Marine shales release gas under isotherm gas flow at low-temperatures 1. Gas is released discontinuously, in distinct aperiodic episodes that continue over time. It is nonlinear kinetic behavior resembling chaotic catalysis by transition metals 2. Trace levels of oxygen suppress gas emission, and gas compositions reflect equilibrium control. The recent disclosure of natural gas at thermodynamic equilibrium and catalytic gas from marine shales similarly constrained, strongly suggests natural catalysis as the source of natural gas 3.

The origin of natural gas remains controversial, however. Many believe that thermal cracking is the source and cite various pyrolysis simulation experiments to support this view 4567891011121314. Recent hydrous pyrolysis experiments would seem to rule out natural catalytic activity in general 15, and activity by transition metals 16 in particular. This conclusion was based on the premise that any natural catalytic activity that might exist would express itself under hydrous pyrolysis conditions. However, natural activity is a low-temperature phenomenon that is not observed at pyrolysis temperatures (> 300°C) 1.

Here we address catalytic gas generation under open and closed conditions to distinguish it from alternative explanations, desorption and thermal degradation in particular. Gas compositions change very little over time in thermal degradation experiments under open and closed conditions. Methane concentrations remain within a narrow range between 20 and 60% vol (C1-C5) irrespective of kerogen type, temperatures, or reaction conditions 1718192021. Desorption under isothermal gas flow follows first-order kinetics with the lighter hydrocarbons eluting before the heavier hydrocarbons 22, as typically seen in gas chromatography. Both processes should give characteristic compositions over time, easily distinguishable from catalytic generation.

Rates of catalytic reactions are controlled by substrate concentrations at catalytic sites. They can vary in molecular weight if mass transport controls hydrocarbon transit to active centers. This is particularly the case in heterogeneous systems where active sites can be isolated from the hydrocarbon pools surrounding them. Mass transport can then control concentrations at active sites and thus product compositions. Under these circumstances, product compositions can be very different under gas flow where mass transport is active 22 and static conditions where it is not. In our systems, we would expect gas flow to carry higher hydrocarbons to active sites while static conditions will limit these sites to the hydrocarbons in-place. The purpose of this research was to see if there were dramatic differences in the gasses emitted under gas flow and static conditions consistent with catalytic generation as opposed to desorption or thermal degradation.

The kinetics of generation and desorption are different and their products reflect the differences. Desorption is first order and therefore yields characteristic exponential curves over time 22. Rates are proportional to k_a* [a], where k_a is the first order rate constant and [a] is the concentration of hydrocarbon a in the rock's kerogen and bitumen. Under isothermal gas flow, the gas-phase ratio of two desorbing hydrocarbons, (a) and (b), will be proportional to k_a* [a]/k_b* [b], where (a) and (b) denote concentrations in the effluent gas and [a] and [b] denote concentrations in solution (kerogen and bitumens). If a is the lighter hydrocarbon (k_a > k_b), the ratio (a)/(b) emerging from the rock will fall exponentially over time as the ratio of their concentrations in solution ([a]/[b]) falls exponentially over time. The ratio (a)/(b) will change with [a]/[b], but the exponential fall over time will not. In first order desorption, the ratio (a)₁/(b)₁ at any point in time t₁ is greater than (a)₂/(b)₂ at dt: [(a)₁/(b)₁]* [(1-k_a)/(1-k_b)] = (a)₂/(b)₂. This will be true at all points in time, from the onset of desorption where [a] and [b] are high, to infinite time when they are in trace amounts. Thus, the fall in (a)/(b) is independent of [a]/[b] and time. It is therefore independent of sample history and preparation. It shall make no difference how much a and b are lost or retained in sample preparation. Desorption can only be the major source of a and b if (a)/(b) falls exponentially over time.

