Produced Gas and Condensate Geochemistry of the Marcellus Formation in the Appalachian Basin: Insights into Petroleum Maturity, Migration, and Alteration in an Unconventional Shale Reservoir

Abstract

The Middle Devonian Marcellus Formation of North America is the most prolific hydrocarbon play in the Appalachian basin, the second largest producer of natural gas in the United States, and one of the most productive gas fields in the world. Regional differences in Marcellus fluid chemistry reflect variations in thermal maturity, migration, and hydrocarbon alteration. These differences define specific wet gas/condensate and dry gas production in the basin. Marcellus gases co-produced with condensate in southwest Pennsylvania and northwest West Virginia are mixtures of residual primary-associated gases generated in the late oil window and postmature secondary hydrocarbons generated from oil cracking in the wet gas window. Correlation of API gravity and C₇ expulsion temperatures, high heptane and isoheptane ratios, and the gas geochemical data confirm that the Marcellus condensates formed through oil cracking. Respective low toluene/nC₇ and high nC₇/methylcyclohexane ratios indicate selective depletion of low-boiling point aromatics and cyclic light saturates in all samples, suggesting that water washing and gas stripping altered the fluids. These alterations may be related to deep migration of hot basinal brines. Dry Marcellus gases produced in northeast Pennsylvania and northcentral West Virginia are mixtures of overmature methane largely cracked from refractory kerogen and ethane and propane cracked from light oil and wet gas. Carbon and hydrogen isotope distributions are interpreted to indicate (1) mixing of hydrocarbons of different thermal maturities, (2) high temperature Rayleigh fractionation of wet gas during redox reactions with transition metals and formation water, (3) isotope exchange between methane and water, and, possibly, (4) thermodynamic equilibrium conditions within the reservoirs. Evidence for thermodynamic equilibrium in the dry gases includes measured molecular proportions (C₁/(C₁ − C₅) = 0.96 to 0.985) and δ¹³C₁ values significantly greater than δ¹³C_KEROGEN. Noble gas systematics support the interpretation of hydrocarbon–formation water interactions, constrain the high thermal maturity of the hydrocarbon fluids, and provide a method of quantifying gas retention versus expulsion in the reservoirs.

1. Introduction

The Middle Devonian Marcellus Formation in the Appalachian basin of North America is an iconic stratigraphic interval. It is the most prolific hydrocarbon play in the basin, the second largest producer of natural gas in the United States, and, in combination with the Ordovician Utica shale, the largest source of natural gas and natural gas liquids (NGL) in the world [1,2]. Proven natural gas reserves for the Marcellus Formation total more than 129 tcf and proven lease condensate reserves total 247 million barrels [3]. The Marcellus is also a prospective CO₂ sequestration reservoir and a potential seal for Carbon Capture and Underground Storage (CCUS) and NGL storage [4,5]. It is a potential source of critical minerals [6]. In contrast to its natural resource utility, the Marcellus is a potential source of stray gas contamination in ground water supplies and building structures as well as fugitive emissions of atmospheric methane [7,8,9,10].

Several critical geologic factors define the Marcellus play in the Appalachian basin. These factors are thermal maturity, pressure, thickness, porosity and permeability, gas-in-place, natural fracturing, mineralogy, depth, structural setting, rock mechanics, target landing issues, diagenesis, and hydrocarbon alteration [11,12,13]. Thermal maturity appears to define specific wet gas and dry gas production in the Appalachian basin [1] (Figure 1). Condensate production occurs in the western part of the basin where wet gases with approximately 1400 MMBtu/Mcf are recovered from reservoir depths of −1067 to −1524 m (−3500 to −5000 ft.) subsea with reported vitrinite reflectance (VR_o) of 1.0 to 2.8%. Dry gas with 1000 to 1080 MMBtu/Mcf is produced from reservoir depths of approximately −914 to −1981 m (−3000 to −6500 ft.) subsea with reported VR_o from 2.0 to >3.0%.

In this contribution, integrated from twenty-four years of study of Marcellus natural gas/condensate chemical and isotope data, I argue that regional differences in fluid chemistry reflect variations in thermal maturity, hydrocarbon alteration, and geological processes active in the petroleum system. These differences not only define specific wet gas and dry gas production in the basin BUT predict relative productivity as well. Marcellus Formation hydrocarbon composition, isotope, and noble gas geochemistry provide a quantitative methodology independent of rock data, choke management, and completion design for defining and mapping economic production limits. Noble gas systematics provide a method of quantifying hydrocarbon maturity and gas retention versus fluid expulsion or loss from the reservoirs. The data and geochemical trends documented in this report can also facilitate the identification of sources of stray gas contamination and provide fundamental baseline measurements for comparison with data derived from recent atmospheric measurements conducted to monitor fugitive emissions of atmospheric methane [7,8,9,10].

2. Geological Framework and Current Status of Marcellus Formation Research

The Appalachian basin and its foreland represent mountain building throughout most of the Paleozoic on the former eastern convergent margin of North America [14]. The Devonian Appalachian basin was a retroarc foreland basin that developed adjacent to the Acadian orogenic belt, a mountain chain built by oblique collision of the North American continental margin with Avalon terrane [14,15,16,17]. Crustal loading by the resultant fold and thrust belt caused subsidence of an elongate foreland basin which was mostly filled with siliciclastic sediments eroded from the uplifted orogen. Sediment supply from the uplifting and eroding Appalachian Mountains generally exceeded subsidence during the late Devonian Acadian and Pennsylvanian-Permian Alleghenian orogenies resulting in mostly clastic terrestrial deposits that spread across the basin and foreland in a broad, westward coalescing coastal plain [18]. Periods of high eustasy and/or reduced sediment supply resulted in marine incursions and both clastic and carbonate deposition across the region. The consequent Middle to Upper Devonian sedimentary section comprises the Catskill Delta succession of which the Marcellus Formation is an important component [19] (Figure 2).

The Marcellus Formation includes the oldest of a dozen or so regionally extensive organic-rich petroliferous black shale deposits associated with the Catskill delta in eastern North America [23]. The Marcellus Formation is the basal unit of the Hamilton Group (Figure 2). It is about 305 m (1000 ft.) thick in central Pennsylvania, but thins to the north, the west, and the south and feathers out of the sequence in the subsurface of eastern Ohio, western West Virginia, and southwestern Virginia [23]. Throughout much of the basin, the Marcellus Formation comprises two black shale members separated by a sequence of limestone, shale, and lesser sandstone of variable thickness [24] (Figure 2). In eastern West Virginia, however, these two shale members are only mappable to the point where the medial Cherry Valley Member limestone can be recognized [25]. The Cherry Valley Member then becomes very thin and loosely defined moving from northeast to southwest across most of West Virginia [25]. The Marcellus Formation encompasses two third-order transgressive–regressive sequences, the thicknesses of which reflect the interplay of Acadian thrust-load-induced subsidence, short-term base-level fluctuations, and recurrent basement tectonics [24]. Sea-level changes, which influenced clastic dilution and organic carbon production and preservation, were the principal control for organic matter accumulation in the black shales [26]. Sea-level rise led to sediment starvation and organic carbon concentration in distal basin sediments.

Marcellus Formation reservoir shales are dominated by nine principal mineral phases including quartz, muscovite, illite, pyrite, chlorite, albite, calcite, dolomite, and barite [12,27]. Variable amounts of anhydrite, kaolinite, and apatite are reported in some cores [12]. Four distinct mineralogical facies within the Marcellus shales reflect depositional influences on the variability of total organic carbon (TOC) between the upper and lower black shale members [27]. The medial Cherry Valley Member limestones contrast lithologically and petrophysically with the bounding organic-rich shales and comprise three distinct carbonate depositional facies [28].

Organic matter in the Marcellus Formation is dominated by originally hydrogen-rich unstructured (amorphous) bituminite and unicellular alginite [29,30,31,32]. Minor amounts of mostly oxidized and recycled humic material occur in most of the samples directly studied for this investigation. In the wet gas-producing region, the originally amorphous organic matter is severely degraded and converted to approximately equal amounts of solid hydrocarbon and pyrobitumen (Figure 3). VR_o in the wet gas region explicitly measured during this investigation is between 1.59 and 1.78%. Pyrobitumen dominates the organic matter in the dry gas-producing region where VR_o specifically measured during this study is between 2.35 and 4.61% (Figure 4). Pyrobitumen in the highest maturity Marcellus rocks (>3.0% VR_o) is highly aromatic and has undergone partial graphitization [12,30]. The Marcellus Formation in the subsurface of northeastern Pennsylvania is metagenetic and characterized by prehnite–pumpellyte to incipient greenschist metamorphic facies [12,30].

Porosity in Marcellus Formation reservoirs is a function of compaction, cementation, and organic matter thermal maturation through the occlusion of primary void space and the development of secondary porosity [12,29,30]. Organic matter-hosted porosity provides the most significant portion of effective porosity and free-gas storage capacity in the reservoirs [12,29] (Figure 5). Matrix porosity is also present within the clay mineral aggregates and at the interface between rigid clasts and clay minerals [12,29].

Modeled maximum burial of Marcellus Formation strata in the wet gas-producing region was between 3.4 and 4 km (11,220 and 13,400 ft.) followed by uplift and erosion of 1.04 to 2.4 km (3419.6 to 8049.8 ft.) of Carboniferous and Permian overburden which was removed during post-Alleghenian uplift [33,34]. Modeled maximum burial in the dry gas-producing region was between 4.89 and 4.95 km (16,050 and 16,250 ft.) with subsequent uplift and erosion of 2.6 to 2.9 km (8550 to 9650 ft.) of overburden [33,34]. Higley and Enomoto [33] assigned the following rates of erosion of overburden sediments based on modeled thicknesses of erosion of nine wells in the study area: (A) 8.78 to 11.3 m/m.y. (28.8 to 37.1 ft./m.y.) from 260 Ma to Present; (B) 12.8 to 21.3 m/m.y. (42 to 70 ft./m.y.) from 140 to 90 Ma; and (C) 50.3 to 83.8 m/m.y. (165 to 275 ft./m.y.) from 23.5 to 5.3 Ma. The latter two rates (B and C) represent two time periods of increased uplift and erosion.

