Chemical Geology 339 (2013) 194–204

Contents lists available at SciVerse ScienceDirect

Chemical Geology

j ourna l homepage: www.e lsev ie r .com/ locate /chemgeo

Isotope reversals and universal stages and trends of gas maturation insealed, self-contained petroleum systems

Barbara Tilley ⁎, Karlis MuehlenbachsDepartment of Earth and Atmospheric Sciences, 1-26 Earth Sciences Building, University of Alberta, Edmonton, Alberta, Canada T6G 2E3

⁎ Corresponding author. Tel.: +1 780 492 2827; fax:E-mail address: btilley@ualberta.ca (B. Tilley).

0009-2541/$ – see front matter © 2012 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.chemgeo.2012.08.002

a b s t r a c t

a r t i c l e i n f o

Article history:Accepted 8 August 2012Available online 17 August 2012

Keywords:Isotope reversalShale gasMaturationClosed systemRollover

Isotope geochemistry is now a tool for shale gas exploration, largely due to the association of isotope reversalswith mature, highly productive shale gas. Its utility, however, depends on an understanding of the isotope sys-tematics for the particular region of interest, as well as for shale gas maturation in general. This paper reviewsand re-examines isotope data from four published papers that include shale gas from the Barnett and FayettevilleShales (Rodrigez and Philp, 2010; Zumberge et al., 2012), and gas from fractured reservoirs in the Appalachians(Burruss and Laughrey, 2010) and the Foothills of the Western Canada Sedimentary Basin (WCSB) (Tilley et al.,2011). New shale and tight sandstone gas data are also presented for theWCSB. Comparisons of these data showthat the progression through three stages of gas maturation (pre-rollover zone, rollover zone and post-rolloverzone) is universal in sealed, self-contained petroleum systems and that each zone has characteristic isotopic re-lationships and trends that are seen in all areas examined.Gases in the pre-rollover zone are isotopically normal (δ13Cmethanebδ13Cethanebδ13Cpropane) unless mixing of gasesfromdifferent sources has occurred. In the rollover zone, δ13Cethane and δ13Cpropane become progressivelymore neg-ative as δ13Cmethane becomes less negative, and ethane andmethane are reversed (δ13Cethanebδ13Cmethane) only to-wards the most mature portion of the rollover zone. At the beginning of the post-rollover zone, δ13Cethanebδ13Cmethane, but as δ13Cethane and δ13Cpropane become increasingly less negative at varying rates, ethane and pro-pane may or may not be reversed with respect to each other at the highest maturities. δ2Hmethane in gases of thepost-rollover zone approaches isotopic equilibrium with local formation water, generally becoming more nega-tive with increasing maturity. Correct assignment of maturity stage could be of importance because the rolloverzone may represent the peak of high productivity shale gas whereas the post-rollover zone may represent a de-cline in productivity (Burruss and Laughrey, 2010).

© 2012 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1952. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1953. Re-interpretation of published data with new data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

3.1. Appalachian fractured reservoirs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1973.2. Appalachian shale gas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1973.3. Barnett and Fayetteville Shale gas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1983.4. New data for Doig and Montney Fm. Tight gas and Horn River Shale gas in the WCSB . . . . . . . . . . . . . . . . . . . . . . . . 1993.5. WCSB Foothills fractured reservoirs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200

4. Comparison of gas maturation in different basins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2024.1. Pre-rollover/rollover zone transition and rollover zone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2024.2. Post-rollover zone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203

5. Conclusions and implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204

+1 780 492 2030.

rights reserved.

-50 -45 -35-40

-30

-45

-35

-40

-25

conventional

reversed

δ13

C methane (per mil)

conventional maturation

rolloverzone

reversed gases

‘normal’ gases

pre-rollover zone

ethane reversal

δ13C2<δ13C1

δ13C1<δ13C2

δ13C1<δ13C2

increasing maturity

Fig. 2. δ13Cmethane (C1) versus δ13Cethane (C2) cross plot illustrating the distinction be-tween the ethane reversal and the rollover zone (gray shaded area outlined in black) inwhich gases can be normal (area outlined in blue) or reversed (triangular area outlinedin red). Curved black arrows indicate increasing maturity. Data are from the BarnettShale (Zumberge et al., 2012).

195B. Tilley, K. Muehlenbachs / Chemical Geology 339 (2013) 194–204

1. Introduction

Until recently, reports of isotopically reversed thermogenic gases(δ13Cpropanebδ13Cethanebδ13Cmethane) were rare in the literature (Jendenet al., 1993; Dai et al., 2004; Seewald and Whelan, 2005). Their originwas poorly understood and was generally attributed to mixing of differ-ent gases. However, in the past three years, four papers that each presentsignificant sets of data featuring isotopically reversed gases from NorthAmerican sedimentary basins (Fig. 1) have beenpublished. These includeisotope data for shale gas from the Barnett and Fayetteville Shales(Rodriguez and Philp, 2010; Zumberge et al., 2012), and gas from frac-tured reservoirs in the Appalachians (Burruss and Laughrey, 2010) andthe Foothills of the Western Canada Sedimentary Basin (WCSB) (Tilleyet al., 2011). Though no longer oddities, the mechanism that createdthe reversed gases is still not well understood and is controversial.Burruss and Laughrey (2010) postulate mixing of gases combined withRayleigh fractionation during redox reactions at late stages of diagenesis(250–300 °C) to explain the isotopic reversals that they observe in theAppalachians. Zumberge et al. (2012) support Tang and Xia's (2011)‘water reforming of residual organic matter in shale’mechanismwhere-by isotopically light ethane (and presumably propane) is created by thereaction of the resulting H2 and CO2. Tilley et al. (2011) emphasize that(1) isotopically reversed gases (except where created by mixing,Jenden et al., 1993) are restricted to sealed, self-contained hydrocarbonsystems (‘shale gas’) where gas is a mixture resulting from conventionalmaturation of kerogen and cracking of residual oil; and (2) isotopicallyreversed gases in fractured reservoirs may be migrated mature shalegas, in which case, these gases can be interpreted to be related to shalegas.

Isotope geochemistry is now a tool for shale gas exploration. Itsutility however depends on an understanding of the isotope system-atics for the particular region of interest as well as for shale gas mat-uration in general. The purpose of this paper is to review and comparedata from the aforementioned papers, along with new tight gas andshale gas data from the WCSB, to derive a general model for the iso-topic evolution of ‘shale gas’ from low to high maturity. The graphicalpresentation of the data given here, which is different from that in the

Mi.600

600 Km.

BarnettShale

FayettevilleShale

Appalachians

Utica Shale, Quebec

Horn River Basin

WCSB FoothillsFractured Reservoirs

Montney/Doig Fms.

N

Fig. 1. Generalized map of North America showing relative locations of the sedimenta-ry basins with isotopically reversed gases that are examined in this study. WCSB =Western Canada Sedimentary Basin.

previous papers, will clearly illustrate that there are three distinctstages of shale gas maturation that each have distinct isotope rela-tionships and trends.

2. Terminology

The terms rollover, isotope reversal, and isotopically reversed gashave been used interchangeably when discussing shale gas. However,these terms are not synonymous and a clarification may be useful.

