Radiogenic 4 He as a conservative tracer in buried-valley aquifers

Transcription

1 WATER RESOURCES RESEARCH, VOL. 41,, doi: /2004wr003857, 2005 Radiogenic 4 He as a conservative tracer in buried-valley aquifers Stephen J. Van der Hoven, R. Erik Wright, and David A. Carstens Department of Geography-Geology, Illinois State University, Normal, Illinois, USA Keith C. Hackley Illinois State Geological Survey, Champaign, Illinois, USA Received 30 November 2004; revised 3 August 2005; accepted 16 August 2005; published 16 November [1] The accumulation of 4 He in groundwater can be a powerful tool in hydrogeologic investigations. However, the use of 4 He often suffers from disagreement or uncertainty related to in situ and external sources of 4 He. In situ sources are quantified by several methods, while external sources are often treated as calibration parameters in modeling. We present data from direct laboratory measurements of 4 He release from sediments and field data of dissolved 4 He in the Mahomet Aquifer, a well-studied buried-valley aquifer in central Illinois. The laboratory-derived accumulation rates ( mcm 3 STP kg water yr ) are 1 2 orders of magnitude greater than the accumulation rates based on the U and Th concentrations of the sediments ( mcm 3 STP kg water yr ). The direct measurement of accumulation rates are more consistent with dissolved concentrations of 4 He in the groundwater. We suggest that the direct measurement method is applicable in a variety of hydrogeologic settings. The patterns of accumulation of 4 He are consistent with the conceptual model of flow in the aquifer based on hydraulic and geochemical evidence and show areas where in situ production and external sources of 4 He are dominant. In the southwestern part of the study area, Ne concentrations are less than atmospheric solubility, indicating gases have been lost from the groundwater. Available evidence indicates that the gases are lost as groundwater passes by pockets of CH 4 in glacial deposits overlying the aquifer. However, the external flux from the underlying bedrock appears to dominate the accumulation of radiogenic 4 He in the aquifer in the southwestern part of the study area, and the loss or gain of helium as groundwater passes through the overlying sediments is minor in comparison. Citation: Van der Hoven, S. J., R. E. Wright, D. A. Carstens, and K. C. Hackley (2005), Radiogenic 4 He as a conservative tracer in buried-valley aquifers, Water Resour. Res., 41,, doi: /2004wr Copyright 2005 by the American Geophysical Union /05/2004WR Introduction [2] The helium 4 ( 4 He) isotope is a powerful tool for investigating hydrogeologic processes. Helium is a noble gas that does not participate in chemical reactions, and therefore is a conservative tracer. The sources of helium in groundwater are relatively well understood, and can generally be identified and separated by the distinct 3 He/ 4 He ratios of each source. In many groundwater systems, dissolved 4 He concentrations increase along a flow path. This increase can be due radiogenic production and release from the aquifer materials, from fluxes external to the aquifer, or a combination of the two. Disagreement or uncertainty in quantifying internal and external sources is often the main limitation for the use of 4 He as a hydrogeologic tracer. [3] Dating is theoretically possible if in situ production and release can be quantified, and if there is no external flux, or the external flux is known. The age of water is the total amount of radiogenic 4 He divided by the rate of accumulation from production within the aquifer. While this seems conceptually straightforward, practical applications of 4 He as a dating tool have been limited by uncertainty related to in situ production and external flux sources. In settings where the accumulation of 4 He in groundwater is thought to be solely from in situ production, the 4 He accumulation rates are often calculated by dating the water with another technique ( 3 H/ 3 He, CFC, 14 C, 81 Kr) [Andrews and Lee, 1979; Torgersen and Clarke, 1985; Solomon et al., 1996; Lehmann et al., 2003; Carey et al., 2004]. Aeschbach-Hertig et al. [2000] provide an example of the use of 4 He as a stand-alone dating technique. However, their 4 He ages correlate well with inferred changes in noble gas recharge temperatures, d 18 O, and chloride as the climate warmed at the end of the last glacial maximum. [4] It has been demonstrated in many aquifers that there is a external flux of 4 He into the aquifer. In this paper, we define an external flux as being from geologic units adjacent to the aquifer, from deeper in the crust, or from the mantle. 