2.2.2. Hydrocarbon Migration

Basin sediments are heated as they are buried and move downward through thermal gradients typically 20 to 30 ◦C km−1. Organic material is thermally cracked, first to oil (between 90–110 ◦C), and then to progressively dry (pure CH4) gas (between 110–130 ◦C). Oil and gas under basin conditions are lighter than water, and both will thus rise buoyantly through the sediments. If their rate of generation is slow compared to the rate at which they rise, they will rise as isolated rivulets of oil or in gas chimneys. If their rate of generation is fast compared to the rate at which they can rise, they will displace the pore water and completely fill the pores with oil or gas. The former is the conventional oil and gas scenario where the rising hydrocarbons are ponded beneath low permeability strata forming oil and gas reservoirs. The latter is the unconventional oil and gas scenario that has been of much interest recently. We'll return to this latter case later. First consider conventional hydrocarbon migration.

If hydrocarbons mature in a basin with extensive sand layers that are more permeable than ~1 millidarcy (10−<sup>15</sup> m2), vertically migrating hydrocarbons can be intercepted and diverted laterally over hundreds of kilometers. The oils migrate along only the upper centimeters of the carrier beds and fill reservoirs where the carrier beds are folded or o ffset by faults. The North Sea presents many examples of this kind of lateral migration and trapping, and the process is well described in [10]. Even a slight change in the tilt of the North Sea carrier sands can be important, and Løtviet et al. [11] show how tilting by glacial isostatic adjustment can redistribute oil in the North Sea.

The Gulf of Mexico is a currently active hydrocarbon generating area where there are no strata permeable enough to laterally divert the rising hydrocarbons in a very substantial fashion. In the 1990s we selected a portion of the Gulf of Mexico basin we thought would be large enough to capture the processes operating there. We called this ~200 × 100 km area of o ffshore Louisiana the "GRI Corridor" after our funding source, the Gas Research Institute. The corridor tells quite a remarkable story [12–17]. Briefly (elaboration follows): the oils in reservoirs in the northern half of this Corridor were sourced by Jurassic carbonate and Eocene shale strata, but in the south only by Jurassic carbonates. The deeper Jurassic oils matured and migrated first. The Eocene oils matured and migrated later, mixing with the earlier Jurassic oils in the north but not in the south. The Jurassic carbonates then generated gas which altered the oils by "gas washing" in a fashion that allows both the depth of washing and the amount of washing gas to be determined. This small area of one basin generated more hydrocarbon resources than have been extracted and consumed by humans over the entire petroleum era. Ninety percent of the generated oil was either expelled into the ocean or retained in the source strata. Discovered reservoirs in the Corridor constitute less than a fifth of the hydrocarbons currently migrating within it. The current reservoirs were filled recently (all perhaps in the last 100,000 years). In this flow-through hydrocarbon system "the present is the key to the present" (statement by Glen Gatenby, 2001).

Perhaps the most immediately obvious and remarkable feature of the GRI Corridor is the regular N-S change in the chemistry of its reservoir oils (Figure 3) [14]. Over 90 wt% of the +10 n-alkane component of the oils in the Tiger Shoals field at the north end of the Corridor have been removed by gas washing. At the South Marsh Island Block 9 field (SMI 9) slightly to the south, 50% have been removed. At South Eugene Island Block 330 15% have been removed. At the Jolliet field at the south end of the transect, the oils have not been washed at all, and 0 Wt% of the n-alkanes have been removed.