Isotope rollover refers to the change in the direction of the maturitytrend of isotope ratios of one gas component (e.g. ethane) from becom-ing less negative with increasing maturity (or decreasing wetness) asoccurs in conventional gas systems to becomingmore negative with in-creasingmaturity (or decreasing wetness). In other words, a reversal inthe direction of the maturation trend of the isotope ratios for ethaneoccurs at the rollover. In this paper, we will refer to the rollover as azone in which the isotope ratios of ethane become more negativewith increasing maturity. Fig. 2 illustrates the rollover zone on a crossplot of methane and ethane isotope ratios (reported in per mil with re-spect to the VPDB zero point on the conventional carbon isotopic scale),where increasing maturity is represented by increasingly less negative

0 5 10 15 20 25

-45

-40

-35

-30

-25

Wetness (ΣC2-C5)/(ΣC1-C5) %

Barnett Shale gas1,2

Fayetteville Shale rollover gas1

Fayetteville Shale post-rollover gas1

Appalachian gas3

Appalachian post-rollover gas3

WCSB Triassic Montney-Doig gas5

WCSB Horn River Shale gas5

WCSB Triassic-Permian Foothills gas4

pre-rollover zone

rollover zone

post-rolloverzone

δ13C

eth

ane

(per

mil)

Fig. 3.Wetness versus δ13Cethane plot with data from areas described in this paper. Yel-low shading denotes the pre-rollover zone, gray— the rollover zone, and green squaresand stars are post-rollover zone gases in the Fayetteville Shale and the Appalachians,respectively. Triangles and solid black circles at δ13Cethane>−40‰ also representpost-rollover zone gases.Data sources: 1Zumberge et al. (2012); 2Rodriguez and Philp (2010); 3Burruss andLaughrey (2010); and 4Tilley et al. (2011), 5Table 1.

δ13

C ethane (per mil)

δ13C

pro

pane

(pe

r m

il)

normal

reversed

1NY Blackriver

1PA Blackriver1OH Blackriver

1PA Baldeagle

1KY Trenton1WV Trenton

1OH Clinton-Medina1PA Clinton-Medina

1PA Utica Shale

2Quebec mudgas

3PA Marcellus Shale

3Quebec Utica Shale (wellhead)

δ13

C methane (per mil)

δ13C

eth

ane

(per

mil)

reve

rsed

nor

mal

δ13

C methane (per mil)

δ 2H

met

hane

(pe

r m

il)

4SW PA Marcellus Shale4NC PA Marcellus Shale

1NY Blackriver

1PA Blackriver1OH Blackriver

1PA Baldeagle

1KY Trenton1WV Trenton

1OH Clinton-Medina1PA Clinton-Medina

1PA Utica Shale

3NC PA Marcellus Shale

4NC PA Marcellus Shale well1 4NC PA Marcellus Shale well2

2Quebec mudgas 2Quebec Utica Shale 3Quebec Utica Shale (wellhead)

1NY Blackriver

1PA Blackriver

1OH Blackriver

1PA Baldeagle

1KY Trenton1WV Trenton

1OH Clinton-Medina1PA Clinton-Medina

1PA Utica Shale

5NY Marcellus Shale

5SW PA Marcellus Shale

5NY Marcellus Shale

5SW PA Marcellus Shale

-40 -35 -30

-40

-35

-30

-25

-45

-45 -40 -35 -30 -25

-40

-35

-45 -40 -35 -30 -25

-130

-150

-170

-190

-210

-30

(b)

(c)

(a)

196 B. Tilley, K. Muehlenbachs / Chemical Geology 339 (2013) 194–204

197B. Tilley, K. Muehlenbachs / Chemical Geology 339 (2013) 194–204

methane isotope ratios. The identification of a gas as a rollover gas is notalways trivial. Isotopic analysis of a suite of gas samples from one unit inone basin must be followed by a comparison of the isotope ratios ofthese gases based on their maturity.

An isotopically reversed gas has δ13Cpropanebδ13Cethanebδ13Cmethane

which is the reverse of the relationship between the isotope ratiosof these gas components in a conventional thermogenic gas whereδ13Cmethanebδ13Cethanebδ13Cpropane. A gas may be in the rollover zone,but may not be a reversed gas (Fig. 2). Also, as will be seen later inthis paper, a gas may be reversed, but may not be in the rollover zone.The identification of an isotopically reversed gas requires only the isoto-pic analysis of the methane, ethane (and propane) components of onegas sample.

Gases may also be partially reversed where either

(a) δ13C ethanebδ13C methane (reversed) but δ13Cethanebδ13Cpropane(normal); or

(b) δ13Cpropanebδ13Cethane (reversed) but δ13C methanebδ13Cethane(normal).

Partially reversed gases are relatively common and can be createdby processes such as mixing and biodegradation at a range of matu-rities. It is the fully reversed gases that are difficult to understand.

3. Re-interpretation of published data with new data

3.1. Appalachian fractured reservoirs

Isotopically reversed thermogenic gases were first noted in theAppalachians by Jenden et al. (1993) who at that time interpreted thereversal to be the result of mixing of low and high maturity gases.Since then, Burruss and Laughrey (2010) have presented a very detailedcarbon and hydrogen isotope study of low to very high maturity gasesfrom fractured reservoirs throughout the Appalachian Basin. Theirresults are replotted in Fig. 3 (together with data from other basins)and in Fig. 4 to highlight the isotopic relationships and trends in threegas maturation zones that we propose are characteristic of shale gas.The wetness versus δ13Cethane plot in Fig. 3 and the δ13Cmethane versusδ13Cethane plot in Fig. 4a clearly illustrate the three maturation zones.There is a zone of conventional maturation (or the pre-rollover zone;indicated as the yellow zone in Figs. 3 and 4a) for the low maturitygases where δ13Cethane becomes less negative as maturity increases(gas becomes drier and δ13Cmethane becomes less negative). For driergases and at slightly less negative δ13Cmethane (gray zone on Figs. 3and 4a), δ13Cethane apparently reverses and becomes more negative asδ13Cmethane becomes less negative. This mimics the rollover zone ofshale gases (Zumberge et al., 2012) but is interpreted by Burruss andLaughrey (2010) as being due to mixing of migrated gases.

The most significant contribution of the Burruss and Laughrey(2010) study in terms of our work is the recognition of an extremelyhigh maturity zone (>250 °C in their study) beyond the rolloverzone, where the stability limit of hydrocarbon gases may have beenexceeded. This extremely high maturity zone is referred to here asthe post-rollover zone (green symbols in Fig. 3 and shaded green inFig. 4a). In the post-rollover zone, δ13Cmethane is less negative than−33‰ and δ13Cethane reverses again and becomes less negativewith increasing maturity (drier gas and less negative δ13Cmethane).The gases in this post-rollover zone have δ13Cethanebδ13Cmethane. Aplot of δ13Cmethane versus δ2Hmethane for the Burruss and Laughrey(2010) data (Fig. 4b) is interesting when considered in terms of therollover and post-rollover zones. In the rollover zone (or perhapspseudo-rollover zone may be a more appropriate term for these data),

Fig. 4. Isotope cross plots for Appalachian data. (a) δ13Cmethane versus δ13Cethane cross plot; (bpropane cross plot. Colored shading denotes the maturation stage: yellow— pre-rollover zone;tral, PA = Pennsylvania State, NY = New York State, OH = Ohio State, KY = Kentucky StaData sources: 1Burruss and Laughrey (2010); 2Molgat et al. (2011); 3our unpublished data; 4B

δ2Hmethane becomes less negative as δ13Cmethane becomes less negative,the trend expected for gases of increasing maturity. In contrast, thepost-rollover zone is characterized by δ2Hmethane becomingmore nega-tive as δ13Cmethane becomes less negative. As Burruss and Laughrey(2010) state, this δ2Hmethane trend suggests isotopic exchange betweenmethane and formation water.