4 He has been used as a conservative tracer in modeling to quantify the external flux of 4 He as well as to constrain the flux of water through an aquifer or sedimentary basin [Castro et al., 1998; Bethke et al., 1999; Cserepes and Lenkey, 1999; Castro and Goblet, 2003] In Situ Production [5] 4 He is produced in minerals by radioactive decay of U and Th. The 4 He produced will eventually leave the 1of13

2 VAN DER HOVEN ET AL.: HELIUM 4 IN BURIED-VALLEY AQUIFERS crystal lattice. The release of 4 He from minerals and its accumulation in groundwater is a complex function of mineralogy, grain size, temperature, and geologic history. Three different methods have been developed to quantify the rate at which in situ produced helium accumulates in groundwater. [6] The first method that was developed, and is still widely used, calculates the accumulation of 4 He in groundwater based on the concentrations of U and Th and their rate of radioactive decay. The production rate within the aquifer materials is calculated using the formula (modified from Andrews and Lee [1979]): P 4He ¼ ðc U P U ÞþðC Th P Th Þ ð1þ where P 4He is the production rate (cm 3 STP g solid yr ), C U and C Th are the concentrations of U and Th in the solids (mg g ), and P U and P Th are 4 He production rates from the decay of U and Th ( and cm 3 STP mg solid yr, respectively). Use of this method assumes that the rate of in situ production is equal the rate at which 4 He is accumulating in water, thus the transfer efficiency is 1. The production rate per unit mass of aquifer is converted to an accumulation rate per unit mass of water using the formula: A 4He ¼ P 4He r solid =rwater L 4He ðð1 nþ=nþ ð2þ where A 4He is the accumulation rate (cm 3 STP g water yr ), r solid is the density of the aquifer solids, r water is the density of water, L 4He is the transfer efficiency, and n is porosity. [7] The U/Th production method is simple, requiring only knowledge of U and Th concentrations. This method is most appropriate for use in sediments or sedimentary rocks that are >10 6 years old. These aquifer materials are likely to have released any 4 He that accumulated in the rocks prior to erosion, and they are most likely to have reached equilibrium between production and release from the crystal lattice. Potential complications with this method include preferential retention/release of 4 He [Tolstikhin et al., 1996] and high concentration external sources that may result in diffusion of 4 He into aquifer materials [Tolstikhin et al., 1996; Lehmann et al., 2003]. [8] A second method for calculating the release of 4 He from aquifer materials uses a diffusion model to estimate release [Solomon et al., 1996; Sheldon et al., 2003; Dowling et al., 2004]. To use a diffusional release model, the total 4 He and 21 Ne concentrations in aquifer sediments, the 4 He diffusion rate, and grain size(s) of sediments must be measured, and erosion time must be known. The erosion time is the period since the sediments were first generated by erosion of the bedrock protolith. The initial 4 He in aquifer sediments is calculated based on the 4 He/ 21 Ne production ratio and the amount of 21 Ne produced. The 4 He release rate (as a function of grain size) is then modeled based on the initial 4 He concentration, the 4 He diffusion rate, and the erosion time. [9] Disadvantages of the diffusion model methods is the number of chemical and physical properties that must be measured. Since diffusion is a function of mineralogy, detailed knowledge of mineralogy and diffusion from various minerals must be known. As with the U/Th production method, a potential complication for using this method are preferential retention/release of the 4 He. In addition, if the geologic history of the sediments is not well known, the erosion time may be difficult to estimate. [10] A third method involves direct measurement of the release rates. For this method, bulk sediment or separates are