WaterRF Project: Background

State of Knowledge

One of the important approaches being considered for mitigating greenhouse gases (GHGs) is sequestration of CO₂ in deep geologic formations. Injection of CO₂ into brine formations is being considered highly favorably because of the large storage capacity estimated to range globally from 350 to 1,000 Gt of CO₂ (Holloway, 2001; IPCC, 2005). The widespread occurrence of brine formations in the U.S. greatly increases the likelihood that such formations are located adjacent to major CO₂ sources (Figure 1). According to the 2005 U.S. Government Energy Information Administration (EIA), Texas is the eighth-largest emitter of CO₂ in the world, emitting 0.6 billion m3 of CO₂. The large emissions result from Texas being a leading producer of energy, as well as supporting large petrochemical industries. However, the Gulf Coast has the potential to serve as the leading carbon sequestration sink in the U.S., with 4.5 billion m3 of storage in Gulf Coast brine formations (Figure 2).

Geologic sequestration literature, such as the IPCC Special Report (2005) and the U.S. Department of Energy (DOE) Carbon Sequestration Road Map (DOE, 2007), divides sequestration sites into three main types: depleted hydrocarbon reservoirs, brine-bearing "saline" formations, and unminable coal seams. Unminable coal seams are clearly a separate type of storage, having a distinctive trapping mechanism and requiring their own types of capacity and risk assessment. However, hydrocarbon reservoirs are a subset of regional brine formations, representing the parts of the formation in which geometry and fluid migration coincided so that economically significant hydrocarbon was able to accumulate. Hydrocarbon reservoirs are formed by the same types of rocks and have the same types of seals as their host brine formations; CO₂ will move through and be trapped in essentially the same manner in both.

One of the critical issues with respect to CO₂ sequestration in brine formations is the potential impact of such storage on fresh groundwater in overlying aquifers. One way in which CO₂ sequestration in deep brine formations can impact aquifers is by perturbing the flow field in the deep aquifer, which could result in brines or brackish water being pushed updip into unconfined sections of the same formations or into the capture zone of fresh-water wells. This issue was examined in the Gulf Coast system by Nicot (2008) using groundwater modeling. Results of this analysis indicate that injection of water, equivalent to 50 million tons of CO₂/yr for 50 yr resulted in an average water-table rise of ~1 m and no significant increase in groundwater salinity (Nicot, 2008).

Geologic sequestration can also impact fresh-water aquifers by leakage of CO₂ occurring through the caprock or through preferential pathways, such as faults and wells, especially abandoned wells. Upward leakage of CO₂ from deep geologic sequestration reservoirs can impact groundwater quality by transporting deep brines and associated naturally occurring contaminants into the aquifer and/or by mobilizing contaminants within the aquifer through related groundwater pH reduction (Keating et al., 2009). These processes can alter many aspects of groundwater chemistry, resulting in exceedances of primary and/or secondary Environmental Protection Agency (EPA) Maximum Contaminant Levels (MCLs). Many of the studies that evaluate potential impacts of CO₂ leakage into aquifers are based on modeling analyses. Wang and Jaffe (2004) suggested that pH reduction from CO₂ leakage would result in large concentrations of Pb as a result of dissolution of galena. The negative impacts of CO₂ leakage are greatly reduced in aquifers with high alkalinity and pH buffering capacity such as those containing calcite. Wang and Jaffe (2004) also suggested that high trace-metal concentrations in an aquifer may be used as an indicator of CO₂ leakage above deep geologic sequestration sites. However, information on mineral dissolution kinetics, which strongly controls simulated trace-metal concentrations, is limited. Reactive transport modeling of CO₂ input into an aquifer for 100 yr by Zheng et al. (in press) indicates that CO₂ leakage could mobilize As and Pb through mineral dissolution. This study used 38,000 groundwater chemical analyses from aquifers throughout the U.S. and identified potential mineral hosts containing hazardous trace metals. However, As concentrations rarely exceeded the EPA MCL of 10 ug/L and Pb concentrations never exceeded the MCL. An increase in leakage rate leads to increases in partial pressure of CO₂ and, consequently, lower pH and higher maximum contaminant concentrations in the aquifer.

Apps et al. (2009) used the same data set and found that the most serious problem from CO₂ leakage resulted from enhanced pyrite dissolution and release of As. At the highest p(CO2) used in the simulations (1 bar), Ba, Pb, and Zn also approached or exceeded regulatory limits. The simulations indicated that Cd and Zn are unlikely to be exceeded and Hg, Se, and U concentrations are unaffected by CO₂ leakage. Modeling analyses by Carroll et al. (2009) indicate that CO₂ leakage can be detected by pH changes and carbonate chemistry in aquifers. The ability to detect CO₂ leakage from a storage reservoir into overlying dilute groundwater depends on CO₂ solubility, leakage flux, CO₂ buoyancy, and groundwater flow. However, Zheng et al. (in press) pointed out that all modeling analyses are subject to considerable uncertainty inherent in geochemical parameters and that quantitative predictions of CO₂ leakage impacts would require integration of modeling with detailed laboratory and field studies to test model assumptions and parameters. Most modeling studies attribute trace-element mobilization related to CO₂ input into aquifers to mineral solubilization; however, evaluation of trace-element distributions in natural aquifers suggests that the primary mechanism for releasing trace elements is desorption from Fe or Mn (oxy)hydroxides (Smedley and Kinniburgh, 2002; Smedley et al., 2005; Scanlon et al., 2009). It is not clear that the actual processes causing trace-element mobilization in aquifers are being evaluated in the modeling studies.

A variety of approaches can be used to assess impacts of CO₂ leakage from deep reservoirs on overlying groundwater quality in aquifers, including natural analogs, laboratory and field tests, and numerical modeling. Each approach has advantages and disadvantages. Aquifers adjacent to volcanic aquifers or that receive CO₂ from geothermal sources may provide suitable natural analogs for CO₂ leakage (Federico et al., 2004; Keating et al., 2009). One of the problems with a natural analog is that it may be difficult to distinguish upward flow of contaminants in brine that is moving with the CO₂ from contaminants mobilized within the aquifer as a result of CO₂ injection. Studies at a sandstone aquifer in New Mexico indicate that pH changes resulting from CO₂ influx were buffered by high alkalinity and carbonate mineral dissolution, whereas increased As, Pb, and U were attributed to leakage of brine with the CO₂ based on high correlations between these elements and Cl (Keating et al., 2009). However, in other areas it may be difficult to distinguish impacts of upward brine migration from in situ mobilization of contaminants in the aquifer.

Batch experiments have been conducted in the laboratory to evaluate potential impacts of CO₂ on water chemistry. Lu et al. (in press) examined impacts of CO₂ on rock samples from major aquifers in Texas, including the Ogallala and Gulf Coast aquifers. Both carbonate and silicate systems were examined. Results indicate that all major and trace elements increase as a result of CO₂ injection; however, the increase represents a transient pulse. In carbonate systems pH increases are buffered by carbonate mineral dissolution, whereas in silicate systems, pH decreases sharply because it is not buffered. Silica increases in both carbonate and silicate systems. In carbonate systems, Ca and Mg concentrations level off after a time. Mineral dissolution rates are much slower in silicate systems.

To date, specific field experiments have not been conducted to evaluate impacts of CO₂ leakage on groundwater quality. Field tests have been conducted to assess impacts of deep geologic sequestration. Single-well push-pull tests were conducted in deep basalts in New York and showed that carbonic acid from CO₂ was neutralized by carbonate dissolution and cation exchange (Assayag et al., 2009). ​