3. Materials and Methods

Sixty-seven production gases were collected at the wellhead or separator in pressurized stainless-steel cylinders or Isotubes^® and shipped to Isotech Laboratories in Champaign, IL, USA for processing and analyses (

Table 1. Marcellus Formation produced gas samples, chemical composition, and isotope ratios. nd—not detected. nm—not measured. * Concentration too small to measure isotope compositions. Sample data in bold blue print are time-lapse data discussed in the text. DG—dry gas. WG—wet gas. WV—West Virginia. PA—Pennsylvania.

Well ID/Group	O2 Mole%	CO2 Mole%	N2 Mole%	C1 Mole%	C2 Mole%	C3 Mole%	iC4 Mole%	nC4 Mole%	iC5 mole%	nC5 Mole%	C6+ Mole%	δ13C1 ‰	δDC1 ‰	δ13C2 ‰	δ13C3 ‰
DG1	0.099	0.02	0.47	97.45	1.87	0.0755	0.0017	0.0047	0.0013	0.0008	0.0076	−27.08	−158.5	−33.2	−36.3
DG2	0.086	0.039	0.42	97.38	1.97	0.0865	0.0033	0.0065	0.0027	0.0014	0.009	−28.69	−157.7	−34.74	−36.79
DG3	0.13	0.019	0.5	97.5	1.79	0.0561	0.0007	0.0024	nd	nd	0.0018	−27.85	−158.2	−34.1	−36.8
DG4	0.16	0.018	0.57	97.35	1.84	0.0594	0.0008	0.0027	nd	nd	0.001	−27.85	−159.3	−34.13	−36.9
DG5	0.071	0.018	0.4	97.54	1.9	0.0659	0.001	0.0033	0.0001	0.0001	0.0004	−27.95	−158.2	−34.22	−36.9
DG6	0.13	0.023	0.53	97.39	1.86	0.063	0.0009	0.0028	nd	nd	nd	−28.5	−157.8	−34.96	−37.4
DG7	0.06	0.021	0.47	97.79	1.61	0.0486	0.0006	0.0022	nd	nd	nd	−26.09	−159.1	−32.52	−35.1
DG8	0.015	0.023	0.4	97.82	1.69	0.0491	0.0006	0.0017	nd	nd	nd	−25.96	−159.9	−32.48	−35
DG9	0.16	0.017	0.73	97.32	1.71	0.0607	0.0009	0.0032	0.0001	0.0001	nd	−25.41	−158.4	−32.19	−34.9
DG10	0.037	0.016	0.57	97.62	1.69	0.0639	0.001	0.0033	0.0001	0.0001	nd	−24	−158.6	−30.8	−33.9
DG11	0.36	0.022	2.04	96.12	1.41	0.0425	0.0006	0.0017	0.0001	0.0001	0.0004	−24.04	−163.9	−29.9	−32.5
DG12	nd	0.061	0.57	97.87	1.42	0.032	0.0004	0.0008	nd	nd	nd	−23.9	nm	−29.86	−31.4
DG13	0.025	0.009	0.48	97.69	1.74	0.0571	0.0008	0.0025	0.0001	0.0001	nd	−25.27	−161.5	−31.58	−34.8
DG14	0.094	0.019	0.6	97.6	1.63	0.0513	0.0007	0.0022	nd	nd	nd	−25.03	−160.3	−31.28	−34.2
DG15	0.16	0.031	0.68	97.37	1.7	0.0565	0.0008	0.0025	nd	nd	nd	−25.36	−161.1	−31.72	−35
DG16	0.19	0.033	0.7	96.74	2.22	0.108	0.0021	0.0067	0.0003	0.0002	nd	−26.23	−158.3	−32.79	−35.2
DG17	nd	0.011	0.39	97.52	1.93	0.035	0.0003	0.0011	nd	nd	nd	−36.2	nm	−42.11	−43
DG18	nd	0.017	0.38	97.05	2.41	0.093	0.0016	0.0058	0.0002	0.0002	nd	−31.41	nm	−38.73	−40.9
DG19	0.043	0.065	27.06	55.27	0.02	nd	nd	0.0001	nd	nd	0.0009	−59.58	nm	−32.7	*
DG20	nd	0.028	0.23	97.57	2.09	0.057	0.0007	0.002	nd	nd	nd	−29.34	nm	−35.46	−37.8
DG21	0.06	0.021	0.4	97.02	2.37	0.0868	0.0014	0.0047	0.0002	0.0001	nd	−30.8	nm	−37.74	−39.9
DG22	0.016	0.042	0.28	97.33	2.21	0.0716	0.001	0.0033	0.0001	0.0001	nd	−35.91	nm	−41.13	−43
DG23	0.027	0.01	0.38	96.4	2.63	0.101	0.0016	0.0054	0.0002	0.0001	nd	−33.57	nm	−39.94	−42.1
DG24	0.056	0.029	0.42	96.99	2.37	0.0948	0.0016	0.0055	0.0002	0.0002	nd	−28.78	nm	−35.95	−37.5
DG25	0.11	0.037	0.61	96.78	2.33	0.0926	0.0015	0.004	0.0001	0.0001	nd	−26.44	nm	−33.4	−34.6
DG26	0.015	nd	0.37	94.45	2.48	0.105	0.0018	0.0062	0.0003	0.0002	nd	−28.36	nm	−34.97	−37.6
DG27	0.039	0.012	0.37	96.94	2	0.0505	0.001	0.0067	0.0004	0.0003	nd	−31.12	nm	−38.13	−40.8
DG28	0.038	0.009	0.69	96.46	2.31	0.0812	0.0013	0.0045	0.0002	0.0001	0.0049	−33.98	nm	−40.16	−42.3
DG29	0.2	0.014	1.02	96.44	2.11	0.0639	0.0009	0.0023	nd	nd	nd	−27.29	nm	−33.52	−35.9
DG30	0.055	0.031	0.55	97.56	1.71	0.0454	0.0005	0.0014	nd	nd	nd	−24.94	nm	−31.16	−33.3
DG31	0.045	0.026	0.49	97.62	1.74	0.0447	0.0005	0.0014	nd	nd	nd	−25.13	nm	−31.44	−33.6
DG32	0.048	0.03	0.62	97.81	1.39	0.0336	0.0004	0.0011	nd	nd	nd	−23.53	nm	−29.86	−32
DG33	0.052	0	0.74	96.78	0.546	0.0048	nd	nd	nd	nd	nd	−23.59	nm	−29.59	*
DG34	0.034	0.015	0.96	97.42	1.44	0.0385	0.0005	0.0012	nd	nd	0.0054	−24.63	nm	−30.89	−33.1
WVDG1	nd	nd	nd	95.8	2.26	0.1329	0.0003	0.0064	1.82	0.013	0.003	−39.1	nm	−41.7	−38.9
WVDG2	nd	nd	nd	97	1.845	0.0579	0.0003	0.0034	1.074	0.0045	0.0069	−39.2	nm	−41.9	−40.7
WG1	0.016	0.15	0.4	80.85	12.95	3.68	0.471	0.817	0.22	0.187	0.236	−43.7	−189.2	−32.74	−28
WG2	nd	0.17	0.51	75.4	15.61	5.49	0.573	1.4	0.272	0.317	0.223	−45.5	−200.8	−35.28	−30.34
WG3	0.017	0.15	0.48	78.94	13.72	4.26	0.493	1.02	0.255	0.255	0.351	−45.73	−201.4	−34.34	−29.63
WG4	0.015	0.14	0.48	76.41	14.88	5.2	0.614	1.37	0.299	0.319	0.243	−45.41	−201.5	−34.41	−29.7
WG5	0.013	0.17	0.43	77.02	14.83	4.87	0.556	1.21	0.281	0.3	0.284	−44.97	−196.6	−34.1	−29.39
WG6	0.011	0.18	0.47	79.62	13.36	4.1	0.528	0.964	0.255	0.225	0.247	−44.94	−191.1	−33.67	−29.04
WG7	0.015	0.17	0.54	79.21	13.28	4.26	0.541	1.07	0.277	0.274	0.328	−45.21	−195.4	−34.01	−29.48
WG8	0.011	0.16	0.44	78.77	13.87	4.32	0.508	1.05	0.255	0.259	0.31	−45.25	−197.5	−33.98	−29.13
WG9	0.011	0.14	0.46	78.84	13.71	4.38	0.538	1.09	0.274	0.268	0.246	−45.38	−198.2	−33.94	−29.17
WG10 WG11	0.015 0.013	0.16 0.16	0.44 0.44	78.77 78.94	13.85 13.83	4.31 4.27	0.506 0.495	1.05 1.02	0.258 0.245	0.267 0.25	0.328 0.289	−45.22 −45.24	−198.4 −198.3	−33.9 −33.91	−29.08 −29.11
WG12	0.012	0.14	0.5	76.61	14.69	5.12	0.618	1.35	0.317	0.326	0.288	−45.5	−203.1	−34.63	−29.87
WG13	0.095	0.16	0.71	69.37	16.68	6.86	0.854	2.33	0.578	0.81	1.53	−45.23	−197.6	−35.12	−30.15
WG14	0.012	0.15	0.5	71.89	16.14	6.14	0.847	1.84	0.381	0.469	0.83	−45.19	−200.9	−35.14	−30.2
WG15	0.014	0.21	0.34	90.49	7.56	1.04	0.118	0.115	0.0288	0.0137	0.0235	−42.25	−174.4	−31.2	−25.67
WG16	0.32	0.14	1.4	92.27	5.27	0.457	0.0333	0.0356	0.0058	0.0029	0.004	−40.15	−167.2	−33.34	−28.18
WG17	nd	0.15	0.54	76.52	14.52	5.11	0.632	1.4	0.304	0.269	0.456	−46.74	−210.6	−35.87	−31.05
WG18	nd	0.18	0.42	84.51	10.64	2.41	0.358	0.507	0.184	0.133	0.568	−42.98	−179.5	−32.15	−27.18
WG19	nd	0.2	0.33	90.96	7.35	0.876	0.082	0.0897	0.0202	0.0108	0.0456	−40.83	−172.1	−31.38	−26.33
WG20	nd	0.18	0.43	80.35	13.42	3.75	0.473	0.805	0.197	0.159	0.192	−44.06	−186.6	−33.11	−28.54
WG21	nd	0.15	0.44	79.2	13.84	4.23	0.498	0.998	0.213	0.205	0.185	−45.27	−192.5	−33.9	−29.13
WG22	nd	0.15	0.39	82.48	12.38	3.1	0.392	0.599	0.157	0.112	0.205	−43.21	−179.4	−32.37	−27.56
WG23	nd	0.16	0.48	83.37	11.49	2.98	0.379	0.624	0.164	0.124	0.177	−44.53	−184.9	−33.17	−28.48
WG24	nd	0.16	0.47	83.14	11.41	3.1	0.427	0.691	0.196	0.143	0.204	−44.61	−186.7	−33.28	−28.75
WG25	nd	0.15	0.49	74.93	16.48	5.58	0.549	1.25	0.198	0.221	0.125	−44.97	−200.4	−34.66	−29.78
WG26	0.014	0.15	0.52	81.85	12.29	3.32	0.359	0.598	0.164	0.175	0.502	−45.29	−196	−33.89	−29.18
WG27	nd	0.12	0.39	73.07	15.61	6.01	0.868	1.73	0.369	0.409	0.813	−44.5	−242.8	−33.82	−29.25
WV28	0.015	0.13	0.39	73.49	15.21	5.65	0.852	1.62	0.378	0.413	1.18	−44.36	−197.7	−33.69	−28.96
PAWG1	nd	0.086	0.57	70.64	16.42	7.81	0.792	2.2	0.387	0.555	0.49	−45.23	−212.3	−35.84	−31.22
PAWG2	nd	0.085	0.49	77.24	14.48	5.24	0.544	1.15	0.215	0.243	0.249	−43.82	−193.5	−33.83	−29.37
PAWG3	nd	0.084	0.47	79.79	12.67	4.32	0.519	1.1	0.261	0.287	0.434	−42.19	−185.4	−33.15	−28.46

