Dolomitization (AAPG Course Notes 24)

Dolomitization

Presented at the 1982 AAPG Fall Education Conference in Denver, Colorado.

Education Course Note Series #24 Lynton S. Land University of Texas at Austin

fflPG

Dolomite It is probably safe to state that in 1982 no single model of dolomitization unequivocally accounts for all aspects of any massively dolomitized ancient limestone. All models have significant flaws, and our understanding of the dolomitization process and its relation to other diagenetic processes (silicification, stylolitization, organic maturation, etc.) is imperfect. Rather than advocate one solution over another, I will try to summarize some of the strengths and weaknesses of several of the models which have been proposed. As a starting point I will review several important aspects of dolomite mineralogy and chemistry that place constraints on all models and that are sometimes overlooked. Mineralogy Dolomite is a rhombohedral carbonate with the ideal formula CaMg[C03)2 in which calcium and magnesium occupy preferred sites. In the ideal mineral, planes of C03 anions alternate with planes of cations with the c-axis of the crystal perpendicular to the alternating stacked anion and cation planes. Ordering occurs by the additional alternation of cation planes containing only calcium with cation planes containing only magnesium (Fig. 1). It is possible to conceive of a mineral having the same composition as ideal dolomite ((Cao.6Mg0 6)C03) in which all cation planes are alike, containing equal numbers of calcium and magnesium ions. Such a mineral is not dolomite. Such a disordered arrangement of ions occupies more volume than that of the ideal dolomite structure and is unstable with respect to an ordered phase. Perhaps surprisingly, the two compounds just described, ideal dolomite and a disordered 1-to-l ratio Ca-Mg carbonate, are both rare in sedimentary rocks. Ideal dolomite rarely comprises ancient dolomitic sediments and never modern sediments, and the completely disordered polymorph does not occur at all. The dolomite which does occur in sedimentary rocks is commonly Ca-rich, having compositions which range from about Ca(Cao.16Mgog4)(C03)2 to ideality, and/or exhibits weak, diffuse, X-ray diffraction, suggesting considerably less structural order than its composition should dictate. With respect to ideal dolomite, all such naturally occurring dolomite is metastable, and the capacity exists for reactions to occur toward a more stable (more stoichiometric or better ordered) phase. The term protodolomite was defined by Graf and Goldsmith (1956) as "single-phase rhombohedral carbonates which deviate from the composition of the dolomite that is stable in a given environment, or are imperfectly ordered, or both, but which would transform to dolomite if equilibrium were established." Gaines (1977) modified the definition to include only ordered phases. I recommended (1980) that the term be dropped altogether, since almost all sedimentary dolomite is really protodolomite by Gaines' definition. What is important is not what we call these natural materials, but what they really are. 1

o

J^ CARBONATE

MAGNESIUM

CALCIUM

Figure 1 — Schematic representation of the crystal structure of dolomite showing the alternation of cation and anion (carbonate) planes, and the alternation of calcium and magnesium planes. 2

Hydrothermal experiments (Graf and Goldsmith, 1956; Goldsmith and Heard, 1961), extrapolated to low temperature, demonstrate that calcite and dolomite are essentially ideal in composition at 25 °C (Fig. 2). In other words, any double carbonate crystal of Ca and Mg at 25 °C which is not essentially pure dolomite is either metastable or unstable with respect to a mixture of pure calcite plus pure dolomite. The same thing is true with respect to ideal dolomite plus magnesite. The composition of phases which we observe at Earth's surface define the range of metastability. Unstable phases are only observed as transient states in the laboratory. In the case of dolomite, few phases containing more than about 8% excess calcium (on a molar basis) have been reported to date, although the data are admittedly sparce. Reeder (1981) has shown that the structure of various kinds of dolomite revealed by transmission electron microscopy and electron diffraction can be classified into at least three types. All structures are ordered, although the degree of order is variable and difficult to quantify. The first, characteristic only of Holocene dolomite, consists of irregular "mosaics" on a scale of tens or hundreds of Angstroms. The crystals are characterized by extremely high densities of crystallographic faults and dislocations, and can be thought of as an aggregate of "micro-crystals" whose compositions may vary, forming a very discontinuous lattice. This leads to many unsatisfied or strained chemical bonds and to X-ray diffraction patterns with broad, generally weak reflections. This kind of dolomite is also characterized by large trace element substitutions, especially strontium (Behrens and Land, 1972), and sodium (Land and Hoops, 1973). Qualitative data suggest that this material is extremely soluble compared to better ordered forms of dolomite. My attempts to beneficiate samples composed of mixtures of this kind of dolomite and aragonite (for example, supratidal crusts from Florida and the Bahamas) by slow leaching in acetic acid resulted in only slight concentration of the dolomite by selective solution of aragonite. C0 2 for isotopic analyses of Holocene dolomite is evolved much faster than from finely ground ancient dolomite. All evidence suggests that Holocene dolomite is a unique, highly soluble material. It is clearly a metastable phase, unknown (in an unmodified form) in ancient rocks. The second and most common kind of sedimentary dolomite exhibits a lamellar or "tweed" structure when examined by transmission electron microscopy and electron diffraction, which Reeder (1981) has interpreted as a structural and/or compositional modulation on a scale of several hundred Angstroms (Fig. 3). At present this kind of dolomite is thought to consist of two intimately intergrown lamellar domains parallel to the rhomb face with slightly different structures and/or compositions. The texture resembles spinoidal decomposition, or solid state unmixing on a scale of a few hundred angstroms from a single homogeneous precursor. The exact structure and composition of the two domains or lamellae is not known, although one must be more stable (and presumably more magnesium rich) than the other. This type of dolomite is clearly metastable, but continued stabilization cannot proceed spontaneously because it is limited by solid state diffusion. Continued stabilization can occur as a result of solution-reprecipitation processes however, and it has been demonstrated that bulk Ca-rich dolomites dissolve more rapidly than ideal dolomite (Busenberg and Plummer, 1982). Continued stabilization toward a more stoichiometric dolomite would presumably be promoted if pore fluids in the rock changed to enable dissolving out of the less stable, Ca-rich domain. Porosity could easily increase under these conditions.