Figure 3B illustrates how the n-alkane depletion is measured. Unaltered oils have a linear logarithmic decrease in the n-alkane mole fraction with carbon number, called the Kissen slope. This unaltered trend is shown by the black slanted line that melds to the measured n-alkane mole fractions of the un-depleted oil (red squares) at carbon numbers greater than 24. The mass depletion (in this case 90 wt%) is the purple shaded area. The depletion results from dry gas (methane) interacting with the oils. At basin pressures, methane can dissolve a lot of oil, and the lower carbon numbers are more soluble than the higher ones, which accounts for the greater depletion of the low carbon numbers (the rollover of the mole fraction curve). The depletion caused by gas washing can be distinguished from other kinds of alteration such as bacterial degradation. Modeling [18–21] shows that the break number at which the oil composition departs from the Kissen trend measures the depth at which the

oil was washed, and the degree of n-alkane depletion measures the amount of gas that washed the oil. The reservoir oils in the transect appear to have been gas washed in the first sand layer encountered by the hydrocarbons rising to fill the reservoirs.

**Figure 3.** Systematic changes in the n-alkane chemistry of 138 oils from N to S across a 202 km (N-S) by 125 km (E–W) transect in the offshore Louisiana Gulf of Mexico. (**A**) The transect is red rectangle in the insert. Below the insert the sample locations are shown on the left, and the percent n-alkane depletion on the right. (**B**) Illustration of how the n-alkane depletion is calculated (purple area) with reference to the un-washed n-alkane distribution (bounding straight line melding onto the measured n-alkane mole fractions indicted by red squares). The illustration is for Tiger Shoals. Figure simplified from [18].

Extensive modeling was carried out to determine if the washing could have been produced by the oil and gas generated in the Corridor [15]. First the McKenzie stretching factor was determined from the sediment thickness and water depth along a 1050 km section near the Corridor that was contributed by Exxon. The section ran from the Lousiana border with Mississippi to the Sigsbee Knolls in the Gulf of Mexico. The evolution of the basin was reconstructed by back-stripping and decompaction. Heat flow, temperature, and vitrinite reflectance were calculated using a finite element grid tied to deposited strata and extending to the 150 km depth. The mantle heat flux determined from the stretching factor was applied there. Radiogenic heat production in the crust and sediments, changes in the surface temperature with water depth, the cooling due to sediment deposition, and the effect of compaction on lithology-specific thermal conductivity were taken into account. This modeling matched Exxon's measured heat flow data as well as temperature and vitrinite reflectance depth profiles using unadjusted literature parameters. Heat flow is about half normal (30 mW m<sup>−</sup>2) near the shelf edge due to cooling by rapid sedimentation there, and about half this low heat flow is due to radiogenic heat production in the basin sediments. Despite the low heat flow, the temperature gradient is ~20 ◦C km−<sup>1</sup> because of the low thermal conductivity of shales in the vertical direction. Agreement between the model and observations is excellent, and identification of the sources of heat is instructive, but grea<sup>t</sup> care had to be taken to include and properly specify all the important parameters in the model.

Heat flow at the base of the sediment section was then extracted from the portion of the Exxon line corresponding to the Corridor, and temperature calculated for the evolution of a much more realistic stratigraphy. Radiogenic heat production in the sediments was included, and the stratigraphic evolution included inversion of the Louann salt to a surface sill which was then buried. Salt diapirism

and mini-basin formation was simulated. Maturation of a 100 m thick 5 wt% TOC (total organic carbon) carbonate Jurassic source strata (Type II kerogen) across the full section, and a 30 m thick 4 wt% TOC Type III Eocene coal across the northern half of the section was then calculated. The matured hydrocarbons were moved vertically out of the source strata once the saturation in the source strata exceeded 20 vol%, and moved upward thereafter when the pores of an element were filled to a specified migration saturation.

