The hydrocarbons in natural gas are believed to come from two sources, one biological ('biogenic gas'), and the other from thermal cracking, 'primary thermal gas' from kerogen cracking and 'secondary thermal gas' from oil cracking 12. Although there is general agreement on the source of biogenic gas, disagreement persists over the origin of thermal gas. One point of controversy is that thermal cracking does not produce a gas resembling natural gas. Oil and kerogen pyrolysis typically give between 10 and 60% wt methane (C₁–C₄) 3456789 while natural gas contains between 60 and 95+% methane 10. None of the explanations that have been offered to explain this 1112131415 are satisfactory 16. Catalysis by reduced transition metals can, in theory, explain high-methane in natural gas 17, and this hypothesis is supported by experimental results. Crude oils and n-alkanes decomposed over reduced nickel and cobalt oxides produce gas resembling natural gas in molecular and isotopic compositions 18. There is, however, no evidence of metal activity in sedimentary rocks and therefore no compelling reason to question thermal cracking theory. Moreover, recent hydrous pyrolysis experiments with metal-rich Permian Kupferschiefer shale showed little evidence of catalytic activity, seemingly dismissing the possibility of a catalytic path to natural gas 19.

This changed with the recent disclosure of natural catalytic activity in marine shales at temperatures 300° below thermal cracking temperatures 20. Shales generated gas under anoxic gas flow at 50°C, nearly five times more gas than the same shale would generate at 350°C through thermal cracking. Although there was only indirect evidence for transition metals as the active catalysts, it nevertheless established natural catalytic activity in source rocks believed to be major sources of natural gas. There are, therefore, two possible paths to natural gas, a thermogenic path operating almost exclusively at high temperatures, and a catalytic path operating at much lower temperatures. The latter redefines the time-temperature dimensions of gas habitats opening the possibility of gas generation at subsurface temperatures previously thought impossible.

Thermodynamic equilibrium could shed light on which of these paths might dominate in nature. Thermal reactions are generally under kinetic control and their products removed from thermodynamic equilibrium while catalytic reactions are often under equilibrium control and their products near equilibrium. Hydrocarbons can achieve equilibrium through metathesis where homologues interconvert (Reaction 1).

Olefin metathesis 2122 is a well-known catalytic reaction shown in Reaction 2 for propylene. It is an extraordinary catalytic process because it breaks and makes carbon-carbon double bonds and reshuffles olefinic carbons distributing them randomly to new partners. Catalyzed by a variety of transition metals, it proceeds to equilibrium at low temperatures with conservation of π and σ bonds 23. Metathesis of methane, ethane, and propane is illustrated in Reaction 3, referred to here as 'gas metathesis'. Although hypothetical, it bears a strong resemblance to olefin metathesis in stoichiometry (1) and to low-temperature gas generation in marine shales 20. Hydrocarbons decompose over reduced nickel and cobalt oxides to C₁–C₃ compositions near equilibrium, possibly through catalytic gas metathesis 18. Equilibrium between hydrocarbons in natural environments is not limited to metathesis, however. Metastable equilibria have been reported from the interaction of hydrocarbons, water, and authigenic mineral assemblages 242526.

Gas metathesis without the aid of a catalytic agent is highly unlikely. Hydrocarbon cracking generates methane, ethane, and propane removed from thermodynamic equilibrium 327, and their extraordinary thermal stabilities 2829 preclude equilibrium over geologic time. Thus natural gas at metathetic equilibrium (3) would implicate catalytic assistance.

Here we address metathetic equilibrium in shale gas generation and in natural gas deposits, and discuss the genetic implications.

Methane or propane tends to dominate the hydrocarbons emerging from marine shales under anoxic isothermal gas flow 20. Fig. 1 illustrates two examples, a methane-rich gas from a high-maturity Mississippian Barnett shale (a) and a propane-rich gas from a lower-maturity Devonian/Mississippian New Albany shale (b).

The possibility that these are preexisting hydrocarbons desorbed under isothermal gas flow is inconsistent with the order in which hydrocarbons are released over time. Desorption under isothermal gas flow is a first order process where light hydrocarbons (C_x) will desorb before heavy hydrocarbons (C_y) with concentrations [C_x] and [C_y] in the effluent gas stream falling exponentially over time 30. In our analysis of exponential desorption, ratios ([C_x]/[C_y]) will also fall exponentially over time irrespective of the concentrations adsorbed on surfaces or in solution. Thus, the relative amounts of heavy and light hydrocarbons adsorbed will effect the ratio ([C_x]/[C_y]), but not its exponential fall over time with isothermal gas flow. It would not be possible for ([C_x]/[C_y]) to remain constant or increase over time, for example.