3

Ordered Dolomite

1000 Dolomite + Magnesite

800

600

-

TEMP. (°C)

Lower limit of experimental data

400 -

Ranges of metastable phases observed in nature (

10

20

30

40

50

MOLE % MgC03

Figure 2 — Stability relations in the system CaC03 • MgC03

60

Figure 3 — Dark field transmission electron micrograph of a calcian dolomite (Caj i2Mg0 88(C02)2) of Eocene age. The prominent modulated structure is typical of sedimentary dolomite, and such crystals are metastable with respect to ideal stoichiometric dolomite. Photograph by Richard Reeder. A third kind of dolomite is nearly ideal in composition, and when examined by transmission electron microscopy and electron diffraction is observed to be homogeneous, consisting of large single domains. This kind of dolomite is presently known mostly from ancient, deeply buried sequences and from metamorphic rocks. The philosophy that, like limestone, the diagenesis of dolomite is dominated by the stabilization of metastable dolomitic phases, is relatively new. There is no question that calcium-rich dolomite has the capacity to react to form crystals with a more stoichiometric composition, but many important questions remain. What kinds of diagenetic environments promote the reaction? Does stabilization to ideal dolomite take place all at once or in several stages? How far from ideality must a phase be before it is prevented from further reaction for kinetic reasons? Many of these questions must be answered both by laboratory work and by careful mineralogical analysis of particular dolomites under investigation before thinking can advance much further. 5

Aqueous solution equilibria Several lines of evidence have been used to determine the solubility of dolomite at sedimentary and early burial conditions. The data are complicated by the mineralogical variations in dolomite already discussed. All metastable phases must be more soluble than ideal dolomite, and variations in the degree of metastability can obviously occur. Data have been derived from two sources, (1) high temperature experiments and (2) natural dolomite aquifers. Of interest is the equilibrium constant, K, for reactions between the ideal solids, 2CaC0 3 + Mg + + ^

CaMg(Co3)2 + Ca ++

K = (Ca ++ )/(Mg ++ )

or the calcium-to-magnesium activity ratio of a solution at equilibrium with calcite + dolomite (as a function of temperature). Solutions more magnesium-rich than the equilibrium solution should cause dolomitization of calcite, while solutions more calcium-rich should cause dedolomitization. Dolomite is easily synthesized hydrothermally at about 300 °C, with reaction times of only a few days. Progressively slower reaction is observed at lower temperatures and below about 100°C very long experiments are required. Nobody has yet synthesized dolomite at Earth-surface conditions (although a Dalmatian has!, Mansfield, 1980). Experimental data are in reasonable agreement around 300°C, and the molar Ca/Mg ratio of a solution in equilibrium with calcite and dolomite is about 15. In other words, as temperature increases, dolomite becomes increasingly less soluble than calcite. Any solution with a molar Ca/Mg ratio of less than 15 is capable of dolomitizing at 300 °C (Fig. 4)! At lower temperatures, experimental data become more conflicting, the reason being, I suspect, that metastable Ca-rich phases are much more easily formed. Kinetic experiments (Land, 1967) have shown that the formation of a Ca-rich (metastable) dolomite rather than the ideal phase is favored (within the stability field of dolomite) by (1) higher Ca/Mg ratio of the solution, (2) lower solution concentration, and (3) lower temperature. Metastable Ca-rich phases are more soluble and therefore will coexist with more magnesium-rich fluids (Helgeson et al, 1978). Hsu (1963), Holland et al, (1964), Barnes and Back (1964) and Langmuir (1971) all studied the Ca/Mg ratio of natural dolomite aquifers, reasoning that equilibrium with dolomite would eventually be reached as water recharged a dolomite aquifer and moved downdip at rates typical for groundwater flow. Langmuir's compilation is plotted on Figure 4. The extrapolation of Rosenburg and Holland's (1964) data to intercept Langmuir's low temperature data is not too unreasonable if one accepts that the lower temperature hydrothermal data points of Rosenburg and Holland may be displaced toward magnesium-rich compositions because of formation of a non-ideal (more soluble) phase. The reasonable agreement between low temperature and high temperature data ignores non-ideal solution behavior, which is significant in the saline solutions Rosenburg and Holland used. But at 300°C, experiments at 2M, 1M and 0.5M solutions all yield similar results, suggesting the effects are not large. Further support for extrapolation between the two types of data was 6

Temp. (°C) 3.5

— 10

Langmuir, 1971

-

— 25 CALCITE

— 50 3.0

-

DOLOMITE — 100 2CaC03 + Mg++ 2.5

CaMg(C03)2 + Ca++

^

-

— 150

1000 T(°K) t 2.0

\

— 200

Rosenberg and Holland, 1964 — 250

-

Gaines (pers. comm.) — 300 Land. Rosenberg, Burt and Holland

1.5 -0.2

0.0

0.2

0.4

0.6

0.8 I 5

1.0 I 10

1.2

1.4 logCa/Mg I 25 Ca/Mg

Figure 4 — Aqueous solution compositions presumed to be in equilibrium with calcite plus dolomite.