placed in a evacuated, sealed container, and held at constant temperature for some period of time [Solomon et al., 1996]. Since diffusional release of the 4 He from sediments is a function of temperature, the release experiments are performed at multiple temperatures. An Arrhenius-type plot is then used to calculate the release rate as a function of temperature (i.e., the groundwater temperature). [11] Direct measurement of release rates do not require detailed knowledge of mineralogy, grain size, and geologic history. Large bulk samples (100 g) used for release rate experiments minimize the uncertainty due geologic heterogeneity that can be a problem with the small sample size (100 mg) used in the other two methods. The application of direct measurement to quantify 4 He release assumes that the release rate has remained constant since the time of groundwater recharge. This assumption may not be valid for very young sediments that have stored 4 He. Models of diffusional release indicate that there is a rapid exponential decrease in release rate for up to several hundred thousands years after erosion [Solomon et al., 1996; Dowling et al., 2004] External Sources [12] The observation that 4 He groundwater ages often appear anomalously old [Andrews and Lee, 1979; Bottomley et al., 1984] lead to the conclusion that there is a flux of 4 He into aquifers from external sources. The external fluxes have been attributed to degassing of the deep crust [Torgersen and Clarke, 1985; Castro et al., 1998; Cserpes and Lenkey, 1999] and from geologic units (commonly shale) adjacent to an aquifer [Tolstikhin et al., 1996; Solomon, 2000; Lehmann et al., 2003]. [13] External fluxes of 4 He are most often quantified using numerical transport models. In studies of large sedimentary basins, the deep crustal flux is specified crossing the lower boundary, and is often treated as a calibration parameter. The crustal fluxes can be uniform [Torgersen and Clarke, 1985; Castro et al., 1998; Cserepes and Lenkey, 1999] or spatially variable [Torgersen and O Donnell, 1991; Zhao et al., 1998]. In local or regional aquifer settings, the external flux can be modeled by specifying the 4 He concentration on the lower boundary [Solomon et al., 1996; Sheldon et al., 2003] Present Investigation [14] In this paper we report radiogenic 4 He concentrations in groundwater and direct measurements of 4 He release from sediments from the Mahomet Aquifer, a buried-valley aquifer in central Illinois, USA (Figure 1). While direct measurement of release rates has been reported in several recent studies [Solomon et al., 1996; Sheldon et al., 2003], to our knowledge this technique has not been applied to a regional groundwater flow system. We suggest that direct measurement is a simple method that is widely applicable because it has fewer disadvantages than the other two methods. 2of13

3 VAN DER HOVEN ET AL.: HELIUM 4 IN BURIED-VALLEY AQUIFERS Figure 1. Boundaries of the Mahomet Bedrock Valley and potentiometric contours of the Mahomet Aquifer. Cross section A-A 0 is shown on Figure 2. [15] To date, 4 He has been successfully applied as a quantitative tracer primarily in relatively simple hydrogeologic settings. The Mahomet Aquifer, while well studied, is hydrogeologically complex. The patterns of accumulation of 4 He are consistent with the current understanding of flow and groundwater age in the aquifer, and provide important insight into interactions with the overlying glacial units and the underlying bedrock. Because of complex carbonate chemistry in the Mahomet Aquifer, 14 C ages of the groundwater are uncertain. Our data suggest that 4 He can be used as a dating technique in some parts of the aquifer, and as a calibration parameter in models of the aquifer. 