and

Table 2. Additional isotope ratios for selected samples. nm—not measured.

Well ID/Group	δDC2 ‰	δ13CO2 ‰	δ15N2 ‰
DG1	−188.2	nm	nm
DG2	−190.6	nm	nm
DG3	−191.6	nm	nm
DG4	−191.5	nm	nm
DG5	−192	nm	nm
DG6	−189.7	nm	nm
DG7	−188.9	nm	nm
DG8	−192.6	nm	nm
DG9	−191.3	nm	nm
DG10	−190.4	nm	nm
DG11	−189.4	nm	−10.223
DG12	nm	−26.5	nm
DG13	−192.9	nm	nm
DG14	−189.4	nm	nm
DG15	−190.9	nm	nm
DG16	−190.9	nm	nm
DG19	nm	−31.8	nm
PAWG1	−171.4	nm	−13.9
PAWG2	−162.7	nm	−13.3
PAWG3	−162.4	nm	−12.7

). Thirty-four of these produced gas samples were collected in the dry gas region of northeast Pennsylvania (Bradford, Sullivan, Susquehanna, and Wyoming Counties). Two additional dry gas samples were collected in Upshur County, West Virginia. Twelve additional dry gas samples were collected from Marcellus Formation pressure core recovered from a well in Susquehanna County, Pennsylvania (

Table 3. Pressure core isotope ratio and cumulative degassing volume results. Nm—not measured.

Cumulative Degassing Volume %	δ13C1 ‰	δ13C2 ‰	δ13C3 ‰
21.77	−26.3	−31	nm
22.93	−26.88	−32.7	−34.8
33.12	−27.64	−33.15	−33.5
25.47	−27.31	−32.93	−35.1
28.45	−27.49	−32.92	−32.3
37.52	−27.62	−33.39	−33.57
40.69	−27.68	−33.45	−35
53.53	−27.03	−33.5	−35
66.03	−26.25	−32.94	−34.8
75.35	−25.63	−33.45	−34.7
71.81	−25.69	−33.24	−34.8
100 (crushed gas)	−22.8	−29.1	−34.7

). These 12 gas samples were collected during direct gas-in-place resource assessment in the laboratory. Thirty-one wet gas samples were collected in northwest and northcentral West Virginia (Tyler, Richie, Wetzel, Doddridge, and Harrison Counties) and southwest Pennsylvania (Washington County).

All gas samples were analyzed for chemical composition using a Shimadzu 2010 gas chromatograph. Quantification of fixed gases was done by thermal conductivity detector. Compositional data were processed on GC Solutions software and PC. Isotopic analyses were performed using traditional offline sample preparation techniques followed by dual inlet mass spectrometric ¹³C/¹²C measurement on a Finnigan MAT Delta S Isotope Ratio mass spectrometer. ²H/¹H measurements were performed on a Finnigan Delta Plus XL isotope ratio mass spectrometer. Sample results were compared with accepted reference standards (NGS #1, #2, or #3); isotope ratio determinations involved multiple direct comparisons of the sample to the reference standard (generally at least 6 comparisons). Stable carbon and hydrogen isotope compositions are reported as the difference between the ratios of the two isotopes of interest in the sample and the ratio in a primary reference standard. That is,

X_sample = [(R_sample − R_standard)/R_standard] × 1000,

(1)

where X represents the isotope of interest, in this case ¹³C and ²H or deuterium (D), and R represents the ratio of ¹³C/¹²C or ²H/¹H. The value is expressed in terms of per mil (‰), or parts per thousand.

Noble gas measurements were made on second splits of selected produced dry gas samples collected in stainless-steel pressurized cylinders at the same time as the gas samples analyzed at Isotech Laboratories (

Table 4. Noble gas concentrations and isotope ratios of selected Marcellus Formation dry gas samples. Concentration prefixes are p = pico (10⁻¹²), n = nano (10⁻⁹), µ = micro (10⁻⁶). R/R_a is the ratio of helium isotopes in the sample divided by the ratio in air = (³He/⁴He)_measured/(³He/⁴He)_air. Selected characteristic noble gas ratios for water and crust are included for comparison. Note that the water ratios are determined from the relative solubility of atmospheric noble gases in contact with water. The ⁸⁴Kr and ¹³²Xe are fission products of uranium and their concentrations are dependent upon the uranium content of the crustal rocks. ²¹Ne is generated by the α particle produced by uranium decay reacting with oxygen and magnesium in the rock. ⁴⁰Ar is produced by the radioactive decay of K.

Well	3He pcc/cc	4He µcc/cc	20Ne µcc/cc	21Ne µcc/cc	22Ne µcc/cc	Ne µcc/cc	36Ar µcc/cc	38Ar µcc/cc	40Ar µcc/cc	Ar µcc/cc	84Kr ncc/cc	Kr ncc/cc	132Xe ncc/cc	Xe ncc/cc
DG19	39.67	501.6	6.53	2.609	93.25	9.14	9.65	1.97	3088.4	3100.0	132.14	231.9	3.720	13.8
DG17	4.56	298.2	0.45	0.002	0.06	0.46	0.29	0.06	108.6	109.0	6.63	11.6	0.164	0.6
DG12	41.17	774.3	0.07	0.001	0.05	0.07	0.22	0.04	90.9	91.1	6.33	11.1	0.335	1.2
DG20	27.42	276.2	1.46	0.005	0.18	1.46	0.67	0.13	216.7	217.5	64.00	112.3	4.713	17.5
DG18	17.23	300.1	0.23	0.001	0.03	0.23	0.30	0.06	93.9	94.2	9.46	16.6	0.377	1.4
Well	R/Ra	20Ne 22Ne	21Ne 22Ne	38Ar 36Ar	40Ar 36Ar	82Kr 84Kr	83Kr 84Kr	86Kr 84Kr	128Xe 132Xe	129Xe 132Xe	130Xe 132Xe	131Xe 132Xe	134Xe 132Xe	136Xe 132Xe
DG19	0.057	9.597	0.0289	0.204	320.100	0.194	0.248	0.297	0.146	0.973	0.148	0.828	0.394	0.354
DG17	0.011	8.223	0.0332	0.189	373.200	0.190	0.247	0.294	0.135	0.985	0.155	0.802	0.452	0.422
DG12	0.038	9.674	0.0311	0.189	404.100	0.194	0.242	0.289	0.181	0.992	0.160	0.821	0.411	0.348
DG20	0.071	9.472	0.0290	0.200	324.600	0.190	0.252	0.295	0.114	0.965	0.134	0.800	0.396	0.365
DG18	0.041	9.597	0.0341	0.198	315.690	0.176	0.224	0.277	0.157	0.965	0.136	0.799	0.398	0.346

and

Table 5. Select sample noble gas ratios compared to water and crustal ratios.

	R/Ra	20Ne 22Ne	21Ne 22Ne	40Ar 36Ar	84Kr 36Ar	132Xe 84Kr	4He 40Ar*	21Ne* 40Ar*	20Ne 36Ar	84Kr 36Ar
Water	1	9.8	0.029	295.5	0.038	0.021
Crust	0.02	0.09	0.450	3000	0.029	0.138
DG19	0.057	9.597	0.0289	320.100	0.0137	0.028	196.7	0	0.677	0.014
Dg17	0.011	8.223	0.0332	373.200	0.0228	0.025	1.12	0.43	1.559	0.023
DG12	0.038	9.674	0.0311	404.100	0.0281	0.053	1.60	0	0.294	0.028
DG20	0.071	9.472	0.0290	324.600	0.0959	0.074	6.343	0	2.180	0.096
DG18	0.041	9.597	0.0341	315.690	0.0318	0.040	4.42	43.3	0.765	0.032

). Noble gas measurements were performed using a ThermoFisher Helix Split Flight Type noble gas mass spectrometer at Smart Gas Sciences laboratories in Columbus, OH, USA.

Twenty-three condensate samples, co-produced with the wet gases described above, were collected as dead oil (i.e., collected at atmospheric pressure and lacking volatiles) at the separator in glass jars with Teflon lids and shipped to Stratum Reservoir Laboratories in Houston, Texas USA (

Table 6. Condensate data for selected wells.

