 S values are reported relative to the CDT standard.
Chemical preparation for the Pb isotope analyses involved dissolution of handpicked galena, followed by electrodeposition of lead on a Pt electrode. A Nu Plasma II multicollector ICP-MS instrument in the Vegacenter facility at the Natural History Museum in Stockholm, Sweden, was used to measure the isotopic ratios. For these analyses, Pb was stripped off from the Pt electrode and then diluted in 0.5 N HNO3 to an estimated concentration of 30 to 50 ppb Pb, with Tl added to allow for an internal correction of mass bias. Measured intensities were corrected for background and Hg interference on mass 204, and the Pb isotope ratios of the unknowns were normalized to 203 Tl/ 205 Tl = 0.418918. The obtained numbers for the standard are within error of those given by Todt et al. (1993), and the external reproducibility is estimated to be between 0.04 ( 206 Pb/ 204 Pb) and 0.08% ( 208 Pb/ 204 Pb).

Structural geology
The Lovisa mine is located on the eastern limb of a moderately SSW plunging anticline, here termed the "Lovisa anticline" (Figs. 3,5). The anticline is part of a parasitic fold with Z symmetry on the western limb of the Guldsmedshyttan syncline (Fig. 3) and is correlated with the first-generation folds (F1) of Carlon and Bleeker (1988). F1 is associated with the most conspicuous NNE-trending tectonic fabric in the mine, making up a steeply E dipping spaced foliation, which is commonly accentuated by later spaced fractures. Mesoscale F1 fold hinges in the mine generally plunge ca. 30° to 40° toward the south-southwest.
Cross folding related to the subsequent F2 phase of folding is subordinate in the mine but is locally evidenced by NNWto W-striking disharmonic folds with a poorly developed foliation and undulating fold axes that refold the more NNEtrending fabrics. A strong stretching lineation plunging ca. 60° toward the south is defined by stretched quartz phenocrysts and fragments in metavolcanic rocks and may correlate with the regional L2 lineation (cf. Stephens et al., 2009).
Several steeply SE to ESE dipping shear zones with associated chlorite-coated fractures caused stacking of the stratigraphy, including the massive sulfide ore bodies. The most important in the mine is here termed the "central shear" (Figs. 5,6). Mineralization displays attenuation, local folding as well as local brecciation, and gouge proximal to these zones. Hinges of folds in ore plunge ca. 30° south-southwest and indicate reverse displacement with a dextral component, despite locally yielding apparent sinistral offsets on plan view (e.g., Fig. 5).
These shear zones and associated folds are subparallel to the axial surface and hinge of the Lovisa antiform, although they display local crosscutting relationships with F1 and F2 structures. The shear zones offset rare, thin, gently dipping pegmatite veins that record no ductile deformation except where adjacent to the shear zones.
The structures described above are crosscut by several subvertical brittle faults trending northwest to north-northeast. These commonly have negligible, mainly dextral displacements seldom exceeding 1 m (Fig. 5). However, a prominent NNW-trending fault, termed the "Lovisa fault," forms a southwest limit to the mine at present, although an offset portion of the ore has been found south of it (Carlon, 1987;Fig. 5). This fault corresponds with an apparent dextral displacement on plan view, although the true displacement vector is unknown.
A prominent NNE-trending fault east of Lovisa was inferred during exploration drilling (Engvall and Persson, 1985) and verified during recent exploration drilling from the mine (Figs. 3, 6). The fault contains both younger clay gouge and older mylonite. In drill hole 1507, the fault is intruded by a dolerite dike with wide chilled margins. Based on its orientation, the dike can be correlated with the Sveconorwegian dolerite dike suite, which was emplaced at around 0.98 to 0.95 Ga (cf. Söderlund et al., 2005). The dike was unaffected by mylonitization but was altered to clay gouge on the eastern contact. Displacement along the fault is at least 500 m based on stratigraphic correlation of the Håkansboda limestone and an overlying magnetite occurrence (Fig. 6).

