Composition and fate of gas and oil released to the water column during the Deepwater Horizon oil spill

Results and Discussion

Owing to its collection from within the LMRP, sample MW-1 contained predominantly petroleum hydrocarbons (gas and oil) and a minor amount (approximately 5% vol/vol) of an aqueous fluid of seawater composition. In contrast, MW-2 contained approximately 23% seawater, presumably due to turbulent mixing of ascending oil, gas, and water at the collection site a few meters above the LMRP. The observation that both samples contained only seawater indicates that significant amounts of saline formation waters were not being released from the well along with gas and oil (14). Chemical analysis of both samples yielded comparable results and revealed that the total C₁-C₅ hydrocarbon components (Table 1) are composed predominately of methane, comprising approximately 80 mol %. The stable carbon and hydrogen isotope ratios of the gases exhibit increasing ¹³C and ²H content with increasing carbon number, indicating a thermogenic origin (16), consistent with production from the deep petroleum reservoir tapped by the Macondo well.

The contents of sample MW-1 enabled determination of the gas-to-oil ratio (GOR, defined as the standard cubic feet per petroleum barrel at 15.6 °C and 1 bar) for fluids flowing from the Macondo well LMRP. Although chemical and isotopic analyses of MW-1 and MW-2 yield nearly identical compositions for the gas and oil fractions, measured GORs differed between the samples. MW-1 yielded a GOR of 1,600, whereas MW-2 yielded a value of 2,470 (Table 1). We suspect that the GOR determined for MW-2 was modified by phase segregation during the few meters of ascent to the point of collection and may have been further biased by a modified ROV thruster that directed a high velocity jet of water at the Top Hat #4 vent in order to clear well fluids from the field of view during sample collection. Accordingly, MW-1 likely represents a more accurate representation of the fluid exiting the well. The MW-1 value of 1,600 is similar to the GOR predicted from thermal maturity indicators in the oil that indicate a value of 1,730 (Table 1) (15). Ambient hydrostatic pressure and temperature (150 bar, 5 °C) at the Macondo well leak site suggest that ethane and higher hydrocarbons would be found predominantly in liquid oil phase, whereas methane is chiefly in the gas phase (17).

Previously reported GOR values for fluids released from the well are limited to estimates from oil-recovery vessels after hydrocarbons captured at the seafloor were pumped to the sea surface. Fluids collected from the choke line of the blowout preventer between June 17 and 23, 2010, on the Q4000 recovery vessel were characterized by GOR values that ranged from 1,760 to 1,965 (mean = 1,810 ± 70) (18), close to the value determined in this study on June 21. On June 24, 2010, however, the GOR for fluids recovered on the Q4000 underwent an abrupt increase to values near 2,400 and remained at this level until July 16, 2010 (Fig. 2) (18).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Plots of GOR as a function of (A) date and (B) daily oil collected by the recovery vessels Discoverer Enterprise and Q4000 (19).

Analysis of Q4000’s daily GOR values from the two periods indicates that their means and distributions are statistically different with a greater than 99% confidence level. In contrast, hydrocarbons simultaneously captured on Discoverer Enterprise recovery vessel using the Top Hat #4 across both periods do not exhibit this abrupt GOR change, indicating that the apparent GOR variability recorded by these surface vessels is attributable to the collection process itself, not variability in the endmember GOR. Indeed, daily GOR values for fluids recovered from the Top Hat #4 reveal an inverse correlation with increasing oil collection rate (Fig. 2). Extrapolation of these trends to higher oil collection rates would yield decreased GOR values that are consistent with the MW-1 endmember GOR that was also collected from within Top Hat #4. Similar trends exist for fluids collected before and after the step change on the Q4000.

Using the federally estimated net liquid oil release of 4.1 million barrels to the Gulf of Mexico (19), the net total C₁-C₅ hydrocarbons released to the water column was 1.7 × 10¹¹ g. Comparison of our calculations to other studies (Table S1) reveals differences that are primarily due to GOR values employed. For example, Valentine et al. (3) used a GOR of 3,000 that yields total methane, ethane, and propane releases that are higher than our values by a factor of two. By comparison, Joye et al. (20) found fluxes nearly four times greater than ours for their high-end estimate.