The dynamics of generation are different. The concentrations of a and b in the emerging gas now become proportional to concentrations of precursors that generate a and b. These could be free hydrocarbons or hydrocarbon appendages to kerogen. The ratio (a)/(b) now becomes a function of the average molecular weight (MW) of the substrates feeding the reaction. If the catalytic rate constants are about equal, (a)/(b) = [S_a]/[S_b], where S_a and S_b are light and heavier substrates, respectively. Substrate MW will diminish over time as higher MW substrates are converted to lower MW substrates. Thus, [S_a]/[S_b] will increase as [S_b] → [S_a]. We anticipate two possibilities for catalytic gas generation. With mass transport under gas flow, [S_a]/[S_b] should remain relatively constant (steady-state) as higher hydrocarbons are delivered to active sites. Under static conditions where active sites are limited to the hydrocarbons in-place, average MW will decline over time and [S_a]/[S_b] will increase. Thus, the ratio (a)/(b) should increase under static conditions were mass transport is minimal and remain relatively constant under flow conditions where mass transport is active.

We analyzed gas compositions (% mol C₁-C₅) from three shales under isothermal gas flow using a procedure described elsewhere 1. Figure 1 shows the ratio (C₃)/(n-C₄) increasing under He flow at 50°C (Floyd shale). A similar curve obtains from New Albany shale at 100°C, and a slightly declining curve is seen from Mowry shale at 50°C (Table 1). The propane and butane released under isothermal gas flow is therefore generated under gas flow. The fact that it occurs at 50°C, is suppressed by oxygen, and is episodic 1, points to catalytic generation, as opposed to thermal generation.

Gas compositions (% vol C₁-C₅) changed only moderately as Fig. 1 and Table 1 illustrate. The Mississippian Floyd shale (Black Warrior Basin, MISS) and the Devonian/Mississippian New Albany shale (Illinois Basin, ND) gave similar compositions (Figs. 2 &3) while the Cretaceous Mowry shale (Colorado) gave a distinctly different composition (Fig. 4). Figs. 2, 3, 4 (Table 1) displays three distinct compositions, but each remains relatively constant over time under isothermal gas flow.

To replicate flow conditions without mass transport, samples were subjected to multiple closed reactions. Products were removed between reactions by syringe at ambient temperatures. This generated a series of products representing gas generation unaffected by gas-flow mass transport. The two procedures (gas flow and sequential closed reactions) should give similar products if mass transport makes no contribution to product compositions and substantial differences if mass transport controls product compositions. Seven sequential products from Floyd shale at 50°C are shown in Fig. 5 (Table 2) and six sequential products from Mowry shale at 100°C are in Fig. 6 (Table 3).

The two shales emit entirely different gases under flow and closed conditions with two distinctions outstanding. First, overall compositions remained distinct throughout both procedures. The Floyd product was dominated by ethane under gas flow (Fig. 2) and propane and butane under closed conditions (Fig. 5). Mowry gas under flow conditions (Fig. 4) bears no resemblance to Mowry gas under closed conditions (Fig. 6). Secondly, gas compositions under closed conditions progressed to lighter hydrocarbons over time. The effect was subtle but clear in the Floyd experiment (Fig. 5), and dramatic in the Mowry experiment (Fig. 6). It is definitive evidence against desorption as the source of hydrocarbons in these experiments.

The stark differences between gas flow and closed conditions are also illustrated in data variance. Average variance (% of mean over time) in Floyd data was 14% under gas flow (Table 1) and 28% in closed reactors (Table 2). The average variance in Mowry data went from 33% under gas flow (Table 1) to 740% under closed conditions (Table 3). It is noteworthy that the product in Fig. 4 bears no resemblance to that in Fig. 6 for the same shale at the same temperature under flow and static conditions, respectively.

Mass transport accounts for these differences. Gas flow transports hydrocarbons at rates greatly exceeding diffusion rates 2223. We can therefore anticipate a uniform steady-state composition of hydrocarbons at catalytic sites reflecting equivalent rates of hydrocarbon delivery and removal and changing compositions at catalytic sites under static conditions. The relatively uniform distribution of products under gas flow and the changes in hydrocarbon compositions under static conditions are consistent with this interpretation.