Well ID/Group	°API Gravity	C7 Temp °C	Pristane/Phytane	Toluene/nC7	nC7/MCH	Heptane Ratio	Isoheptane Ratio
WG1	65.5	129.27	1.84	0.17	1.37	31.3	9.16
WG2	74.6	131.88	1.94	0.11	1.83	33.56	8.55
WG3	69.8	131.09	1.48	0.16	1.63	32.93	9.18
WG4	69.6	131.47	2.05	0.14	1.87	34.05	9.98
WG5	69.4	130.17	1.42	0.13	1.74	33.95	9.92
WG6	70.9	131.23	1.35	0.21	1.12	29.85	5.71
WG7	57.8	132.03	1.35	0.2	1.16	30.72	5.09
WG8	65.3	130.05	1.58	0.15	1.64	33.91	9.75
WG9	68.3	130.86	1.7	0.16	1.67	32.99	10.05
WG10	66.5	130.04	1.62	0.15	1.67	33.91	9.93
WG11	64.8	129.79	1.61	0.16	1.65	34.07	9.86
WG12	73	132.21	1.94	0.14	1.81	33.36	9.26
WG13	-	131.2	1.91	0.11	1.88	34.29	8.94
WG14	70.8	130.7	1.92	0.106	1.95	34.8	9.25
WG18	60.1	129.23	1.88	0.2	1.17	30.45	7.34
WG20	57.4	130.1	1.38	0.22	1.14	30.16	6.34
WG21	70.4	130.9	2.21	0.14	1.76	33.36	10.43
WG23	-	130.5	1.61	0.232	1.21	28.86	8.58
WG24	58.4	131.55	1.43	0.2	1.12	27.33	8.15
WG25	53.2	131.1	1.89	0.12	1.97	34.45	9.77
WG26	-	131.6	1.38	0.41	1.16	30.9	5.41
WG27	-	131.5	2	0.12	1.78	33.92	9.46
WV28	-	131.7	1.82	0.17	1.71	32.32	9.97
PAWG1	74.6	-	-	-	-	-	-
PAWG2	53.2	-	-	-	-	-	-

). High-Resolution Gas Chromatography (HRGC) was performed using an Agilent 6890 GC-FID instrument. API gravity was derived from measurement of specific gravity by hydrometer (ASTM D1298).

Excepting the pressure core gas samples, all of the fluid samples were collected from horizontal wells stimulated by hydraulic fracturing. The Marcellus reservoir samples represent the landing zones of the wells and are not necessarily from the exact petroleum source rock intervals of the produced gas and condensate.

4. Results and Discussion

4.1. Bulk Composition and Stable Isotopes of Produced Gases

Samples collected in the wet gas-producing region exhibit bulk compositions and isotope ratios characteristic of oil-associated thermogenic gases [35,36,37,38] (Figure 6). Methane (C₁) concentrations range from 69.37 to 92.27 mole%, with δ¹³C₁ ranging from −46.74 to −40.15‰ VPDB and δD_CH4 from −243.8 to −167.2‰ VSMOW. Ethane (C₂) and propane (C₃) concentrations are between 5.27 and 16.68 and 0.5 and 7.81 mole%, respectively.

All but one of the production gas samples collected in the dry gas-producing region exhibit bulk compositions typical of high-maturity thermogenic gas [35,36,37,38] (Figure 6). The gases are composed primarily of methane, with small volumes of ethane and trace contributions of propane and butane+ (C₄₊). Methane concentrations range from 94.45 to 97.87 mole%. One anomalous sample contains only 55.27 mole% C₁ due to a high nitrogen (N₂) content related to air contamination as confirmed by a N₂/Ar ratio similar to that of air. Isotope ratios of hydrocarbon gases are consistent with a thermogenic origin, with δ¹³C₁ ranging from −39.2 to −23.53‰ VPDB and δD_CH4 from −163.9 to −157.7‰ VSMOW (Figure 6b). Both ranges indicate a thermogenic origin of methane without any contribution from a microbial source [33,34,35,36] (Figure 6). The one produced gas sample with only 55.25 mole% C₁ is an exception: it has a δ¹³C_CH4 of −59.58‰ VPDB and a C₁/(C₂ + C₃) ratio of 2763.5 indicating mixing with microbial methane [32,33] (Figure 6a).

The Marcellus Formation gases from both the dry gas-producing region and the wet gas-producing region show a general trend of enrichment in ¹³C as the gases become richer in methane (Figure 6a). However, there is considerable scatter in ¹³C₁ of the dry gas samples compared to the wet gas samples due to the shifting fields for kerogen type on the plot of ¹³C₁ against C₁/(C₂ + C₃). The shift from kerogen type II to kerogen type III exhibited by the dry gas samples in Figure 6a is a function of increasing thermal maturity accompanied by decreasing atomic H/C in the parent organic matter, and not due to variation in organic facies. Hydrogen Index (HI) and Oxygen Index (OI) decrease across the basin from average respective values of 529–582 mg hydrocarbon/g TOC and 15–34 mg CO₂/g TOC in thermally immature to mature Marcellus samples [31,32] to 6–31 mg hydrocarbon/g TOC and 7–9 mg CO₂/g TOC in overmature Marcellus samples [12]. Respective H/C and O/C ratios in the latter are as low as 0.40 and 0.06 [12].

The respective δ¹³C values of ethane and propane in the wet gas samples range between −35.87 and −31.2‰ and −31.22 and −25.67‰ (Figure 7a). The wet gases exhibit normal kinetic isotope trends on the natural gas plot shown in Figure 7a, with δ¹³C₁ < δ¹³C₂ < δ¹³C₃. Deviations from a straight line in Figure 7a indicate a mixture of source inputs, heterogeneity in the organic source material, and/or secondary hydrocarbon gas alteration effects [39,40]. By contrast, the ethane δ¹³C of dry gas samples ranges from −42.11 to −29.59‰ and the propane δ¹³C of the dry gases ranges from −43 to −31.4‰ (Figure 7b). All the samples collected from the dry gas-producing region exhibit full carbon isotope reversals with δ¹³C₁ > δ¹³C₂ > δ¹³C₃ (Figure 7b). The δD ratios of the dry gas samples are reversed as well with δD_CH4 > δD_C2 (Figure 8). Wet gases, by comparison, exhibit normal hydrogen isotope trends with δD_CH4 < δD_C2 (Figure 8). The carbon and hydrogen isotope reversals observed in the dry gas samples are considered in detail in Section 4.6 below.

Figure 9 is a plot of gas wetness versus δ¹³C₂ for all the samples from both the wet and dry gas-producing regions. Beginning at measured VR_o = 1.59%, the wet gases plot along the expected thermal maturity trend defined by decreasing wetness and increasing δ¹³C₂. However, the most thermally mature wet gas sample does not follow the expected maturity trend. It “rolls over”, i.e., exhibits decreasing wetness accompanied by decreasing δ¹³C₂. Rollover occurs at a wetness of about 5.9% and VR_o = 1.78%. Dry gases with full isotopic reversals (Figure 7 and Figure 8) exhibit rapidly decreasing wetness (~4 to 1.4%) and sharply increasing δ¹³C₂ between measured VR_o values of 2.35 and 4.61%.

Non-hydrocarbon gases produced from the Marcellus reservoirs are present in much smaller amounts, with CO₂ concentrations up to 0.21 mole% in the wet gas-producing region and 0.061 mole% in the dry gas-producing region. δ¹³CO₂ was measured for two dry produced gas samples. δ¹³CO₂ = −26.5‰ in the DG12 gas and −31.8‰ in the DG19 gas (Table 2). These values indicate an organic, high maturity thermogenic origin for the CO₂ [37]. Excepting the one air-contaminated sample (DG19), N₂ concentrations in the dry gas region are between 0.4 and 2.04 mole%. δ¹⁵N was measured in the DG12 sample and is −10.2‰ (Table 2). Nitrogen concentrations in the wet gas region are between 0.33 and 1.4 mole%. δ¹⁵N measured in three of the wet gas samples is between −13.9 and −12.7‰. These isotopically light δ¹⁵N values indicate a thermogenic organic nitrogen source [37].

4.2. Noble Gas Isotopic Analyses of Dry Gas Samples

Helium concentrations in a subset of gas samples collected from the dry gas-producing region range from 276.2 to 774.3 µcc/cc for ⁴He and 4.56 to 41.17 pcc/cc for ³He (Table 4). Helium isotope ratios are reported relative to the atmospheric ratio R_a = 1.4 × 10⁻⁶ and range from 0.011 to 0.071R_a (Table 3 and Table 4 and Figure 10a). ⁴He/²⁰Ne ratios are between 76.8 and 11,704.3, which by comparison with the atmospheric ⁴He/²⁰Ne ratio (0.188) shows that atmospheric helium contributions are negligible. Helium isotope ratios can therefore be considered as two-endmember mixing between crustal radiogenic helium (0.02R_a) and minute contributions of ³He-enriched mantle helium, for which sub-continental lithospheric mantle ³He/⁴He is 6.1R_a [44,45,46]. ³He/⁴He ratios > 0.1R_a are attributed to mantle fluid contributions [44]. Most of helium in the Marcellus Formation dry gases is derived from crustal radiogenic production (Figure 10a).

Neon isotope ratios in the dry Marcellus Formation gases show a slight excess of radiogenically produced ²¹Ne and/or ²²Ne relative to air, for which atmospheric ²⁰Ne/²²Ne is 9.81 and ²¹Ne/²²Ne is 0.029 [45] (Figure 10b). ²⁰Ne/²²Ne in the Marcellus samples ranges from 8.223 to 9.674. ²¹Ne/²²Ne is between 0.0289 and 0.0341. The samples mostly follow the crustal neon production-air mixing line, but scatter among some samples possibly reflects mass-dependent fractionation effects [44,46] (Figure 10b). Atmospheric ²⁰Ne concentrations range from 0.07 to 6.53 µcc/cc. The highest value is in the air contaminated DG19 sample and is considered anomalous and non-representative of Marcellus fluids.

Argon isotope ratios (⁴⁰Ar/³⁶Ar) show a modest excess of radiogenically produced ⁴⁰Ar, ranging from 315.69 to 404.1, compared with the air ratio of 298.6 [44,45,46,47]. Atmospheric ³⁶Ar is introduced into the subsurface dissolved in groundwater and in sediment pore water, and its concentration in the gas phase is controlled by solubility-dependent partitioning upon water-gas interaction. ⁴⁰Ar/³⁶Ar values in excess of the atmospheric ratio are due to contribution of radiogenic ⁴⁰Ar (designated ⁴⁰Ar*) produced by ⁴⁰K decay within the crust. The Marcellus ⁴⁰Ar/³⁶Ar ratios indicate variable additions of radiogenic ⁴⁰Ar* to atmospheric, groundwater-derived Ar (Figure 11a). The ⁴He/⁴⁰Ar* ratios of the dry Marcellus gases indicate crustal production ratios reflecting very high thermal stress for the source material of the produced hydrocarbons (Figure 11b) [45,46,47,48,49].