Gas is venting actively at hundreds of locations along the Corridor and there are many discovered hydrocarbon reservoirs. The modeling indicates that for gas to vent, the pore saturation of migrating hydrocarbons must be much less than 0.5%. The oils in the northern half of the section contain oleanane, a biomarker from plants that evolved in mid-Cretaceous time. The northern oils are also low in sulfur. Both indicate these oils came from the silicate Eocene source rock. For the Eocene oils to be dominant in the northern half of the Corridor, the migrating hydrocarbon pore saturation must be <0.05%. For a migration pore saturation of 0.025%, the oils are 85% Eocene at Tiger Shoals and 50% Eocene at the middle of the Corridor close to the end of the Eocene section, and there is su fficient late-generated gas to wash the oils as observed. Overall, the Corridor generated 184 Btoe (billion tons of oil equivalent) of hydrocarbons, mainly from Jurassic source beds, 37 Btoe were retained in the source rocks, 15 Btoe are in migration pathways between the source strata and the surface, and 131 Btoe (~1000 billion barrels- about 20% more than humans have consumed across the entire petroleum era) have been expelled into the ocean. Most of the oil has either been expelled (71%) or retained in the source strata (20%). The hydrocarbons discovered in the Corridor (1.4 Btoe) constitute ~9% of the hydrocarbons expelled from the source but not ye<sup>t</sup> vented from the Corridor. The migration and filling is ongoing and the reservoirs were all filled recently, some very recently. The Jolliet reservoirs are hosted in 0.6 to 1.8 Ma strata. The model hydrocarbon flux across the 0.95 Ma horizon at the Jolliet location, assuming draw from a 40 km diameter mini-basin size area, is 190 and 110 tons of oil and gas per year respectively. Thus the Jolliet reservoirs (4400 kt oil and 3100 kt gas) could have been filled in 23 and 29 ka, respectively. The gas venting rate at Bush Hill near the Joliet field is estimated to be 900 t per year, which suggests the filling rate could be even more rapid [22].

### 2.2.3. Nature of Fluid Flow in Basins

### Steady Expulsion: The Kupferschiefer Deposits in Germany and Poland

Base metal enrichment of the Kupferschiefer shale in Germany and Poland provides one of the most spectacular and best-documented examples of basin base metal mineralization.

The Kupferschiefer is the lowermost unit of the Zechstein evaporate sequence that extends from the England across the North Sea to southeastern Poland. It is a thin (usually 30 to 60 cm, but sometimes up to 1m thick), 258 Ma old pyrite and organic rich (~6% carbon) black shale, enriched in Cu, Pb, Zn, Au and other metals, and mined since at least 1199 AD. The technical challenges of mining made the Germans early leaders in technology. The deposits in Poland were discovered in 1957.

In the early Carboniferous (~350 ma) the 1700 × 500 km area later covered by the Zechstein sediments was the site of Hercynian clastic foreland basin sedimentation. In the Autunian (295–285 Ma), wrench tectonics extended the area and heated the lithosphere. Oxidized Rotliegende sediments accumulated in a basin and range topography with sill injection and some mafic volcanism. Erosion in a semi-desert setting of seasonal rivers and playa lakes leveled the topography. The peneplane was then partly covered by white aeolian dune sands (the Weissliegendes) which were later partially reworked into beach sands. As the lithosphere cooled, the area that would become the Permian Basin subsided in a broad down warp to 100 s of meters below sea level, but remained subaerial until ~259 Ma when it was suddenly flooded [23]. The laminated shale/shaly limestone or dolostone Kupferschiefer could thus be deposited throughout the basin below wave level in calm, shallow waters. The waters in this restricted basin were saline, the organic productivity high, and the bottom anaerobic [24]. The reduced, ubiquitous Kupferschiefer starkly contrasted with the oxidized sediments

below. Four to five carbonate-anhydrite-salt cycles were then deposited in a shallow marine setting on top of the Kupferschefer. By the end of the Cretaceous up to 8 km of sediments had accumulated over the Kupferschiefer [6,23–29].