The isotopic trends of δ13Cethane and δ13Cpropane in each of thethree maturation zones are shown for the Burruss and Laughrey(2010) data in Fig. 4c. In the pre-rollover zone, both δ13Cethane andδ13Cpropane become less negative as maturity increases (yellow zoneson Fig. 4c). In the pseudo-rollover zone (gray), δ13Cethane and δ13Cpropaneboth become more negative as maturity increases. Note that δ13Cethaneand δ13Cpropane are reversed (i.e. δ13Cpropanebδ13Cethane) only in themost mature portion of the pseudo-rollover zone. In the post-rolloverzone (green), δ13Cethane and δ13Cpropane again both become less negativewith increasing maturity. In most of the post-rollover zone, δ13Cethaneand δ13Cpropane are reversed, but the most mature gas is normal againwith respect to δ13Cethane and δ13Cpropane (i.e. δ13Cethanebδ13Cpropane).

3.2. Appalachian shale gas

Themain targets for shale gas in theAppalachians are theOrdovicianUtica Shale and the Devonian Marcellus Shale. Limited isotopic datafrom these shale gases are available from a variety of sources. Molgatet al. (2011) presented a mudgas isotope profile from a Quebec wellthat includes Ordovician Utica Shale gas. Baldassare (2011) providedtwo depth profiles of methane and ethane isotope data for DevonianMarcellus Shale gas in north-central Pennsylvania (Bradford County)and a cross plot of hydrogen and carbon isotope data for methanefrom Marcellus Shale gas in southwest and north central Pennsylvania.We have estimated values from Molgat et al.'s (2011) and Baldassare's(2011) diagrams and presented the resulting data in Fig. 4 for compar-ison with the Burruss and Laughrey (2010) data. Data for three samplesof Marcellus Shale gas in New York and southwest Pennsylvania fromOsborn and McIntosh (2010) have also been included in Fig. 4a and b.In addition, data from one Utica Shale gas sample from Quebec andtwo Marcellus Shale gas samples from north-central Pennsylvaniafrom our unpublished database are also plotted in Fig. 4a and c.

Mudgas data for the Quebec well (Molgat et al., 2011) mimic theshape of the Burruss and Laughrey data on the δ13Cmethane versusδ13Cethane diagram (Fig. 4a). The trends of increasing to decreasing to in-creasing δ13Cethane values are the same but the inflection points occur atmore positive and more negative δ13Cethane values and δ13Cmethane isshifted to more negative values. The Utica Shale gas in the Molgat etal. (2011) mudlog occurs at the beginning of the post-rollover zonebased on the trends of the δ13Cethane values (Fig. 4a). A Utica Shale gasfromQuebec sampled at thewellhead (our unpublished dataset) is rep-resented by the blue cross that occurs on the mudgas trend in the roll-over zone on Fig. 4a. A Utica Shale gas from Pennsylvania in the Burrussand Laughrey (2010) dataset (dark blue star on Fig. 4a) occurs at a verylate stage in the post-rollover zone. Unfortunately, there are no δ2Hdata provided by Molgat et al. (2011), but the Utica Shale gas fromPennsylvania (Burruss and Laughrey, 2010) fits in the post-rolloverzone based on its δ2Hmethane and δ13Cmethane (Fig. 4b). The mudgas datafor the Quebec well are also shown as a dashed line on the δ13Cethaneversus δ13Cpropane plot (Fig. 4c) and again mimic the pre-rollover androllover trends of Burruss and Laughrey (2010). No data were providedfor δ13Cpropane of the Utica Shale gas in the Molgat et al. (2011) mudgas.The Utica Shale gas from our database (light blue cross on Fig. 4c) hasδ13Cethanebδ13Cpropane (normal) and lies along the rollover zone isotopetrend for ethane versus propane (Fig. 4c). The Utica Shale gas from

) δ13Cmethane (VPDB) versus δ2Hmethane (VSMOW) cross plot; (c) δ13Cethane versus δ13C-gray— rollover zone; and green— post-rollover zone. SW= southwest, NC = north cen-te, and WV = West Virginia State.aldassare (2011), 5Osborn and McIntosh (2010).

δ2 H m

etha

ne (

per

mil)

δ13C methane (per mil)

pre-rollover

rollover

post-rollover Barnett 1

Barnett (low maturity)2

Barnett (high maturity)2

Fayetteville1

δ13C

pro

pane

(pe

r m

il)

δ13C ethane (per mil)

conventional

reversed

-50 -45 -40 -35-220

-200

-180

-160

-140

-120

-100

-45

-40

-35

-30

-25

-20

-30-45 -40 -35

-50 -45 -40 -35

-45

-40

-35

-30

-25

δ13C methane (per mil)

δ13C

eth

ane

(per

mil)

pre-rolloverrollover post-rollover

conventional

reversed

Barnett 1

Barnett (low maturity)2

Barnett (high maturity)2

Fayetteville1

Barnett Shale gaspre-rolloverrolloverpost-rollover

Fayetteville Shale gasrolloverpost-rollover

pre-rollover

rollover

post-rollover

(a)

(b)

(c)

Fig. 5. Isotope cross plots for the Barnett Shale, Texas and the Fayetteville Shale, Arkansas.(a) δ13Cmethane versus δ13Cethane cross plot; (b) δ13Cmethane versus δ2Hmethane cross plot;(c) δ13Cethane versus δ13Cpropane cross plot. Colored shading denotes the maturationstage: yellow — pre-rollover zone; gray — rollover zone; and green — post-rollover zone.Arrows indicate directions of increasing maturity.Data sources: 1Zumberge et al. (2012) and 2Rodriguez and Philp (2010).

198 B. Tilley, K. Muehlenbachs / Chemical Geology 339 (2013) 194–204

Pennsylvania (Burruss and Laughrey, 2010) has δ13Cpropanebδ13Cethane(i.e. reversed) and matches the post-rollover zone.

Osborn and McIntosh's (2010) Marcellus Shale gas data fromNew York and southwest Pennsylvania are shown as purple stars en-closed in a diamond and circles, respectively, in Fig. 4a and b. In theδ13Cmethane–δ13Cethane diagram (Fig. 4a), the Marcellus Shale gas fromNew York (in particular) plots very close to the pre-rollover portion of

Molgat et al.'s (2011) Quebec mudgas curve. All three of the MarcellusShale gases from the Osborn and McIntosh (2010) dataset have signifi-cantly less negative δ13Cethane values than any of the lower maturityfractured reservoir gases from the Burruss and Laughrey (2010) datasetand appear to define a pre-rollover zone that extends to less negativeδ13Cethane values. On the δ13Cmethane versus δ2Hmethane plot of Fig. 4b,the Osborn and McIntosh (2010) Marcellus Shale gas samples areconsistent with the pre-rollover zone gases of Burruss and Laughrey(2010).

Marcellus Shale gas data from the two depth profiles of Baldassare(2011) are shown in Fig. 4a. The depths of these gases range from 1341to 3094 m in well1 and 2133 to 3291 m in well2. Although there is nosystematic variation of carbon isotope ratios with depth in either well,on average based on δ13Cmethane values, well2 appears to have less vari-able andmoremature gas in theMarcellus Shale thanwell1 This suggeststhat gases fromwell2 are post-rollover zone gases, whereas gases fromwell1 span the range from rollover zone to post-rollover zone. Thewidevariation of carbon isotope values and lack of systematic variation withdepth most likely reflect variations in lithology, permeability, distribu-tion of organic matter and degree and timing of gas migration.