2. Geologic and Hydrogeologic Setting [16] Prior to the onset of Pleistocene glaciation in North America, a drainage system had eroded into bedrock across Illinois. In eastern and central Illinois, this drainage system is known as the Mahomet Bedrock Valley (Figure 1). Successive periods of glaciation over the past >1 Ma have completely filled this paleovalley, and buried it with up to 100 m of a complex sequence of glacial sediments. In Illinois, the glacial deposits are divided into the Wedron and Mason Groups, the Glasford Formation, and the Banner Formation [Kempton et al., 1991; Hansel and Johnson, 1996]. For the purposes of this manuscript, we combine the Mason with the Wedron Group Wedron Group [17] The Wedron Group are sediments deposited during the Wisconsin Episode, between 25 and 12 ka. This unit is composed primarily of low-permeability diamicton units, but also contains thin and discontinuous outwash sand and gravel units, lacustrine, and loess deposits. The Wedron Group sediments are up to 30 m thick (Figure 2) Glasford Formation [18] Sediments of the Glasford Formation were deposited during the Illinoisan Episode, between 180 and 125 ka. The Glasford has been divided into three low-permeability diamicton units, and is up to 40 m thick. In the study area, there are widespread and laterally continuous outwash or ice contact sand and gravel units greater than 6 m thick within the diamictons [Kempton et al., 1991] (Figure 2). These sand and gravel bodies are thought to represent a 3of13

4 VAN DER HOVEN ET AL.: HELIUM 4 IN BURIED-VALLEY AQUIFERS Figure 2. A cross section running along the thalweg of the buried valley. The location of the cross section is shown on Figure 1. Circles represent the screened interval of wells used in constructing the cross section. The light shading represents clay-rich diamicton, and the dark shading represents sand and gravel units. Note the thickness of the sand and gravel units in Ford and Champaign counties, where the main recharge area of the aquifer is thought to be located. major glacial meltwater drainage way during the Illinois Episode Banner Formation [19] The pre-illinoisan Banner Formation represents the oldest glacial sediments in the study area. The age of the Banner formation is uncertain, but it is thought to be 700 ka. The Mahomet Sand Member of the Banner Formation has a sand facies that occurs in the main valley and a silt facies that is restricted to tributary valleys [Kempton et al., 1991]. The sand facies of the Mahomet Sand Member is commonly referred to as the Mahomet Aquifer (MA), and this terminology will be used throughout the rest of this article. [20] The MA consists of coarse-grained sand and fine- to medium-grained gravel of glaciofluvial origin, and may have been deposited in multiple episodes. The sand is predominantly quartz, with lesser amounts of feldspar [Willman and Frye, 1970]. Manos [1961] reported that heavy minerals, including hornblende, garnet, epidote, and hypersthene, make up a minor fraction of the sediments. The thickness of the MA ranges up to 45 m in the deepest part of the valley (Figure 2) Bedrock Geology [21] The Silurian, Devonian, Mississippian, and Pennsylvanian bedrock in the eastern part of the study area is primarily of dolomite and limestone with some shale and coal [Willman et al., 1967]. The Pennsylvanian bedrock in the western part of the study area is primarily shale with some sandstone and coal. The La Salle Anticlinorium is a major structural feature trending northwest-southeast through the study area, and results in a series of anticlines, synclines, and monoclines that intersect the bedrock valley [Panno et al., 1994] Groundwater Flow [22] Recharge to the MA is thought to occur primarily as vertical leakage of precipitation through the overlying glacial sediments and by upward leakage through the bedrock [Kempton et al., 1991; Panno et al., 1994]. The main recharge area for the eastern and central part of the aquifer appears to be in the vicinity of the potentiometric high centered around the town of Paxton [Wilson et al., 1998] (Figure 1). Relatively thick and extensive sand and gravel deposits in the Glasford Formation in this region appear to allow significant recharge to the MA (Figure 2). From the potentiometric high, groundwater flows to the north, the east, and the southwest. The cities of Urbana and Champaign obtain their water supply from the MA, and a major cone of depression has developed to the west of these cities (Figure 1). Geochemical evidence indicates that significant recharge from the underlying bedrock occurs in several parts of the study area [Panno et al., 1994; Hackley, 2002]. 