Krypton and xenon abundances in the Marcellus Formation gas samples range from 11.1 to 231.9 ncc/cc and 0.6 to 17.5 ncc/cc, respectively (Table 4). Krypton and xenon ratios of gases produced from the DG17 and DG19 wells are indistinguishable from air and are slightly elevated relative to air in the DG12 and DG18 samples (Table 4 and Table 5). In sample DG20, Kr abundance is enriched relative to Ar when compared to air or ASW values (Figure 12). ⁸⁴Kr and ¹³²Xe are fission products of uranium and their concentrations are dependent upon the uranium content of the crustal rocks. Previous studies suggest that excess Kr and Xe may be sourced from organic-rich shales themselves due to preferential sorption of the heavier noble gases onto clay or kerogen [47,48,49]. Figure 12 is a plot of ²⁰Ne/³⁶Ar against ⁸⁴Kr/³⁶Ar for dry Marcellus gases. Most of the gas samples lie near to the air and air-saturated water (ASW) values for ⁸⁴Kr/³⁶Ar indicating gas/water phase equilibrium. Three of the samples show variably high fractionation in ²⁰Ne/³⁶Ar suggesting that equilibration of groundwater with an earlier oil phase may have increased ²⁰Ne/³⁶Ar in the water phase [45,47]. One gas sample (DG20) exhibits high ²⁰Ne/³⁶Ar and extreme fractionation of ⁸⁴Kr/³⁶Ar (Figure 12). This may reflect release of atmosphere-derived Kr originally adsorbed and trapped in organic-rich sediment but might also indicate non-equilibrium diffusion or multiple stages of exsolution and dissolution [48,49].

4.3. API Gravity and High-Resolution Gas Chromatography (HRGC) Analyses of Condensate Samples

The API Gravity of produced fluids measured in the wet gas region ranges from 57.8 to 74.6° (Table 6). All the samples are reported as condensate (>45° API) [52]. The condensates are dominated by light hydrocarbons mostly in the C₆ to C₁₅ range. Figure 13 shows two HRGC chromatograms of condensates produced in the wet gas region. The WG6 condensate exhibits a unimodal n-alkane distribution with a maxima at nC₆ and resolvable hydrocarbons out to nC₄₀. The WG2 condensate also exhibits a unimodal n-alkane distribution, with maxima at nC₅, but GC peak responses are much diminished beyond nC₁₂ and the resolvable hydrocarbons only extend out to nC₃₀. All the produced condensate samples listed in Table 6 exhibit unimodal n-alkane distributions that vary between these two extremes with maxima between nC₅ and nC₇ and resolvable hydrocarbons between nC₄ and nC₄₀. Differences in the molar slopes of the light hydrocarbons (nC₆–nC₁₂) and the black oil C₁₅₊ hydrocarbons (nC₁₇–nC₂₇) observed on the HRGC chromatograms of the Marcellus samples (Figure 13 and Figure 14) reflect various primary and secondary processes that affected the produced fluids such as maturity and thermal alteration, expulsion and migration, mixing, gas stripping, water washing, and plausibly production fractionation [53,54].

Thompson [55] developed several C₆–C₇ ratios that are useful for describing light hydrocarbon distributions by compound class. A cross-plot of two of these ratios, the heptane and isoheptane ratios, is useful for assessing light oil maturity [56], including maturity of the Marcellus fluids (Figure 15a). The heptane and isoheptane ratios reflect the relative abundance of C₇ structural isomers in the fluid [55]. The heptane ratio of the WG6 condensate is 29.85 and its isoheptane ratio is 5.71 (Table 6). The respective heptane and isoheptane ratios of the WG2 condensate are 33.56 and 8.55 (Table 6). These ratios suggest that the condensates are so-called supermature light oils. The supermature classification indicates that all of the light Marcellus oils were thermally cracked [55]. The fluids experienced protracted thermal transformation, i.e., ring opening, and substantial gasification (Figure 15a).

BeMent et al. [57] and Mango [58] concluded that the ratio of 2,4-dimethylpentane/2,3-dimethylpentane in light oil corresponds to the temperature of petroleum generation and expulsion from organic matter in a source rock:

°C7_TEMPERATURE = 140 + 15((ln [2,4-dimethylpentane/2,3-dimethylpentane]).

(2)

C₇ expulsion temperatures in the Marcellus Formation condensates range from 130 to 132 °C (Figure 15b) indicating late-mature hydrocarbons [52]. These temperatures are less than those indicated by post mature VR_o values in the wet gas reservoirs (Figure 3 and Figure 9). Accurate C₇ temperatures based on 2,4-dimethylpentane/2,3-dimethylpentane ratios are limited to a range of about 95 to 135 °C, the temperatures associated with intense generation and expulsion; subsequent thermal decomposition of oil expelled from parent kerogen into the pore space of the rock is unimportant [57,58]. I conclude that (1) primary Marcellus oils were mostly generated in the late oil window at burial temperatures of approximately 130 to 132 °C and subsequently (2) thermally cracked and altered to yield lighter hydrocarbons and gas. The API values of most condensates correlate with C₇ expulsion temperatures (Figure 15b).

The two plots of toluene/nC₇ against nC₇/methylcyclohexane (MCH) shown in Figure 16 suggest additional alteration processes in the Marcellus condensates. Toluene/nC₇ is an aromaticity ratio and nC₇/MCH is a paraffinicity ratio. Unaltered marine oils have toluene/nC₇ between about 0.2 and 0.6 and nC₇/MCH between 0.4 and 1.5 [55,56]. Toluene/nC₇ in the Marcellus condensates is between 0.11 and 0.21 and nC₇/MCH is between 1.12 and 1.87 (Figure 16). Most of the Marcellus Formation liquids produced in the wet gas region appear to have been stripped by gas, i.e., migrating gas dissolved light end alkanes into itself [54]. Furthermore, the plots in Figure 16 shows systematic depletion of soluble toluene accompanied by increasing nC₇ which suggests the hydrocarbons were affected by hydrodynamic flow and water washing [55]. Low concentrations of water-soluble benzene (0.62 to 1.19%) in the liquids support this interpretation [59]. Very low TR7 (1-trans−3-dimethylcyclopentane/1,1-dimethylcyclopentane) and TR8 (2-methylhexane + 3-methylhexane/2,2-DMP +2,3-DMP +2,4-DMP + 3,3-DMP + 3-ethylpentane) suggest migration and secondary alteration effects [59].

Figure 17 is a plot of C₇ expulsion temperature versus molar slope A (nC₁₇–nC₂₇) for the Marcellus condensates. The plot provides a comparison of the thermal maturities of the light hydrocarbons (nC₆–nC₁₂) and the C₁₅₊ hydrocarbons in the oils. While a few samples follow the expected thermal maturity trend for normal unaltered oils, most samples plot up and to the left of that trend indicating mixing with higher maturity oil [54]. Most Marcellus Formation condensates produced in the wet gas region are mixtures of primary late-mature light oils cracked from kerogen and secondary liquids cracked from residual oils in the shales.

4.4. Variations in Thermal Maturity of Marcellus Formation Hydrocarbon Gases

As stated earlier, condensate production in the basin is from reservoirs with reported VR_o of 1.0 to 2.8%. Dry gas production is from reservoirs with reported VR_o from 2.0 to >3.0%. While VR_o is considered a critical parameter for maturity assessment in the Marcellus play [1,11], it is a difficult value to accurately measure in these rocks [32]. Published vitrinite reflectance data and maps show VR_o contours ranging from 0.5% in western New York, northwestern Pennsylvania, and eastern Ohio to between 2.5 and 3.5% in eastern New York, northeastern Pennsylvania, western Maryland, and central/eastern West Virginia [2,32,33]. On these maps, wet gas production in western Pennsylvania and northwest West Virginia is bound by VR_o = 1.0 to 1.6% and dry gas production is bound by VR_o = 2.0 to 3.5% [2]. This contrasts significantly with the carefully sampled and measured VR_o completed for this study (Figure 3 and Figure 4) and in published site-specific investigations [29,30,31,32]. Published burial history reconstructions indicate modeled VR_o of 1.3 to 1.8% in the wet gas region and 2.3 to 3.2% in the western part of the dry gas region [33,34]. Interestingly, reported measured VR_o in samples collected from the same modeled wells are 0.75 to 1% in the wet gas region and ±2.0% in the dry gas region [33]. These large reported VR_o discrepancies are due to numerous sources of error including sample type (core, cuttings, outcrop) and handling, contamination, lithology, and organic matter identification [32]. Organic matter in thermally post mature and overmature Marcellus Formation reservoirs consists mostly of a solid bitumen/pyrobitumen organic matrix with sparse inertinite (Figure 3 and Figure 4) [12,30]. Vitrinite is largely absent or difficult to recognize and oil-prone kerogens have been converted to hydrocarbons (Figure 4 and Figure 5) [30]. Consequently, many workers have endeavored to identify and utilize different geochemical proxies for rapid and practical maturity assessment during Marcellus reservoir development [12,30,32,60,61]. These include produced gas composition and stable gas isotopes [62,63,64].

Figure 18a is a popular maturity model from Whiticar [65] based on mathematical models published by Faber [66] and Whiticar [67]. This model grossly underestimates the actual thermal maturity of the Marcellus gases due secondary effects discussed below [65]. However, this plot of δ¹³C₂ versus δ¹³C₃ exhibits strong congruence for the gas isotopic values suggesting that the data do reflect thermal maturity [65,68,69]. Calculating distance between δ¹³C₂ and δ¹³C₃ of the Marcellus gas samples (Figure 18b) provides calibration of the carbon isotope correlation trends to measured and modeled maturity parameters.

Figure 19a is a plot of methane δ¹³C versus ethane δ¹³C. The trends for the wet and dry gas samples are again highly congruent although some scatter in δ¹³C₁ values suggests mixing of gases cracked from kerogen and gases cracked from oil (discussed further below). In this model, the wet gas maturities do approximate those observed in the Marcellus Formation organic matter. The dry gas data, however, shift away from the modelled maturity trends and lie within the field of a so-called super mature Type II kerogen source.