obtained by Pakhomov and Kisson (1973) (reproduced in Carpenter, 1980) who plotted the Ca/Mg ratio of saline formation water from the Russian platform versus temperature. Despite the fact that they totally ignored rock composition (calcite plus dolomite may not both have been present to control the solution composition), and obtained considerable scatter, their regression line essentially connects Rosenburg and Holland's and Langmuir's data! Until further experimental work is conducted (which must include characterization of the dolomite phase) the data presented in Figure 4 are all that are available. They are consistent both with a gross oversaturation of seawater with dolomite, and the Mg-depleted nature of most saline formation water. The reason for the gross oversaturation of seawater with respect to dolomite ultimately lies in the kinetic problem of nucleating and growing the ordered crystal (Goldsmith, 1953). The molar Ca/Mg ratio of seawater (0.19) is apparently incapable of causing dolomitization at observable rates. By either decreasing the molar Ca/Mg ratio of seawater (say by gypsum precipitation) or decreasing the activity Ca/Mg ratio the kinetic constraints can be overcome, at least to the point of being able to nucleate and grow a poorly crystalline Ca-rich phase. The activity Ca/Mg ratio of seawater is 0.18 (Berner, 1971), and can be decreased by dehydrating the Mg+ + ion (Usdowski, 1968) or by removing components which form strong ion pairs with Mg+ + (for example, S0 4 = , Baker and Kastner, 1981). These factors do not alter the equilibrium relations (Fig. 4) and only provide the kinetic "push" to form the initial phase. The early-formed phase can then stabilize by further reaction. Another variable in the dolomitization process which needs additional confirmation is the role of organic material, particularly dissolved organic acids. Dissolved organic acids are known to control the kind of calcium carbonate which precipitates from solution. Increased organic acid content favors Mg-calcite over aragonite precipitation (Kitano and Kanamori, 1966). Although algal processes have been invoked as being able to cause dolomitization (Gebelein, 1973), the "organic gremlin" is neither proven nor disproven. Stable Isotopic Geochemistry Most current evidence supports the contention that sedimentary dolomite is enriched in 180 about 3 to 4 ppt with respect to a co-existing calcite in the range of sedimentary and burial diagenetic temperatures of normal interest (Land, 1980). Little evidence exists for dolomite replacement of calcite without change of isotopic composition (Katz and Matthews, 1977). The fact that many ancient dolomites are significantly depleted in 180 is best explained by stabilization of an earlier-formed phase during burial (Fig. 5). The isotopic composition of the dolomite comprising sedimentary rocks is controlled both by the chemistry of the latest recrystallization (stabilization) event and by the chemistry of the precursor (aragonite, Mg-calcite, calcite and/or dolomite). Dolomite rarely recrystallizes homogeneously in an open aqueous chemical system, accurately recording the conditions of recrystallization, just as it almost never accurately retains the chemistry of the precursor. Recrystallization may be incomplete, leaving an inhomogeneous rock, and the composition of the replaced phase may "contaminate" the replacing phase (Land, 1980). The practical problem of analyzing intimate mixtures of dolomite of slightly different compositions is not yet solved. 8

200

160-

Temp. (°C)

12 o

80

40-

<5180 Dolomite (PDB scale)

Figure 5 — Oxygen isotopic composition of dolomite as a function of temperature and 5180 water (curved lines). The histogram at top was constructed from published analyses (Land, 1980) and shows the wide range of conditions under which dolomite formed and/or stabilized. 9

Trace Element Geochemistry Trace element partitioning is complicated by kinetic factors. The ratio of the concentration of a trace element in a crystal to the concentration of the element for which it substitutes (say Sr/Ca) is dependent on the concentration ratios of the elements in the solution from which the crystal forms, on temperature, on pressure (usually ignored), and on other variables such as the rate of crystal growth. The distribution coefficient "D" in the following equation is thus a function of variables which are not always easy to define: (Sr/Ca)crygtal = Dx(Sr/Ca) solution

Modern marine and hypersaline dolomite has an Sr content of about 600 ppm (Behrens and Land, 1972), yet few ancient dolomites contain more than 200 ppm Sr, even when presumed to be initially of hypersaline origin. Although it was once assumed that removal of the trace elements by flushing with a low Sr (meteoric) water was required (Land, 1973), this is no longer acceptable for all ancient dolomite. As an example of this problem, Bein and Land (1982) studied Permian San Andres dolomite from the subsurface in north Texas, where dolomite beds are intimately interbedded with bedded halite and anhydrite. Both halite and anhydrite display sedimentary structures indicating a primary subaqueous origin, and both contain trace elements (Br in halite and Sr in anhydrite) indicative of primary precipitation. It seems clear that thin dolomite beds intimately interbedded with and "entombed" by primary evaporites could never have formed from or been modified by low Sr (meteoric) water. Yet the dolomites all contain less than 200 ppm Sr. Bein and Land suggest that although the original dolomite may have resembled Holocene analogs (about 600 ppm Sr), during burial it stabilized to a more ordered structure, expelling Sr to form celestite. In other words, at least two distribution coefficients apply to this situation, one for the formation of the original phase, and a second (lower) for the stabilization reaction to a more ordered, stoichiometric phase. Because of these kinetic problems which plague other sedimentary phases as well — anhydrite (Kushnir, 1980), halite (Holser, 1979), trace element analyses of dolomite are of limited practical value today. Hopefully, more experimental work will rectify this situation. Mechanisms of Dolomitization Clearcut petrographic evidence indicates that most dolomite initially forms by replacing a precursor carbonate. That is, a fluid simultaneously imports Mg++, dissolves the precursor phase, precipitates dolomite, and exports Ca++. Of course, the situation is actually more complex due to the import and export of other components such as other trace elements and their isotopes (for example, 87Sr/86Sr), carbon and oxygen isotopes, C02, etc. Because of considerable compositional differences between dolomite and any presumed precursor (calcite, aragonite, or Mg-calcite), considerable fluid transport is required. Advection (fluid flow) must accomplish most of the transport, although diffusion may play an important part on a local scale. Models for dolomitization are therefore basically hydrologic models. Before discussing 10