The mineralization occurred as post-Kupferschiefer sediment deposition loaded and compacted the underlying sediments. Oxidized brines, expelled from the compacting pores, forced their way through the Kupferscheifer and immediately adjacent strata, and pushed through in greater quantities where these strata were less resistive to flow. The metal-bearing oxidized brines were reduced as they moved through the reducing strata, and base metals were deposited. Where there was su fficient brine throughput, the Kupferschiefer was completely oxidized (called Rote Fäule or red fooling rock because it is barren of mineralization and red because the iron-bearing sulfides have been replaced by hematite) [24,27,30]. This process of metal deposition is not unusual. It also occurred in the midcontinent Proterozoic rift in the upper Peninsula of Michigan, for example, where reduced siltstones and shales of the Nonesuch Formation were oxidized by brines expelled from the underlying Copper Harbor conglomerate [6,31–33]. What is remarkable about the Polish Kupferschiefer mineralzation is the documentation of its metallization by 774 drillholes (50,000 analyzed samples) across all of Poland [34] (now there are 1700 drill holes [26]). The metallization maps allow us to estimate the volume of brine required to produce the metal enrichment. The estimated volume is so large that a large fraction of the brine expelled from the basin must have participated in the mineralization.

Figure 4 shows the copper surface density in kg/m<sup>2</sup> from one of the maps in the Metallogenic Atlas of the Zechstein Copper-bearing Series in Poland [34]. Similar maps in the Atlas show the zinc and lead metal surface density. The metal surface density shown in Figure 4D is the kilograms of metal in the Kupferschiefer and adjacent strata under each m<sup>2</sup> of plan area. The entire mineralized portion of the Lower Zechstein was analyzed down to a cuto ff grade of 0.1%. The maximum thickness of the mineralized interval was 123 m, but typically the thickness analyzed was between 10 and 60 m. The metal density contours can be integrated to obtain the total metal added. I did this by tracing and summing the area of each metal density interval (e.g., 1–5, 5–10, 10–50 kg m<sup>−</sup>2, etc.) in the Atlas maps, and multiplying by the log average metal density added in each interval, as shown in Table 1. For example, the metal added between the 1 and 5 kg m<sup>−</sup><sup>2</sup> contours equals the area of this contour interval times 2.24 (= 100.5(log 1 + log 5)). The metal added between the 5 and 10 kg m<sup>−</sup><sup>2</sup> contours equals the area of this contour interval times 4.83. The 4.83 kg m<sup>−</sup><sup>2</sup> metal added equals the metal under this contour (7.07 = 100.5(log 5 + log 10)) minus the 2.24 kg m<sup>−</sup><sup>2</sup> log average surface density of the first contour interval. Table 1 shows that 824, 927, and 1523 million tons of Cu, Pb and Zn, respectively, were added to the Kupferschiefer and immediately adjacent sediments in the area covered by the Atlas maps. The reserves of the Lubin, Polkowice-Sieroszosice and Rudna mining districts (near the black mining symbol at the north border of the red-outlined Fore-Sudetic Block in Figure 4D) are 30.4 Mt Cu [26]. The mine reserves thus constitute only 3.6% of the metallization.

The 9th row of Table 1 (labeled Brine in column 1) indicates the concentration of metals that would have to precipitate from the brine to account for the metallization in the 8th row of the table, if the brine volume expelled through the Kupferschiefer were 200,000 km3. For this volume of brine, 4.1, 4.6 and 7.6 ppm of Cu, Pb, and Zn respectively would have to precipitate for the observed metal tonnages to be deposited. Metal concentrations more than those required have been measured in oil field brines (last 3 rows in Table 1). The 200,000 km<sup>3</sup> volume of brine is very large. It could be supplied if compaction reduced the porosity of a 4 km thick strata covering a 500 × 500 km area (the area of Poland) by 20%. Such a reduction of porosity is possible (see discussion below), so the hypothesis of mineralization-by-brine-expulsion is plausible. Much could be discussed (e.g., the magnitude of the porosity change early and later in burial, how much basement rocks might compact, how much brine was expelled unrecorded through the Rote Fäule vents, etc.). The important point made by this rough calculation is that the brine expulsion recorded by the metallization mapped in Poland constitutes a significant portion of the brine that could be expelled by compaction from the sub-Kupferschiefer eastern portion of the Southern Permian Basin.