The natural progression of wet gas to dry gas is rarely reflected in thermal cracking simulations. This is partly because ethane and propane will not crack at temperatures typically used in thermal cracking experiments 456789101112131415. Dry gas generation in the laboratory has been reported 27, but only at temperatures above 450°C where these hydrocarbons decompose. With half-lives in the tens of millions 25 to billions of years 26 at temperatures under 200°C, it is very unlikely that thermal cracking could be the source of dry gas in sedimentary basins.

Wet gas to dry gas is replicated here for the first time. It proceeds through natural catalytic activity carried from the subsurface. It is paleoactivity that has persisted over geologic time as opposed to artificial activity created under laboratory conditions. Shales express activity at ambient temperatures without chemical or thermal activation and lose activity only under artificial conditions, temperatures over 300°C or exposure to oxygen 1. Marine shales and their counterparts should generate gas in the subsurface as they do in the laboratory, and therefore propose natural catalysis as the source of natural gas in sedimentary basins. It explains high methane concentrations in natural gas, methane enrichment over geologic time, and natural gas at thermodynamic equilibrium 3. We reject the thermal cracking hypothesis because it cannot explain these observations.

Hydrocarbons released from Floyd, New Albany, and Mowry shales are products of a natural catalytic process under fluid dynamic control. Desorption of preexisting hydrocarbons makes no significant contribution to the products in our experiments and thermal degradation at these temperatures would be unprecedented and highly unlikely.

The sharp distinctions in yield and compositions between closed reactions and gas flow reactions reflect relatively uniform substrate compositions under gas flow and changing compositions in closed reactions.

These results add to the evidence 13 that the natural catalytic activity in carbonaceous sedimentary rocks is the source of natural gas.

The Mississippian Floyd shale and Devonian/Mississippian New Albany shales are described elsewhere 1. The Cretaceous Mowry shale is whole core (2500 m) from an unknown well in Colorado. Rock-Eval: S1 = 2.61; S2 = 9.33; S3 = 0.15; Tmax = 439. Total Organic Carbon (Leco) = 2.5. The experimental procedures used in sample preparation and product analysis are described elsewhere 1. The experimental procedure under gas flow is also described in that publication. Closed experiments were carried out in 5 ml glass vials fitted with PTFE/SIL septa purchased from Cole-Parmer (Vernon Hills, IL, USA). Rocks were prepared for analysis by grinding to 60 mesh at ambient temperatures in glove bags under argon, placed in vials under argon, and sealed with open caps fitted with septa. The caps were then wrapped tightly with plastic electrical tape at the vial-screw cap interface to prevent leakage under heating. The vials proved to be leak-proof for the duration of our experiments at 100°C. In sequential experiments, a charged vial containing about 1 gm sample was heated for a time segment, usually 1 hour at constant temperature (± 5°C), then cooled. Product gas was removed by syringe and analyzed as previously described. The vial was then purged of all hydrocarbons by flowing argon in and out of the vial through two needles inserted through the septa (15 min, ~5 ml/min). The purged vials were then reheated for a period of time, usually one hour, and purged again. This procedure was repeated several times. Hydrocarbon compositions in Tables 2 &3 reflect the gas generated in the time segments indicated. The Floyd shale was heated four times at 50°C for one hour, a fifth time for 19 hours at 50°C, and a final time for 19 hours at 50°C. The experiments were duplicated with another sample of Floyd shale (Duplicate Reaction, Table 2). The product from the 3^rd heating was inadvertently lost and therefore not analyzed. Duplicate reactions of Mowry shale were carried out in four sequential one-hour reactions at 100°C and a fifth reaction at 100°C for 24 hours (Table 3). Duplicate experiments did not use aliquots of 60 mesh shale. Different samples from the same source were subjected to the same experimental procedures: grinding in argon, sieving, and so forth. The variations in yield and product compositions shown in Tables 2 &3 therefore reflect heterogeneity in samples as well as the variance in our analytical procedure.

The authors declare that they have no competing interests.

FM formulated theory, and both authors contributed to the experimental work and the final the paper.