3. Methods 3.1. Dissolved Gas Sampling [23] Dissolved gas samples were collected using the diffusion sampler method [Sheldon et al., 2003]. Diffusion samplers consist of 1 cm OD copper tubing that is sealed at one end using a cold weld tool. The other end is attached to silicon tubing that is plugged at its distal end. Prior to deploying the sampler, the well is purged to remove stagnant water. The diffusion sampler is lowered into the well to the elevation of the screened interval. The gases in the sampler, which are originally atmospheric, exchange with gases dissolved in the groundwater by diffusion through the silicon tubing. The sampler is left in the well for at least 24 hours so that the gases inside the sampler reach equilibrium with the gases in the groundwater. When the sampler is retrieved, the open end of the copper tube is quickly sealed with the cold weld tool. After well purging, the total dissolved gas pressure (TDG) is measured using a Hydrolab MiniSonde 4a Analysis of Dissolved Gases [24] The analysis of gases in the samplers was performed at the University of Utah s Noble Gas Laboratory. The mole fractions of N 2, 20 Ne, 40 Ar, 84 Kr were determined on a Stanford Research quadrupole mass spectrometer, with 4of13

5 VAN DER HOVEN ET AL.: HELIUM 4 IN BURIED-VALLEY AQUIFERS Table 1. Analytical Results of Dissolved Gas Analysis and Calculated Radiogenic 4 He Concentrations a Well Name Geologic Unit Screen Depth, m N 2 20 Ne 40 Ar 84 Kr 3 He 4 He Recharge Temperature, C DNe, % Residual R/R air 4 He radiogenic, mcm 3 STP kg water ASW b 1.59E E E E E E D Wedron E E E E E E E S Wedron E E E E E E E D Wedron E E E E E E E D Wedron E E E E E E E Dug Well Wedron E E E E E E E CHM94B Glasford E E E E E E E G Glasford E E E E E E E VER94B Glasford E E E E E E E FRD94B Glasford E E E E E E E North Shallow Glasford E E E E E E E East Shallow Glasford E E E E E E E B Glasford E E E E E E E West Shallow Glasford E E E E E E E East Deep Banner E E E E E E E M (CHM95B) Banner E E E E E E E CHM96C Banner E E E E E E E IRO95A Banner E E E E E E E VER94A Banner E E E E E E E CHM94A Banner E E E E E E E CHM96A Banner E E E E E E E IRO96A Banner E E E E E E E FRD94A Banner E E E E E E E CHM95D Banner E E E E E E E North Deep Banner E E E E E E E West Deep Banner E E E E E E E White Heath Banner E E E E E E E Decatur S Banner E E E E E E E Decatur W Banner E E E E E E E ISGS1 Banner E E E E E E E ISGS2 Banner N/A 1.69E E E E E E E Decatur C Banner E E E E E E E A Banner E E E E E E E a Read 1.59E-02 as b ASW is air-saturated water, 5 C, no excess air. analytical precision of ±3%. While CH 4 can be accurately quantified on the quadrupole, the system as currently designed does not collect the data necessary to separate potential interferences with the main CH 4 peak (i.e., mass 16). As a part of another investigation, separate samples for CH 4 analysis were collected from a limited number of wells and analyzed by gas chromatography by the Illinois State Geological Survey. [25] Mole fractions of 3 He and 4 He were determined on a Mass Analyzer Products sector field mass spectrometer operated in static mode. Typical analytical precision is ±1.0% for both 3 He and 4 He Calculation of Radiogenic 4 He [26] The concentration of gases in each well was calculated using the measured mole fraction, the measured TDG and temperature in the well, and the temperature-dependent Henry s coefficient. Since Illinois is a region of low topographic relief, the atmospheric pressure of recharge was assumed to be a constant 0.97 atm, the average atmospheric pressure at an elevation of 250 m. The recharge temperature and amount of excess air were estimated using an iterative optimization routine (the closed system equilibrium model) similar to that of Aeschbach-Hertig et al. [2000]. Radiogenic 4 He was calculated by subtracting the equilibrium solubility and excess air components from the total 4 He measurement [Solomon, 2000]. The 3 He/ 4 He ratios (Table 1) indicate that there is no contribution from mantle helium The 4 He Release From Sediments [27] The 4 He release rates were determined from the sediments of all three Quaternary units in the study area. Sediment samples were taken from discrete intervals within a single borehole. Approximately 100 g of sample was put into a copper flask, and the flasks were sealed under high vacuum. Step heating experiments were performed by holding the samples at a constant temperature for a period of time, and then measuring the helium released. Afterward, the flask was evacuated again, and held at a higher temperature. The temperature and time period increments were 20 C for 200 days, 100 C for 5.75 days, and 200 C for 0.08 days. The samples were not fused to measure total helium. 