Noble gas data also constrain the thermal maturity of dry gases collected in northeastern Pennsylvania (Figure 11b and Table 5). Hunt et al. [50] demonstrated that post mature to overmature gases in the northern Appalachian basin have δ¹³C₂–δ¹³C₁ ratios of −7.10 to 7.20 and ⁴He/⁴⁰Ar* ratios of 12 to 65. The δ¹³C₂–δ¹³C₁ ratios of most Marcellus Formation hydrocarbons produced in the dry gas region are between −7.32 and −5.33. Most Marcellus ⁴He/⁴⁰Ar* ratios range from 1.6 to 6.34, values within or below crustal noble gas production ratios due to high-temperature release of He, Ne, and Ar from the rocks [50]. ²¹Ne/⁴⁰Ar* ratios (1.06 × 10⁻⁵ to 2.18 × 10⁻⁵) are below crustal production ratios as well [50].

4.5. Secondary Effects

Several secondary effects have altered the chemical composition and isotopic ratios of the produced Marcellus Formation gases. The documented range of carbon isotopic compositions of Marcellus organic matter suggests that some part of the range of isotopic compositions of the produced gases might be due to mixing of gases generated from kerogen with variable δ¹³C (Figure 7). However, a plot of gas dryness against the per-mil difference between kerogen δ¹³C and methane δ¹³C (Figure 19b) indicates that all the wet gas samples and several of the dry gas samples were altered by incipient to extensive oil cracking in the wet gas window. This is consistent with the heptane and isoheptane ratios in the condensates discussed above (Figure 15a). Gases derived from oil cracking are isotopically heavier than gases derived directly from kerogen cracking and become heavier with increasing maturity due to Rayleigh fractionation [68]. Gas derived from oil cracking mixes with earlier-formed gas derived from kerogen cracking which remained dissolved in the oil undergoing cracking [68]. Thus, the methane δ¹³C of the wet gas samples is a weighted average of the isotope ratio of the gas cracked from oil and earlier-formed gas cracked from kerogen [68,69]. Wet gas mixing and maturation trends are linear and highly congruent on the plots of methane δ¹³C versus methane δD (Figure 20a) and methane concentration (Figure 20b). The C₂/C₃ ratios and difference between δ¹³C₂ and δ¹³C₃ in these samples are 2.1 to 4.4 and −4.62 to −4.97‰, respectively, indicating secondary oil cracking [70]. Rayleigh-type fractionation should cause the isotope ratios of individual hydrocarbon components to be a linear function of the natural logarithm of the concentration as shown in Figure 21 [71,72]. Ethane isotope ratios of the wet gases plot on a straight line as a function of the inverse of concentration, indicating progressive depletion of δ¹³C₂ with decreasing concentration (Figure 21). Wet Marcellus gases are mixtures of residual associated gases generated in the late-oil window and post-mature hydrocarbons generated from oil cracking in the wet gas window.

Most of the samples collected from the dry gas-producing region fall within the genetic field for methane cracked from refractory kerogen at VR_o equivalents of approximately 1.5 to ≥ 2.5% on the plot of C₁/∑C₁-C₅ versus the per-mil difference between methane and total source kerogen (Figure 19b). Actual thermal stress in these Marcellus rocks is much higher based on carefully measured VR_o, illite crystallinity, H/C and O/C ratios, noble gas data, and calibrated stable gas isotope ratios (Figure 3, Figure 4, Figure 11 and Figure 18). Published mineral and fluid inclusion data indicate burial temperatures of 250–300 °C [30,72,73,74,75,76,77]. Lewan and Kotarba [78] experimentally demonstrated that primary gas generation from oil-prone and gas-prone kerogen and associated bitumen ends by a VR_o of 2.0% because no alkyl moieties remain to source hydrocarbons. Secondary gas generation from oil cracking occurs primarily between VR_o 1.7 and 2.9% [78]. Hydrocarbon gases already generated and trapped, however, can remain in the shales to much higher temperatures [78,79,80]. The productivity of Marcellus Formation reservoirs in the dry gas region is a function of shale reservoir properties (porosity, permeability, fluid saturations, rock mechanics) and hydrocarbon retention versus expulsion; productivity is not a function of thermal maturity [12,30,81,82]. The noble gas data listed in Table 3 support this interpretation. In Figure 22, mixing is evident between two atmosphere-derived end members, ²⁰Ne and ³⁶Ar. ²⁰Ne is highly fractionated. ³⁶Ar is non-fractionated. Atmospheric ³⁶Ar exhibits a negative correlation with ²⁰Ne/³⁶Ar (Figure 22). Gas collected from economically productive wells have high ²⁰Ne/³⁶Ar. Gases sampled from poor producing wells with high water cut have low ²⁰Ne/³⁶Ar similar to unfractionated groundwater. Decreasing gas productivity and decreasing gas retention are correlated with decreasing ²⁰Ne/³⁶Ar (Figure 22).

Like the wet gases described above, the dry gas isotope ratios are cumulative so mixing of wet gas components generated by extensive oil cracking and methane derived from postmature refractory kerogen cracking is implied by the plot in Figure 19b [68]. The dry gas samples exhibit depletion in ²H as methane becomes progressively enriched in ¹³C (Figure 20a). This uncommon trend is also observed in deeper non-associated Silurian through Ordovician conventional and unconventional reservoir gases in the study area [72]. Burruss and Laughrey [72] suggested that isotopic exchange with formation water is one mechanism that could explain ²H depletion in methane produced from these deeper reservoirs. Isotopic exchange with formation water is a possible mechanism for ²H depletion in the dry Marcellus gases as well.

The isotopically heaviest methane measured in the Marcellus Formation to date, sampled from mud gases recovered from non-productive shale wells drilled in Luzerne County, Pennsylvania, has δ¹³C₁ between −18.9 and −23.8‰ and δD_C1 between −186 and −166‰ [12]. The dry gas mixing and maturation plots in Figure 20 are reflective of mixing with this isotopically heavy, postmature late-stage methane derived from refractory kerogen. C₂/C₃ ratios in the dry gas samples are between 20.5 and 44.4, and the difference between δ¹³C₂ and δ¹³C₃ is between 2.41 and 1.4‰ indicating secondary gas cracking [70]. The dry gas samples exhibit Rayleigh fractionation trends consistent with ethane and propane destruction during secondary gas cracking accompanied by redox reactions [72] (Figure 21).

One of the dry gas region samples (DG19) has an unusually light methane δ¹³C of −59.58‰ indicating a microbial gas component (Figure 6, Figure 7 and Figure 19b). Ethane δ¹³C is −37.2‰. The gas is a mix of microbial and thermogenic hydrocarbons. This sample has unusually high crustal ⁴⁰Ar (3088.4 µcc/cc) and relatively high atmospheric ³⁶Ar (9.65 µcc/cc) compared to other noble gas concentrations in this region (Table 4 and Figure 22). The high ³⁶Ar concentration suggests recent interaction with atmospheric gases [44,45,46,47]. This sample has a ²⁰Ne/³⁶Ar ratio of 0.677 which approximates the upper limit for single-stage equilibrium for fractionation between groundwater and gas [44,45,46,47] (Figure 22). The sample also contained 17.2 mole% hydrogen suggesting hydrolysis and casing corrosion.

4.6. Isotopic Reversals: Origin and Significance

Experimental and field evidence indicates that most terrestrial thermogenic hydrocarbons have so-called normal δ¹³C and δD alkane patterns such as those exhibited by Marcellus Formation samples collected from the wet gas-producing region [37,38,39,40] (Figure 7a and Figure 8). This normal pattern, with δ¹³C₁< δ¹³C₂< δ¹³C₃ and so on, and δD_C1< δD_C2 < δD_C3 and so on, is the result of kinetic isotope fractionation in which the lighter carbon and hydrogen isotopes are more reactive than their heavier counterparts and take part in petroleum-generating reactions more readily [39,40]. Terrestrial hydrocarbons generated at high temperatures and pressures deeper within the crust, however, can have reversed δ¹³C and δD patterns [62,63,72,83,84,85,86,87,88,89,90,91,92]. Marcellus gases in the dry gas-producing region of the basin exhibit reversed patterns (Figure 7b and Figure 8).

The origin of isotopic reversals in dry gas is a controversial topic [85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100]. Reversed thermogenic δ¹³C and δD alkane patterns have been interpreted as diagnostic evidence for:

• Mixing of different sources of alkanes or of hydrocarbons generated from the same source at different levels of thermal maturity [62,63,83,84,85,86,87];
• A combination of mixing with either Rayleigh fractionation (for carbon isotopes) or the exchange of methane hydrogen with formation water [72];
• Destruction and subsequent reformation of C₂₊ alkanes via gas-phase radical recombination reactions at high temperatures and pressures [90];
• Depletion of ¹³C in residual ethane and propane as their molar fractions decrease during thermal decomposition [94];
• Diffusive gas leakage [95];
• Water reforming followed by Fischer–Tropsch synthesis [96];
• Carbon exchange at high temperature [97];
• Aerobic and anaerobic microbial oxidation of select alkane components [98];
• Thermochemical sulfate reduction [99];
• Isotopic fractionation during desorption from uplifted and de-pressurized late-mature shale resulting in isotope reversals in residual gas [92]; and
• Hydrogen and carbon isotope distributions reflecting thermodynamic equilibrium either at the conditions of gas formation or during reservoir storage [100].

Two of these proposed mechanisms for generating isotopic reversals, microbial oxidation and thermal sulfate reduction, can be ruled out for the dry Marcellus gases based on geologic setting. Diffusive gas leakage is eliminated as a possible mechanism because the Marcellus reservoirs produce both isotopically heavy hydrocarbons and enriched light noble gases [50]. The Marcellus ⁴He/CH₄ ratios (283.1 to 907.6) are consistent with in situ Marcellus reservoir gases [8]. While C_2–4 alkane synthesis from methane during gas-phase radical recombination reactions at high temperatures and pressures, resulting in isotopic reversals, has been demonstrated experimentally and for some abiogenic hydrocarbon reservoirs [88,89,90,91], maximum burial temperatures for the Marcellus Formation in northeast Pennsylvania fall short of those needed for these reactions [91]. Any argument for water reforming followed by Fischer–Tropsch synthesis remains equivocal because supporting carbon isotope measurements for CO₂ are mostly lacking in this dataset. The δ¹³CO₂ of the DG12 and DG19 gas samples (−26.5‰ and −31.8‰, respectively) suggest that hydrocarbon oxidation might have affected these gas samples so might support such an interpretation [38,96]. However, mud gas δ¹³CO₂ of wells in the study area are between −19.7‰ and −10.2‰, values more indicative of kerogen decarboxylation and high thermal stress [36,37,38].