^ ^r

seawater

STORM RECHARGE

EVAPORATIVE DRAWDOWN

Figure 6 — Only two models of sabkha hydrology can apply if seawater is the hydrologic baselevel, and simultaneously the source for magnesium. Elevation of seawater onto the sabkha surface by storms, or by lowering the baselevel, provides the elevation head to move water back to the sea (reflux). Evaporative drawdown can only be a transient condition as the depression is rapidly filled with salts.

these, we should not completely ignore' 'primary" dolomite. Replacement is a dissolution-precipitation process, albeit on a sub-microscopic scale. The dolomite which replaces a precursor is a precipitate in every sense of the word, and is primary in the sense that it was not there before. It is not primary only in the sense that it occupies space previously occupied by another solid phase. "Primary dolomite" is usually defined as crystals which nucleate from solution and either accumulate as primary sediment or precipitate into megascopic pores as cement, and so displace only fluid as they grow. Dolomitic rocks which contain pore-filling dolomite cements do, in every sense of the word, consist in part of primary dolomite. Dolomitic rocks are often very inhomogeneous as a result of containing several generations of dolomite. Primary dolomite micrite may not be as uncommon as current thinking presumes. Baffin Bay, Texas, provides a possible example. Essentially pure, laterally extensive dolomite beds up to 4 cm thick occur in early Holocene subtidal laminated terrigenous muds. Cores from the 11

upper part of the Bay sequence, from the sea floor to about 7 m below, are extremely well laminated, documenting alternating periods of aragonite (and rarely Mg-calcite) precipitation and terrigenous deposition, which occurred during and after storms (often hurricanes). The bay is normally hypersaline except after hurricanes, and so the deposition of chemical precipitates during hypersaline periods and the deposition of terrigenous material accompanying runoff accounts for the laminations, and the hypersalinity for their preservation. The middle part of the sequence, from about 7 to 13 m below sea floor, formed in about 5 m of water about 3500 -1000 years ago (Behrens, 1974), and is texturally similar except for the presence of dolomite beds. Very little terrigenous material is present within the dolomite beds, ruling out any kind of a mixing model since fresh water would have contributed terrigeneous mud. Interstitial water analyses of the very impermeable sediments, obtained by hydraulic squeezer, have a relatively uniform chlorinity (36 ppt), molar Ca/Mg ratio (0.15) and <5180 (+ 2 ppt), very similar to Baffin Bay water during normal summers. Only small amounts of interstitial gypsum are present in the sediment and no beds of gypsum occur. Clearly a 4 cm-thick dolomicrite could not have formed after burial because the pore water shows no strong depletion in magnesium. The beds must have formed at the sea floor, either by primary precipitation or by complete replacement of some precursor (Mg-calcite?), prior to being buried by terrigeneous influx associated with storms. Although it is probable that the early Baffin Bay was silled, and sulfate reduction in the (stratified?) bottom water may have been important in reducing the Ca/Mg activity ratio of the water (Baker and Kastner, 1980), the possibility that the dolomite is a primary precipitate cannot be excluded. The possible importance of primary dolomicrite should not be ignored. Three hydrologic models for dolomitization are currently "in vogue." All require a potential field to move fluid through the rocks and an "inexhaustible" source of magnesium. The following calculation serves to illustrate the magnitude of fluid flow and magnesium required. Assume a typical carbonate sediment is to be dolomitized. A typical sediment contains about 6.3 mole percent MgC03 (Land, 1973, Table 2), and has about 40% porosity. As a place to start, assume that seawater is to be the dolomitizing agent. A cubic meter therefore contains: a) 400 liters of seawater x 1.025 Kg water/liter water = 4.1 x 102 kg seawater; and b) 600,000 cu cm of limestone consisting of: 570,885 cu cm CaC03 or 1.545 x 104 moles CaC03 (36.94 cu cm/mole) 29,115 cu cm MgC03 or 1.039 x 103 moles MgC03 (28.02 cu cm/mole). 4.1 x 102 kg of seawater contains 4.26 moles of Ca+ + and 22.1 moles of Mg++. If the sediment reacts with the water to reach equilibrium (calcite -I- dolomite + a solution having a Ca/Mg ~ 1, Fig. 4), then the interstitial water will provide 8.92 additional moles of magnesium and the rock will contain 6.7% dolomite of ideal composition. 99% of the dolomite is derived from the magnesium originally in the Mg-calcites and slightly less than 1 % is derived from the magnesium in the interstitial seawater. To completely dolomitize the remaining CaC03,7.19 x 103 moles of Mg+ + must be added. E ach pore volume of "new'' seawater can provide 8.92 moles of Mg+ + for dolomitization (the water can only provide magnesium until the Ca/Mg ratio is increased to about 1—Fig. 4—at which point it reaches equilibrium with calcite + dolomite). 12