Fig. 4b includes δ2Hmethane data forMarcellus Shale gas samples fromsouthwest Pennsylvania compared to north central Pennsylvania(Baldassare, 2011). Unfortunately, the Baldassare (2011) gas samplesanalyzed for δ13Cmethane and δ13Cethane (Fig. 4a) are not the same asthose analyzed for δ13Cmethane and δ2Hmethane (Fig. 4b), so there is nodirect link between their samples in the two diagrams. The MarcellusShale gas samples from southwest Pennsylvania (Baldassare, 2011)have the most negative δ13Cmethane and δ2Hmethane values and definea pre-rollover trend on Fig. 4b similar to that of the Burruss andLaughrey (2010) data. Some of the more mature Marcellus Shalegases from north-central Pennsylvania (Baldassare, 2011) appear tooccur in the transition between the rollover and post-rollover zones.Other Marcellus Shale gas samples from the Baldassare (2011) datasetwith less negative δ13Cmethane perhaps show a hint of the tendency to-wards a trend of decreasing δ2Hmethane with increasing δ13Cmethane,characteristic of the post-rollover zone observed in the Burruss andLaughrey (2010) data.

Marcellus Shale gases from Tioga County in north-centralPennsylvania (our unpublished database) are shown in Fig. 4a and cas open multi-pointed purple stars. It is instructive that comparedonly to the Burruss and Laughrey (2010) fractured reservoir gas trendsfor methane, ethane and propane, these Marcellus Shale gases wouldbe classified as rollover zone or transition to post-rollover zone gases.However, compared to the Quebec mudgas data and the MarcellusShale gas data from Baldassare (2011), these gases appear to be post-rollover zone gases.

3.3. Barnett and Fayetteville Shale gas

A re-examination of published data for the Barnett and FayettevilleShale gases provides an opportunity to see if the isotope trends ob-served in the Appalachians apply to other basins. Fig. 5 shows plotsfor these shale gases that are comparable to Fig. 4 Appalachian plots.Wetness versus δ13Cethane data for the Barnett and Fayetteville Shalegases are included in Fig. 3. The Fayetteville Shale gases are generallymore mature (vitrinite reflectance equivalent values of Ro=2–3%)than the Barnett Shale gases (Ro=1–2%) (Rodriguez and Philp, 2010;Zumberge et al., 2012). Figs. 3 and 5a show that the majority of theBarnett Shale gases are in the pre-rollover zone and the rollover zone.The Barnett Shale gases in the pre-rollover zone (yellow) are conven-tional in that δ13Cmethanebδ13Cethanebδ13Cpropane. Although the widespread of the data generally makes it difficult to assign definitiveboundaries to the maturation zones, the boundary between thepre-rollover and rollover zones in Fig. 5a is based largely on the ethanerollover observed in the wetness versus δ13Cethane plot for the BarnettShale data (Fig. 3). The rollover zone (gray) represents the broad

Table 1Carbon isotopic data for ‘Shale’ gases from the WCSB.

Sample Rock unit Lat. Long. Depth(m)

Wetnessa

(C2+ %)δ13Cmethane

b

(per mil)δ13Cethane

(per mil)δ13Cpropane

(per mil)

MD1 Montney Fm. 55.6 −117.1 1034 7.7 −45.4 −36.5 35.6MD2 Doig Fm. 55.4 −119.2 2021 17.4 −46.2 −34.6 −33.5MD3 Montney Fm. 55.9 −120.8 2258 11.5 −41.3 −30.0 −27.0MD4 Montney Fm. 55.9 −120.8 2253 11.6 −40.4 −29.7 −26.4MD5 Montney Fm. 55.9 −120.8 2233 13.1 −42.2 −30.4 −28.2MD6 Montney Fm. 55.9 −120.8 2261 10.8 −39.6 −29.0 −25.5MD7 Montney Fm. 55.9 −120.8 2284 11.3 −37.4 −28.5 −26.0MD8 Montney Fm. 55.8 −120.5 2386 −39.5 −26.8 −20.5MD9 Montney Fm. 55.9 −120.9 2411 9.5 −41.4 −28.8 −25.9MD10 Doig Fm. 55.8 −120.9 2302 1.7 −36.3 −35.1 −26.8MD11 Montney Fm. 55.8 −120.9 2300 8.1 −36.0 −27.0 −23.6MD12 Doig Fm. 55.9 −121.1 2500 3.4 −37.4 −27.2 −21.7MD13 Doig Fm. 55.7 −120.9 2585 1.0 −37.5 −38.8 −33.4MD14 Montney Fm. 55.7 −120.8 2680 4.8 −36.4 −25.7 −19.8MD15 Montney Fm. 55.8 −120.9 8.0 −41.8 −27.9 −22.6MD16 Doig Fm. 55.5 −120.7 3100 0.5 −36.2 −42.2 −41.8MD17 Doig Fm. 55.5 −120.7 3111 0.6 −37.5 −41.4 −43.2MD18 Montney Fm. 55.5 −120.6 3097 0.7 −35.5 −41.5 −40.5HR1 Horn River Sh. 59.4 −122.1 0.2 −27.6 −33.8HR2 Horn River Sh. 59.4 −122.1 0.2 −32.1 −34.9 −38.8HR3 Horn River Sh. 59.4 −122.1 0.2 −31.3 −34.1 −37.3HR4 Horn River Sh. 59.3 −122.0 0.2 −31.2 −32.0 −35.5HR5 Horn River Sh. 59.3 −122.0 0.2 −30.7 −34.4 −36.9HR6 Horn River Sh. 59.4 −122.1 0.1 −34.5 −34.4 −37.4

a Gas wetness (C2+ %) defined as %Σ(C2−C5)/Σ(C1−C5).b With respect to the VPDB zero point.

199B. Tilley, K. Muehlenbachs / Chemical Geology 339 (2013) 194–204

zone where the general trend is δ13Cethane becoming more negative asδ13Cmethane becomes less negative. This trend is observed in both theBarnett and Fayetteville Shale gas datasets. It is interesting that the roll-over trend occurs at similar δ13Cmethane values in both datasets, but theFayetteville Shale gases have δ13Cethane values that are significantlymore negative than any of the Barnett Shale gases. It is also noteworthythat the Fayetteville Shale gases in the rollover zone are isotopicallyreversed with respect to methane and ethane whereas only the mostmature Barnett Shale gases are reversed.

Gases from the post-rollover zone are shaded green in Figs. 5a, b andc. The post-rollover zone was selected to correspond to the possibletrend of decreasing δ2Hmethane with increasing δ13Cmethane (green inFig. 5b). Many of the Fayetteville Shale gases and a few of the BarnettShale gases are in the post-rollover zone.

Several features of the data are notable.

1. Barnett Shale gases encompass the full range of maturation stagesfrom pre-rollover zone to post-rollover zone. Only gas samples atthe latest stage of the rollover zone are isotopically reversed withrespect to methane and ethane, and two samples at the lateststage of the post-rollover zone are not reversed.

2. Fayetteville Shale gases represent the rollover and post-rollovermaturation trends. All gases are isotopically reversed with respectto methane and ethane.

3. The post-rollover stage of maturity appears to have been attained bya few of the Barnett Shale gases and many of the Fayetteville Shalegases in these datasets. In both cases, post-rollover zone gasesshow trends towards less negative δ13Cethane (Fig. 5a) andmore neg-ative δ2Hmethane (Fig. 5b) as δ13Cmethane becomes less negative.

