Article pubs.acs.org/est
Laboratory Investigations of Weathering of Soils from Mammoth Mountain, CA, a Naturally CO2‑Impacted Field Site Helen Sanchez,† Gustavo Menezes,† Andre Ellis,‡ Claudia Espinosa-Villegas,§ and Crist Khachikian*,†,∥ †
Department of Civil Engineering, ‡Department of Geosciences and Environment, and §Department of Technology, California State University, Los Angeles, 5151 State University Drive, Los Angeles, California 90032, United States ABSTRACT: The potential impacts of CO2 leakage from a natural subsurface reservoir on soil and water quality were studied. Field measurements of soil pore CO2 concentrations and visual inspection of plants at Mammoth Mountain, CA, allowed the demarcation of tree-kill and non-tree-kill zones, with CO2 concentrations >100,000 ppm and ∼1,000 ppm, respectively. Soils collected from six sites along a transect stretching from the center of the tree-kill zone to an equidistant point into the non-tree-kill zone were analyzed for surface area and organic carbon content. Batch and column leaching tests were conducted to determine the extent of weathering induced by the presence of CO2 in the aqueous solution. Soils deep into the tree-kill area exhibited significantly higher surface areas (10.67 m2/g vs 2.53 m2/g) and lower organic carbon content (9,550 mg/kg vs 35,550 mg/kg). Batch results indicated that lower pH values (∼2) released higher concentrations of Mg, Si, Fe, and As, while, for soils in the tree-kill zone, longer-term batch results indicated higher releases at the higher pH of 5.5. Column experiments were used to compare the effects of pH adjusted using HCl vs CO2. For pore volumes (PV) < 100, CO2 enhanced trace element release. For 100 < PV < 10,000 concentrations of elements in the two systems were equivalent and steady. At PV > 10,000, after a drop in pH in the CO2 system, larger amounts of Fe and As were released, suggesting a CO2-induced dissolution of Fe-silicates/clays and/or reductive dissolution of Fe3+ that releases Fe-bound arsenic. The specific role of pore water-dissolved CO2 on the release of trace elements is hitherto unknown. However, interactions of pore-water CO2 and the minerals in the Mammoth Mountain soils can cause the release of environmental pollutants.
■
INTRODUCTION Climate change, caused primarily by anthropogenic CO2(g), is one of today’s most significant global environmental challenges.1 Technologies that keep the current energy utilization framework while immediately mitigating the effects of CO2(g) emissions are being actively pursued. One promising technology is Carbon Capture and Storage (CCS), which captures and compresses CO2(g) from point sources for storage in deep geological formations.2−4 Despite the promise of CCS, the feasibility of its large-scale implementation and its potential long-term environmental risks are debated. A major question concerning the integrity of carbon storage sites is whether CO2(g) could potentially leak and have adverse environmental effects. It is imperative to determine the risks involved, because CO2(g) leakage can alter the biogeochemistry of soil, groundwater, and surface waters.5 Recent studies have shown that a pH drop of 2 units are possible upon CO2(g) injection, with a corresponding increase in concentration of dissolved ions.6,7 These studies also showed differences in the behavior of the ion concentrations, attributed to variable mobility of different aqueous species, carbonate concentrations, and redox conditions. Few field studies have examined the effects of elevated CO2(g) on soil chemistry and mineralogy, including specific © 2014 American Chemical Society
surface area, solution pH, TOC, cation exchange capacity, composition of soil gas, etc.8−13 Stephens and Hering8,9 studied soils obtained from Mammoth Mountain, where CO2(g) from magmatic sources leaked to the surface pursuant to earthquakes in 1989.14 In their study, soils exposed to higher levels of CO2(g) had lower pH values (5.0 vs 5.6) and higher soil moisture content and surface areas than soils from control areas. Organic material in the high CO2(g) soil was less than in the control soil. Several others have studied natural analogue systems, observing changes in pH, trace elemental concentration, or saline transport. For example, trace elements have been mobilized upon injection of CO215 with further analysis16 indicating an increase in As, attributed to desorption from Ferich phases. In other studies,17,18 pH dropped from 7 to 5.6 and trace elemental concentrations increased, leading Zheng18 to conclude that 1) trace element concentration increase was driven by Ca2+ exchange chemistry; 2) anion release was due to competition for sites from bicarbonate; and 3) Fe mineral Received: Revised: Accepted: Published: 12056