The composition of the MW-1 oil fraction contained 74% saturated hydrocarbons, 16% aromatic hydrocarbons, and 10% polar hydrocarbons (Table 1). Along with other results, these data are consistent with a moderately mature, light sweet crude oil, with no evidence of subseafloor biodegradation (Figs. S2 and S3, Table 1, and Table S2). The polar fraction (10%) is comprised of molecules containing oxygen, nitrogen, and sulfur spanning a wide molecular weight range. Many of these compounds are resistant to evaporation, biodegradation, and photolysis. Hence, they have the potential to remain in the environment long after other oil components have been removed or degraded (21, 22). Because many of these polar compounds are not typically analyzed in field samples, the resulting 0.41 million barrels of polar hydrocarbons released into the Gulf of Mexico may be overlooked in studies examining the fate of oil released during this spill.

In addition to bulk analyses, the gas and oil fractions were analyzed for select compounds (Table S2). We have used our results from the analysis and relative amounts of gas and oil fractions to calculate the total amount released and the composition of the “reconstituted” reservoir fluid on a mass/mass basis. This exercise indicates that the most abundant compound released from the Macondo well on a mass basis was methane at 0.15 g g^-1 of reconstituted fluid. The total C₁ to C₅ hydrocarbons were 0.24 g g^-1, and the other 140 hydrocarbon compounds were 0.24 g g^-1 of the total mass of reconstituted fluid. This indicates that traditional molecular-level methods to characterize oil can account for only half of the material that flowed from the Macondo well.

To constrain the fate of hydrocarbon components released to the deep sea, several convolved processes need to be considered. Petroleum emitted from the LMRP rapidly partitions into four phases in the deep water column: a gas phase, a liquid oil phase, an aqueous phase, and a hydrate phase. Once in the water column, these phases may physically segregate during buoyancy-driven ascent of light liquid oil droplets and gas bubbles toward the sea surface and descent of heavier liquid oil droplets, enriched in dense components such as long-chain n-alkanes (23), toward the seafloor. Moreover, in near-field environments where separate oil, gas, and aqueous phases may coexist, partitioning of components between phases will evolve continuously due to depth-dependent changes in pressure and temperature. These processes were likely influenced by the seafloor injection of dispersants that may enhance aqueous dissolution and the stabilization of oil droplets. Current-driven advective transport of hydrocarbon-rich plumes provided an opportunity for continued aqueous dissolution and microbial degradation of bioavailable components.

Quantitative compositional data for gas and oil exiting the Macondo well provides an opportunity to examine chemical, physical, and biological processes affecting their abundance during transport through the water column. Here, we assess the role of aqueous dissolution as a possible driver for the previously reported formation of deep water hydrocarbon-enriched plumes of neutrally buoyant water at 1,100-m depth. The abundance of low molecular weight n-alkanes and aromatic compounds observed in the hydrocarbon-rich plume at 1,100 m are far below their aqueous saturation values at ambient conditions, suggesting that gas and oil droplets did not reach complete equilibrium with the aqueous phase, and partitioning into the deep water column is a kinetically controlled process. However, the rates at which hydrocarbon compounds dissolve from gas bubbles and oil droplets are likely influenced by their aqueous solubilities (11). To investigate the relationship between aqueous solubility and abundance in the 1,100-m plume, we evaluated the fractionation of hydrocarbon components into the deep water column relative to two highly water-soluble oil components: methane and benzene.

Abundant evidence suggests that methane released at the seafloor was quantitatively trapped at 1,100-m depth. Aqueous methane was the most abundant hydrocarbon component in deep water plumes, reaching values as high as 183 μmol kg^-1 in a mid-June 2010 survey within an approximately 10-km radius of the well (3). Methane was nearly absent at more shallow depths, suggesting that methane bubbles have dissolved completely upon reaching 1,100-m depth (1, 3). This is further confirmed by observed methane concentrations at natural background levels in the atmosphere overlying the sea-surface oil slick (24) and measured sea-to-air fluxes indicating that approximately 0.01% of the methane released from the reservoir was released into the atmosphere (25). The observed near-complete dissolution of methane in deep waters is consistent with previous field and modeling studies in systems of similar water column depth (26, 27).

Assuming that all of the released methane resides in deep water plumes, quantities of ethane and propane retained can be estimated by examining their abundance relative to that of methane. For this purpose, we define the fractionation index for a given compound as [1]where F_i,methane is the fractionation index for species i, C_{i,water column} is the observed molal concentration of component i in the water column, C_{methane,water column} is the molal concentration of methane in the water column, C_i, MW-1 is the molal concentration of component i in the original oil (MW-1 sample), and C_{methane, MW-1} is the molal concentration of methane in the MW-1 sample. The resulting index is an indicator of the chemical fractionation of component i in the deep water column relative to methane. A value of F_i,methane equal to unity corresponds to material that is channeled completely into deep water plumes to the same extent as methane. An F_i,methane value of zero corresponds to material that is completely retained within the ascending oil and likely reaches the sea surface.