4. Compared to the Barnett Shale gases, Fayetteville Shale gases havesignificantly more negative δ13Cethane for comparable δ13Cmethane

values (Fig. 5a). This apparently reflects the greater maturity ofthe Fayetteville Shale gases (relative to the Barnett Shale gases).

5. The post-rollover zone may begin at more negative δ13Cmethane

values in the Fayetteville Shale gases (−39.2‰) than in the BarnettShale gases (about−37.2‰) (Figs. 5a, 5b).

6. All Barnett Shale gases show the conventional relationship betweenδ13Cethane and δ13Cpropane regardless of level of maturation (Fig. 5c).However, in the pre-rollover zone, δ13Cpropane and δ13Cethane becomeless negative together, whereas in the rollover zone they becomemore negative together.

7. Fayetteville Shale gases have more negative δ13Cethane and δ13Cpropanethan the Barnett Shale gases (Fig. 5c). The post-rollover gases generallyhave less negative δ13Cethane and δ13Cpropane than the rollover zonegases. Some of the Fayetteville Shale gases from both rollover andpost-zones are isotopically reversed with respect to ethane andpropane.

The above observations for the Barnett and Fayetteville Shalegases are similar to those made for the Appalachian data and suggestthat the isotopic trends and the maturation stages are universal toshale gases.

3.4. New data for Doig and Montney Fm. Tight gas and Horn River Shalegas in the WCSB

The TriassicMontney and overlyingDoig formations in northeasternBritish Columbia comprise a thick marine package of tight, very finegrained sandstone to siltstone embedded within a shale matrix(Fitzmartyn, 2009). Shoreline deposits occur to the east in northwestAlberta, where oil and gas recoveries have been by conventionalmeans. Marine depositional fans and lower shoreface deposits are char-acteristic to the west in northeast British Columbia where the sedimentpackage thickens and deepens (Potter et al., 2008; Fitzmartyn, 2009). Inthe past few years, these resources in British Columbia have beenexploited by horizontal drilling and considered to be a tight gas play.Table 1 includes a list of wetness and carbon isotope data for gasesfrom the Montney and Doig formations. All samples are from tight gasplays in northeastern British Columbia with the exception of samplesMD1 and MD2 which are conventional oil-associated gases from west-ern Alberta. Wetness versus δ13Cethane data are included in Fig. 3. Fig. 6shows isotope ratios of methane, ethane and propane for Montney and

δ13C ethane (per mil)

δ13C

pro

pane

(pe

r m

il)

-45 -40 -35 -30 -25

-45

-40

-35

-30

-25

-20

R4

R1

R2

Devonian Horn River Shale

Triassic Fractured Reservoirs, B.C.

Triassic Fractured Reservoirs, Alberta

Permian Fractured Reservoirs, B.C.

Triassic Doig Fm.Triassic Montney Fm.

normal

reversedpre-rollover

rollover

post-rollover

-45 -40 -35 -30 -25-45

-40

-35

-30

-25

δ13C methane (per mil)

δ13C

eth

ane

(per

mil)

R1 R2

R3 R4

Devonian Horn River Shale

Triassic Doig Fm.

Triassic Fractured Reservoirs, B.C.

Triassic Fractured Reservoirs, Alberta

Permian Fractured Reservoirs, B.C.

Triassic Montney Fm.

normal

reverse

d

pre-rollover

rollover post-rollover

(a)

(b)

Fig. 6. Isotope cross plots of new data (Table 1) for the Triassic Montney/Doig Fm. and the Devonian Horn River Shale gas in the Western Canada Sedimentary Basin (WCSB).(a) δ13Cmethane versus δ13Cethane cross plot; numbers beside data points are vertical sample depths. (b) δ13Cethane versus δ13Cpropane cross plot. Solid arrows indicate directions ofincreasing maturity for Montney–Doig Fm. gases. Samples linked by dashed arrows and labeled R1 through R4 are selected gases from fractured reservoirs. Dashed arrows indicatedirections of increasing maturity based on structural relationships (R1 through R4) of Triassic and Permian fractured reservoirs in the Foothills of the WCSB (Tilley et al., 2011).Maturity relationships for these gases are discussed in the text. Colored shading denotes the maturation stage: yellow — pre-rollover zone; gray — rollover zone; and green —

post-rollover zone. B.C. = British Columbia.

200 B. Tilley, K. Muehlenbachs / Chemical Geology 339 (2013) 194–204

Doig Fm. gas samples that include the conventional oil-associated gasfrom Alberta to the east and non-conventional tight gas in BritishColumbia to the west, representing a downdip distance of more than200 km. The two eastern-most gases (Montney Fm. at 1034 m andDoig Fm. at 2021 m depths, Fig. 6a, Table 1) occur at the most negativeend of the pre-rollover zone on both Fig. 6a and b. Other gases inthe pre-rollover zone are from British Columbia, downdip to the west.δ13Cmethane, δ13Cethane, and δ13Cpropane values become progressivelyless negative along the pre-rollover trend as gas maturity increasesdowndip. Doig and Montney Fm. gases that are the furthest downdip,at present depths of 2302 m, 2585 m and up to 3110 m (Fig. 6a), repre-sent the rollover zone. These gases havemore negative δ13Cethane valuesas maturities (dryness, depth and δ13Cmethane values) increase.

Data for a few samples of Devonian Horn River Shale gas innortheast British Columbia are listed in Table 1 and plotted inFig. 6 as open squares. These gases are fully isotopically reversed(δ13Cpropanebδ13Cethanebδ13Cmethane, Fig. 6a and b), δ13Cmethane

values are less negative than those of the Doig and Montney Fm.

rollover zone gases (Fig. 6a, Table 1), and these gases are very dry(C2+ %b0.2, Fig. 3, Table 1). As all these features suggest very highmaturity, we suggest that the Horn River Shale gases represent thepost-rollover zone of very mature gases (green in Fig. 6).

The pre-rollover and rollover trends for Doig and Montney Fm.tight gas are similar to those already observed for the Appalachiangases and the Barnett/Fayetteville Shale gases. The position of theHorn River Shale gases in Fig. 6 is consistent with post-rollover zonegases observed in the Appalachians and the Barnett/Fayettevilleshales. We apparently see the full spectrum of maturation zones intight gas and shale gas plays in the WCSB. Unfortunately, we do notyet have sufficient δ2H data available for these gases for meaningfulinterpretation.

3.5. WCSB Foothills fractured reservoirs

Gases from the WCSB Foothills fractured reservoirs are described inTilley et al. (2011). Here, we wish to re-examine these gases to see how

-45 -40 -35 -30 -25-45

-40

-35

-30

-25

δ13C methane (per mil)

δ13C

eth

ane

(per

mil)

δ13C ethane (per mil)

δ13C

pro

pane

(pe

r m

il)

-45 -40 -35 -30 -25

-45

-40

-35

-30

-25

-20

U.Triassic Sukunka, B.C.

U.Triassic Narraway Area, Alberta

Permian Sukunka, B.C.

M.Triassic Narraway Area, Alberta

Mississippian Minnow Area, AlbertaU.Triassic Minnow Area, Alberta

Permian Narraway Area, Alberta

U.Triassic Sukunka, B.C.

U.Triassic Narraway Area, Alberta

Permian Sukunka, B.C.