4. Results 4.1. Release Rate Experiments [28] Release rate experiments were performed on 5 samples, one from the Wedron Formation and two each from the Glasford and Banner Formations. Samples were sieved to remove the gravel fraction (>2 mm). Grain size analyses (Figure 3) show that most samples are predominantly fine to medium sand (0.125 to 0.5 mm). One sample each from the Wedron and Glasford Formations have a considerable 5of13

6 VAN DER HOVEN ET AL.: HELIUM 4 IN BURIED-VALLEY AQUIFERS Figure 3. Grain size analyses for the sediments used in the release rate experiments. fraction >2 mm (42% and 28%, respectively). Removal of the gravel fraction may result in an over estimation of the release rate for these samples. No attempt was made to determine how release rate varies by mineral type. Since groundwater in the MA passes through all three heterogeneous Quaternary units, it is impossible to estimate the percentage of each mineral type that is contributing radiogenic 4 He. We consider the release rate experiments to be representative of the bulk release rates of radiogenic 4 He. [29] The noble gas concentrations measured during these experiments are shown in Table 2. During each experiment, air leaked into the containers, and 3 He/ 4 He ratios were used to make a blank correction. The 3 He/ 4 He ratios provide an accurate correction since the ratio of atmospheric He ( ) is very different than the released He ( ). The ratio of the released He was estimated by two methods. Figure 4 plots the evolution of He ratios as the groundwater evolves from atmosphericdominated to subsurface-dominated He. The y intercept of the best fit line ( ) represents 100% He from subsurface production. The other method looks at the average 3 He/ 4 He ratio ( ) for the two highertemperature release rates (where the blank correction is 15%). The field-derived release ratios agree well with the laboratory-derived release ratios. Note that for two of the samples (Box 32 and Box 46), there are data for only two heating steps. These samples developed large leaks during a heating step, and were unusable for that heating step. [30] Table 2 shows the total 4 He released. Note that for the 20 o C temperature step, the blank makes up 75 90% of the total 4 He. This is a result of the slow leakage over 7 months that the samples were stored at this temperature step. In the future, the storage time and atmospheric blank could be decreased by increasing the sample mass by an order of magnitude. The atmospheric blank for the highertemperature steps and shorter time periods is 15%. [31] If diffusion controls the release of 4 He from the sediments, then the Arrhenius relationship predicts that Table 2. Results of 4 He Laboratory Release Rate Experiments a Sample 20 Ne Concentration, cm 3 STP g solids 40 Ar 3 He 4 He Measured R/R air Fraction of Air b 4 He Released, b cm 3 STP kg solids 4 He Release Rate, b cm 3 STP kg yr 20 C for 200 days NIWC-1 Box feet 1.84E E E E E E-07 NIWC-1 Box feet 1.68E E E E E E-07 NIWC-1 Box feet 1.98E E E E E E-07 NIWC-1 Box feet 1.81E E E E E E-07 NIWC-1 Box feet 1.93E E E E E E C for 5.75 days NIWC-1 Box feet 7.28E E E E E E-04 NIWC-1 Box feet 3.26E E E E E E-04 NIWC-1 Box feet 4.16E E E E E E-04 NIWC-1 Box feet 2.82E E E E-10 c N/A 2.93E-07 c 1.86E-05 c 200 C for 0.08 days NIWC-1 Box feet 8.90E E E E E E-01 NIWC-1 Box feet 4.61E E E E E E-02 NIWC-1 Box feet 4.12E E E E E E-01 NIWC-1 Box feet 6.71E E E E E E-01 a The release rates have been corrected for the atmospheric blank. b Fractions of air contamination and released 4 He are blank-corrected on the basis of 3 He/ 4 He ratios as discussed in the text. c Suspected analytical error in the 4 He measurement; the blank correction is made using 20 Ne. 6of13