July 18, 2014 September 15, 2014 September 16, 2014 September 16, 2014 dx.doi.org/10.1021/es501829z | Environ. Sci. Technol. 2014, 48, 12056−12062
Environmental Science & Technology
Article
dissolution caused the increase in Fe concentrations. In a recent study with a controlled release of groundwater with dissolved CO2(g), pH values decreased by ∼3 units with an initially fast release of trace elements, decreasing to background levels.19 Keating et al.20 concluded that the upward flow of CO2(g)infused water transported the trace elements and that changes in shallow water quality is due to CO2. The current study focuses on field measurements at Mammoth Mountain as well as laboratory studies on soils from that site. Magmatic CO2(g) is continuously released to the surface, rendering this site a natural analogue of a potential leak at CCS sites; at this site, pore-gas CO2(g) concentrations > 100,000 ppm are common. Large-scale ecosystem degradation is visible, with large zones of high tree mortality. In order to understand how CO2(g) escaping from deep geological reservoirs impacts weathering of soil at the surface, we present results of 1) physicochemical analyses of soils from the area and 2) laboratory batch and column experiments conducted to measure the effects of elevated CO2(g) on the mobilization of select trace elements from these soils.
Figure 1. Colinear sample locations (A through F) and distances are shown here. The “transition zone” was approximately 60 m wide, delineated by an initial sampling phase where soil CO2(g) was measured (data shown in plot with open circles representing data from transition zone) and by visual observations of tree health in the field. Site F was located approximately at the center of the tree-kill- zone; Site A was chosen to be an equal distance away in the opposite direction of the transition zone from Site F. Sites C and D were chosen to be near the edge on either side of the transition zone. Site B was 100 m and site A was 200 m away from site C and into the tree-kill zone. Site E was 100 m and site F was 200 m away from site D and into the non-tree-kill zone.
■
MATERIALS AND METHODS Study Site. The site is located at Mammoth Mountain, California, at the southwest edge of the Long Valley Caldera, having been formed by a series of eruptions between 110 and 40 kiloannum.21 Starting in 1989, a series of small earthquakes caused fissures that connected the surface to an underground CO2(g) reservoir formed from degassing magma.14 One such fissure is at Horseshoe Lake (the study site), south of Mammoth Mountain. It is where elevated releases of CO2(g) as high as 1,200 tons/day can be measured in zones visibly affected with high tree mortality (herein referred to as the “treekill” zones) that cover approximately 360,000 m2.21,22 Field CO2 Measurements and Sampling. Trees in the high CO2(g) region overlying the fault line near Horseshoe Lake are visibly white, ashy, and lack any signs of healthy growth, caused by CO2(g) degassing.23 In order to delineate the tree-kill and the non-tree-kill zones, visual observations of plant health and soil-gas CO2(g) measurements were used. Initially, boundaries were delineated by visual observation to encompass the extreme tree-kill zone using a hand-held GPS unit. Rogie et al.23 found higher fluxes of CO2(g) within the center of the treekill zone (8,000 g/m2-day) than at the edges (125 g/m2-day). A 400-m southwest transect was sampled to include both the visible tree-kill and non-tree-kill zones (Figure 1). Soil-gas CO2(g) readings (Vernier CO2(g) probe, saturation point of 100,000 ppm) helped define the transition zone. A chamber was inverted at the soil surface and at a depth of 0.3 m, and CO2(g) concentrations were recorded for a period of 5 min as the soil gas filled the chamber. From initial CO2(g) readings (ranging from 870 ppm in the non-tree-kill zone to 2,665 ppm in tree-kill zone) and field observations of tree health, a transition zone of 60 m was chosen. Six colinear sampling sites were chosen, three of which were located in the non-tree-kill zone (sites A, B, and C), and three were in the tree-kill zone (sites D, E, and F). CO2(g) concentrations were measured at the surface soils of each site (after manually removing ∼0.5 cm of topsoil, herein designated as the “surface” measurements) and at a depth of 0.3 m. Soil samples (1 kg at 0.3 m depth) at each site were collected in triplicate (i.e., three distinct sampling locations, 1 m apart). Caution should be exercised at inferring mechanisms or causality from data from sites C and D (nearest the transition zone). Based on Rogie et al.23 and data presented