M.Triassic Narraway Area, Alberta

Mississippian Minnow Area, AlbertaU.Triassic Minnow Area, Alberta

Permian Narraway Area, Alberta

normal

reversed

normal

reve

rsed

pre-rollover

rollover

post-rollover

pre-rollover

rollover

post-rollover

(a)

(b)

Fig. 7. Isotope cross plots for gases from Triassic and Permian fractured reservoirs in the Foothills of theWCSB. (a) δ13Cmethane versus δ13Cethane cross plot; (b) δ13Cethane versus δ13Cpropanecross plot. Colored shading represents the data for the Montney/Doig Fms. and Horn River Shale plotted in Fig. 6a and 6b. Solid arrows indicate directions of increasing maturity forMontney–Doig Fm. gases. Dashed arrows indicate directions of increasing maturity for fractured reservoir gas data.Data from Tilley et al. (2011).

201B. Tilley, K. Muehlenbachs / Chemical Geology 339 (2013) 194–204

they fit into the scheme of maturation stages and trends that we havedeveloped based on gases from other sedimentary basins. We havestructural data that provide information about maturity relationshipsbetween the numbered pairs of gases linked by the dashed lines inFig. 6a and b. In the deformation zone of the WCSB, deformation oc-curred first to the west and progressively moved eastwards. Accordingto our hypothesis described in Tilley et al. (2011), gas was releasedfrom shale during the deformation event andwas stored in the adjacenttight fractured reservoirs. Therefore, shale gas would have been re-leased and stored in fractured reservoirs in thrust sheets to the westearlier than shale gas would have been released from shales in thrustsheets further to the east. Since shale gas in thrust sheets to the eastwould have had more time to mature in situ in the shale than shalegas to the west, we might expect gas stored in fractured reservoirs ofthrust sheets to the east to bemoremature than gas stored in fracturedreservoirs of thrust sheets to the west.

In a series of structural cross sections across the central Foothills ofthe WCSB, Tilley et al. (2011, their Fig. 11) illustrated the spatial

relationships of sampled wells (and the isotope ratios of the producedgases) relative to the structural positions of the various thrust sheetsin which the fractured reservoirs occur. The best example for our pur-pose here of determining maturity relationships was shown in theirFig. 11c from the Narraway field in Alberta. The gas in the Triassic res-ervoir in the westernmost thrust sheet has δ13Cmethane=−35.2‰,δ13Cethane=−40‰, δ13Cpropane=−41.7‰, whereas the more maturegas in the same Triassic layer several thrust sheets to the east hasδ13Cmethane=−33.7‰, δ13Cethane=−35.8‰, δ13Cpropane=−29.9‰.These gases are labeled as relationship R1 in our Fig. 6a and b andsuggest that as maturity increases, δ13Cethane and δ13Cpropane becomeless negative. Similarly, relationships R2, R3 and R4 on Fig. 6a and bare for the Sukunka–Bullmoose area in eastern British Columbia asfollows:

R2 gases in Triassic reservoirs in thrust sheets illustrated in Tilleyet al. (2011, their Fig. 11a), (δ13Cmethane=−32.8‰, δ13Cethane=−38.8‰, δ13Cpropane=−35.2‰ to the west and δ13Cmethane=−31.1‰,δ13Cethane=−35.1‰, δ13Cpropane=−27.0‰ to the east);

-45

-40

-35

-30

-25δ13

C e

than

e (p

er m

il)

-45 -40 -35 -30 -25

δ13C methane (per mil)

reve

rsed

conv

entio

nal

WCSBQuebec mudgasAppalachiansFayetteville ShaleBarnett Shale

pre-

rollo

ver

post-

rollo

ver

rollover

Fig. 8. Generalized summary plot of δ13Cmethane versus δ13Cethane showing a compari-son of data from all the basins examined in this paper. Arrows indicate the directionsof increasing gas maturity for each set of gas data.

202 B. Tilley, K. Muehlenbachs / Chemical Geology 339 (2013) 194–204

R3 gases in Permian reservoirs in thrust sheets and folds illustratedin Tilley et al. (2011, their Fig. 11b), (δ13Cmethane=−27.5‰, δ13Cethane=−31.2‰, δ13Cpropane=−32‰ to the west and δ13Cmethane=−27.9‰,δ13Cethane=−26.3‰, no detectable Cpropane to the east); and

R4 gases in Permian reservoirs located on a map in Tilley et al.(2011, their Fig. 9b), (δ13Cmethane=−27.1‰, δ13Cethane=−32.7‰,

rollo

ver

-45

-40

-35

-30

-25

δ13C

pro

pane

(pe

r m

il)

-20

δ13C ethane (per mil)-30-45 -40 -35

reversed

-25

-45

-40

-35

-30

-25

δ13C

pro

pane

(pe

r m

il)

-20

δ13C ethane (per mil)

-30-45 -40 -35

conventional

reversed

-25

post-rollover

rollover

conventional

rollo

ver

pre-rollo

ver

post-rollover

rollo

ver

rollo

ver

(a)

(b)

Fig. 9. Generalized summary plots of δ13Cethane versus δ13Cpropane showing a comparisonof data from all the basins examined in this paper. a) Pre-rollover and rollover zonegases. Shaded yellow areas indicate the pre-rollover zones for each dataset. (b) Rolloverand post-rollover zone gases. Shades of green indicate the post-rollover zones for eachdataset. Arrows indicate the directions of increasing gas maturity for each set of gas data.

δ13Cpropane=−30.7 to the west and δ13Cmethane=−24.4‰, δ13Cethane=−29.5‰, δ13Cpropane=−35.6‰ to the east).

It is possible that the trend of δ13Cethane versus δ13Cpropane shown inR4may not be valid because of analytical uncertainties in the δ13Cpropanevalue, due to the extremely low abundance of propane in this sample. Ingeneral, thematurity relationships of these gases in the fractured reser-voirs are typical of the trends we have observed in other sedimentarybasins for the post-rollover zone.

Fig. 7 shows the same cross plots as Fig. 6, but with all the gasanalyses from the fractured reservoirs of Tilley et al. (2011). Usingthe maturity relationships shown in Fig. 6 and described in the previ-ous paragraph as a guide, we can postulate that the majority of theadditional samples in Fig. 7 follow similar post-rollover trends, partic-ularly those gases with δ13Cmethane less negative than−35‰. Some ofthe Alberta samples with δ13Cmethane more negative than −35‰ liewithin the rollover zone of the Montney/Doig Fm. gases (Fig. 7a),suggesting either that these are more immature gases or the isotoperatios have been complicated by local mixing of gases from differentsources as these fractured reservoirs were filled.

4. Comparison of gas maturation in different basins

Figs. 8, 9 and 10 are generalized summary diagrams that show thethree stages of maturation that are universally present and character-ized by similar isotope trends in each basin examined here. Specific iso-tope relationships vary from basin to basin and this variation could bethe result of a complex interplay of such factors as (a) distributionand composition of source kerogen, (b) burial history, (c) gas expulsionhistory and (d) the presence of mineral catalysts (Seewald, 2003;Mango et al., 2010) and/or water (Lewan, 1997; Seewald, 2003). Fol-lowing is a discussion of the observed inter-basinal variations in thewetness (Fig. 3) and the range of isotope values in each maturationstage.

4.1. Pre-rollover/rollover zone transition and rollover zone

The pre-rollover to rollover zone transition is characterized by thechange from a trend of δ13Cmethane, δ13Cethane and δ13Cpropane all be-coming less negative with increasing maturity to a reversed trendwhere δ13Cethane and δ13Cpropane become more negative with increas-ing maturity. In the early stages of the rollover zone, gases are not yetisotopically reversed. It is only as ethane and propane progressivelybecome lighter as methane continues to become heavier, that eventu-ally, in themoremature portion of the rollover zone, the gas can easilybe recognized as an isotopically reversed gas with δ13Cethane morenegative than δ13Cmethane (Fig. 8).

