Abstract The failure to precipitate dolomite experimentally at low temperatures or from seawater in which it is both supersaturated and the most thermodynamically favoured carbonate phase, together with its unequal distribution through geological time relative to limestone, are all aspects of the ‘dolomite problem’, a subject of continuing controversy. A plethora of physicochemical models has been invoked to explain sedimentary dolomite formation, none of which satisfactorily addresses the basic problem of how kinetic barriers are overcome. These barriers are related to the disproportionate distribution of the component ions of dolomite, cation hydration and ion complexing in seawater. Competing claims for the effectiveness of sulphate as an inhibitor to dolomite formation further confuse the debate, although there are many reports of modern dolomite associated with bacterial sulphate reduction. The uppermost sediments in some lakes of the Coorong region of South Australia comprise almost 100% dolomite, and afford an ideal opportunity to study this association. Samples of lake waters taken during late evaporative stages of several shallow hypersaline dolomitic lakes showed high initial sulphate concentrations, high pH and high carbonate alkalinities. Pore waters from unlithified lake sediment cores directly below the lake-water sample sites showed a substantial and progressive decrease in sulphate concentrations with depth, coupled with an exponential increase in carbonate concentrations, through the sulphate-reduction zone. By the end of the evaporative cycle, sulphate was entirely removed. High bacterial counts on cultures from the sediment cores, and sulphur isotope values consistent with ‘bacterial’ fractionation in lake waters, indicate that the chemical changes in ambient water chemistry can be related to active bacterial sulphate reduction. Laboratory experiments using sulphate reducers cultured from the lake sediments and simulating the anoxic microbiogeochemical environment of the lakes, have resulted in the precipitation of dolomite, demonstrating that bacterial sulphate reduction in the Coorong lakes modifies lake-water and pore-water chemistry so that dolomite precipitation is kinetically favoured. Given the wide spatial and temporal distribution of sulphate-reducing bacteria, and their frequent association, both past and present, with cyanobacteria, it is likely that this process was more widespread in the geological past when dolomite was found in far greater abundance than limestone. Bacterial sulphate reduction may thus have played an important role in dolomite formation throughout the geological record.

Abstract Despite intensive research over more than 200 years, the origin of dolomite, the mineral and the rock, remains subject to considerable controversy. This is partly because some of the chemical and/or hydrological conditions of dolomite formation are poorly understood, and because petrographic and geochemical data commonly permit more than one genetic interpretation. This paper is a summary and critical appraisal of the state of the art in dolomite research, highlighting its major advances and controversies, especially over the last 20–25 years. The thermodynamic conditions of dolomite formation have been known quite well since the 1970s, and the latest experimental studies essentially confirm earlier results. The kinetics of dolomite formation are still relatively poorly understood, however. The role of sulphate as an inhibitor to dolomite formation has been overrated. Sulphate appears to be an inhibitor only in relatively low-sulphate aqueous solutions, and probably only indirectly. In sulphate-rich solutions it may actually promote dolomite formation. Mass-balance calculations show that large water/rock ratios are required for extensive dolomitization and the formation of massive dolostones. This constraint necessitates advection, which is why all models for the genesis of massive dolostones are essentially hydrological models. The exceptions are environments where carbonate muds or limestones can be dolomitized via diffusion of magnesium from seawater rather than by advection. Replacement of shallow-water limestones, the most common form of dolomitization, results in a series of distinctive textures that form in a sequential manner with progressive degrees of dolomitization, i.e. matrix-selective replacement, overdolomitization, formation of vugs and moulds, emplacement of up to 20 vol% calcium sulphate in the case of seawater dolomitization, formation of two dolomite populations, and — in the case of advanced burial — formation of saddle dolomite. In addition, dolomite dissolution, including karstification, is to be expected in cases of influx of formation waters that are dilute, acidic, or both. Many dolostones, especially at greater depths, have higher porosities than limestones, and this may be the result of several processes, i.e. mole-per-mole replacement, dissolution of unreplaced calcite as part of the dolomitization process, dissolution of dolomite due to acidification of the pore waters, fluid mixing (mischungskorrosion), and thermochemical sulphate reduction. There also are several processes that destroy porosity, most commonly dolomite and calcium sulphate cementation. These processes vary in importance from place to place. For this reason, generalizations about the porosity and permeability development of dolostones are difficult, and these parameters have to be investigated on a case-by-case basis. A wide range of geochemical methods may be used to characterize dolomites and dolostones, and to decipher their origin. The most widely used methods are the analysis and interpretation of stable isotopes (O, C), Sr isotopes, trace elements, and fluid inclusions. Under favourable circumstances some of these parameters can be used to determine the direction of fluid flow during dolomitization. The extent of recrystallization in dolomites and dolostones is much disputed, yet extremely important for geochemical interpretations. Dolomites that originally form very close to the surface and from evaporitic brines tend to recrystallize with time and during burial. Those dolomites that originally form at several hundred to a few thousand metres depth commonly show little or no evidence of recrystallization. Traditionally, dolomitization models in near-surface and shallow diagenetic settings are defined and/or based on water chemistry, but on hydrology in burial diagenetic settings. In this paper, however, the various dolomite models are placed into appropriate diagenetic settings. Penecontemporaneous dolomites form almost syndepositionally as a normal consequence of the geochemical conditions prevailing in the environment of deposition. There are many such settings, and most commonly they form only a few per cent of microcrystalline dolomite(s). Many, if not most, penecontemporaneous dolomites appear to have formed through the mediation of microbes. Virtually all volumetrically large, replacive dolostone bodies are post-depositional and formed during some degree of burial. The viability of the many models for dolomitization in such settings is variable. Massive dolomitization by freshwater-seawater mixing is a myth. Mixing zones tend to form caves without or, at best, with very small amounts of dolomite. The role of coastal mixing zones with respect to dolomitization may be that of a hydrological pump for seawater dolomitization. Reflux dolomitization, most commonly by mesohaline brines that originated from seawater evaporation, is capable of pervasively dolomitizing entire carbonate platforms. However, the extent of dolomitization varies strongly with the extent and duration of evaporation and flooding, and with the subsurface permeability distribution. Complete dolomitization of carbonate platforms appears possible only under favourable circumstances. Similarly, thermal convection in open half-cells (Kohout convection), most commonly by seawater or slightly modified seawater, can form massive dolostones under favourable circumstances, whereas thermal convection in closed cells cannot. Compaction flow cannot form massive dolostones, unless it is funnelled, which may be more common than generally recognized. Neither topography driven flow nor tectonically induced (‘squeegee-type’) flow is likely to form massive dolostones, except under unusual circumstances. Hydrothermal dolomitization may occur in a variety of subsurface diagenetic settings, but has been significantly overrated. It commonly forms massive dolostones that are localized around faults, but regional or basin-wide dolomitization is not hydrothermal. The regionally extensive dolostones of the Bahamas (Cenozoic), western Canada and Ireland (Palaeozoic), and Israel (Mesozoic) probably formed from seawater that was ‘pumped’ through these sequences by thermal convection, reflux, funnelled compaction, or a combination thereof. For such platform settings flushed with seawater, geochemical data and numerical modelling suggest that most dolomites form(ed) at temperatures around 50–80 °C commensurate with depths of 500 to a maximum of 2000 m. The resulting dolostones can be classified both as seawater dolomites and as burial dolomites. This ambiguity is a consequence of the historical evolution of dolomite research.