Depletion of ¹³C in residual ethane and propane during thermal decomposition [94] is a plausible hypothetical explanation for the carbon isotopic reversals observed in the Marcellus dry gas samples. However, the hydrogen isotope reversals (Figure 8) and variable, but significant, concentrations of atmospheric noble gas components (²⁰Ne, ³⁶Ar, ⁸⁴Kr) in the produced gases provide strong supporting evidence for the role of water-involved reactions in generating the isotope reversals [44,45,46,47,50,51,72,100].

Carbon and hydrogen isotopic reversals in the dry Marcellus Formation gases resemble those described by Burruss and Laughrey [72] for deeper Silurian and Ordovician conventional and unconventional reservoirs in northeastern and north-central Pennsylvania and south-central New York state. As in those deeper reservoirs, mixing combined with Rayleigh fractionation for the carbon isotopes (Figure 20 and Figure 21) and the exchange of methane hydrogen with formation water for the hydrogen isotopes provides a plausible model for the Marcellus isotope reversals shown in Figure 7b and Figure 8. Methane δD in the dry Marcellus gas samples ranges from −163.9 to −157.7‰ (Figure 8). Produced Marcellus brines in the dry gas region have reported δD between −49.1 and −41.5‰ [101]. Methane δD is depleted by approximately −118.8‰ relative to co-produced water. Methane equilibrated with formation water of the reported isotopic composition should have δD compositions of about −183‰ to −160‰ assuming maximum burial temperatures of 250 to 300 °C [72,102]. Measured δD in the dry Marcellus gas samples has this composition (Figure 6b and Figure 20a; Table 1). The noble gas data discussed above indicate gas/water phase equilibrium for most Marcellus dry gases (Figure 12). Whyte and Darrah [103] argued that hydrocarbon isotope reversals are generated in relatively closed systems that retain greater amounts of ASW noble gas components as observed in the Marcellus gases and in the deeper Silurian and Ordovician fluids in the study area [72]. Finally, experimental work published by Gao et al. [104] demonstrated that water facilitates secondary cracking of C₂₊ hydrocarbons to methane, possibly via a hydrated surface with catalytic activity, thus supporting the Burruss and Laughrey [72] model for isotopic reversals.

Milkov et al. [92] examined a comprehensive global data set of shale-gas composition and isotope data and observed that samples with carbon isotope reversals generally occur in onshore shale plays with overmature organic matter that underwent significant tectonic uplift. They hypothesized that isotope fractionation during desorption induced by significant basin inversion and depressurization of overmature shale source rocks leads to carbon isotope reversals. They concluded that the amount of uplift and associated depressurization is more important to creating isotopic reversals than thermal maturity in shale plays. Milkov et al. [92] noted that produced Marcellus gases with isotopic reversals occur in areas with 3 to 5 km of uplift and VR_o between 2 and 4% while produced Marcellus gases without reversals occur in areas with 1 to 2 km of uplift and VR_o between 1 and 2%. Milkov et al.’s samples without reversals are from wells completed along the up-dip margins of the Appalachian basin in western and central New York and in northwestern and western Pennsylvania [86,92,105]. The Marcellus wet gas samples without reversals in this paper’s data set (Figure 7a) are from deeper locations that experienced ~1.04 to 2.4 km of erosion and uplift, have maturities of 1.59 to 1.78% VR_o, and are within the boundaries of the Rome Trough, a northeast-trending graben that is one of the major Cambrian interior rift system elements that extends into the Appalachian foreland basin [106]. In addition to burial temperatures, the thermal history of Devonian black shales within the Rome Trough was affected by excess heating connected to hot brine migration within the structure [107]. Marcellus dry gas samples with reversals in this paper’s data set are from locations that experienced ~2.6 to 2.9 km of erosion and uplift and have maturities of 2.35 to 4.61% VR_o. The Marcellus gases without reversals in this paper’s data set fall within the maturity parameters of such gases as defined by Milkov et al. [92] but experienced greater uplift. The Marcellus gases with reversals in this paper’s data set are within the maturity range of such gases as defined by Milkov et al. [92] but experienced somewhat less uplift than those classified as reversed by those authors. However, uplift in the dry gas region was more rapid than in other parts of the basin and was accompanied by more rapid post-Alleghenian cooling and unroofing [108].

The Milkov et al. [92] hypothesis is compelling, particularly because it is based on a large global data set. However, the Marcellus produced gas isotope data do not entirely support their conclusion that uplift and associated depressurization is more important than thermal maturity for generating isotopic reversals in shale plays. In Figure 7b, the gases produced from Marcellus Formation shale with VR_o = 2.37% exhibit partial reversals with δ¹³C₁ > δ¹³C₂ < δ¹³C₃. Uplift and erosion amounted to ~2.6 km at this well location (WVDG1 and WVDG2 in Upshur County, WV) [33,34]. All of the remaining dry gas samples exhibit full carbon isotopic reversals, with δ¹³C₁ > δ¹³C₂ > δ¹³C₃, with the δ¹³C of each gas component essentially increasing in parallel with increasing VR_o without any significant change in slope (Figure 7b). Furthermore, gas accumulations in Middle and Upper Devonian conventional sandstone reservoirs sourced by the Marcellus shale exhibit partial to full carbon isotopic reversals in several different parts of north central and western Pennsylvania, most of which experienced ≤2 km of uplift and erosion [7,62,109,110]. Petroleum generation and expulsion in Marcellus source rocks, and migration/accumulation in these conventional reservoirs occurred between ~359 and 225 Ma [33] suggesting that the isotopic compositions of the conventional reservoir gases are cumulative [66,69]. Combined oil and gas geochemical systematics of Marcellus-Upper Devonian sandstone petroleum system elements in Bradford County, Pennsylvania show carbon isotope reversals generated in-situ in a conventional Lock Haven Formation reservoir due to mixing of primary hydrocarbons and secondary gases from oil cracking [111]. Finally, comparisons of δ¹³C_1–3 variations observed during laboratory degassing of Marcellus pressure core with time-lapse production gas isotopic measurements (discussed below) contradict the hypothesis that molecular and isotopic fractionation during desorption after depressurization (from uplift and production fractionation) is the principal mechanism causing isotopic reversals in the Marcellus dry gas reservoirs.

Thiagarajan et al. [100] proposed that C₂₊ n-alkane gases are initially produced by irreversible cracking chemistry, but, as thermal maturity increases, the isotope distribution of these species approaches thermodynamic equilibrium, either at the conditions of gas formation or during reservoir storage. They argue that at higher maturities the conversion of C₂₊-rich wet gas to methane-dominated dry gas is controlled by thermodynamic conditions (temperature, fO₂, and fH₂). In this model, the spatial and temporal distributions, and chemical/isotopic compositions, of major gas components and their breakdown products are controlled by their relative thermodynamic stabilities. Highly mature gases (VR_o > 2%) such as the dry Marcellus fluids attain thermodynamic equilibrium. Turner et al. [112] concurred showing that (1) thermogenic gases may form in hydrogen isotopic equilibrium with co-occurring formation water and (2) that CH₄-CO₂ carbon isotopic equilibrium is a function of temperature in thermogenic shale-gas settings. Isotopic reversals observed in both unconventional and conventional reservoirs may be a function of quasi-equilibrium chemistry [100,112]. Modeled gas mixing in combination with Rayleigh fractionation and methane hydrogen exchange with formation water (Figure 20 and Figure 21) is useful for describing isotopic enrichment or depletion as material moves between reservoirs in an equilibrium process and is consistent with these recently proposed thermodynamic equilibrium models.

Although the actual mechanism (or mechanisms) responsible for isotopic reversals in natural gases are debatable and consensus remains elusive, plots of δ¹³C ratio distributions in reversed gases such as those shown in Figure 7b have pragmatic utility for identifying and mapping economic production limits in deep basins [72].

4.7. Stable Isotope Systematics of Marcellus Formation Shale-Gas during Pressure Core Degassing and Production

Unconventional shale-gas and shale-oil core delivered to the surface in pressurized sealed containers, which maintain formation pressure, facilitate direct gas-in-place measurements in the laboratory [113,114]. The cores are drilled under reservoir pressure and lifted to the surface without any loss of reservoir gas during transit. Controlled release of pressurized core samples in the laboratory permits accurate measurements of gas volumes extracted from natural high-pressure mudrocks. This laboratory procedure also provides an opportunity to collect a time-lapse series of reservoir gas chemical and isotopic analyses to document geochemical changes that occur during degassing of the core samples. Degassing is defined as expulsion of both gas phase and adsorbate phase gas molecules from the rock matrix [115]. These data reveal diagnostic geochemical trends that systematically vary with degassing time and cumulative hydrocarbon yield. Should such laboratory data anticipate gas compositional and isotopic variations during production, then they might provide useful predictive information about well performance over the operational life of the well [113].

The Marcellus pressure core samples listed in Table 3 provided a time-lapse data set of carbon isotopic measurements that document changes that occurred during degassing of the rock. Methane δ¹³C measured during direct gas-in-place analysis of the Marcellus pressure core shows changes that vary systematically with cumulative degassing volume (Figure 23a). Initially, the methane became depleted in ¹³C with increasing degassing time with a corresponding negative shift in ¹³C of 1.38‰ at ~34% cumulative degassing volume. This early fractionation was due to the drop in partial pressure and the diffusivity difference of the isotopolouges [115]. Adsorption/desorption within the reservoir pores was coupled with diffusion and delayed mass transport. ¹²C was preferentially desorbed. This trend was followed by enrichment in methane ¹³C beginning at approximately 40% of the cumulative degassing volume. Enrichment in ¹³C₁ continued with increasing degassing time with a positive isotope shift of 2.05‰ by 75.35% of the cumulative degassing volume. δ¹³C increased towards the carbon isotope composition of crushed gas, i.e., gas liberated from the shale by mechanical crushing at the end of the degassing experiment. Methane concentration of the crushed gas sample was 0.794 mole% and the final observed positive shift in ¹³C towards the crushed gas was 4.88‰.