here, it appears that the center of the tree-kill zone has remained unchanged, while the transition zone is transient, resulting in unsteady environmental conditions. Physicochemical and Environmental Characteristics of the Sediment. Particle Size Distribution (PSD), Surface Area, Carbon Content. PSDs were obtained by sieve analysis using sieve #’s 4 (dp = 4.75 mm) to pan. Triplicate samples from sites A, B, C, D, E, and F were sieved, from which triplicate subsamples were subjected to surface area analysis, using a Micromeritics TriStar Surface Area Analyzer (Norcross, GA) and procedures outlined previously.24 At sites A and F, surface areas were obtained for all sieve fractions. At all other sites, surface areas were obtained for the +80−100 (0.149 mm < dp < 0.177 mm) and +100−200 (0.074 mm < dp < 0.149 mm) soils. Also, total organic carbon content was obtained for all +80−100 and +100−200 mesh soils using a PerkinElmer Series II 2400 CHNS/O Analyzer (PerkinElmer, Waltham, MA). Batch Experiments. Polypropylene tubes were filled with 5 g of #4 sieve (20 tubes) and pan (20 tubes) soils from site A and 40 similar tubes were filled with site F soils. Each tube was filled with artificial groundwater solution simulating Mammoth Mountain groundwater (1,191 μg/L MgSO4, 382 μg/L NaCl, 8,096 μg/L Ca(HCO3)2, and 1,256 μg/L KHCO3), bringing the total volume to 15 mL. pH values of 2 and 5.5 were chosen to elucidate the role of pH in the release of ions. These pH values were chosen to mimic field conditions (i.e., soil porewater pH of 5.5) and to create a condition that would yield results with significant differences. The soil particles in the reactors were mixed constantly for 14 days. Every 24 h, pH levels were adjusted using 10% HCl. Samples were collected with 10-mL syringes and were filtered (Fisherbrand, 25 mm Syringe 0.45-μm nylon filter) and preserved with 2% HNO3. Samples were analyzed for Al, Fe, Mg, Si, and As 12057
dx.doi.org/10.1021/es501829z | Environ. Sci. Technol. 2014, 48, 12056−12062
Environmental Science & Technology
Article
concentrations using ICP-MS by an external laboratory (Activation Laboratories, Canada). Column Leaching Experiments. Two reservoirs were filled with artificial groundwater. The first was continuously infused with CO2(g) using a stone nozzle. The second was maintained at the same pH as the CO2-infused reservoir by adding appropriate amounts of 10% HCl. Both reservoirs were loosely capped and maintained as open systems. The pH of each reservoir was monitored continuously; pH fluctuated between 4.0 and 4.2 during the experiments. Columns (Spectrum Chromatography Model EW-34697-00, ColeParmer) with dimensions of 0.9 cm inner diameter, 15 cm in length, cross-sectional area of 6.36 cm2. and a total volume of 9.4 mL were used in all studies. Six columns were filled with non-tree-kill soils retrieved from site A to obtain data on the effects of CO2-rich pore-water on soils not previously exposed to such pore-water. The columns were packed with +80−100 and +100−200 mesh soils because of their relatively high reactivity, lower potential for clogging end filters, and the creation of ideal flow conditions (e.g., appropriate flow rate for sampling and mass transfer considerations). Wet packing was performed to ensure saturation and to reduce short-circuiting in the column. Soil porosity measurements were used to determine the pore volume of the columns (∼3.14 mL). The average dp of 0.14 mm was used to determine the optimal flow rate of 1.4 mL/min (i.e., laminar flow; Reynolds number ∼0.05), while ensuring mass-transfer limitation at the particlefluid interface (i.e., Peclet number ∼50). The pump (Master Flex model no.7523-80) fitted with L/S 13 Tygon tubing was used to supply the solutions to the columns from the bottom to maintain saturated flow. Effluent samples (5 mL) were collected at regular intervals and preserved with 2% HNO3 to be analyzed for Al, Si, Mg, Fe, and As concentrations with an ICP-MS using EPA Method 200.8.