7 VAN DER HOVEN ET AL.: HELIUM 4 IN BURIED-VALLEY AQUIFERS Figure 4. Helium isotopic ratio versus fraction of helium from atmospheric equilibrium. Theoretically, dissolved helium should evolve from only atmospheric helium (the atmospheric equilibrium point) to the only subsurface produced helium (the y intercept of the linear regression). there will be a linear relationship between natural log of diffusion rate and temperature (Figure 5). Linear regressions of individual samples sets on this Arrhenius-type plot yield high correlation coefficients (R 2, Table 3). One sample (Box 41, 100 C heating step) appears to have anomalously low 4 He. While the 20 Ne and 40 Ar in this sample are similar to others in this heating step, the 4 He is about 1 order of magnitude lower. Although we believe that anomalously low 4 He represents an analytical error, including it in the linear regression yields the lowest R 2 of Several other samples (Box 32 and Box 46) have only two points, and their R 2 can only be 1.0. [32] The linear relationship for each sample can be used to predict the release rate at the in situ groundwater temperature of 12 C (Table 3). The rate of release from sediments is converted to the accumulation rate in the groundwater by equation (2), a sediment density of 2.6 g cm 3, and a porosity of The accumulation rates range from 0.13 to 0.91 mcm 3 STP kg water yr, and are similar for all three geologic units. [33] The release and accumulation rates of radiogenic 4 He were also calculated based on the U and Th concentration of the samples used in the step heating release experiments. U concentrations ranged from 0.41 to 1.03 mg kg, and Th concentrations ranged from 1.36 to 2.35 mg kg (Table 3). Using equations (1) and (2), the U/Th accumulation rates ranged from to mcm 3 STP kg water yr. reasonable fit was defined as recharge temperatures between 5 C and 10 C, excess air between 10 and 50% of the equilibrium solubility (DNe), and normalized residuals between modeled and actual value of <1% (Table 1). In the western part of the study area, the fit between modeled and actual data was less satisfactory. Recharge temperatures were often >10 C, excess air was often zero, and normalized residuals were >10%. We suspect that subsurface production of gases (CH 4 and N 2 ) is responsible for the loss a atmospheric gases from the groundwater, resulting in the poor model fit. We will discuss our hypotheses for stripping in the later sections Wedron Formation [36] A limited number of shallow wells in the Wedron were sampled to evaluate current recharge conditions. These wells are screened at or below the water table, at depths between 5 and 14 m below the land surface. [37] Most groundwater samples from the Wedron Formation had little radiogenic 4 He, and several had negative values (Table 1). Several samples have negative DNe values, indicating loss of atmospheric gases. The widespread application of N fertilizers in Illinois may result in production of N 2 through denitrification of NO 3 in the groundwater. It is also possible that CH 4 produced in the underlying till is transported to the surface. One well had 27 mcm 3 STP kg water of radiogenic 4 He, which may indicate an enhanced release rate (discussed in more detail in section 5.1) Glasford Formation [38] Groundwater recharging the MA must also pass through the Glasford Formation. In some parts of the study area, sand and gravel units in the Glasford are thick and may be well connected to the MA (Figure 2). In other parts of the study area, Glasford sand and gravel units are confined by fine-grained diamicton, and are poorly connected to the MA. All of the wells screened in the Glasford Formation were adjacent to wells screened in the MA, and had downward hydraulic gradients. [39] Radiogenic 4 He concentrations in the Glasford Formation ranged from 18 to 516 mcm 3 STP kg water (Table 1) Radiogenic 4 He in Groundwater [34] As discussed in section 3.3, the amount of radiogenic 4 He was calculated using an optimization routine that varies the recharge temperature and amount of excess air. This routine provides a theoretical original amount of 4 He from equilibrium solubility and excess air. The difference between the original 4 He and the total measured 4 He is assumed to be radiogenic 4 He. [35] For most of the samples collected in the central and eastern part of the study area there is a reasonable fit between the modeled and actual data. For the MA, a 7of13 Figure 5. Arrhenius-type plot of the natural log of 4 He release rate versus inverse temperature for laboratory release rate experiments.