Ethane and propane δ¹³C showed very different variations during direct gas-in-place measurements of Marcellus pressure core mostly due to increasing adsorption affinity from C₁ to C₃. The pattern of ethane δ¹³C variation was, at first, similar to that of methane (Figure 23a,b). The initial ethane δ¹³C of −31‰ approximated the median δ¹³C₂ of production gases in the region, but rapidly decreased to −33.5‰ by 53.53% of the cumulative degassing volume (Figure 23b). This decrease was followed by a small positive δ¹³C shift of 0.56‰ at 66.03% of the cumulative degassing volume before it resumed a decreasing δ¹³C trend with increasing cumulative degassing volume. However, the ethane δ¹³C of the crushed gas sample was −29.1‰, a positive shift of 4.14‰ from the final degassing sample measured at 71.81% cumulative degassing volume. Propane δ¹³C initially decreased by 0.3‰ between 22.93 and 25.47% of the cumulative degassing volume before exhibiting significant ¹³C enrichment between 28.45 and 37.52% of the cumulative degassing volume, with a positive shift of 2.8‰ (Figure 24a). Propane δ¹³C then decreased to −35‰ at 40.69% of the cumulative degassing volume followed by steady ¹³C enrichment until the end of the degassing experiment. The crushed gas δ¹³C of −34.7‰ is consistent with the average propane δ¹³C of 34.9‰ in the immediate field vicinity.

Diffusion and desorption are the principal processes affecting the carbon isotope ratios of the Marcellus pressure core samples during laboratory degassing measurements. The variable direction and magnitude of isotope fractionations in the shale-gas samples during degassing are a function of these secondary processes. The plots in Figure 23 and Figure 24 support the model predictions of Xia and Tang [115] who postulated that large isotopic fractionation due to gas transport would be observed under laboratory conditions with significant degassing.

The natural gas plot of the pressure core carbon isotope results reflects these secondary processes as well as those inherent in the Marcellus reservoir prior to drilling. The plot exhibits both partial and full carbon isotope reversals, with mostly non-linear trends (Figure 24b). The fully reversed natural gas plots of most of the pressure core degassing samples align with those of production gases in the study area (Figure 7b). However, there are three exceptions to this generalization. First, the ethane δ¹³C of the initially released expansion canister gas is distinctly enriched in ¹³C compared to all the other pressure core samples. There was insufficient propane in this sample for isotopic analysis. Second, propane in three of the degassing samples is enriched in ¹³C compared to the other samples and consequently exhibits partial carbon isotopic reversals with δ¹³C₁ > δ¹³C₂ < δ¹³C₃ (or approximately equal to δ¹³C₃; Figure 24a). Third, the δ¹³C of methane and ethane in the crushed gas sample is significantly enriched in ¹³C relative to the other degassing samples. This is not surprising because the ratio of adsorbed gas to free gas should be largest in the crushed gas samples resulting in larger observed fractionations [115]. The natural gas plot for the crushed gas is like that of the most thermally mature (VR_o ≥ 4.5%) and marginally productive Marcellus shale-gas wells in the basin (Figure 7b).

Unlike the pressure core degassing samples, limited time-lapse geochemical analyses of Marcellus Formation natural gases in the northern Appalachian basin do not reveal significant or systematic changes of stable carbon isotope ratios that vary with production (samples noted in bold dark blue in Table 1). In the dry gas-producing region, for example, methane δ¹³C of mud gas samples collected from the vertical and horizontal legs of the DG11/DG12 well on 2/21/2011 ranged between −25.0 and −22.8‰. Methane δ¹³C of production gas collected from the well on 10/19/2011 was −24.04‰ (DG11 in Table 1). Methane δ¹³C of production gas collected on 12/18/2014 was −23.90‰ (DG12 in Table 1). Ethane and propane carbon isotopic data exhibit similar invariance. The δ¹³C₂ of mud gases collected on 2/21/2011 ranged between −30.3 and −28.9‰ and the δ¹³C₃ was between −32.6 and −30.8‰. Ethane and propane δ¹³C of production gas collected from the well on 10/19/2011 were −29.9 and −32.5‰, respectively. On 12/18/2014, the production gas δ¹³C₂ was −29.86‰ and the δ¹³C₃ was −31.4‰. Similarly, in the wet gas-producing region, the respective methane δ¹³C values of gas samples WG2/WG13/WG14 (same well) collected in April and May 2019 and January 2020 were −45.5, −45.23 and −45.5‰ (Table 1). Ethane δ¹³C of gas collected on these dates was −35.28, −35.12 and −35.14‰. Respective propane δ¹³C of wet gas collected in April 2019 and in May 2019 was −30.34 and −30.34‰. Propane δ¹³C was −30.2 in January 2020. These observations are consistent with the model predictions of Xia and Tang [115]. Although δ value variations of up to 10‰ can occur during laboratory degassing experiments, depending on the cumulative degassing volume, production gas flowing from shale-gas reservoirs should be dominated by advective flux through stimulated fractures which has little fractionation effect [115].

4.8. Relevance to Environmental Issues in the Marcellus Play of the Appalachian Basin

In addition to energy production (heating and electricity generation), natural gas provides critical feedstock for a variety of necessities in modern economies such as basic chemicals, plastics, synthetic fibers, resins, dyes, foams, adhesives, explosives, synthetic rubber, paints and coatings, industrial chemicals, and numerous medical supplies. NGL directly contribute to the manufacture of solar panels and wind turbines. The mining of critical minerals needed to manufacture electric batteries and fuel cells is dependent on the energy supplied by fossil fuels. Natural gas extraction will continue as a fundamentally important societal resource well into the 21st Century. The Marcellus Formation in the Appalachian basin will continue to serve as one of the world’s most significant natural resources for some time to come [5]. However, the commercial success of the Marcellus unconventional play has led to public apprehension about the safety of shale-gas and shale-oil extraction. Concerns about the safety of petroleum development in the region are focused on stray gas migration to shallow groundwater aquifers and wells, buildings, and the atmosphere, as well as the potential for contamination from toxic materials in produced brine or hydraulic fracturing fluid during drilling, transport, and disposal [7,8,9,10,116,117,118,119,120]. Hydrocarbon migration into shallow groundwater aquifers in the study area occurs both naturally [117,118,119,121] and as a consequence of oil and gas well operations [116,120]. The geochemical data and interpretations provided in this report comprise a significant contribution to the available regional background dataset against which stray gas investigations may be conducted [7,118,119,121].

Methane is the third most important greenhouse gas (after water vapor and CO₂) and thus an important contributor to global climate change [120]. Fossil fuels represent an obvious and important potential source of increasing atmospheric methane concentrations due to growing global production. Milkov et al. [122] utilized a global dataset of δ¹³C₁ measurements from >1600 shale-gas wells to constrain the contribution of shale-gas emissions to the observed atmospheric increases in the global CH₄ load and concluded that increases in atmospheric CH₄ since 2008 are not dominated by emissions from shale-gas operations. Their conclusions are supported by the observation that increasing CH₄ concentrations are accompanied by ¹³C depletion. Milkov et al.’s [122] conclusions are consistent with studies by Schaefer et al. [123] and Schwietzke et al. [124] who argue that the recent increase in atmospheric CH₄ concentrations is a result of increasing microbial methane emissions with lesser to no significant increased contributions from fossil fuel emissions [125]. Howarth [126] challenged this interpretation and claimed that emissions from shale-gas, and probably shale-oil, production make up more than half of the total increase in fossil fuel emissions since 2008. He based his interpretation on the erroneous claim that the δ¹³C₁ of shale-gas is notably lighter than that of conventionally produced hydrocarbons. Lewan [125] thoroughly disputed and dismissed Howarth’s [126] argument by reviewing several shale-gas datasets, including Marcellus Formation mud gas data published by Baldassare et al. [7]. The production gas isotopic dataset provided in this report support and refine the arguments of Milkov et al. [122] and Lewan [125]. The dataset for this report should also help supplement and facilitate ongoing research related to atmospheric methane and ethane emissions in the study area [10,127,128].

5. Conclusions

Regional differences in Marcellus thermal maturity and hydrocarbon alteration effects define specific wet gas and dry gas production in the Appalachian basin. Thermal maturity and hydrocarbon alteration are best constrained by fluid chemistry. Marcellus hydrocarbons recovered in the wet gas region are mixtures of residual primary-associated gases generated in the late-oil window and post-mature secondary hydrocarbons generated from oil cracking in the wet gas window. Condensate production in the wet gas region is constrained by increasing maturity in conjunction with oil cracking and gasification. Most Marcellus liquids also have been stripped by gas and altered by hydrodynamic flow and water washing. Gas stripping results in enhanced expulsion and migration in some wells resulting in reduced liquid reserves related to depletion of more volatile hydrocarbon components [129]. Water washing also removed more water-soluble compounds to varying degrees within the sample set discussed in this report. Although water washing in low permeability mudrocks may seem unlikely to many readers, several investigators have published evidence that supports this interpretation. Repetski et al. [130] speculated that a westward bulging salient in Devonian conodont alteration indices mapped in north central West Virginia reflets the migration of hot, basin-derived brines. Dorobek [131] published fluid inclusion evidence for hot basinal fluid migration through Devonian rocks in the study area. Evans [132] and Evans et al. [133] published evidence for the influx of low-salinity, CH₄-saturated brines during the Alleghenian orogeny. Tamulonis and Carter [134] recently published unequivocal evidence for hydrothermal alteration in Devonian shales in the Rome Trough. And Atwah et al. [135] documented water washing in the analogous Woodford Shale of the Anadarko basin.

Marcellus hydrocarbons recovered in the dry gas-producing region are mixtures of residual methane cracked from overmature refractory organic matter and hydrocarbons generated by extensive oil and wet gas cracking. Production in the dry gas region is largely constrained by relative gas retention versus expulsion as controlled by regional and local structure, and by seal integrity. Noble gas systematics reveal gas retention versus migration trends in the produced fluids. Thermal maturity also influences productivity where declining well performance is correlated with increasing relative proportions of late-stage residual methane and decreasing amounts of gas generated by oil/wet gas cracking. This trend is accompanied by increased aromatization, graphitization, and loss of the interconnected organic porosity within a pyrobitumen network [30] (Figure 25).

The data provided by this report also facilitate the identification of sources of stray gas contamination and provide new baseline measurements for comparison with atmospheric greenhouse gas measurements conducted to monitor fugitive emissions of atmospheric methane.