**Figure 4.** Key aspects of the Kupferschiefer mineralization in Poland. (**A**) Pyrite in the Kupfereschiefer is replaced first by Zn, then Pb, then increasingly copper rich sulfides and finally flushed completely of metals (Rote Fäule). (**B**) The metallization and Cu enrichment and depletion is zoned around sites of more intense brine discharge near the Fore-Sudetic Block (red outline in D). Figure from [34]. (**C**) The Zechstein (blue cross hatched) and its basal Kupferschifer shale are underlain by the Rotliegend sediments (orange), Carboniferous sediments (blue), and Variscan (light brown), Caledonian (red), and pre-Cambrian (pink) basement rocks. Section from [28]. (**D**) Copper surface density determined by 774 drill holes through the Kupferschiefer [34]. Red outlines the Fore-Sudetic Block. The Lubin, Polkowice-Sieroszosice and Rudna mines are all adjacent to this block at the location marked by the black crossed rock picks.

**Table 1.** Integration of metal surface densities in Metallogenic Atlas of Poland [34]. The first column indicates the metal surface density interval in the Atlas maps, and the second the log average metal content added by the interval, as discussed in the text. The next 3 columns show the area covered by each surface density interval, and the last 3 columns the additional metal introduced in each surface density interval, (e.g., the metallization) in millions of tons of Cu, Pb, and Zn. The 7th and 8th rows (labeled Sum and Brine in columns 1 and 2) show the total Cu, Pb, and Zn introduced in the area covered by the map, and the ppm that must have been extracted from a brine volume of 200,000 km<sup>3</sup> to account for the introduced metals. The last 3 rows show metal concentrations observed in oil field brines.


Another remarkable aspect of the Kupferschiefer mineralization, illustrated schematically in Figure 4A,B, is a regular regional scale metal zonation. As shown by a color-coded metal dominance map in the Atlas (but not reproduced here), the northeast of the map area in Figure 4D is largely unenriched in base metals and the original Kupferschiefer pyrite dominates. To the west of this pyrite zone lies an NE-SW trending band of about equal area where Zn is dominant. To the west of this lies a smaller band where Pb is dominant, to farther still to the lies an area where Cu is dominant, and still further west is an area of barren Rote Fäule.

What this suggests is that the oxidized brines have forced their way through the Kupferschiefer. Where only a little brine was forced through, sphalerite is the dominant mineral and Zn dominates. With more brine throughput Zn is flushed out, and Pb is the dominant metal. With still more throughput, successively more Cu-rich sulfides (chalcopyrite, then bornite, then chalcocite) dominate. With still more brine throughput, all the sulfides are oxidized, all the base metals are flushed out, and only hematite remains (Rote Fäule). Gold and platinum-group metals precipitated at the base of the zone of copper enrichment and the top of the encroaching Rote Fäule [38]. The metal enrichment process involves oxidative titration of the reduced Kupferschiefer shale by brine throughput. This story is a common one and applies to other types of base metal deposits, for example Kuroko-type volcanogenic massive sulfide deposits [39].

The progressive westerly increase of brine movement through the Kupferschiefer in Poland suggests the brines below the Kupferschiefer moved to the west. Near the Fore-Sudedetic Block the leakage through the Kupferschiefer was optimal for metal enrichment and the mines are located there, as indicated in Figure 4. Why was leakage easier through the Kupferschiefer in western Poland, and why does the brine throughput per unit area change so regularly? The brines must have had remarkably equal access to the base of the Kupferschiefer. Perhaps this is not surprising. The first Zechstein deposition was a thin basal limestone which could be very permeable and the Variscan sediments could be permeable. If the leakage was slow, uniform access to the base of the Kupferschiefer could be expected provided only that the underlying rocks were dramatically more permeable, as seems very likely. What is important here is that the metal zoning is remarkably coherent, so brine access must have been relatively uniform, and, in this case at least, the regular increase in leakage can be followed westward to the locations where it produced mineral deposits.