Stratigraphy
The current study focuses on the stratigraphic interval between the upper part of the Håkansboda limestone and the Mårdshyttan Formation (Figs. 4,6). All rocks in the area, excluding dolerite dikes and pegmatites, were subjected to regional metamorphism. Descriptive premetamorphic rock names are used for brevity and to emphasize stratigraphic features, except for calc-silicate rocks (skarns) and strongly altered rocks where precursors are uncertain. These are named according to their current metamorphic mineralogy. Whole-rock lithogeochemical data for the volcaniclastic units are provided in Table 1, and data on representative samples of Fe formations are provided in Table 2.
Stratigraphic footwall: Ductile deformation and lack of drill holes intersecting the lower contact of the Håkansboda limestone prevent determination of true stratigraphic thickness for this unit, although it is estimated to be at least 200 m thick. The marble is dolomitic, most commonly massive, and hosts centimeter-sized, dark-green crystals of dark-green calcic clinoamphibole (Fig. 7A). Subordinate interbeds of rhyolitic ash-siltstone occur locally. Biotite-tremolite schists commonly formed at the dolomite-siltstone interfaces and, where the interbeds are thin, completely replaced them.
The stratigraphic upper 20 m of the Håkansboda limestone consists of a zone of interbedded pink-red rhyolitic siltstone, calcitic limestone, and bedded calc-silicate rock (epidoteactinolite ± quartz). These rocks grade upward into an interval of massive magnetite with a gangue of dolomite, calcite, actinolite, serpentine, and chlorite (Fig. 6). Minor pyrite, pyrrhotite, and chalcopyrite are locally present and are completely mantled by magnetite. Actinolite crystals of variable molar Mg/Mg + Fe (0.76-0.88) form complex intergrowths with serpentine, commonly occurring as rounded aggregates rimmed by magnetite in a carbonate matrix. These intergrowths most likely reflect retrograde alteration products after former Banded-laminated sphalerite (Sphalerite ore) Sphalerite-galena ball ore (Main ore) Py-impregnated grey-red rhyolitic silt-/sandstone  olivine porphyroblasts (Fig. 8A). This Mg-rich magnetite Fe formation can be correlated with the magnetite occurrence previously mined at the Lekeberg mine north of Lovisa ( Fig.  3) and is herein referred to as the Lekeberg Fe formation. It can also be correlated with magnetite Fe formation directly stratigraphically above the Håkansboda limestone east of the Engvall fault (Fig. 6). The Lekeberg Fe formation is overlain by a 20-to 50-m succession of gray, weakly stratified rhyolitic siltstone and finely feldspar + quartzphyric sandstone with subordinate bedded calc-silicate interbeds of calcic clinoamphibole ± magnetite with euhedral-subhedral garnet porphyroblasts. The feldspar + quartzphyric texture is locally concealed by millimeter-sized metamorphic porphyroblasts of pale-pink garnet and sillimanite needles (Fig. 7B).
The lower gray rhyolitic ash-siltstone is overlain by a 50to 70-m-thick unit of coarse-grained volcaniclastic rock. The lower half of the unit is a massive, weakly feldspar + quartzphyric sandstone, whereas the upper half contains abundant phyllosilicate-altered former pumice clasts (Fig. 7C). These clasts contain 5 to 10 vol % 1-mm feldspar phenocrysts and 1 to 4 vol % 1-to 3-mm quartz phenocrysts, whereas the matrix contains fewer crystals. The larger pumice clasts are aligned along a foliation that is subparallel to bedding and overprinted by S1 (Fig. 8C). Locally, the unit is considerably more crystal rich, containing 20 to 25 vol % 2-mm white feldspar and 5 to 10 vol % 1-to 2-mm blue-gray quartz crystals in a gray fibrous groundmass with tiny pumice fragments and ca. 10 vol % biotite. Few stratigraphic younging indicators have been observed except in drill hole 1502, where thin normal    Abbreviations: ICP-ES = inductively coupled plasma-emission spectroscopy, ICP-MS = inductively coupled plasma-mass spectrometry 1 Half of the lower detection limit was used during the calculation of median and MAD when element concentrations were below detection limit 2 Only AcmeLabs samples at <2 wt % Zn  7. Photographs of rock types in the stratigraphic footwall to the stratiform Zn-Pb ore. A: Massive dolomite marble with minor dark-green actinolite. B: Dark-gray finely feldspar + quartzphyric rhyolitic sandstone with minute porphyroblasts of sillimanite and garnet. C: Sericite-altered and compacted large pumice clasts (fiamme) in the footwall feldspar + quartzphyric rhyolitic pumice breccia. The former pumice clasts are aligned parallel to S0 and are crenulated by S1. D: Banded magnetiteactinolite-epidote-garnet rock from the lower Lovisa iron formation. E: Massive K-feldspar + quartzphyric rhyolite. F: Pink rhyolitic ash-siltstone with interbands of sericite-chlorite schist parallel to S0. The sericite-chlorite schists have a subparallel foliation defined by alignment of mica, which is crenulated by a spaced S1 foliation, which is here clearly visible in the ash-siltstone. The mica bands are rich in ca. 2-to 3-mm-round quartz phenocrysts. Pink rhyolitic ash-siltstone with streaks and blotches of clinochlore-pyrite-K-feldspar. The pink ground mass also carries tiny crystals of pyrite. G: Pink rhyolitic ashsiltstone with pyrite-porphyroblastic clinochlore patches.
The Lovisa Fe formation can be subdivided into a lower and an upper part, separated by an interbed of gray, garnetporphyroblastic rhyolitic siltstone. A variable content of magnetite, ferrohornblende, calcic clinopyroxene, biotite, quartz and garnet, minor epidote, and microcline and trace apatite defines a banding parallel to bedding, which is less conspicuous in thin section (Fig. 8B). The biotite is rich in Ba (3.18 wt %) and Cl (1.56 wt %). Barite has not been observed. The amphibole displays alteration to micaceous meshes, but the amphibole crystal shapes are retained, indicating that the alteration was postkinematic and retrograde. Garnet occurs as up to centimeter-sized porphyroblasts (Sps30-32Alm37-39Grs20-24Adr7-9Prp1-2) and as anhedral, skeletal crystals (Sps38-42Alm18-23Grs22-26Adr11-17Prp2-3) arranged along grain boundaries and forming rims around magnetite crystals (Fig.  8B). Despite slight compositional variations, there are no systematic zonation patterns in garnet; the compositional anhedral garnet reflects variations between different crystals in the same sample. Bands of massive magnetite in the lower part of the Lovisa Fe formation contain abundant disseminated galena with minor sphalerite and accessory Ag tetrahedrite, pyrite, chalcopyrite, and pyrrhotite (Fig. 8C). Overall, the Fe sulfide content in the Lovisa Fe formation is low.
East of the Engvall fault, the Nya Backadahl magnetite deposit (Fig. 3), which is interpreted as stratigraphically equivalent to the Lovisa Fe formation, is interbedded with a magnetite-free graphitic pelite with thin bands of sphalerite and galena. Electron microprobe analyses by Bleeker (1984) identified minor Ag tetrahedrite, allargentum, and dyscrasite.
The upper part of the Lovisa Fe formation is more distinctly banded and is immediately overlain by a meter-wide interval of very finely laminated gray siltstone, wherein laminae are defined by variable proportions of muscovite, quartz, microcline, plagioclase, and accessory minute tourmaline crystals. East of the Engvall fault, the equivalent interval hosts a graygreen, fine-grained calcite-altered mafic unit not present at Lovisa. It has not been possible to determine whether it is intrusive or extrusive, and the paucity of data in this area precludes determination of its distribution and geometry.
The laminated gray siltstone is overlain by rhyolitic sandstone, varying from nearly aphyric to containing ca. 3 to 4 vol % diffuse 1-to 2-mm feldspars and a few stretched 1-to 3-mm quartz phenocrysts (Figs. 7E, 8D). The metamorphic grain size is on average ca. 0.1 mm and the main minerals are quartz, feldspars, muscovite, and chlorite.
Lenses of sericite schist up to 0.5 m wide with abundant 2-mm quartz phenocrysts occur locally (Fig. 7F). The lenses are predominantly concordant but locally crosscutting. They may constitute either altered clasts of quenched rhyolite (peperite) or large pumice clasts (cf. Allen et al., 1996). The latter interpretation is favored, based on lack of evidence for coherent rhyolitic rocks in this part of the Guldsmedshyttan syncline.
Minute (ca. 0.5-1 mm) pyrite porphyroblasts are common in the rhyolitic sandstone unit and increase in abundance stratigraphically upward toward the Lovisa Zn-Pb deposit, where the unit gradually fines to silt size and attains a distinct pinkred color. This unit is almost exclusively composed of feldspar and quartz in addition to minor chlorite and local 1-to 3-mm magnetite porphyroblasts. The feldspar is mainly albite-oligoclase, although there are local zones with complexly growthzoned celsian or barian microcline, which is intergrown with pyrite and clinochlore (pyknochlorite-ripidolite) in disseminations, patches, and veins (Figs. 8E, 9A, B). Celsian occurs as inclusions in pyrite (Fig. 8E). Barite has not been observed.
Stratigraphic hanging wall: The Lovisa Zn-Pb deposit is overlain by a succession of gray-pink ash-siltstone (Fig. 10A), which is dominated by plagioclase and quartz. Lithogeochemical data suggests that the precursor rock was of rhyolitic composition and that the plagioclase is largely secondary in origin. The ash-siltstone is sharply overlain by a ca. 80-m-thick, finingupward succession of coarse-grained rhyolitic volcaniclastic rocks (Fig. 10B). Sharp and gradational transitions in grain size suggest the succession consists of several normal graded subunits. The lowermost unit displays normal grading over ca. 10 m from basal breccia with lithic clasts of rhyolitic ash-siltstone, to rhyolitic pumice breccia with 10 to 15 vol % 2-mm feldspar and 1 to 3 vol % 1-to 2-mm quartz crystals, to crystal-bearing sandstone and uppermost, indistinctly laminated rhyolitic siltstone. The overlying units contain similar matrix-supported pumice clasts but contain 2-to 4-mm quartz phenocrysts (Fig.  10B). The phenocrysts are mostly confined to former pumice clasts. Biotite is the mafic phase and is commonly variably chloritized. At the base, plagioclase is the only feldspar, but the proportion of microcline increases stratigraphically upward.
A gradational transition is observed from the coarse-grained succession into an overlying massive gray to pink rhyolitic siltstone, with rare thin interbeds of banded limestone. Approximately 150 m above the Lovisa ore horizon, a thin BIF is present, consisting of centimeter-thin bands of magnetite regularly interbedded with white calcite limestone, chloriterich bands, and massive, red, jasper-like rhyolitic ash-siltstone (Fig. 10C). The rhyolitic ash-siltstone is dominated by quartz and microcline, and the red coloration is likely due to the presence of submicron hematite crystals. Magnetite is mainly present in the stratigraphically lower part of this subunit but gradually diminishes stratigraphically upward, whereas the interbedding of limestone and red siltstone prevails. Minor amphibole, epidote, and garnet formed at the contacts between compositionally distinct bands and as rims around feldspar crystals in carbonate-rich matrix (Fig. 8F), but the calc-silicate content is generally low except in and around faults (Fig. 10C). The garnet is predominantly andradite rich (Adr68Grs19Sps9Alm4) and ranges from anhedral to subhedral. The andradite-rich garnet is discontinuously overgrown by more alumina-rich garnet (Adr19Grs31Sps27Alm21).