Figure 2. Concentrations of CO2 measured at a depth of 0.3 m at each sampling site (identified in the legend). Sites C and D are near the transition zone and share the square symbol. Sites A and F are farthest from the transition and well within the non-tree-kill (site A) and treekill (site F) zones. Concentrations exceeding 100,000 ppm are outside the useable range of the detector and are, therefore, not included in the figure.
detector saturation limit. CO2(g) measurements for sites E, D, C, B, and A reached approximately 99,845 ppm, 4,939 ppm, 9,815 ppm, 1,921 ppm, and 1,431 ppm, respectively. Thus, CO2(g) seepage increases from site A, the healthiest site, toward site F. This is corroborated by visual observations of increased tree and plant mortality at site F. Particle Size Distribution of Non-tree-kill and Tree-Kill Zone Soils. Generally, the non-tree-kill soils (A and B) are well-graded (with coefficient of uniformity,27 Cu = 16.11 and coefficient of curvature,27 Cc = 1.26 for Site A and Cu = 10.43 and Cc = 1.56 for Site B) and the tree-kill soils (E with Cu = 5.37 and Cc = 0.47 and F with Cu = 5.75 and Cc = 0.51) are poorly graded (Figure 3). The transition zone soils are well-
■
RESULTS AND DISCUSSION Mineralogy. Mammoth soil is composed of 50−60% volcanic glass and less than 10% for each of the following: plagioclase, K-feldspar, quartz, hornblende, magnetite, and biotite.8,26 For our study, we chose to focus on the elements Mg, Fe, Al, and Si based on results from previous work from our group26 (see Table 1). Table 1. Composition of Minerals Found in Mammoth Mountain (Stephens, 2002) mineral
composition
silica plagioclase feldspar K-feldspar quartz hornblende biotite
SiO2 ((Na, Ca) AlSi3O8) (AlSi3O8) (SiO2) crystalline ((Na, Ca)2(Mg, Fe, Al)5(Si,Al)8O22(OH)2) (K(Mg, Fe)3Si3(Al,Fe)O10(OH,F)2)
Field CO2(g) Measurements. Surface CO2(g) measurements for sites A, B, and C were 1,111 ppm, 491 ppm, and 1,218 ppm, respectively and for sites D, E, and F were 1,620 ppm, 1,143 ppm, and 2,665 ppm, respectively. Measurements obtained at a depth of 0.3 m were higher than the surface measurements (Figure 2). Also, CO2(g) measurements increase in the direction of the tree-kill zone. Specifically, CO2(g) measurements for site F, approximately at the center of the tree-kill zone,23 are highest, reaching a value of 100,000+ ppm, exceeding the
Figure 3. Particle size distributions for soils collected at a depth of 0.3 m at each sampling site (identified in the legend). Soils from sites A, B, and D are well-graded, and those from C, E and F are poorly graded. Sites A and F are farthest from the transition and well within the nontree-kill (site A) and tree-kill (site F) zones. 12058
dx.doi.org/10.1021/es501829z | Environ. Sci. Technol. 2014, 48, 12056−12062
Environmental Science & Technology
Article
graded (at site D with Cu = 11.83 and Cc = 1.01) and poorly graded (at site C with Cu = 8.66 and Cc = 0.41). For the nontree-kill soil, the mode of the mass retained (26%) is found in the +4−10 mesh size (2 mm < dp < 4.76 mm), and, for the treekill soil, the mode of the mass retained (21%) is found in the +20−40 mesh size (0.42 mm < dp < 0.841 mm). This is reasonable since we hypothesize that the CO2-laden soils will have experienced greater weathering or less organic agglomeration, which may have caused the break up of soil particles to smaller sizes. Surface Area of Non-tree-kill and Tree-Kill Zone Soils. The data here represent triplicate runs of soils at a depth of 0.3 m at each location (3) at each site (i.e., at total of 9 samples per site). Surface soils were not analyzed because it is often difficult to obtain a representative sample. Surface areas were measured for soils from sites A and F. Generally, surface areas increased with decreasing particle size and were largest for the soils collected in the sieve pan (Figure 4). Pan soils from Site F
Figure 5. Surface area measurements for Sieve #100 and #200 soils obtained from all sites. Within error, soils from the healthy areas exhibit similar surface area values, regardless of sieve number. Highest surface area values were obtained for soils from Site F, where the Sieve #100 soils had surface areas three times higher than those from Sites A−C and the Sieve #200 samples had surface areas that were 5 times higher than those from Sites A−C. Error bars are smaller than data values.