A couple of brief comments: First, transgressive reduced black shales lying between evaporates and red beds of continental origin are commonly enough mineralized that a "Kupferschiefer" sub-type of the sediment-hosted statiform copper deposit class has been distinguished [40]. Second, the estimate of the volume of brine expelled is for the eastern portion of the Southern Permian Basin. An equally large area of the western part of the basin would be required to produce the German Kupferschiefer deposits. Third, the few mm thick horizontal chalcocite and other veins in the Kupferschiefer indicate the brines forcing their way through were nearly lithostatically over-pressured, and that flow persisted through the Kupferschiefer for protracted periods of time. Near lithostatic pressures were required to jack the horizontal veins open, and time was required to fill them with mineralization. Over 80% of the mineralization is disseminated (replacement of framboids) which appears to have occurred early in the mineralization history, but the veins indicate overpressure. Fourth, convective flow is not responsible for the metal zoning. The Kupferschiefer is underlain by very permeable strata (the Weissliegend sands and conglomerates) which preclude the possibility of horizontal pressure gradient even vaguely large enough for brine convection to drive flow horizontally through the Kupfersciefer. Flow was driven by fluid overpressures (pressures greater than hydrostatic) produced by compaction, or possibly positive volume change hydrocarbon maturation reactions. Overpressured fluids were driven vertically (with perhaps slight, few meter scale, lateral diversions) through the Kupferschiefer. The metal zoning is due to the amount of brine throughput, as discussed above. Finally, the Kupferschiefer metallization records regional scale brine flow and documents leakage through a sealing shale capped by evaporites. Episodic Expulsion: Mississippi Valley Type Pb-Zn Deposits

The sediment-hosted base metal deposits, of which the Kupferschiefer is a sub-type, seem to have formed by quiet, steady expulsion of over-pressured brine. Mississippi Valley-Type (MVT) Pb-Zn deposits formed by short, sudden pulses of brine expulsion. Ore deposition is by cooling rather than reduction.

From the homogenization temperature of sphalerite fluid inclusions we know that ore deposition in MVT deposits occurred at T > 80 ◦C (Figure 5A), but the low maturation of conodonts indicates the sites could have been heated, cumulatively, for only a short period of time (<200,000 yrs; [41]). The deposits formed within a kilometer of the surface. Figure 5B shows that for the near subsurface to be even slightly warmed, a million years of compactive brine expulsion must occur in a few years. Temperature constraints thus sugges<sup>t</sup> that the ore deposition occurred in short pulses. Other observations support this conclusion. For example, 8 pulses of mineralization have been documented in the Buick mine in the Viburnum Trend (carbonate reef escape hatch; [42]). Eight episodes of chalcopyrite deposition, 6 of sphalerite, 5 of galena and quartz have been documented in the Tri-state district [43]. Thermal pulses are indicated by cathodoluminescent banding in hydrothermal dolomite that is coherent over 275 km south of the Viburnum Trend [44]). The ore minerals show corrosion between pulses of deposition, and the cathodoluminescent bands show unconformities, as would be expected if the hydrothermal discharge were pulsed, and between pulses, cool meteoric water incurred and partly dissolved the mineralization.

**Figure 5.** (**A**) Mississippi Valley-Type (MVT) lead-zinc deposits formed at shallow depths and temperatures >80 ◦C. From [45]. (**B**) To warm the near-surface significantly requires a million years of compactive expulsion to occur in <10 years. ( **C**) Cross basin hydrologic flow warms the discharge margin but also cools the basin. Temperatures at 1 km depth can be raised only to half the maximum in the basin. From [46].

Hydrologic flow across the basin is too steady to easily explain pulses in mineralization (topography changes slowly), cross basin flow would flush the brines before the margins are warmed, and cross basin flow at the rates required to warm the margins will cool the basin as shown in Figure 5C (see reviews in [6,41]). What caused the pulses of brine expulsion? For over 30 years I could find no good explanation, but now there is a mechanism that might just work: gas suddenly introduced to the brine-filled aquifer following the failure of a capillary seal, as discussed below.