Mineralization
Zinc ± lead mineralization is confined to a 5-to 10-m-thick stratigraphic interval that contains sphalerite laminae, sphalerite-galena veins, and sphalerite-galena with a durchbewegung texture in strongly banded, gray to pink-red, cherty rhyolitic ash-siltstone with lesser interbeds of clinochlore schist (pyknochlorite-ripidolite). The term "durchbewegung texture" refers Fig. 9. Backscattered electron images: A: Clinochlore (Chl)-celsian (Cls) alteration patch in pyritic rhyolitic silt-sandstone (same sample as in Figure 8E). B: Close-up of area in box in A, revealing complex zonation patterns in celsian, defined by variable contents of Ba. Sample LK20150081. The difference in gray mainly reflects difference in the Ba content in the feldspar.
to highly ductile deformed, very fine grained massive sulfide mineralization with subrounded to rounded clasts of vein quartz and host rock (Marshall and Gilligan, 1987).
Mining specifically targets two, 0.1-to 1-m-thick, beddingconcordant zones of massive to semimassive sulfide. The lower "sphalerite ore" consists of bedding-parallel layers of sphalerite-dominant sulfide interbedded with rhyolitic ashsiltstone and chlorite schist (Fig. 11A), whereas the upper "main ore" is a tabular body of massive galena-sphalerite accompanied by peripheral laminated sphalerite + galena crosscut by sphalerite + galena veins. Pyrrhotite is rare in both ore types. The calc-silicate content in this stratigraphic interval is very low. The main ore is present throughout the mine whereas the sphalerite ore only occurs in the southern part. The sphalerite is commonly yellow to beige in color and Carlon (1987) reported Fe contents are relatively low, varying from 1.6 to 2.2 wt % and averaging 2.0 wt %.
Rhyolitic interbeds in the sphalerite ore are commonly boudinaged and define Z-symmetric folds (when viewed downward). The folds are locally disharmonic as the contacts may be planar and lack folding in the same exposure, suggesting that they result from layer-parallel shearing (Fig. 11A). A thin zone of strongly pyrite impregnated chlorite schist interbedded with pink rhyolitic siltstone is locally present at the lower contact of the sphalerite ore (Fig. 11B). Pyrite porphyroblasts locally form S 0-parallel trains. The sphalerite ore has been traced to the hinge and western limb of the Lovisa anticline, where it is discontinuous and less endowed in base metals (Fig. 5). It gradually thins out and disappears at the north end of the mine.
The gangue minerals in the sphalerite ore are mainly K-feldspar, clinochlore (pyknochlorite-ripidolite), quartz, muscovite, with minor clinozoisite, magnesiohornblende, calcite, millimeter-sized garnet porphyroblasts (Alm36-41Sps28-33Grs21-29Prp3-4Adr0-2), and accessory tourmaline and allanite (Fig. 11C). The compositional range in garnet reflects a core-to-rim zonation, involving decreasing spessartine and almandine and increasing grossular. Although Carlon (1987) reported diopside in the ore zone, no pyroxene was found in the ore zone during the current study. Minor porphyroblastic overgrowths of pyrite and arsenopyrite on sphalerite produce a poikiloblastic texture and were seemingly unaffected by a tectonic foliation defined mainly by micas. Stratification is defined by interbedded rhyolitic ash-siltstone and yellow to beige sphalerite, millimeter-scale variability in sphalerite abundance, and locally by trains of pyrite porphyroblasts.
The main ore is commonly characterized by 1-to 10-mmround fragments of wall rock and dark-gray vein quartz in a fine-grained, gray, sphalerite-galena matrix, i.e., a durchbewegung texture (Fig. 11D, E). Gangue minerals include K-feldspar, chlorite, accessory graphite, and rare tourmaline. The main ore is generally subparallel to S 0 and its strike similarly curves from northeast to a north-northeast direction when traced from north to south in the mine (Fig. 5). The main ore dips 60° to 70° toward the east-southeast but flattens out at depth (Figs. 6, 9D, F). A peripheral zone of laminated, yellow sphalerite occurs adjacent to the ore with a durchbewegung texture (Fig. 11F). The laminated sphalerite facies locally displays complex sheath folds (0.5-1 m wide) that are truncated by ore with durchbewegung texture. Veins of coarsely crystalline sphalerite and galena peripheral to the ore with durchbewegung texture locally crosscut the laminated facies.
The composition of the two ore layers, based on representative ore grade samples, is provided in Table 3. All varieties of Lovisa ore are characterized by a dominance of Zn over Pb yet vary in terms of their Zn/Pb ratio (Fig. 12). The sphalerite ore contains minor Cu and Pb (Zn/Pb averages 43), reflecting a low content of chalcopyrite and galena. It is compositionally much more homogeneous than the main ore. The ore with durchbewegung texture has heterogeneous Zn and Pb contents and exhibits significantly lower Zn/Pb (around 3.8) than the sphalerite ore. The laminated main ore facies is similarly highly variable in composition, but it is always more Pb rich than the sphalerite ore (Fig. 12). The various ore facies in the main ore are also generally more Ag rich, even though Ag/Pb is higher in the sphalerite ore. Data are insufficient to model the metal zonation in the two main massive sulfide beds in detail. However, if the Lovisa Zn-Pb ore zone is considered in bulk, the gradational disappearance of the sphalerite ore from the southern to the northern part of the mine means that there is a general decrease in the sphalerite content from south to north. All types of ore in the Lovisa Zn-Pb deposit have distinctly higher Zn/Pb than the Pb-Ag mineralization in the lower Lovisa Fe formation and the correlative Nya Backadahl Fe formation (Fig. 12), and they lack magnetite.

Hydrothermal alteration
The stratigraphic footwall between the Håkansboda limestone and the pyrite-porphyroblastic pink-red rhyolitic siltstone-sandstone is commonly weakly sericite-chlorite altered and locally silicified (Fig. 13A). Minor porphyroblasts of garnet, amphibole, and sillimanite are common in this part of the stratigraphy and are similar in appearance to metamorphic porphyroblasts formed by regional metamorphism of hydrothermally altered volcanic rocks in many other parts of Bergslagen (e.g., Jansson et al., 2013;Kampmann et al., 2017) Most samples display anomalously high K2O/Na2O compared with the igneous spectrum of Hughes (1973) (Fig.  13B) (Large et al., 2001). This shows that relative to other parts of the Lovisa stratigraphy, the deeper footwall rocks are slightly more chloritized. Furthermore, these rocks are the only ones that display a clear association between silicification and an increase in AI and CCPI (Fig. 13A, C).
Increasing albite and whole-rock Na 2O and decreasing K2O/ Na2O define a zone of Na-enrichment stratigraphy upward toward the red pyritic siltstone (Fig. 13A). This pattern continues in the immediate stratigraphic hanging wall to the Lovisa ore zone, where Na enrichment extends into the lower part of the hanging-wall rhyolitic pumice breccia. An identical pattern of Na enrichment in the ore-equivalent stratigraphic interval also occurs on the western limb of the Lovisa anticline. Thus, this alteration style is stratabound, restricted to the Lovisa ore zone, and must predate folding. In detail, the massive sulfide layers occur in discrete intervals of high K2O/Na2O, reflecting the presence of K-feldspar within this broader alkali-altered zone (Fig. 14A). Many of the samples of red pyritic rhyolitic ash-siltstone directly stratigraphically below the Lovisa Zn-Pb deposit have also been silicified, as evidenced by anomalously high SiO2 contents ( Fig. 13A; Table 1).
In the upper hanging-wall rhyolitic ash-siltstones, silicification is generally patchy in nature and commonly developed as bleached patches, locally containing a few garnet porphyroblasts. Similar to the deeper part of the stratigraphic footwall, many samples have anomalously high K2O/Na2O (Fig. 13B), yet here there is no obvious positive correlation between K2O/ Na2O and SiO2.

Chemostratigraphy
The hydrothermal alteration overprint of the rocks precludes using the contents of major oxides for further stratigraphic subdivision (e.g., Fig. 14A). Plotting ratios of the commonly immobile Zr, Al2O, and TiO2 across stratigraphy reveals that the mapped stratigraphic intervals contain distinct negative and positive excursions in these ratios (Fig. 14B-D). These distinct compositional breaks corroborate the stratigraphic division. The coarser-grained volcaniclastic units are compositionally homogeneous, which most likely reflects their origin as rapidly emplaced, syneruptive mass flow deposits of pyroclastic debris derived from discrete volcanic events. In binary plots of Zr, Al2O3, and TiO2, samples of these coarsegrained units lie along linear arrays anchored at the origin ( Fig. 15A-C). These arrays are "alteration lines" in the sense of Barrett andMacLean (1994, 1999), i.e., trends resulting because Zr, Al2O3, and TiO2 were immobile during alteration but decreased or increased in abundance due to a net mass gain or mass loss, respectively, of mobile components.
The narrow range in immobile element ratios indicates homogeneous original composition of the mass flow deposits.
The rhyolitic sandstone unit directly stratigraphically below the ore zone is similarly compositionally homogeneous. Based on this observation and textural evidence, it is likely that this unit is also associated with a discrete rhyolitic volcanic pulse in the basin. Consequently, at least three compositionally distinct rhyolitic volcanic units are present. These three volcanic units have overlapping Al2O3/TiO2 but can be distinguished based on Zr/Al2O3 and Zr/TiO2. The volcanic unit directly underlying Lovisa has the highest Zr/TiO2 and Zr/Al2O3.
The immobile element ratios of the rhyolitic ash-siltstones are more compositionally heterogeneous and only partly overlap with those of the main volcanic units. The larger spread most likely reflects the more reworked character of these units and possibly indicates that they are composed of a distal ash component of variable provenance (cf. Allen et al., 1996).
Except for the Lovisa Fe formation, Zn contents are generally low (<100 ppm) below the massive rhyolitic sandstone of the stratigraphic footwall (Fig. 16A). The massive rhyolitic sandstone and the overlying pyritic red rhyolitic siltstone have slightly elevated Zn contents relative to other metavolcanic rocks in the succession. A sharp increase in Zn content is not encountered before the appearance of the sphalerite ore. Similar weak Zn anomalies extend up to the hanging-wall rhyolitic pumice breccia, above which Zn values return to <100 ppm levels, with the exception of a thin marble interbed below the hanging-wall BIF. The distribution of Pb and S follows a broadly similar pattern (Fig. 16B, C).
The Lovisa Zn-Pb deposit occurs in the only stratigraphic interval without carbonate, bedded calc-silicate rocks, or Fe oxide that shows positive Mn anomalies. However, the Fe formations, associated calc-silicate rocks, and limestone units display more pronounced positive anomalies (Fig. 16D). There are similarly discrete positive anomalies in P2O5 and V associated with the ore zone and the Fe formations (Fig. 16E, F).
Barium is generally <1,000 ppm except in the Lovisa Fe formation and in the immediate footwall of the ore zone (Fig.  16G). Sporadic B anomalies exist in the stratigraphic footwall up to the Lovisa Fe formation (Fig. 16H), but above this B is mostly below the 10-ppm lower limit of detection, despite the presence of trace tourmaline in the sphalerite ore.