Figure 4. Surface area measurements for soils obtained from sites A and F. Data represent results from triplicate runs (i.e., three distinct subsamples) for sieved soils. As particle size decreases (i.e., sieve number increases), surface areas generally increase. Also, surface areas for soils from site F (in the tree-kill zone) are significantly higher across all particle sizes, except for Sieve #4 where it is slightly higher. Error bars are smaller than data values. Figure 6. Total organic carbon measurements of soils from all sites. Except for a large organic carbon content at Site D for the Sieve #100, soils from the tree-kill zone (Sites D−F) contained significantly lower amounts of organic carbon than soils from the non-tree-kill zone. Error bars are smaller than data values.
exhibited significantly larger surface areas than those at Site A (10.67 m2/g vs 2.53 m2/g). For site F, soil surface areas increased with decreasing particle size (Figure 4), as expected (see Gregg and Sing25). Surface areas measured for soils in the +100−200 mesh and +200-Pan mesh at site F were higher than anywhere else (Figure 5). Surface areas for the +200-Pan soils form sites D and E were slightly higher than the relevant surface areas of soils form the non-tree-kill zones. Larger surface areas generally result in higher rates of weathering and stronger soil-solute interactions (i.e., anion exclusion, ion exchange, adsorption, etc.). Soils in the tree-kill zone exhibit higher surface areas, possibly due to chemical or biological weathering induced by the presence of CO2(g). Carbon Content. Results presented here represent triplicate runs of soils from each of the three samples taken at the sampling locations (Figure 6; data represent averages from 9 samples). Generally, the organic carbon content of soils from the non-tree-kill sites A, B, and C were higher than their counterparts in sites D, E, and F (except for the 100-mesh soils for site D). Measured soil organic carbon decrease as the sampling sites move from the non tree-kill zone of site A to the
zone closest to the tree-kill zone, site C. This is consistent with the hypothesis that the carbon content in soils that are supporting trees and soil microorganisms should be higher than that in the tree-kill zone sites. Batch Leaching Experiments. The highest amount of leaching was observed for pan soils from sites A and F incubated at pH = 2 (Table 2). When compared to pH 5.5 data, batch results in site A yielded concentrations that were 10−90 times higher for the elements tested. Similar results are found for soils from site F. After 14 days, leaching was greatest for all elements in site F at pH = 5.5, whereas at site A, leaching was the greatest for all elements at pH = 2. Al, Si, and Fe increased in concentration irrespective of grain size or pH. In general, the lower the pH, the higher the concentration of leached elements. The effects of 12059
dx.doi.org/10.1021/es501829z | Environ. Sci. Technol. 2014, 48, 12056−12062
Environmental Science & Technology
Article
Table 2. Batch Test Leaching Results of Elemental Concentrations for Mg, Al, Si, Fe, and As in μg/L for Day 1 and Day 14 for pH Levels 2 (Acidic Condition) and 5.5 (Open System Analogue) Using Soils from Sites A and F site A pH = 2
Day 1 Day 14
pH = 5.5
Day 1 Day 14
Mg, ppb
Al, ppb
Si, ppb
Fe, ppb
As, ppb
sieve pan sieve pan sieve pan sieve pan
384 1310 925 27,600 210 835 320 732 Mg, ppb
10,200 13,500 17,900 >20,000 204 1860 555 1180 Al, ppb
3800 4600 9100 78,900