The compositional evolution of methane, ethane, and propane at 1,100 m suggests that phase partitioning processes influenced these low molecular weight hydrocarbons differently. Valentine et al. (3) used the spatial variations in the isotopic and chemical compositions of C₁-C₃ hydrocarbons in plume water samples to demonstrate the preferential microbial degradation of propane in the plume. The isotopic compositions for C₁-C₃ hydrocarbons in our sample MW-1 collected at the LMRP is nearly identical to the near-field plume values of Valentine et al. (3) (Table 1), consistent with an absence of biodegradation in the region of the plume nearest the Macondo well. However, the relative distributions of methane, ethane, and propane for these nonbiodegraded plume samples show small but noteworthy differences relative to the MW-1 sample. For example, the molal methane/ethane ratio of 10.85 consistently observed in nonbiodegraded plume samples (3) is slightly higher than the value of 9.9 observed in the pristine petroleum. These concentration ratios produce an F_{ethane,methane} value of 0.91, indicating that most of the ethane emitted from the Macondo well was retained in the deep sea, but to a slightly lesser extent than methane. Similarly, the F_{propane,methane} value of 0.78 for nonbiodegraded samples indicates that most of the propane released from the Macondo well resided in the subsurface, but to a lesser extent than ethane and methane. These results imply that a sizeable portion of ethane (9%) and propane (22%) was retained in the buoyant liquid oil phase that continued ascent to the sea surface. The rapid dissolution of methane, ethane, and propane is consistent with their high aqueous solubilities (Table S3) at ambient conditions in the deep water column (100–150 bar and 4–6 °C). Because the aqueous solubility of short-chain n-alkanes decreases with increasing chain length, the observed trend suggests that aqueous dissolution governed the fractionation of these gases into the deep water column.

Joye et al. (20) measured C₁ to C₅ gases at 1,100 m depth from 2 to 11 km southwest and west of the Macondo well from May 25 to June 6, 2010, and did not observe the inverse correlation of methane/ethane and methane/propane ratios with total hydrocarbon abundance documented by Valentine et al. (3). They observed methane/ethane and methane/propane ratios of 9.9 and 14.7, respectively, values that are remarkably similar to the MW-1 sample (Table 1). An explanation for the differences in the data of Valentine et al. (3) and Joye et al. (20) remains elusive but may be related to numerous activities at the well in late May and early June that likely affected the flow and partitioning of released gases. For example, the riser insertion tube was removed on May 25, the “top kill” began on May 26 and ended on May 29, the initial shearing of the riser pipe occurred on June 1 and was followed by a second shearing on June 2, and installation of the Top Hat #4 collection device occurred on June 3 (28). It should be noted that all of the Valentine et al. (3) samples were collected after these events.

Looking beyond hydrocarbon gases, we investigated the fractionation of higher molecular weight petroleum components into the deep water column by comparing their abundances to benzene. Benzene is highly water-soluble, suggesting that it too may have dissolved rapidly in the deep sea. Indeed, benzene concentrations were systematically elevated (0.4–21.7 μg L^-1) in the 1,100-m-depth plume and nearly absent at depths shallower than 1,000 m (1) (Fig. 3, and Table S4). A comprehensive survey of hydrocarbons in air above the Macondo well site revealed that very little benzene reached the sea surface (24). Taken together, water column and surface measurements suggest that benzene was predominantly retained in the deep water column. Replacing C_{methane,water column} and C_{methane, MW-1} in Eq. 1 with the concentrations of benzene in the water (C_{benzene,water column}) and original oil (C_{benzene, MW-1}), respectively, we calculated benzene-normalized fractionation indices (F_i,benzene) for less water-soluble hydrocarbons.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Water column profiles of BTEX collected at (A) 2.3, (B) 6.1, (C) 16.5, and (D) 27 km from the Macondo well. These stations were within the region of the deep water plume identified by Camilli et al. (1) in June 2010 (See Table S4 for actual BTEX values in these water samples). BTEX composed 2.2% of the total oil in MW-1 (Table S2).