Sulfur isotopes
Sulfur isotope data from the Lovisa Fe formation and the Lovisa Zn-Pb deposit are shown in Table 4 and Figure 17. The lowest δ 34 S values are for galena and sphalerite that occur disseminated in the Lovisa Fe formation (δ 34 S = -6.1 to -1.9‰). Pyrite in a vein crosscutting a ductile fabric in semimassive magnetite gives a slightly higher value (δ 34 S = 1.1).
Two analyses of pyrite porphyroblasts from the stratigraphic footwall of the Lovisa Zn-Pb deposit have higher δ 34 S of 3.3 and 4.8‰. Pyrite from the sphalerite ore has similar values (δ 34 S = 1.8-5.4‰), with the sphalerite having marginally higher δ 34 S values than that in the Lovisa Fe formation (δ 34 S = 1.4-4.7‰). Galena has the lowest δ 34 S value (-0.9‰).
Fine-grained, yellow sphalerite in the laminated sphalerite facies of the main ore has lower δ 34 S values of 1.0 to 1.5‰,  whereas coarser, paragenetically later red sphalerite has an even lower value of -0.2‰. The latter falls within the range of δ 34 S values for sphalerite and galena from a directly adjacent vein (-1.3 to 0.5‰) and the ore with durchbewegung texture (-2.1 to 0.2‰). A local zone of semimassive sphalerite and pyrite in the central shear on the western limb of the Lovisa anticline was sampled at the 73-m level of the mine and has δ 34 S values of -2.8 and 0.0‰ respectively. Stratigraphic analysis suggests that this mineralization represents a highly sheared and attenuated equivalent of the ore zone on the eastern limb. Due to the high degree of deformation, this mineralization cannot be placed into the detailed stratigraphic scheme of the eastern limb, so these data points are not shown in Figure 17.
In summary, the highest δ 34 S values in the Lovisa deposit occur in the sphalerite ore. Slightly lower δ 34 S values occur in the laminated sphalerite facies of the main ore, possibly indicating a general decrease in δ 34 S stratigraphically upward in the Lovisa ore zone. Furthermore, isotopic compositions of sulfides from texturally and structurally late parts of the main ore (veins and ore with durchbewegung texture) have slightly lower δ 34 S, which also indicates a possible decrease in δ 34 S with time, although the limited amount of samples and considerable overlap between them must be emphasized.   Romer and Wright (1993) for galena in the Guldsmedshyttan Pb-Ag deposit. However, Romer and Wright (1993) obtained considerably more heterogeneous values involving more radiogenic Pb from chalcopyrite, sphalerite, and glaucodot ([Co,Fe]AsS) from the Håkansboda Cu deposit northeast of Lovisa (Fig. 3).

Structural evolution
The structural relationships at Lovisa are consistent with Carlon and Bleeker's (1988) model for an F1-F2 fold interference pattern. The pre-S1 foliation defined by altered and compacted former pumice clasts is interpreted as a diagenetic compaction foliation, and no tectonic structures related to this foliation have been observed. The Engvall fault crosscuts the F1-F2 fold interference pattern, so it is most likely a post-F2 structure, despite being subparallel to the axial surface of the Guldsmedshyttan syncline.
The dolerite dike at Lovisa shows no ductile fabric and cuts the Engvall fault at an oblique angle. Where the dike and fault intersect, the dike follows the fault trace for about 1 km, suggesting that the dike is younger and intruded along the fault. Clay gouge along the margins of the dike implies fault movement after dike emplacement at ca. 0.98 to 0.95 Ga, thus indicating multiple phases of movement.
The relict laminated sphalerite facies at the margins of the main ore suggests that the ore with durchbewegung texture    Fig. 16. (Cont.) developed by deformation of a preexisting sulfide bed rather than by injection of sulfides remobilized from elsewhere. This is also consistent with the Pb isotope ratios that are similar to those of most 1.89 Ga sulfide deposits in the Bergslagen district (cf. Romer and Wright, 1993;Sundblad, 1994;Jonsson and Billström, 2009). However, the slightly more radiogenic sample 235B KUL N plots along the trend for more evolved Svecofennian Pb observed in western Bergslagen by Jonsson and Billström (2009) and may indicate an addition of secondary Pb in conjunction with formation of the durchbewegung texture.
A higher initial galena content in the main ore relative to the sphalerite ore could explain why the durchbewegung texture developed mainly in the former. This is due to the refractory strength of sphalerite and pyrite, which is higher than that of galena (e.g., Marshall and Gilligan, 1987), whereby an originally more galena-rich bed would have been more favorable for partitioning strain.
It is clear from crosscutting relationships that the durchbewegung texture developed prior to the last offsets along the central shear, the Lovisa fault, and the Engvall fault (Fig. 11F). This is exemplified by minor drag folding of the ore with durchbewegung texture against the central shear. Furthermore, there is no relationship between the intensity of strain in the main ore and proximity to the central shear, since the main ore and shear zones can be separated by relatively well preserved laminated sphalerite ore (Fig. 11F). A young age for the central shear is further suggested by the fact that it crosscuts shallow-dipping pegmatite veins, which cross S1 at a high angle.
Thus, we interpret the durchbewegung texture to have formed along the eastern limb of the Lovisa anticline late during D1 at ca. 1.87 to 1.85 Ga, when strain was partitioned to the most incompetent bed in the succession. The adjacent vein facies may have formed either as piercement veins peripheral to this bed in conjunction with shearing or by later remobilization at the transition to more brittle deformation. The observed offsets of the main ore across the central shear may have similarly resulted from later reactivation of the same structure under more brittle conditions.

Stratigraphic evolution
Deposition of massive to planar, stratified rhyolitic ashsiltstone followed formation of the Håkansboda limestone. Allen et al. (1996Allen et al. ( , 2003 interpreted the rhyolitic ash-siltstone facies in Bergslagen to be reworked pyroclastic ash, deposited below storm wave-base by dilute turbidity currents and suspension settling, distal to the source volcanoes. The common planar stratification and normal grading at Lovisa are consistent with this interpretation. The presence of this facies indicates that broadly contemporaneous, explosive felsic volcanism occurred somewhere in the Bergslagen district (cf. Allen et al., 1996). The Lovisa Fe formation and the Lovisa stratiform Zn-Pb ore are sandwiched between two compositionally distinct intervals of coarser-grained, rhyolitic, pyroclastic debris (Fig.  6). These deposits record discrete felsic volcanic events that punctuate the ambient sedimentation of rhyolitic ash-siltstone in the basin (cf. Allen et al., 1996).
The low Zr/TiO2 and Al2O3/TiO2 in the intervals hosting sulfide mineralization are anomalous relative to volcaniclastic rocks in the succession (Figs. 14, 15). The same geochemical anomalies in Zr/TiO 2 and Al2O3/TiO2 are evident in the Fe formations and limestone, that is, in other units formed at times when the sedimentation rate of rhyolitic ash was low (Fig. 14).
The lower values of Zr/TiO2 could be caused by the ash component attaining a more intermediate composition, cosettling of rhyolitic ash with mafic or intermediate ash, or cosettling of rhyolitic ash and marine clays. The Sc/Th ratio in the stratiform Zn-Pb mineralization is mostly consistent with a clay component (Fig. 18;cf. McLennan et al., 1993). This would imply that chemical, biogenic, and hydrothermal sediments, including sulfide mineralization, could only accumulate when the basin was relatively starved of volcaniclastic sedimentation, i.e., when the fine-grained detrital component became more enriched in Al and Ti (Fig. 18).
Based on the pattern in the Zr/TiO2 vs. Sc/Th plot (Fig.  18), two samples of massive magnetite and associated bedded calc-silicate rock from the Lovisa Fe formation display Sc/Th greatly exceeding that which can be explained by cosettling with a normal clay component, thus indicating a possible presence of mafic to intermediate volcaniclastic material.