Therefore, 807 pore volumes of seawater are required to completely dolomitize 1 cu m of sediment. If seawater diluted 10 times with meteoric water (say in a mixing zone) is utilized, then 8.1 x 103 pore volumes are needed. If seawater having an Mg content of about 8 x 10"1 and a Ca content of about 8 x 10 "moles/Kg is utilized (a typical brine which has precipitated gypsum and evaporated to the point of halite saturation), then only 44 pore volumes are needed. If the various solutions do not reach equilibrium with calcite + dolomite (the Mg/Ca ratio does not fall to 1), or the brine has not reached halite saturation, then proportionately more pore volumes of fluid are required. No porosity reduction has been achieved, and if dolomite cementation occurs, additional fluid flow is required. Reflux Reflux, as defined by Adams and Rhodes (1960) occurs when "hypersaline brines eventually become heavy enough to displace the connate waters and seep slowly downward through the slightly permeable carbonates at the lagoon floor." Examination of Holocene sabkhas has suggested that downward moving water driven solely by potential energy resulting from increased density of the fluid at constant head is probably not as important as the increased head caused by elevation of water onto the sabkha surface by storms. Hsu and Siegenthaler (1964) summarized various ideas of sabkha hydrology. Basically, considering a sabkha which extends relatively far along strike relative to its width (a two dimensional system), the directions of water movement are quite limited. At any point in the sabkha, water can either move up or down, seaward or landward (Fig. 6). It is assumed that an infinite reservoir of magnesium (seawater) is available at some constant level at the margin of the sabkha. Only two processes can move seawater (the source of magnesium) landward in the absence of interaction with an independent underlying aquifer system, namely storm recharge and evaporative drawdown. Evaporative drawdown, or the lowering of the water table by evaporation, can only occur if the landward part of the sabkha is depressed below sea level by subsidence, compaction and/or wind deflation. Unevaporated seawater must be kept from flooding the depression by some sort of sill, either a physical barrier or a long distance. In any case, landward flow of seawater into a depression will result in rapid evaporation and consequent filling of the basin by evaporite minerals, effectively halting flow by eliminating the head difference. The amount of water required to produce 1 cu m of gypsum is about sufficient to completely dolomitize 1 cu m of carbonate sediment. Evaporative drawdown (possibly aided by capillary withdrawal) is, at best, a transient condition and is self-limiting. Storm recharge, however, can continuously (geologically speaking) drive water up onto the sabkha, where it evaporates and flows seaward, driven by elevation head, and aided by its increased density. Such a mechanism dominates modern sabkhas (McKenzie, Hsu, and Schneider, 1980; Amdurer and Land, 1982). In the case of the Trucial Coast of the Persian Gulf, dolomitization takes place only in the storm recharge zone, and the amount of dolomite correlates with the frequency of recharge (Patterson and Kinsman, 1982). Considerable amounts of gypsum may be precipitated as the result of brine evolution. For example, using the figures previously discussed, 44 pore volumes of halite-saturated brine were required to dolomitize 1 cu m of sediment. About 1 cu m of gypsum would have 13

DOLOMITE

— 200

SUPERSATURATED 100 UNDERSATURATED

—

50

ZONE OF DOLOMITIZATION

— I

0

20

1

40

j

60

-

80

100

percent seawater Figure 7 — Percent saturation for mixtures of seawater and a typical meteoric groundwater having a P™ = 102 atmospheres (after Plummer, 1975).

precipitated from the volume of seawater required to generate that much brine, leading to a gypsum-to-dolomite volume ratio of one. Advantages of the reflux mechanism are the rapidity with which dolomite can be formed as documented by Holocene studies, and the relatively smaller volumes of water required due to its magnesium-rich nature (Sears and Lucia, 1980). This mechanism clearly dominates in evaporitic settings. In the absence of evaporites the model is more constrained, barring fluctuations of the Ca/Mg ratio and/or the sulfate content of seawater. The efficient removal of calcium by the formation of surficial algal micrite prior to evaporative concentration can also suppress CaSO< precipitation (Amdurer and Land, 1982). In addition, it is not at all clear that reflux can operate on the regional scale for which it was first proposed. The small fluid 14

potentials caused solely by density differences apparently cannot move water very far through sediments of relatively low permeability. Elevation head is required, and in addition to storm recharge it might easily be accomplished by periodic lowering of the reservoir of seawater either by a local mechanism (say evaporation of a restricted sea) or on a larger scale (eustatic/tectonic), draining of the sabkhas periodically in the same way modern coastal plains were drained during Pleistocene glacial events. Meteoric Mixing In order to account for evaporite-free dolomite sequences, the mixing of meteoric water (providing the driving force through elevation head) with seawater (providing the magnesium) has been advocated (Hanshaw, Back, and Deike, 1971; Land, 1973). Geochemical considerations (Fig. 7) (Plummer, 1975) suggest that the mechanism is plausible even though much longer times are required for dolomitization (Sears and Lucia, 1980). Although examples of Holocene mixing-zone dolomite (mostly as cements!) continue to be found (Magaritz et al, 1980), a major problem with the model is explaining why dolomite is not more common, since mixing of seawater and meteoric water is a ubiquitous worldwide process. The model apparently requires a relatively stable hydrologic setting to establish sufficient continuous recharge for establishment of a mixing cell with seawater over a long period of time to drive the dolomitization reaction. Kinetic problems are overcome by reducing the Ca/Mg activity ratio of the mixture through lowering of the ionic strength. This may not be too much of a problem in a subtropical setting as, say, tidal flats prograde across a shelf leaving behind vast areas for recharge. But in an arid climate the model is difficult to apply unless large adjacent coastal plains provide the recharge zone and evaporites are sealed off from the actively circulating water. The evaporative concentration of continental water accounts for playa-type dolomite including the Coorong examples (von der Borch, Lock and Schwebel, 1975). Burial Diagenesis Dolomite can clearly form as a directly precipitated late cement, as exemplified by studies of sandstone burial diagenesis (Boles, 1978; Land and Dutton, 1978). The dolomite is commonly ferroan, and can approach ankerite in composition, reflecting the large amount of ferrous iron commonly present in the terrigenous system. Although it is true that shales in a sedimentary basin are possible sources for nearly every conceivable component required for any conceivable kind of diagenesis, it is not clear that they are sources for magnesium. In fact, the precipitation of chlorite within the shales may be a local sink for magnesium. Saline formation waters are typically very magnesium-poor, and on the whole commonly approach calcite-dolomite equilibrium (Pakhomov and Kissin, 1973). Supplying large amounts of magnesium from a water nearly in equilibrium with calcite plus dolomite requires vast amounts of water, a definite problem, especially in relatively impermeable rocks. In addition, Figure 4 indicates that if a water initially in equilibrium with calcite plus dolomite moves 15