There is no evidence of desorption in Fig. 1. [C₁]/[C₃] in (a) rises from 20 to over 100 in the first 30 minutes of gas flow, then falls to 60 at the end of data collection. The ratio in (b) is nearly constant over 35 minutes of gas flow (~0.3), then rises sharply to 2.5 at the end of data collection. The remarkable proportionality between ethane and propane sustained throughout both reactions and its independence of methane concentrations are also inconsistent with desorption. Since desorption under isothermal gas flow should not produce [C₂]/[C₃] ratios remaining constant over time and [C₁]/[C₃] ratios increasing over time, it must be dismissed as a possible source of the gases produced in these experiments.

The gases are distinct in one other respect. Barnett gas is near thermodynamic equilibrium while New Albany gas is far removed from equilibrium. Equation 4 shows the equilibrium constant K for methane, ethane, and propane (3) (where C₁ = CH₄; C₂ = C₂H₆; C₃ = C₃H₈).

Log K for equilibrium at a reaction temperature of 200°C is 0.90 at one atmosphere 31. The average composition for the gases in Fig. 1 place the quotient Q, Q = (C₁ *C₃)/(C₂)², at log Q = 0.40 for Barnett gas and log Q = -1.30 for New Albany gas, where C₁, C₂, & C₃ are % vol.

A catalytic reaction can be metathetic, under equilibrium control, and still yield these hydrocarbons removed from equilibrium. All catalytic reactions tend to equilibrium over time in hydrocarbons that interconvert. Product compositions will therefore change over time, removed from equilibrium initially (short residence times), but approaching equilibrium with time. Thus, in flow reactors where residence times are short, gas metathesis could generate these hydrocarbons removed from equilibrium.

The New Albany reaction was repeated under closed conditions for evidence of gas metathesis over longer residence times. Gas flow was continued for 20 minutes to insure active gas generation, then the reactor was closed and its contents allowed to stand for 200 hours at 200°C. Fig. 2 shows the changes in gas composition over time and Fig. 3 shows the approach to equilibrium that attends these changes.

The New Albany shale generates gas removed from equilibrium under flow conditions (Fig. 1b) and approaching equilibrium under closed conditions (Fig. 3). Equilibrium over time is a signature of catalysis. In this instance, it progresses to equilibrium and gas dryness in concert (Fig. 2). Natural gases might also progress to equilibrium and dryness in concert if the natural process is similarly catalytic.

Two methods were used to approximate the equilibrium constant K (eq. 4) in the subsurface. One approximates Gibbs free energies as a function of temperature and pressure 32 and the other log K values at various temperatures for ideal gases at one atmosphere 31. For temperatures between 325 K and 575 K, and pressures between 3 MPa and 150 MPa, log K falls between 0.9 and 1.3 32. The second approximation 31 also places equilibrium limits within the same region: log K = 0.73 (575 K) and 1.4 (325 K). If natural gas compositions are constrained by equilibrium forces, they should have log Q values near these limits relative to the log Q limits for unaltered thermogenic gases.

Figure 4 is a histogram of log Q for offshore Gulf of Mexico gases 33. These gases were chosen because they are mostly free gases not associated with crude oils or other materials that might compromise their thermodynamic properties. The gases are divided into non-microbial and microbial according to dryness (% wt C₁ in C₁–C₄). The non-microbial gases are largely within the approximated equilibrium limits, while the microbial gases are clearly removed from those limits.

Gas products from thermal cracking experiments fall within the region marked 'Thermal Cracking' in Fig. 4327343536. We would expect 'primary thermal gas' from kerogen cracking and 'secondary thermal gas' from oil cracking to fall within this region as well. The displacement of natural gases to the right of this region by two log unit is therefore significant.

Figure 5 shows a similar displacement in 1600 gases from various basins in North America.

Figs. 4 &5 challenge the notion that thermal cracking is the source of natural gas irrespective of thermodynamic equilibrium. How do we explain log Q values displaced from where they should be by two log units? The fact that they are displaced towards thermodynamic equilibrium, in this case metathetic equilibrium, raises the possibility that they were generated catalytically under equilibrium constraints, not thermally under kinetic constraint. It is possible, in other words, that there was no displacement; they were generated where they are.

Figure 6 supports this supposition. It shows a strong correlation between [(CH₄)*(C₃H₈)] and [(C₂H₆)²] consistent with gas compositions constrained to equilibrium. It follows a power function (the solid line) rather than a linear function (lines parallel to the dashed line). The ratio Q ([(CH₄)*(C₃H₈)]/[(C₂H₆)²]) thus varies systematically with concentrations, displaced from equilibrium at high concentrations of C₂ and C₃ (wet gas) and at equilibrium at low concentrations C₂ and C₃ (dry gas). Fig. 7 shows the approach to equilibrium with gas dryness. The line through the data is an equilibrium curve for a reversible reaction approaching equilibrium over time (t): Q = K_equi(1-e^-αt). Time (t) has been replaced with C₁/(C₂+C₃) in Fig. 7, consistent with the generally accepted notion that wet gas converts to dry gas over geologic time 12.