Genesis of the Fe formations
Except for the thin calcite BIF in the hanging wall of the Lovisa Zn-Pb deposit, all Fe formations above the Håkansboda limestone belong to the Mn-rich subtype of Fe oxide deposits in Bergslagen. On a plot of Fe/Ti vs. Al/(Al + Fe + Mn), samples define a trend between Fe oxide deposits and associated calcsilicate rocks on one end and the volcaniclastic rocks on the other end (Fig. 19A). This implies that the precursors of the bedded magnetite-and calc-silicate-rich rocks are mixtures of detrital volcaniclastic/sedimentary material and a hydrothermal Fe (and Mn) component and that increasing Al/(Al + Fe + Mn) suggests an increasing detrital component (cf. Peter et al., 2003). However, presence of a sedimentary carbonate component can also be inferred based on the calc-silicate Most samples from the stratiform Zn-Pb sulfide ore zone plot in the area between the Lovisagruvan volcaniclastic rocks and pelitic rocks. This is consistent with these rocks containing a mixture of rhyolitic ash and a terrigenous clay component, although a contribution from mafic volcaniclastic material cannot be ruled out. Two samples from the Lovisa Fe formation, however, have Sc/Th lower than normal pelitic rocks, implying a contribution from more maficintermediate volcaniclastic material. Bergslagen mafic rocks based on data in Johansson and Hålenius (2013), Bergslagen pelite based on data in Jansson et al. (2017), N. Jansson (unpub. data, 2017), and Post-Archean Australian Shale values from Taylor and MacLennan (1985).  mineralogy and minor interbedded carbonate in, e.g., the Lovisa Fe formation. This complicates interpretation, since carbonates and Fe-bearing minerals cause similar shifts on the mixing line (Fig. 19A). Shale-normalized rare earth element (REE) data can yield substantial information on the genesis of Fe formations. However, the REE pattern reflects a mixture of detrital and hydrothermal material in varying proportion, and the REE pattern of samples with a low hydrothermal component will be dominated by that of the detrital component Bekker et al., 2010;Jansson and Allen, 2011). The Fe formations in the Lovisa area have Al2O3 contents of 2.34 to 6.94 wt %, whereby they invariably contain a detrital volcaniclastic or pelitic sedimentary component.
Exchanging Fe/Ti for SREE/Zr shows that the hydrothermal component in the lower Lovisa Fe formation must contain REEs, since SREE/Zr in the Lovisa Fe formation increases with decreasing Al/(Al + Fe + Mn) to levels that cannot be explained by admixture with carbonates (Fig. 19B). Sample LK20160072 has the lowest measured Al 2O3 content (3.03 wt %) in the Lovisa Fe formation and should best reflect the REE signature of the hydrothermal end member. Normalization of data to Post-Archean Australian Shale (PAAS) yields an irregular pattern in which the main features are enrichment in La and Gd-Y, positive Gd and Y anomalies, and a negative Eu anomaly (Fig. 19C). Using the same approach, the Nya Backadahl Fe formation displays a broadly similar pattern, supporting the correlation (Fig. 19C). The Lekeberg sample is similar in behavior of La and Gd to that in the REE profile, yet differs in a less distinct positive Y anomaly (Fig.  19C). The discrimination of the positive La anomaly from a negative Ce anomaly is achieved by plotting the samples in the Ce/Ce* vs. Pr/Pr* plot of Bau and Dulski (1996), which suggests that no true Ce anomalies are present (Fig. 19D).
Even the samples with lowest Al/(Al + Fe + Mn) contain a detrital component, which impacts the pattern of the normalized REE data. A mass-balance calculation to isolate the hydrothermal component cannot be done, since it would require a homogeneous composition of the detrital component that is identical to that of adjacent detrital rocks. This is not supported by immobile element data ( Fig. 14A-C). The nature of the hydrothermal component can, however, be further evaluated qualitatively by using the mixing relationship outlined in Figure 16A and B. This approach reveals that Y/ Ho shifts from the charge and radius controlled (CHARAC) field toward suprachondritic values (Y/Ho > 27.7; Bau, 1996) with decreasing Al/(Al + Fe + Mn), approaching the Y/Ho ratio of modern seawater (Fig. 19E). The Y/Ho values of the Fe formations are considerably higher than those of hydrogenetic marine ferromanganese crusts (cf. Bau, 1996). A similar, albeit more scattered, trend is observed for Eu/Eu*, yet the majority of the samples project toward Eu/Eu* between 0.8 ± 0.2 with decreasing Al/(Al + Fe + Mn). Thus, positive Eu anomalies are mainly lacking in the hydrothermal endmember component (Fig. 19F). A similar trend is observed for Gd/Gd*, with most Fe oxide samples projecting toward an end member Gd/Gd* =1.3 ± 0.1 (Fig. 19G). Samples of carbonate rock yield consistently higher Gd/Gd* for a given value of Al/(Al + Fe + Mn).
Given the elevated P content of the Fe formations and the Lovisa Zn-Pb deposit relative to adjacent rocks, it is likely that a considerable fraction of the REEs in these intervals are hosted by apatite. The high P in the Fe oxide deposits (up to 0.2 wt % P2O5) and the Lovisa Zn-Pb deposit (0.02-0.07 wt % P2O5) can be further evaluated by plotting TiO2/P2O5 vs. Al/ (Al + Fe + Mn). In the case where the elevated P content reflects presence of a mafic detrital component, the TiO 2/ P2O5 should not be lower than that in an end-member mafic volcanic rock. This is the case for the Lovisa Zn-Pb deposit, whereas in the Lovisa Fe formation anomalously low TiO2/ P2O5 suggests that P2O5 must be derived from a nondetrital component (Fig. 19H). Phosphorous in seawater is typically scavenged by ferric oxyhydroxides suspended in the water column (e.g., Peter, 2003).
The stratiform Fe formations of the Lovisa area differ from other Mn-rich Fe formations with associated base metal sulfide mineralization in Bergslagen in the presence of positive Gd anomalies and the lack of positive Eu anomalies (cf. Jansson, 2011; Jansson et al., 2013). The fact that the Fe formations are otherwise compositionally similar and were analyzed by the same lab using the same methods argues against the REE anomalies at Lovisa being analytical artifacts.
Distinct positive Eu anomalies in the Stollberg ore field in Bergslagen (Fig. 1) were attributed to discharge into seawater of hot (>250°C) slightly to moderately acidic hydrothermal fluids carrying Eu 2+ in solution (Jansson et al., 2013). The absence of positive Eu anomalies indicates the Lovisa Fe formations were precipitated from low-temperature (<250°C) fluids (e.g., Bau and Dulski, 1996;Peter, 2003). However, a hydrothermal origin is indicated by the positive Y anomalies, which, along with the absence of positive Ce anomalies, distinguish them from modern marine hydrogenetic Fig. 19. Geochemical plots of Fe formations (IF) and associated calc-silicate rocks. A: Fe/Ti vs. Al/(Al + Fe + Mn) plot (Boström, 1973), with samples defining a mixing line between metavolcanic rocks and Fe oxide mineralization. Note that marble samples plot along the same line. B: SREE/Zr vs. Al/(Al + Fe + Mn). C: REE data normalized to Post-Archean Australian Shale (PAAS) (Taylor and MacLennan, 1985) for the samples with lowest Al/(Al + Fe + Mn) from the Lekeberg, Nya Backadahl, and Lovisa Fe formations. The REE distribution differs somewhat in the lower and upper part of the Lovisa Fe formation, therefore two samples are shown. D: Plot of Ce/Ce*PAAS vs. Pr/Pr*PAAS, illustrating absence of true negative Ce anomalies in Fe formations from the Lovisa area. The pattern is more consistent with positive La anomalies. Ce/Ce*PAAS = CeN/(LaN × PrN) 0.5 , Pr/Pr*PAAS = PrN/(LaN × NdN) 0.5 , modified after Bekker et al., 2010. E: Y/Ho vs. Al/(Al + Fe + Mn), illustrating that Y/Ho increases with decreasing volcaniclastic component. The ranges for seawater, change and radius controlled (CHARAC), and hydrogenetic marine ferromanganoan crusts are from Bau (1996). F: Eu/Eu*PAAS vs. Al/(Al + Fe + Mn), illustrating the poor development of positive Eu anomalies, even in samples poor in detritus. Due to presence of Gd anomalies, Eu/Eu* was calculated as EuN/(0.67SmN + 0.33TbN) (cf. Bau and Dulski, 1996) ferromanganese crusts that form in oxidized sea floor environments ( Fig. 19C) but are similar to hydrothermal ferromanganese precipitates (Bau and Dulski, 1996). Bau and Dulski (1996) argued that the negative Y anomaly in hydrogenetic marine ferromanganese crusts is due to slow growth rate, resulting in REEs and Y in equilibration with seawater. In contrast, rapidly precipitated hydrothermal ferromanganese precipitates indiscriminately scavenge REEs and Y from seawater too quickly to allow exchange equilibration between dissolved and adsorbed REEs and Y.
Positive La and Gd anomalies have been documented in modern seawater (De Baar et al., 1985;Bau and Dulski, 1996;Alibo and Nozaki, 1999), late Archean-early Proterozoic Fe formations (Bau and Dulski, 1996), phosphorites (Emsbo et al., 2015), and hydrogenous fluorapatite proximal to SEDEX Pb-Zn mineralization (Gadd et al., 2016). According to Bau (1999), fractionation of La, Gd, and Y is not redox controlled but related to the lanthanide tetrad effect. Experiments by Bau (1999) suggest that scavenging of La, Gd, and Y from a solution upon interaction with Fe oxyhydroxides is strongly pH dependent. At low pH, Y was found to become enriched in the solution, whereas at pH >5, fractionation was observed to also cause enrichment of La and Gd in the solution. Thus, the positive La, Gd, and Y anomalies in Fe formations in the Lovisa area could reflect deposition from a near-neutral hydrothermal fluid that had already interacted with Fe oxides along the fluid path (for example, hematite-quartz BIF in the stratigraphic footwall).
Ferric oxyhydroxides, ferrihydrite, green rust, Fe carbonates, and Fe silicates are recognized as likely precursors of magnetite and hematite in Precambrian Fe formations, depending on, e.g., the pH and ƒO 2 of the depositional environment (cf. Peter, 2003;Bekker et al., 2010;Halevy et al., 2017). Whereas hematite is commonly observed as the paragenetically oldest Fe mineral in BIFs stratigraphically beneath the Håkansboda limestone (e.g., Stråssa, Blanka, and Stripa;Geijer, 1927;Geijer and Magnusson, 1944), there is no evidence of former hematite in the Fe formations stratigraphically between the Håkansboda limestone and the Lovisa Zn-Pb deposit. The greater abundance of Fe 2+ -bearing minerals in the latter group of Fe formations is attributed to primary differences and could reflect (1) presence of organic matter, which could drive diagenetic or metamorphic reduction of precursor ferric oxyhydroxides or ferrihydrite, (2) presence of both ferrous (e.g., siderite, silicates) and ferric precursor minerals, or (3) primary mixed Fe 2+ -Fe 3+ precursors, such as green rust.
The first alternative finds support from the local presence of accessory graphite in the magnetite Fe formations and interbedding with graphitic mudstone at Nya Backadahl. However, assuming graphite derivation from organic matter, oxygen-deficient conditions would have been necessary on the sea floor, since the low deposition rate of detrital sediment in these intervals would not be favorable for preservation of organic matter by burial. Occurrence of carbonates and Fe silicates is a more attractive alternative, since these could react with silicates to form calc-silicates during metamorphism (cf. Klein 1973;Jansson andAllen, 2011). French (1970) has furthermore shown that in the absence of quartz, siderite undergoes thermal decomposition to magnetite ± graphite at temperatures exceeding 400° to 450°C, which is lower than the metamorphic grade at Lovisa. Green rust forms by reduction of ferrihydrite and can alter to magnetite, even when still suspended in an anoxic water column (Zegeye et al., 2012). Along with the absence of Ce anomalies, these considerations are most consistent with the precursors to the Fe formations having accumulated in a suboxic to anoxic depositional environment, more reducing than for the hematiterich BIFs stratigraphically below.