δ13C methane (per mil) -50 -45 -40 -35

-220

-200

-180

-160

-140

-120

-100

rollove

r

post-rollover

-30 -25

post-rollover

pre-rollover

pre-

rollo

ver

rollove

r

Fig. 10. Generalized summary plot of δ13Cmethane versus δ2Hmethane showing a compar-ison of data examined in this paper.

203B. Tilley, K. Muehlenbachs / Chemical Geology 339 (2013) 194–204

At the transition from the pre-rollover zone to the rollover zone,the inter-basinal range in methane, ethane, and propane isotopevalues and wetness varies as follows, in the:

Barnett Shale: δ13Cmethane≈−43 to−40‰ , δ13Cethane≈−29.2 to−28‰, δ13Cpropane≈−22‰, wetness≈4 to 5.8%;Quebec mudgas: δ13Cmethane ≈ − 40‰ , δ13Cethane ≈ − 32‰,δ13Cpropane≈−25‰, no wetness data available;Appalachians: δ13Cmethane≈−39.2 to−38.5‰, δ13Cethane≈−34.4‰,δ13Cpropane≈−32‰, wetness≈6%;Montney/Doig Fm.: δ13Cmethane≈−38.5 to −36.5‰ , δ13Cethane≈−25.8‰, δ13Cpropane≈−20‰, wetness≈3.5%.

The inter-basinal relationships for wetness and δ13Cethane at thepre-rollover/rollover zone transition are illustrated in Fig. 3 and forδ13Cmethane, δ13Cethane, and δ13Cpropane by the downward pointingcurved arrows in Fig. 8 and the curved arrows in Fig. 9a. It is interest-ing that the drier the gas at the pre-rollover/rollover zone transition(Fig. 3), the less negative are the δ13Cethane and δ13Cpropane values.This is the expected relationship with increasing maturity in con-ventional maturation where gas is created by cracking of kerogen.δ13Cmethane values for the Montney/Doig Fm. and the Barnett Shalegases are consistent with the relative maturities predicted by thewetness values. TheMontney/Doig Fm. gas was apparently at a higherstage of maturity than the Barnett Shale gas at the pre-rollover/rollovertransition. The lack of correlation between relative δ13Cmethane and rel-ative wetness for the Quebec mudgas may simply reflect a scarcity ofdata that accurately reflects the natural range in values. This couldalso be the case for the lack of correlation in the Appalachians, or alter-natively, it could be the result of the pseudo-rollover effect caused bymixing.

The observed inter-basinal isotopic and wetness relationships raisethe question of what causes the rollover, or what inhibits the rolloverfrom occurring at a fixed maturation level or δ13Cmethane, δ13Cethane, orδ13Cpropane value? If the initiation of the rollover simply reflects theinitiation of oil cracking and the resulting creation of isotopically lightethane and propane, then temperature would be the main control onwhen the rollover begins. It has also been suggested that the presenceof water can delay the initiation of oil cracking (Lewan, 1997; Seewald,2003). It is possible that the δ13Cmethane, δ13Cethane, and δ13Cpropane valuesof the gas and the gas wetness at the transition to the rollover zone re-flect the gas expulsion history, which in turn could vary depending onthe distribution and degree of permeability. Where the permeabilityduring kerogen cracking was sufficient, some of the early gas from ker-ogen cracking could have been lost from a semi-closed system. The gascreated by kerogen cracking and accumulated in situ after that expulsionwould be heavier (isotopically less negative) and drier than other shalegas that had continuously accumulated with no loss due to expulsion(Rooney et al., 1995). At a particular temperature when oil cracking be-gins (and the rollover begins), heavier, drier gas could be the result ofearlier gas loss, rather than the result of overall higher maturity. Bythis argument, the rollover could begin at a universal temperature(depending on the presence or absence of water) in all basins, but theobserved wetness and isotope ratios could vary depending on the rela-tive permeability of the system and the extent of early gas loss. For thedatasets considered here, this would suggest that the Barnett Shale is aless permeable, better sealed system that has retained more originalgas than the silt-richMontney/Doig Fms. In a thick system of irregularlyinterbedded and interlaminated silt and shale, it is possible that therecould be pockets of trapped early gas which would be isotopicallyquite negative, and pockets of residual gas that would be isotopicallyless negative.

The slope of the rollover zone in both the δ13Cmethane versus δ13Cethaneplot (Fig. 8) and the δ13Cethane versus δ13Cpropane plot (Fig. 9a) is fairlyuniform from basin to basin, with the exception of the Burruss and

Laughrey (2010) Appalachian fractured reservoir data. The slope oftheir data in the rollover zone is significantly lower than for data fromother areas. This lends credence to the Burruss and Laughrey (2010) sug-gestion that the isotopic reversals in these gaseswere created bymixing.

4.2. Post-rollover zone

The post-rollover zone is best identified on a δ13Cmethane versusδ2Hmethane plot (Fig. 10) as the zone where δ2Hmethane becomesmore negative as δ13Cmethane becomes less negative (or as maturityincreases). Well-defined post-rollover zones are observed in theAppalachian and Fayetteville Shale δ2Hmethane datasets. δ2Hmethane

for gases at the rollover/post-rollover zone transition differ for thesetwo basins (−122 versus −142‰) with the Appalachians being themore negative. This may not be surprising considering the morenortherly location of the New York and Pennsylvanian gases (mostmature gases in the Appalachian dataset) relative to the FayettevilleShale gas. The Marcellus Shale from Pennsylvania also has gases atthe post-rollover stage of maturation (Fig. 10), with δ2Hmethane valuesonly slightly more negative than the Burruss and Laughrey (2010)New York and Pennsylvanian gases. The trend of decreasing δ2Hmethane

with increasing δ13Cmethane is consistent with a process of local forma-tion water increasingly exchanging its hydrogen with reforming meth-ane (Tang and Xia, 2011). This effect on δ2Hmethane is only observed inthe post-rollover zone, suggesting that water reforming is not the dom-inant process that occurs in the rollover zone.

δ13Cmethane at the rollover/post-rollover zone transition (Fig. 8)varies from −39‰ in the Fayetteville Shale gases, to a range of −38to −35.5‰ which includes the WCSB gases, the Barnett Shale, and theQuebec mudgases, to −33.5‰ in the Appalachian gases. δ13Cethaneand δ13Cpropane are also the most negative for the Fayetteville Shalegases (Fig. 9b) (δ13Cethane=−46.5‰ versus −43.5 to −41.5 for theother areas; δ13Cpropane=−46‰ versus −43 for most of the otherareas). We do not have enough information at this time to speculateas to the reason for these variations.

The post-rollover zone is characterized by δ13Cmethane, δ13Cethane andδ13Cpropane all becoming less negative as maturity increases (Figs. 8 and9b). Presumably, oil cracking has been replaced by gas cracking andwater reforming of methane as the dominant processes. Rayleigh frac-tionation processes may be responsible for the increasingly heavy eth-ane and propane (Burruss and Laughrey, 2010). Fig. 9b shows a widevariation of δ13Cpropane values in the post-rollover zone, even withinone sedimentary basin. Note that at the beginning of the post-rolloverstage, all gases in all basins are isotopically reversed with respect tomethane and ethane (i.e. δ13Cethanebδ13Cmethane), and all, with the pos-sible exception of the Barnett Shale gas, are fully isotopically reversed(i.e. δ13Cpropanebδ13Cethanebδ13Cmethane). As maturation progresses inthepost-rollover zone, the difference between δ13Cmethane and δ13Cethanedecreases until at a very late stage of the post-rollover zone, the gasmayno longer be reversed and may appear to be a conventional, extremelymature gas.