updip (and cools), it becomes undersaturated with dolomite and will either dissolve dolomite or dedolomitize. This exact subsurface reaction has been observed by Land and Prezbindowski (1981) and Budai (1981). Therefore, at the present time, the formation of large amounts of new replacement dolomite is difficult by this mechanism. No large-scale source for magnesium has been identified. Moving magnesium around within a basin without producing any net new dolomite appears to be quite possible, but in this case the "new" replacement dolomite or cement must be balanced by either "new" dedolomite or by secondary porosity somewhere within the basin. The mobility of calcium, magnesium and dissolved carbonate after burial must not be disregarded. Shales are rapidly "decalcified" during burial (Hower et al, 1976) and provide a large-scale source for new carbonate phases. But since calcium loss exceeds magnesium loss by at least a factor of 6, much more calcite than dolomite is involved in the process. Sandstone diagenesis can involve immense quantities of carbonate which is both precipitated and removed (to form secondary porosity). Sandstones can be carbonate-cemented, decemented and then recemented (Milliken et al, 1982), and carbonates probably undergo similar complex histories. Late secondary porosity development in carbonates is known (Moore and Druckman, 1981), and some textures in deeply buried carbonates may be the result of selective dissolution of calcite, leaving the dolomitic component of the rocks as an "insoluble residue" (Wanless, 1979). It is important to "decouple" the process of dolomitization/dedolomitization (controlled by the Ca/Mg of the solution) from cementation/secondary porosity generation (controlled by the acidity of the solution). The dolomitization process is rarely C03=-conservative (Weyl, 1960; Degens and Epstein, 1964). A solution with a low Ca/Mg ratio and capable of dolomitizing can either cause net cementation or net solution, depending on changes occurring in the total dissolved carbonate content of the solution as it moves through the rocks. Addition of C0 2 by organic maturation can cause net solution, whereas loss of C0 2 to adjacent strata of lower carbonate content can cause net precipitation. Thus dolomitization can either result in porosity decrease (by cementation and/or by compaction accompanying recrystallization), or porosity increase (secondary porosity formation). The same is true of the dedolomitization reaction. Classic dolomite reservoirs containing intercrystalline porosity may possibly result from recrystallization of a metastable Ca-rich precursor phase induced by a C02-rich (corrosive) solution. Some or all of the more Ca-rich (more soluble) domains of the metastable phase may be lost to the solution, and additional dolomite may even be dissolved. The less soluble component must recrystallize, and intercrystalline porosity results from the volume loss of the Ca-rich domains. It is possible that such situations may even be "self-reservoiring" in the sense that C0 2 evolved during early maturation may be responsible for creating the reservoir by dolomite recrystallization! Other Possibilities We should be careful about being too actualistic in our approach to dolomitization. Only 25 years ago, we thought that essentially no Holocene dolomite existed (Fairbridge, 1957). Each case of Holocene dolomitization has resulted in considerable over-reaction and 16

"bandwagon-jumping" soon after the discovery. One intriguing possibility, which is gaining considerable support recently, is that "the present is a lousy key to the past because seawater has changed." The observation that the percentage of dolomite in carbonate rocks increases as we go back in geologic time was originally attributed to more time available for dolomitization (the source of magnesium was not specified) (Chilingar, 1956). Changes in the composition of seawater resulting in times of "easier" dolomitization in the past cannot be discounted. Tucker (1982) recently suggested a primary origin for a Pre-Cambrian oosparite (oolitic grainstone) composed entirely of dolomite (including the "spar"!). Changes in salinity, in Ca/Mg ratio, SC%= concentration and PC02 have all been invoked (Sandburg, 1975; Baker and Kastner, 1981; Mackenzie and Pigott, 1981), and sympathetic variation of several components may be particularly effective, and ultimately related to crustal cycles. Conclusions No panaceas exist for dolomitization. Each case must be studied on its own merits, and many scenarios exist. Modern scenarios begin to break down if seawater and/or sediment compositions have evolved with time. Reflux can account for the initial formation of many evaporite-related dolomites but since the poorly ordered phases formed in hypersaline environments are not found in ancient rocks, recrystallization must occur. Mixing zone dolomitization is capable of upgrading early hypersaline phases to a more stable phase, but is not necessary as "isochemical" recrystallization can occur in saline brines. Mixing zones are capable of producing dolomite cements and new replacement phases, given enough time and with sufficient recharge zones. Burial diagenesis can generate dolomite cements (commonly ferroan), induce recrystallization of previously formed, metastable phases, and move previously formed dolomite from place to place. Recrystallization can take place in essentially closed chemical systems, or in partly open systems resulting in gross changes in the chemistry of the dolomite and in the selective removal of either calcite or dolomite from the sequence. Few (if any) carbonate rocks, dolomitized or not, exist as they were originally deposited. Most have resulted from one or more processes of formation, and at least one stabilization (recrystallization) event. Acknowledgements Several students and colleagues critiqued earlier versions of the manuscript and offered valuable corrections, including James Anderson, David Budd, Bob Folk, Donald Miser, and Richard Reeder. Richard Reeder kindly provided Figure 3. Support of the Geology Foundation of the University of Texas at Austin is gratefully acknowledged.