Genesis of the sulfide deposits
We interpret the Lovisa Zn-Pb deposit to be a metamorphosed, stratiform syngenetic-exhalative deposit (laminated mineralization) that has been locally tectonically remobilized (ore with durchbewegung texture and veins). The calc-silicate-poor nature of the deposit contrasts with limestone-and skarn-hosted sulfide deposits in Bergslagen, where massive skarns are commonly abundant and where metasomatic skarn models have been discussed (e.g., Ryllshyttan; Jansson and Allen, 2015). Although garnet compositions in the Lovisa Zn-Pb deposit overlap with those of Zn skarns (cf. Meinert et al., 2005), we emphasize the highly subordinate calc-silicate content in the ore zone and the porphyroblastic texture of garnet, which is similar to that in metamorphic rocks. As in other stratiform deposits in Bergslagen (e.g., Zinkgruvan), we find that the mineralogy and textures can be explained by regional metamorphism of preexisting hybrid rock types containing a variable content of quartz, feldspars, phyllosilicates, Fe oxides, and carbonates (cf. Hedström et al., 1989;Jansson and Allen, 2011;Jansson et al., 2013Jansson et al., , 2017. No obvious feeder zone has been recognized near the Lovisa Zn-Pb deposit, which may indicate that it is distal to a hydrothermal vent. This is consistent with the conclusions of Carlon and Bleeker (1988), who suggested that the Håkansboda Cu-Co deposit ca. 1,500 m northeast of Lovisa (Fig. 3) formed in a vent to stratiform mineralization in the area.
A syngenetic-exhalative origin followed by regional metamorphism is likewise favored for the Pb-Ag mineralization in the Lovisa Fe formation based on (1) absence of Zn and Pb outside the Fe formation, (2) the affinity of the Pb-Ag mineralization to a specific Fe formation over a strike extent of several kilometers, (3) presence of sphalerite-galena laminae in graphitic mudstone at the correlative Nya Backadahl Fe formation, and (4) the asymmetric vertical zonation of Zn and Pb within this unit (e.g., Fig. 16A, B). The vertical zonation is consistent with the accumulation of base metal sulfides during the onset of Fe formation deposition but diminishing with time.
Values of δ 34 S for sulfides from Lovisa similarly cluster close to 0‰, yet the δ 34 S range of the Lovisa Zn-Pb deposit (δ 34 S = -2.8 to 5.4‰) is larger than the examples given above, particularly if the sulfides in the Lovisa Fe formation are included (δ 34 S = -6.1 to 5.4‰). It is uncertain whether the sulfides in the Lovisa Fe formation are genetically related to the Lovisa Zn-Pb deposit. However, the stratigraphic proximity of these two deposits and the similar Pb isotope values support this interpretation. Nevertheless, the relatively wide range in δ 34 S and the stratigraphic variations in δ 34 S over short distances contraindicate the model of Wagner et al. (2005), which does not account for all aspects of the δ 34 S patterns at Lovisa. Instead, mixing of S sourced from different reservoirs (e.g., leached from volcanic rocks and from seawater) or variable degrees of isotopic fractionation of S derived from a single volcanic or seawater S reservoir must be considered. Billström (1991) reported δ 34 S to range from -6.5 to 16.5‰ at the Zinkgruvan stratiform Zn-Pb-Ag deposit based on 260 separates of sphalerite, galena, and Fe sulfides. However, the data yield an average of 2.0‰, and 75% of the data yielded δ 34 S between -3.0 and 3.0‰. Jansson et al. (2017) suggested that Zinkgruvan formed upon venting of oxidized (SO 4 2-> H2S), metalliferous brines at a near-neutral pH into a reduced brine pool. The source of S at Zinkgruvan was interpreted as a mixture of S leached from volcanic rocks (δ 34 S ≈ 0‰) and H2S derived from abiotic or bacterial sulfate reduction of contemporaneous seawater SO 4 2-(δ 34 S ≈ 17‰). This model for the source of S for the Zinkgruvan deposit is attractive for Lovisa, because in order for the ore-forming fluid to transport metals and H2S together, the fluid must have been acidic to transport Zn and Pb in solution efficiently (cf. Cooke et al., 2000). It would then be difficult to explain how the ascending hydrothermal fluid could traverse the Håkansboda limestone without losing most of its base metal content upon neutralization (cf. Jansson et al., 2017). Neutralization of hot, acidic hydrothermal fluids carrying metals and H2S together has been demonstrated as a highly efficient mechanism for forming massive, stratabound limestone-and skarn-hosted polymetallic sulfide deposits elsewhere in Bergslagen, such as at Stollberg (Jansson et al., 2013) and Falun (Kampmann et al., 2017).

Availability of Sulfur and Sources of Sulfur
Sulfide deposition from fluids or in an environment that was H2S deficient is suggested by the generally Fe sulfide poor nature of stratiform sulfide mineralization at Lovisa, locally approaching monomineralic compositions (cf. Ohmoto and Goldhaber, 1997). A similar conclusion was made by Jansson et al. (2017) for the Zinkgruvan deposit. The bulk of the contained S in these deposits is hosted by Fe-poor sphalerite and galena, and consequently whole-rock total S is essentially a function of sphalerite and galena content (Fig. 20). In contrast, the Falun and Stollberg stratabound limestone-and skarn-hosted deposits contain significant pyrite (Falun) and/ or pyrrhotite (Stollberg), such that these deposits are more S rich at a given base metal content (Fig. 20). Since a large part of the mineralization at Stollberg is hosted by Fe-rich rocks including magnetite Fe formation (Jansson et al., 2013), these differences in the Fe sulfide/base metal sulfide content cannot be attributed to variable Fe contents in the deposits.
Thus, aside from differences in the characteristics of exhaled hydrothermal fluids (e.g., T, metal contents) at different times, some of the variation in metal ratios and δ 34 S could reflect variations in H2S availability during sulfide precipitation. Sulfide deposition under H2S-deficient conditions could result if the metalliferous fluids were S poor and exhaled into environments where limited H2S was available to form sulfides. Alternatively, it could have arisen if the hydrothermal fluids were oxidized and mainly carried S as SO 4 2-, and bacterial or thermochemical reduction to H 2S needed to form sulfides was kinetically too slow to build up sufficiently high H2S concentrations to form polymetallic assemblages of galena, sphalerite, and Fe sulfides (Ohmoto and Goldhaber, 1997).
The H2S, often with isotope values relatively close to zero, may have been derived from a number of processes, including bacteriogenic sulfate reduction (BSR) or thermochemical sulfate reduction (TSR) and partly also from leached volcanic S, as further discussed below. However, the limited amount of data and lack of absolute values, e.g., for depositional temperatures, hinders a quantification of their relative importance. As is common for regionally metamorphosed ore deposits, many of the sulfide pairs are out of equilibrium, and T calculations do not yield any consistent values.