Using the benzene-normalized fractionation index, F_i,benzene, we tested the hypothesis that aqueous partitioning controlled the preferential retention of non-gas petroleum components. We examined the relationship between observed F_i,benzene values and compound aqueous solubilities () for a set of 33 detected oil components, including monoaromatic alkylated hydrocarbons, naphthalenes, dibenzothiophenes, and several other alkylated and unsubstituted PAHs (Table S3). (Assuming ideal Raoultian behavior, the aqueous solubility of a given compound in the MW-1 oil is equal to the aqueous solubility of the pure liquid multiplied by the mole fraction in the oil. For compounds that are solids at ambient conditions, aqueous solubility refers to the subcooled liquid aqueous solubility; see SI Text.) Across this compound set, we found that the fractionation index decreased systematically with decreasing aqueous solubility (Table S3). This trend is readily apparent in the representative stations shown in Fig. 4. Similar trends were observed at 20 sampling locations (positions of unique latitude, longitude, and depth) where hydrocarbons were detectable, at distances ranging from 1.7 to 34.6 km from the Macondo well and at water depths ranging from 1,065 to 1,221 m (Fig. S4). With a single exception, n-alkanes were not found at detectable concentrations at sampling locations where benzene was present. Uniquely, the 1,201-m depth sample at station 19 contained significant levels of n-alkanes and other sparingly soluble compounds but was depleted in water-soluble compounds such as BTEX (Fig. S5). This exceptional sample is consistent with the capture of liquid oil droplets that had been partially dissolved into the ambient water, thereby exhibiting enrichment in sparingly soluble components relative to water-soluble components.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Fractionation index (F_i,benzene) as a function of aqueous solubility for 33 different hydrocarbon compounds observed in the southwesterly hydrocarbon plume observed by Camilli et al. (1) at four different stations ranging from 2.3 to 27 km distance from the Macondo well (same stations as shown in Fig. 3). A fractionation index value of 1 corresponds to material that has completely departed the ascending oil, relative to benzene, and contributes to the plume. A low fractionation index corresponds to material that is predominantly retained within the ascending oil. For compounds that are solids at ambient conditions, aqueous solubility refers to the subcooled liquid aqueous solubility; see SI Text for a more complete discussion of this term.

Collectively, the observed relative abundances of C₁-C₃ hydrocarbons and longer-chain petroleum constituents provide compelling evidence that the 1,100-m plume identified by Camilli et al. (1) is comprised of dissolved hydrocarbons. In turn, the results provide strong evidence that advective transport in solution is the principal means by which hydrocarbons are transferred from the well to the deep sea. Evidence for the presence of oil droplets was only observed in one sample. Selective partitioning of water-soluble petroleum components into the deep sea is also consistent with the results of Ryerson et al. (24), who documented a positive correlation between the aqueous solubilities of several hydrocarbon components and their extents of depletion in atmospheric samples collected above the oil spill. These authors suggested that the depleted compounds were likely retained within the water column. The rapid dissolution of highly soluble oil components in the deep water is also consistent with the reported near-quantitative sequestration of the highly water-soluble dispersant component dioctyl sodium sulfosuccinate at 1,100-m depth (29).

The ultimate fate of these water-soluble petroleum hydrocarbons in the deep water plume is unknown, although biodegradation may have been an important process. However, the absolute and relative abundances of BTEX compounds show no systematic spatial trends (Figs. 3 and 4), suggesting that negligible dilution or biodegradation occurred during the 4 d required for their transport in the plume over a distance of 27 km (1). This is consistent with previously reported estimates of respiration rates that indicate hydrocarbon degradation in the plume was unlikely to be in excess of 7 μg L^-1 d^-1, including degradation of hydrocarbon gases and petroleum (1). Because Valentine et al. (3) suggested that the microbial respiration of hydrocarbon gases outpaced petroleum respiration by at least a factor of two, the degradation rate for petroleum hydrocarbons may have been on the order of 2 μg L^-1 d^-1. This degradation rate, in conjunction with the hydrocarbon concentrations reported here, yields a half-life on the order of 1 mo for petroleum hydrocarbons. By contrast, Hazen et al. (2) reported half-lives for n-alkanes on the order of a few days. The absence of observable gradients in the concentrations of BTEX compounds within this plume along with the preceding half-life calculation indicates that water-soluble petroleum hydrocarbons persisted longer than the gas and n-alkane fractions.