Isotope ratios (¹³C/¹²C) in petroleum hydrocarbons are believed to be functions of primary biological inputs and isotope effects in gas generation and decomposition 2. Isotopic equilibrium is another factor that can alter primary biological isotope ratios. Replacing ¹²C with ¹³C lowers the free energy of hydrocarbons because the carbon and hydrogen bonds to ¹³C are stronger than the same bonds to ¹²C. Bond energy enhancement increases with carbon number. Replacing ¹²C with ¹³C in ethane yields more additional bond energy than it does in methane, for example. Thus, at isotopic equilibrium, ¹³C will be distributed preferentially in the higher hydrocarbons according to their respective lower free energies 37. Original ¹³C input will control the amount of ¹³C shared between hydrocarbons at equilibrium, but their respective free energies will control how ¹³C is distributed between them. The distribution of ¹³C at equilibrium will therefore be independent of original input, rates of gas generation and rates of gas decomposition.

The isotopic equilibrium reaction for methane and ethane is shown in Reaction 5 and for methane and propane in Reaction 6 (the position of ¹³C in C₃H₈ is unspecified).

Eqs. 7 and 8 are the isotopic equilibrium equations with K_1,2 (7) the carbon isotopic equilibrium constant for methane and ethane, and K_1,3 (8) the carbon isotopic equilibrium constant for methane and propane: ¹²C_n and ¹³C_n are fugacities; ¹³C₂ = ¹³C¹²CH₆, and ¹³C₃ = ¹³C¹²C₂H₈.

Table 1 shows the molar isotope ratios and the quotients Q for methane, ethane, and propane in 285 natural gases and it includes data for catalytic gases 18 for comparison, to be discussed below. The quotients Q_1,2 and Q_1,3 are very close to theoretical equilibrium values at 400 K: K_1,2 = 2.039; K_1,3 = 3.101 37. All ratios show substantial invariance. The variance in Q_1,2 and Q_1,3 is one half that in the molar isotope ratios within them reflecting the correlations between molar isotope ratios shown in Fig. 8. Table 1 also displays the extraordinary match between catalytic gases and natural gases.

Fig. 9 shows the proximities of natural gases and catalytic gases to isotopic thermodynamic equilibrium on a log K scale.

The approach to equilibrium with dryness (Figs. 6 & 7) mirrors the experimental results shown in Figs. 2 &3. The isotopic results (Figs. 8 &9) reinforce the position that hydrocarbons in natural gas are generated under equilibrium constraints. It is a metathetic equilibrium and therefore a catalytic equilibrium.

Other explanations are less satisfactory. It is difficult to explain dry gas generation through thermal cracking 2829 and harder to explain gas metathesis through thermal stress. Equilibrium requires the facile exchange of carbon atoms between methane, ethane, and propane. Carbon-carbon and carbon-hydrogen σ bonds are broken and reformed with overall bond conservation. This is unprecedented in thermal hydrocarbon reactions and inconceivable without catalytic assistance.

The following results support our position that natural gases are at or near thermodynamic equilibrium:

1) Gas compositions are significantly displaced from thermogenic compositions to equilibrium compositions (Figs. 5 &6).

2) There is a strong nonlinear correlation between [(C₁)*C₃)] and [(C₂)²] in which the quotient Q converges on equilibrium as wet gas progresses to dry gas (Figs. 6) consistent with an approach to equilibrium over geologic time (Fig. 7).

3) The isotopic compositions of methane, ethane, and propane are also constrained to equilibrium compositions (Figs 8 &9).

We propose catalytic gas metathesis as the source of equilibrium in natural gas. The natural catalytic activity in marine shales 20, or some similar activity in other sedimentary rocks, is probably the source of equilibrium in natural gas deposits. This view is supported by the New Albany shale experiment in which Q progressed to metathetic equilibrium over time as wet gas progressed to dry gas (Figs. 2 &3). A mechanistic connection between degradation to methane 20 and metathesis is suggested.

Catalysis by low valent transition metals 161718 may be the source of gas metathesis and degradation in the origin of natural gas. The match in isotope ratios between catalytic gases and natural gases (Table 1) supports this position and the sensitivity of marine shales to oxygen poisoning 20 supports it as well.

The position that thermal cracking adequately explains the origin of natural gas 12 is no longer tenable. It cannot explain the high concentrations of methane in natural gas 16, the distribution of light hydrocarbons 3839, and the associations with thermodynamic equilibrium reported here. Of the two possible pathways to natural gas, the catalytic path 20 appears the more attractive. It is simple, economic, and does not suffer the now mounting deficiencies challenging thermal cracking theory.

The authors declare that they have no competing interests.

FDM formulated theory, experimental design and supervised the experimental work. DJ contributed to the experimental work and helped in shaping the paper. EH was instrumental in early strategy and in writing the ms.