Pb-Ag in the Lovisa Iron Formation
Deposition under H2S-deficient conditions could explain the low δ 34 S and low Zn/Pb of sulfide mineralization in the Lovisa Fe formation, assuming that the hydrothermal fluid transported Fe, Pb, and Zn. The gradational decrease in Zn and Pb and the disappearance of sulfides upward from the base of the Fe formation could be explained if H2S became completely exhausted with time ( Fig. 16A-C). The galena δ 34 S is consistent with a low-T (in the 150°-100°C range) fractionation between galena and H2S at around 0‰, suggesting a volcanic-magmatic source for the S, yet this alternative is not favored since is it more difficult to explain the isotopically light sphalerite (-4.9‰) by this mechanism. Furthermore, the combination of high Ba (Fig. 16G), the very high Fe content, and significant Pb and Zn contents necessitates efficient transport of these elements in solution. Barium is only very weakly soluble in the presence of SO 4 2-, which suggests that the fluid was SO 4 2poor and not highly oxidized. The near-neutral pH in the hydrothermal fluid interpreted above indicates that the fluid could not transport significant amounts of Zn and Pb in the presence of H2S (cf. Cooke et al., 2000). Thus, the most likely scenario is that the metalliferous fluid was S poor and mixed with H2S sourced from seawater sulfate following exhalation.
The generally slightly negative δ 34 S values in sulfides from the Lovisa Fe formation are consistent with BSR of Early Proterozoic seawater sulfate at δ 34 S = 17‰ (Claypool et al., 1980) in a closed system (DH2S-SO 4 2is likely within the range 15-30‰; Ohmoto et al., 1990) where Rayleigh isotopic fractionation processes also might have played a role. The δ 34 S values could also be explained by kinetic fractionation during TSR of seawater sulfate at ca. 100°C (D 34 SSO 4 -H 2 S = 20‰; Machel et al., 1995).

The Lovisa Zn-Pb Ore
The tectonic modification of the main ore makes it uncertain to what extent the decrease in Zn/Pb and δ 34 S stratigraphically upward in the Lovisa Zn-Pb deposit is a primary stratigraphic feature or whether it could be related to postore modification. A primary origin for the variable Zn/Pb and δ 34 S trend is supported by the fact that this pattern is observed in the most well-preserved, laminated sphalerite facies of both ore beds (Fig. 17). Wagner et al. (2005) showed that formation of synmetamorphic, Ag-rich sulfide-sulfosalt veins by remobilization of preexisting polymetallic sulfide mineralization at Hällefors produced insignificant δ 34 S fractionation to slightly lower values. Thus, the δ 34 S of sulfides in the main ore at Lovisa are likely inherited largely from the precursor sulfide mineralization.
The highest δ 34 S values are in pyrite in and directly below the sphalerite ore (3.3-5.4‰), in the same stratigraphic interval in which sphalerite has the highest δ 34 S values (1.4-4.7‰). These values are somewhat high to be accounted for by BSR of seawater sulfate, unless the δ 34 SSO 4 2in the water column was higher than 17‰. They are, however, within the range of δ 34 S values derived by kinetic fractionation during thermochemical reduction of seawater sulfate at ca. 150° to 200°C (D 34 SSO 4 -H 2 S = 10-15‰; Machel et al., 1995). The Fe-poor nature of sphalerite in the sphalerite ore is consistent with formation at relatively low temperatures, since the Fe content of sphalerite decreases with decreasing temperature and increasing aS 2 (e.g., Hannington and Scott, 1989). The paragenetically later red sphalerite in the laminated facies is presumably more Fe rich and likely crystallized at a higher temperature. The overlap in δ 34 with coarse-grained sphalerite and galena in adjacent veins suggests that the red sphalerite may very well have crystallized in conjunction with local remobilization during vein formation.
The pyrite porphyroblasts in the red rhyolitic ash-siltstone below the sphalerite do not define any lamination, suggesting that they formed beneath the sea floor. The coexisting Ba-rich feldspars are texturally similar to complexly zoned hyalophane described in the Early Proterozoic Cuyuna Fe range by McSwiggen et al. (1994). As at Lovisa, textural evidence for replacement of earlier barite is lacking in this area, whereby McSwiggen et al. (1994) attributed the hyalophane to diagenetic or metamorphic dehydration of earlier authigenic Ba-Al silicates formed in a SO 4 2--depleted setting. The stratabound occurrence of pyrite and Ba feldspars in a relatively thin zone of commonly hematite stained volcaniclastic rocks stratigraphically below the sphalerite ore is consistent with pyrite formed when H2S diffused down from the overlying water column and reacted with Fe oxides.
The slight decrease in δ 34 S stratigraphically upward from the sphalerite ore to the main ore is opposite to the trend expected if the H2S in the Lovisa Zn-Pb deposit was exclusively derived from BSR or TSR of seawater SO 4 2undergoing Rayleigh isotopic fractionation in a closed system (cf. Ohmoto and Goldhaber, 1997;Seal II, 2006). Rather, the narrower δ 34 S values close to 0‰ for the main ore could imply that leached volcanic S similar to other sulfide deposits in the Bergslagen district (cf. Wagner et al., 2005) became more important during this final stage of the mineralizing process.
Geochemical modeling by Cooke et al. (2000) suggests that Zn and Pb are equally soluble in the presence of H2S in nearneutral fluids at ca. 225°C but that the Pb solubility decreases more rapidly than Zn with falling temperature. Consequently, the lower Zn/Pb ratio of the main ore compared with the sphalerite ore could imply that the latter formed from a fluid that vented at a lower temperature. However, assuming H 2S derivation from TSR, this is not supported by the lower δ 34 S in the main ore, since D 34 SSO 4 -H 2 S increases with decreasing temperature (Machel et al., 1995). However, the coupled high Zn/Pb and δ 34 S in the sphalerite ore (Fig. 21) could be accounted for if the fluid pulse, which formed the sphalerite ore, had already deposited galena elsewhere in the system (e.g., more proximal to the vent); this would have caused an increase in both Zn/Pb and δ 34 STOT in the fluid prior to forming the sphalerite ore. In this context, the lower Zn/Pb and δ 34 S in the main ore could be explained if the fluid pulse, which formed the main ore, reached the sea floor less fractioned regarding Zn/Pb and δ 34 S.
The apparent vent-distal nature of the sulfide mineralization suggests that only a part of the Lovisa ore system has been discovered and/or preserved. In the absence of a larger study incorporating vent zones and possible source regions, it is difficult to assess fully the cause of the observed differences in metal ratios and δ 34 STOT in the fluid en route to forming the studied sulfide beds. Nevertheless, our results indicate that during local basin evolution, syngenetic-exhalative sulfide deposits accumulated at various times under generally H 2S-deficient conditions. Although the S source cannot be uniquely constrained, the results are permissive of a setting in which early mineralizing phases mainly utilized H2S derived from BSR and TSR of seawater sulfate, whereas leached volcanic S introduced by hydrothermal fluids became more important with time.  Figure 17 and plotted against the average of Zn/Pb of the given mineralization type based on assay data. An average Zn/Pb cannot be provided for the veins in the main ore, which are here grouped with the ore with durchbewegung texture. Although there is a limited number of samples, there is a general trend to higher δ 34 S in, e.g., sphalerite from mineralization types with higher Zn/Pb.