17

REFERENCES Adams, J. E. and M. L. Rhodes, 1960, Dolomitization by seepage refluxion: AAPG Bull., v. 44, p. 1912-1920. Amdurer, M. and L. S. Land, 1982, Geochemistry, hydrology and mineralogy of the sand bulge area, Laguna Madre flats, South Texas: Jour. Sed. Petrology, v. p. Baker, P. A. and M. Kastner, 1981, Constraints on the formation of sedimentary dolomite: Science, v. 213, p. 214-216. Barnes, I. and W. Back, 1964, Dolomite solubility in groundwater: U. S. Geol. Survey Prof. Paper 475-D, p. 179-180. Behrens, E. W., 1974, Holocene sea level rise effect on the development of an estuarine carbonate depositional environment: Memoires de 1'Justitat de Geologie du Bassin d'Aquitaine, No. 7, p. 337-341. and L. S. Land, 1972, Subtidal Holocene dolomite, Baffin Bay, Texas: Jour. Sed. Petrology, v. 42, p. 155-161. Bein, A. and L. S. Land, 1982, San Andres carbonates in the Texas panhandle; sedimentation and diagenesis associated with magnesium-calcium-chloride brines: Austin, Texas, Univ. of Texas, Bureau of Econ. Geology, Rept. of Invest., No. 121,48 p. Berner, R. A., 1971, Principles of chemical sedimentology: McGraw-Hill, 240 p. Boles, J. R., 1978, Active ankerite cementation in the subsurface Eocene of southwest Texas: Contrib. Mineralogy and Petrology, v. 68, p. 13-22. Budai, J. M., 1981, Subsurface dedolomitization of the Madison limestone, Wyoming: Geol. Soc. America Abs. with Programs, p. 419. Busenberg, E. and L. N. Plummer, 1982, The kinetics of dissolution of dolomite in CO2-H20 systems at 1.5 to 65°C and 0 to 1 atm Pco2: Am. Jour. Sci., v. 282, p. 45-78. Carpenter, A. B., 1980, The chemistry of dolomite formation I; the stability of dolomite, in D. H. Zenger, J. B. Dunham, and R. L. Ethington, eds., Concepts and models of dolomitization: SEPM Spec. Pub. No. 28, p. 111-121. Chilingar, G. V., 1956, Relationship between Ca/Mg ratio and geologic age: AAPG Bull., v. 40, p. 2256-2266. Degens, E. T, and S. Epstein, 1964, Oxygen and carbon isotope ratios in coexisting calcites and dolomites from recent and ancient sediments: Geochim. et Cosmochim. Acta, v. 28, p. 23-44. Fairbridge, R. W., 1957, The dolomite question, in R. J. LeBlanc and J. G. Breeding, eds., Regional aspects of carbonate deposition: SEPM Spec. Pub. No. 5, p. 125-178. Gaines, A. M., 1977, Protodolomite redefined: Jour. Sed. Petrology, v. 47, p. 543-546. , 1978, Reply, protodolomite redefined: Jour. Sed. Petrology, v. 48, p. 1009-1011. Gebelein, C. D., 1973, Algal origin of dolomite laminations in stromatolitic limestone: Jour. Sed. Petrology, v. 43, p. 603-613. Goldsmith, Jr., 1953, A "simplexity principle" and its relation to "ease" of crystallization: Jour. Geology, v. 62, p. 439-451. and H. C. Heard, 1961, Subsolidus phase relations in the system CaC03-MgC03: Jour. Geology, v. 69, p. 45-74. 18