The fact that as maturation progresses in the post-rollover zone,the isotopic reversals between the components may be lost and thegas may then appear to be a conventional, extremely mature gas,raises an intriguing question. Are the post-rollover isotope trendsonly a reflection of fortuitous differences in fractionation resultingfrom the destruction of ethane and propane, or do these trends to“normal” distributions reflect isotopic exchange or re-equilibrationbetween methane–ethane–propane?

5. Conclusions and implications

Examination of gases from several sedimentary basins in NorthAmerica has shown that gases from sealed, self-contained petroleumsystems (‘shale gas’ — including shale, very fine grained sandstone orsiltstone, and some fractured, mixed clastic/carbonate reservoirs) share

204 B. Tilley, K. Muehlenbachs / Chemical Geology 339 (2013) 194–204

a common maturation history that differs from conventional opensystem gases. This maturation history is comprised of three distinctstages, pre-rollover zone, rollover zone, and post-rollover zone. Eachzone has characteristic carbon and hydrogen isotopic trends that aresummarized and illustrated in Figs. 3, 8, 9 and 10. Correct assignmentof maturity stage could be of importance because the rollover zonemay represent the peak of high productivity shale gas whereas thepost-rollover zone may represent a decline in productivity (Burrussand Laughrey, 2010).

The easiest distinguishing characteristics of the more advancedmaturation stages of this ‘shale’ gas are the isotopic reversals, whenpresent. The isotopic trends illustrated in Figs. 8, 9 and 10 demon-strate the zones of gas maturation where isotopically reversed gasescan be expected. Gases are only isotopically reversed with respectto methane and ethane in the latest stage of the rollover zone andin the post-rollover zone. Gases are isotopically reversed with respectto ethane and propane only in the latest stage of the rollover zone,and the early stage of the post-rollover zone. Wetter gas near the be-ginning or the middle of the rollover zone, often the optimal produc-tion target, may be difficult to recognize isotopically. This gas mostlikely will not be isotopically reversed at all. However, if sufficientsamples at different maturities are available, gases in this early stageof the rollover zone can be recognized by their trend towards morenegative δ13Cethane values as δ13Cmethane values become more positive.

A gas that is fully isotopically reversed (i.e. δ13Cpropanebδ13Cethanebδ13Cmethane) should be considered carefully. It could represent the ma-ture end of the rollover zone, in which case the maximum amount ofgas may be present. Alternatively, it could represent a gas from thepost-rollover zone, in which case it may have undergone some thermaldestruction and may have matured past the most productive matura-tion stage.

This study has shown that the progression through three stages of gasmaturation is universal in sealed, self-contained petroleum systems andthat each stage is characterized by universal isotopic trends. It is proposedthat Figs. 8, 9 and 10, in addition to Fig. 3, be developed as templates formaturity level determination for gases from shales, tight gas plays andfractured reservoirs, especially in areas where at least some gases are iso-topically reversed. This scheme of isotope maturation trends is unique inthat it places all variations of isotope reversals within a coherent frame-work based on gas maturation. The processes that cause the rollover arenot well understood, but any theory for the origin of the rollover and iso-topic reversals needs to be able to explain the maturation trendspresented here. Hydrogen isotope data could be very important in under-standing the processes that create the rollover and post-rollover zones.

Acknowledgments

The authors would like to thank Dr. A. Schimmelmann and L. Gaofor their careful review of the manuscript.

References

Baldassare, F., 2011. The origin of some natural gases in Permian through Devonian agesystems in the Appalachian basin and the relationship to incidents of stray gas mi-gration. Proceedings of the Technical Workshops for Hydraulic Fracturing Study:Chemical and Analytical Methods, EPA 600/R-11/066 May 2011. http://www.epa.gov/hfstudy/theoriginofsomenaturalgasesinpermianthroughdevonianagesystems.pdf,last accessed June 19, 2012.

Burruss, R.C., Laughrey, C.D., 2010. Carbon and hydrogen isotopic reversals in deepbasin gas: evidence for limits to the stability of hydrocarbons. Organic Geochemistry41, 1285–1296.

Dai, J., Xia, X., Qin, S., Zhao, J., 2004. Origins of partially reversed alkane δ13C values forbiogenic gases in China. Organic Geochemistry 35, 405–411.

Fitzmartyn, R.J., 2009. Montney revisited. http://www.firstenergy.com/research/documents/Focus-I-Montney-2009-01-23.pdflast accessed Feb 20, 2012.

Jenden, P.D., Drazan, D.J., Kaplan, I.R., 1993. Mixing of thermogenic natural gases innorthern Appalachian Basin. AAPG Bulletin 77, 980–998.

Lewan, M.D., 1997. Experiments on the role of water in petroleum formation. Geochimicaet Cosmochimica Acta 61, 3691–3723.

Mango, F.D., Jarvie, D.M., Herriman, E., 2010. Natural catalytic activity in amarine shale forgenerating natural gas. Proceedings of the Royal Society A: Mathematical, Physicaland Engineering Sciences 466, 3527–3537. http://dx.doi.org/10.1098/rspa.2010.0032.

Molgat, M., Chatellier, J.-Y., Rioux, R., Holmes, K., Goodall, C., Smith, R., 2011. Best practicesand applied organic geochemistry to address a surface casing vent flow, examplefrom Talisman 2011 remediation work in Quebec. http://www.forumschiste.com/uploads/images/Poster-QOGA-2011.pdflast accessed June 20, 2012.

Osborn, S.G., McIntosh, J.C., 2010. Chemical and isotopic tracers of the contribution ofmicrobial gas in Devonian organic-rich shales and reservoir sandstones, northernAppalachian Basin. Applied Geochemistry 25, 456–471.

Potter, A., Chan, H., Wilson, S., 2008. What is the potential for North American uncon-ventional resources? UBS Investment Research: Q-Series: North American Oil andGas. 95p. http://thefriendsvillegroup.org/UBS_shaleplays.pdf, last accessed Feb 20,2012.

Rodriguez, N.D., Philp, R.P., 2010. Geochemical characterization of gases from theMississippian Barnett Shale, Fort Worth Basin, Texas. AAPG Bulletin 94, 1641–1656.

Rooney, M.A., Claypool, G.E., Chung, H.C., 1995. Modeling thermogenic gas generationusing isotope ratios of natural gas hydrocarbons. Chemical Geology 126, 219–232.

Seewald, J.S., 2003. Organic–inorganic interactions in petroleum-producing sedimentarybasins. Nature 426, 327–333.

Seewald, J., Whelan, J., 2005. Isotopic and chemical composition of natural gas from thePotato Hills field, southeastern Oklahoma: evidence for an abiogenic origin? AAPGResearch Conference Abstracts, Calgary.

Tang, Y., Xia, D., 2011. Quantitative assessment of shale gas potential based on its specialgeneration and accumulation processes. Search and Discovery Article, 40819.

Tilley, B.J., McLellan, S., Hiebert, S., Quartero, B., Veilleux, B., Muehlenbachs, K., 2011.Gas isotope reversals in fractured gas reservoirs of the western Canadian Foothills:Mature shale gases in disguise. AAPG Bulletin 95, 1399–1422.

Zumberge, J., Ferworn, K., Brown, S., 2012. Isotopic reversal (‘rollover’) in shale gasesproduced from the Mississippian Barnett and Fayetteville formations. Marine andPetroleum Geology 31, 43–52.