Evidence for saline fluids and/or brine pools
Although the metamorphic overprint precludes an estimate of the salinity of the exhaled fluids based on fluid inclusions, there are several indirect indications that the deposits formed from relative saline hydrothermal fluids or that they accumulated in brine pools. Oen and Lustenhouwer (1992) described Cl-rich biotites identical to those that occur in the Lovisa Fe formation from broadly stratigraphically equivalent metaexhalites in the Nora area farther south in the Guldsmedshyttan syncline. These authors attributed the high-Cl content to the passage of high-salinity brines derived from seawater and/or to the former presence of Cl-rich authigenic minerals (halite and sylvite). A high-Cl content in the fluids that formed the Lovisa Fe formation would facilitate the transport of Fe and base metals, even under the near-neutral pH conditions inferred from the REE patterns and limestone in the footwall (cf. Seward and Barnes, 1997;Cooke et al., 2000). Hellingwerf et al. (1994) suggested that high B contents such as those observed in the Lovisa stratigraphy might have been derived from hydrothermal brines sourced from Proterozoic seawater. Jansson et al. (2017) suggested that the brines that enriched the stratigraphic footwall of the Zinkgruvan deposit in K and B might have been derived from the evaporation of seawater.
The best evidence for former brine pools is the presence of concordant zones of anomalous K 2O/(K2O + Na2O) in mainly albite-rich rocks that envelop the Lovisa Zn-Pb deposit. The halo-like distributions of zones with variable K2O/(K2O + Na2O) contrast with normal styles of alkali alteration in Bergslagen, wherein several 100-m-thick, extensive packages of K-feldspar-altered rocks are generally underlain by thick successions of albite-altered rocks (e.g., Lagerblad and Gorbatchev, 1985;Stephens et al., 2009;Jansson et al., 2013).
Concordant haloes of intense alkali enrichment have been described around vent-distal SEDEX deposits, such as the McArthur River deposit (Croxford, 1964;Lambert and Scott, 1973). Sangster (2002) attributed this alteration style to dense, alkali-rich brines that formed pools on the sea floor and sank into the underlying sediments. Alteration of volcanic glass and clays to sodic zeolites, analcime, and adularia has been described from saline-alkaline lakes (e.g., Parker and Surdam, 1971;Ratterman and Surdam, 1981;Remy and Ferrel, 1989). It is conceivable that similar processes could operate in and below brine pools and that medium-grade metamorphism would convert such authigenic minerals to feldspars. Thus, we interpret the Lovisa Zn-Pb deposit to have been deposited in an alkali-rich brine pool.
In detail, the Na-altered zone below the sphalerite ore is pink-red, hematite stained with porphyroblasts of pyrite and lesser magnetite. The alteration scheme of Remy and Ferrell (1989) describes Fe oxide or hydroxide as a byproduct of analcime formation upon interaction of oxidized, Na-rich brine with detrital phyllosilicates. In contrast, the Na-altered zone above the main ore is gray-white, pyrite poor, and devoid of Fe oxides. This may indicate that the brine pool was originally oxidizing but became more reducing with time.
During or shortly before formation of the Lovisa Zn-Pb deposit, a relatively oxidized, Na-rich brine pool may have formed by evaporation of seawater or exhalation of Na-rich brines and with time evolved toward more reducing conditions favorable for sulfate reduction (cf. Sangster, 2002). However, all base metal sulfide-bearing intervals coincide with distinct peaks in high K2O/K2O + Na2O within the albite-altered intervals. This may imply a more complex brine pool history wherein the sulfide deposits formed in conjunction with distinct pulses of exhalation of K-rich fluids into the brine pool, causing perturbations in the brine pool chemistry. Such a scenario would be consistent with the episodic nature of sulfide ore formation as reflected by different sulfide beds with different Zn/Pb and δ 34 S values. A K-rich composition for the metalliferous fluids has been invoked for Zinkgruvan, where the stratiform Zn-Pb-Ag ore is enveloped and underlain by intensely K-feldspar-altered rocks (Hedström et al., 1989;Jansson et al., 2017).

Comparison with Zinkgruvan, VMS deposits, and SEDEX deposits
The regional stratigraphic pattern at Lovisa follows a pattern broadly similar to that at Zinkgruvan (cf. Hedström et al., 1989;Allen et al., 1996;Kumpulainen et al., 1996, wherein the stratiform Zn-Pb-Ag ore occurs in the upper part of a thick felsic volcaniclastic succession deposited at a time of waning volcanic activity close to the transition to an overlying, postvolcanic sedimentary sequence. Similar to Zinkgruvan, stratiform Zn-Pb ± Ag mineralization occurs in a stratigraphic interval in which both a felsic volcaniclastic component and a marine clay component can be identified and where interbedding with former limestone units is common. At Zinkgruvan, Jansson et al. (2017) argued that there is no strong evidence for contemporaneous volcanism and that the fine-grained volcaniclastic component may consist of reworked volcanic detritus. This, along with the earlier mentioned conclusions on fluid chemistry, led Jansson et al. (2017) to consider a SEDEX model for Zinkgruvan.
The Lovisa deposit shares many key features of the SEDEX genetic model, such as (1) the low aspect ratio of the stratiform sulfide mineralization, (2) variable δ 34 S in sulfides, and (3) the inference of saline, alkali-rich, possibly oxidized oreforming brines. Unlike at Zinkgruvan, however, the presence of coarse-grained, pumiceous mass flow deposits clearly indicates that the mineralization at Lovisa formed in an episodically volcanically active basin. In addition, feldspar-destructive alteration is absent at Zinkgruvan, whereas quartz-sericitechlorite alteration has been documented in the stratigraphic footwall near Lovisa (e.g., east of Håkansboda; Fig. 3; Carlon and Bleeker, 1988). The Lovisa area furthermore differs from Zinkgruvan and SEDEX deposits in that the sulfide mineralizations occur in an area rich in stratiform Mn-rich Fe oxide deposits. Manganese-rich Fe oxide deposits are subordinate to Zn-Pb-Ag mineralization at Zinkgruvan, where they occur as discontinuous, stratabound zones in the stratigraphic footwall .
The stratiform sulfide beds at Lovisa possess some features of both SEDEX and VMS deposits. As in VMS systems, thermal perturbations related to volcanism may have been important for driving hydrothermal fluid flow and for generation of thick succession of originally vitreous, pyroclastic debris that could be leached of metals and S. However, as in SEDEX systems, formation of the sulfide deposits appears to have been dependent on the generation and discharge of highly saline brines into relatively deep marine environments favorable for sulfate reduction and accumulation and preservation of sulfides.
Such environments were rare in Bergslagen owing to the almost constant volcanism and resedimentation of volcaniclastic detritus, which generally kept pace with basin subsidence and thus maintained shallow marine environments (Allen et al., 1996(Allen et al., , 2003. In this sense, the significantly higher tonnage and thickness of stratiform Zn-Pb ± Ag mineralization at Zinkgruvan likely reflects a more favorable depositional environment in part related to less volcanic input into the basin. Sulfides at Zinkgruvan accumulated during a postvolcanic stage and were overlain by a metamorphosed organic-rich mudstone facies . These in turn grade stratigraphically upward into a thick succession of migmatized, graphite-bearing pelitic rocks, implying a long-lived, anoxic environment with low sedimentation rates. However, at Lovisa, stratiform mineralization is restricted to relatively thin stratigraphic intervals between beds of proximal-medial volcaniclastic rocks, and hematite-bearing BIF occurs in the stratigraphic hanging wall (Fig. 12C). Thus, formation of sulfide mineralization at Lovisa may be related to the coincidence of discharge of metal-bearing brines, relatively deep water environment, and brief periods of extremely low volcaniclastic sedimentation rate, during which the basin could evolve to more reducing conditions.
There are also differences in the style and scale of associated hydrothermal alteration. At Zinkgruvan, even at the ventdistal parts of the deposit, the mineralization is underlain by extensive volumes of homogeneous K-feldspar-altered rocks, whereas at Lovisa K alteration and other alteration types are less intense and more heterogeneously distributed. These differences may indicate that Zinkgruvan formed from a much larger hydrothermal system. It should, however, be emphasized here that the Zinkgruvan area is better explored.

Conclusions
Despite polyphase ductile deformation, amphibolite facies regional metamorphism, and subsequent brittle faulting, stratiform Zn-Pb ± Ag sulfide and Fe oxide deposits in the Lovisa area have retained evidence of a vent-distal exhalative origin. These deposits accumulated during periods of very low rates of volcaniclastic sedimentation between deposition of pumiceous, rhyolitic mass flow deposits in an actively subsiding basin. Deposition of the Lovisa Fe formation with subordinate Pb-Ag mineralization is attributed to discharge of an Fe-rich, S-poor brine at a near-neutral pH, which mixed with a preexisting H2S reservoir above the sediment-water interface. The low Zn/Pb ratio and low Fe sulfide content is attributed to sulfide deposition under H2S-deficient conditions.
The Lovisa Zn-Pb deposit is inferred to have formed at a later stage in an alkali-rich brine pool, which was periodically recharged with base metal-bearing hydrothermal fluids. The ore-forming system first deposited a nearly monomineralic sphalerite bed with positive δ 34 S values (sphalerite ore) and subsequently a sphalerite + galena bed with δ 34 S close to 0‰ (main ore). The upward decreasing δ 34 S may reflect a decreasing importance of H2S derived from TSR and/or BSR of seawater SO 4 2relative to S sourced from leaching of volcanic rocks. The high content of galena in the main ore led to partitioning of strain in this bed during subsequent ductile deformation, leading to the development of durchbewegung texture and remobilization of sphalerite and galena to veins. Remobilization appears to have only caused minor if any S isotope fractionation.
Our stratigraphic analysis shows that intervals of low sedimentation rate favorable for accumulation of hydrothermal sediments were established at several times during the stratigraphic evolution but that Mn-rich Fe oxide deposits mainly formed during such intervals. This may be because the vented fluids were mainly low-T fluids incapable of significant transport of base metals. However, results from Lovisa suggest that other reasons for the paucity of sulfide deposits in these intervals may have been a low H2S content in the vented hydrothermal fluids and a general H2S deficiency at the depositional sites. The latter may in turn be related to the mainly shallow marine environments due to infilling of the basin by volcaniclastic detritus. Long-lived, calm depositional environments rarely developed.
Whereas we infer that coeval volcanism was beneficial for driving hydrothermal fluid flow and for supplying a reservoir of metals and S that could be leached, we note that Zinkgruvan, which is the largest stratiform Zn-Pb ± Ag deposit in Bergslagen, formed at the stratigraphic boundary between an underlying volcanic and limestone succession and an overlying postvolcanic deep-water sedimentary succession recording long-lived anoxic conditions. Thus, we speculate that the generally high degree of volcaniclastic input and predominance of shallow marine environments in Bergslagen might have been a critical factor as to why large Zn-Pb ± Ag deposits are not more common.