Graf, D. L. and J. R. Goldsmith, 1956, Some hydrothermal syntheses of dolomite and protodolomite: Jour. Geology, v. 64, p. 173-186. Hanshaw, B. B., W. Back, and R. G. Deike, 1971, A geochemical hypothesis for dolomitization by groundwater: Econ. Geology, v. 66, p. 710-724. Helgeson, H. C , et al, 1978, Summary and critique of the thermodynamic properties of rock-forming minerals: Am. Jour. Science, v. 278-A, 229 p. Holland, H. D., et al, 1964, On some aspects of the chemical evolution of cave waters: Jour. Geology, v. 72, p. 36-67. Holser, W. T., 1979, Trace elements and isotopes in evaporites, in R. G. Burns, ed., Marine minerals: Mineralog. Soc. of America Short Course Notes, v. 6, p. 295-346. Hower, R., et al, 1976, Mechanism of burial metamorphism of argillaceous sediment; 1 Mineralogical and chemical evidence: Geol. Soc. America Bull., v. 87, p. 725-737. Hsu, K. J., 1963, Solubility of dolomite and composition of Florida groundwaters: Jour. Hydrology, v. 1, p. 288-310. and C. Siegenthaler, 1969, Preliminary experiments on hydrodynamic movement induced by evaporation and their bearing on the dolomite problem: Sedimentology, v. 12, p. 11-26. Katz, A. and A. Matthews, 1977, The dolomitization of CaC03; an experimental study at 252-295°C: Geochim. et Cosmochim. Acta, v. 41, p. 297-308. Kitano, Y, and N. Kanamori, 1966, Synthesis of magnesian calcite at low temperatures and pressures: Geochem. Jour., v. 1, p. 1-10. Kushnir, J., 1980, The coprecipitation of strontium, magnesium, sodium, potassium and chloride ions with gypsum; an experimental study: Geochim. et Cosmochim. Acta, v. 44, p. 1471-1482. Land, L. S., 1967, Diagenesis of skeletal carbonates: Jour. Sed. Petrology, v. 37, p. 914-930. , 1973, Contemporaneous dolomitization of Middle Pleistocene reefs by meteoric water, north Jamaica: Bull. Marine Science, v. 23, p. 64-92. , 1980, The isotopic and trace element geochemistry of dolomite; the state of the art, in D. H. Zenger, J. B. Dunham, and R. L. Ethington, eds., Concepts and models of dolomitization: SEPM Spec. Pub. No. 28, p. 87-110. and G. K. Hoops, 1973, Sodium in carbonate sediments and rocks; a possible index to the salinity of diagenetic solutions: Jour. Sed. Petrology, v. 43, p. 614-617. and S. P. Dutton, 1978, Cementation of a Pennsylvanian deltaic sandstone; isotopic data: Jour. Sed. Petrology, v. 48, p. 1167-1176. and D. R. Prezbindowski, 1981, The origin and evolution of saline formation water, Lower Cretaceous, south-central Texas, U. S. A.: Jour. Hydrology, v. 54, p. 54-71. Langmuir, D. L., 1971, The geochemistry of some carbonate ground waters in central Pennsylvania: Geochim. et Cosmochim. Acta, v. 35, p. 1023-1045. Mackenzie, E T. and J. D. Pigott, 1981, Tectonic controls of Phanerozoic sedimentary rock cycling: Jour. Geol. Soc. London, v. 138, p. 183-196. Mansfield, C. G., 1980, A urolith of biogenic dolomite - another clue in the dolomite mystery: Geochim. et. Cosmochim Acta., v. 44, p. 829-840. McKenzie, J. A., K. J. Hsu, and J. E Schneider, 1980, Movement of subsurface waters under 19

the sabkha, Abu Dhabi, U. A. E., and its relationship to evaporative dolomite genesis, in D. H. Zenger, J. B. Dunham, and R. L. Ethington, eds., Concepts and models of dolomitization: SEPM Spec. Pub. No. 28, pp. 11-30. Magaritz, M., et al, 1980, Dolomite formation in the seawater-freshwater interface: Nature, v. 287, p. 622-624. Milliken, K. L., L. S. Land, and R. G. Loucks, 1981, History of burial diagenesis determined from isotopic geochemistry, Frio Formation, Brazoria County, Texas: AAPG Bull., v. 65, p. 1397-1413. Moore, C. H. and K. Druckman, 1981, Burial diagenesis and porosity evolution, Upper Jurassic Smackover, Arkansas and Louisiana: AAPG Bull., v. 65, p. 597-628. Pakhomov, S. I., and I. G. Kissin, 1973, Hydrogeochemistry of magnesium in deep aquifer zones: Akad. Nauk SSSR Doklady, v. 209, p. 205-208. (English translation, 1974, American Geological Institute). Patterson, R. J. and D. J. J. Kinsman, 1982, Formation of diagenetic dolomite in coastal sabkha along the Arabian (Persian) Gulf: AAPG Bull., v. 66, p. 28-43. Plummer, L. N., 1975, Mixing of seawater with calcium carbonate ground water: Geol. Soc. of America Mem. 142, p. 219-236. Reeder, R. J., 1981, Electron optical investigation of sedimentary dolomites: Contr. Mineralogy Petrology, v. 76. p. 148-157. Rosenberg, P. E., D. M. Burt, and H. D. Holland, 1967, Calcite-dolomite-magnesite stability relations in solutions; the effect of ionic strength: Geochim. et Cosmochim. Acta, v. 31, p. 391-396. Rosenburg, P. E., and H. D. Holland, 1964, Calcite-dolomite-magnesite stability relations in solutions at elevated temperatures: Science, v. 145, p. 700-701. Sandburg, E A., 1975, New interpretations of Great Salt Lake ooids and of ancient non-skeletal carbonate mineralogy: Sedimentology, v. 22, p. 497-537. Sears, S. 0., and F. J. Lucia, 1980, Dolomitization of northern Michigan Niagara reefs by brine refluxion and freshwater/seawater mixing, in D. H. Zenger, J. D. Dunham, and R. L. Ethington, eds., Concepts and models of dolomitization: SEPM Spec. Pub. No. 28, p. 215-235. Tucker, M. E., 1982, Precambrian dolomites; petrographic and isotopic evidence that they differ from Phanerozoic dolomites: Geology, v. 10, p. 7-12. Usdowski, H. E., 1968, The formation of dolomite in sediments, in G. Muller, and G. M. Friedmann, eds., Recent developments in carbonate sedimentology in central Europe: Springer-Verlag, p. 21-32. von der Borch, C. C , D. E. Lock, and D. Schwebel, 1975, Groundwater formation of dolomite in the Coorong region of South Australia: Geology, v. 3, p. 283-285. Wanless, H. R., 1979, Limestone response to stress; pressure solution and dolomitization: Jour. Sed. Petrology, v. 49, p. 437-462. Weyl, K., 1960, Porosity through dolomitization; conservation-of-mass requirements: Jour. Sed. Petrology, v. 30, p. 85-90. 20