8 VAN DER HOVEN ET AL.: HELIUM 4 IN BURIED-VALLEY AQUIFERS Table 3. Calculations of 4 He Accumulation Rates in Groundwater at 12 C Based on Two Methods: The U and Th Concentration of Sediments and Release Rate Experiments Calculations Based on U and Th Concentrations Calculations Based on Release Experiments Sample Geologic Unit U, mg kg Th, mg kg 4 He Production Rate, mcm 3 STP kg sed yr 4 He Accumulation Rate, mcm 3 STP kg water yr 4 He Release Rate, mcm 3 STP kg sed yr 4 He Accumulation Rate, mcm 3 STP kg water yr R 2 NIWC-1 Box Wedron NIWC-1 Box Glasford NIWC-1 Box Glasford NIWC-1 Box Banner NIWC-1 Box Banner In nearly half of the Glasford/Mahomet well pairs, radiogenic 4 He concentrations were higher in the Glasford (Figure 6). Those with the highest 4 He concentrations also appear to have lost atmospheric gases based on their negative DNe values. Within the Glasford Formation, pockets of CH 4 drift gas are well documented. The organicrich and low-permeability diamictons provide a mechanism for both production and retention of CH 4. The amount of Figure 6. Radiogenic 4 He concentrations in groundwater from the Glasford Formation and the Mahomet Aquifer. Locations where two values are shown represent well pairs screened in the Glasford and Mahomet. The Mahomet value is always shown on top. 8of13

9 VAN DER HOVEN ET AL.: HELIUM 4 IN BURIED-VALLEY AQUIFERS and mol L of CH 4, respectively. In the Mahomet, the North Deep, 1A and ISGS1 wells had , and mol L of CH 4, respectively. The amount of CH 4 in other wells can be estimated from the measured mole fractions of N 2 and Ar (not shown) and the total dissolved gas pressure. Assuming that N 2, Ar and CH 4 are the only gases in solution, the concentration of CH 4 can be calculated. Using this method, CH 4 concentrations ranged from to mol L. For the wells where measured values are available, most of the calculated CH 4 concentrations are within 10% of the measured values. Figure 7. Radiogenic 4 He (diamonds) and DNe (squares) for wells screened in the Mahomet Aquifer are plotted with distance from the center of the main recharge area for the aquifer (the town of Paxton, shown on Figures 1 and 6). Positive distances are along the flow path to the southwest of Paxton, and negative distances are along flow paths to the north and northeast of Paxton. 4 He associated with the drift gas is not known. If the 4 He concentrations in the drift gas are high, groundwater flowing by these gas pockets could be stripped of other atmospheric gases, but gain 4 He. If the 4 He concentrations in the drift gas are low, then the groundwater could have been stripped of 4 He as well as other atmospheric gases. [40] The Glasford samples that do not appear to have been stripped (positive DNe) are located within or close to the main recharge area for the MA centered around the town of Paxton (Figure 6). Since vertical hydraulic gradients are down in this area, it is unlikely that the underlying bedrock contributes radiogenic 4 He to water in these wells. Assuming only in situ production and using the laboratoryderived accumulation rates from section 4.1., the 4 He ages of Glasford groundwater ranges from 20 to 900 years. These ages are conceptually plausible Mahomet Aquifer [41] Radiogenic 4 He concentrations in the MA range from 19 to 3052 mcm 3 STP kg water, and exhibit consistent spatial patterns (Table 1 and Figures 6 and 7). The lowest radiogenic 4 He concentrations are found in the main recharge area for the aquifer around the town of Paxton. To the southwest of Paxton, radiogenic 4 He concentrations increase linearly for about 45 km from Paxton (Figure 7). Beyond 45 km, there is a sharp increase in the accumulation rate with distance. To the north and northeast of Paxton, the accumulation rate of radiogenic 4 He with distance is also relatively high. We believe that the higher rates of accumulation with distance represent an external flux of radiogenic 4 He from the underlying bedrock (discussed in more detail in the next section). [42] There are also spatial patterns for the groundwater that appears to have lost atmospheric gases. Negative DNe are only found in the southwestern part of the study area, beyond 45 km from Paxton (Figure 7). [43] During this investigation, 5 wells in the Glasford and Mahomet were analyzed separately for CH 4. In the Glasford, the North Shallow and 1B wells had Discussion 5.1. He Release [44] From section 4.1, the in situ accumulation rates range between 0.13 and 0.91 mcm 3 STP kg water yr. Several recent investigations conducted in Quaternary glacial sediments using laboratory release rate experiments report accumulation rates similar to ours. Solomon et al. [1996] report an accumulation rate of 0.43 mcm 3 STP yr, and Sheldon et al. [2003] report release rates kg water of 0.2 to 1.17 mcm 3 STP kg sediment yr. [45] We were surprised that the laboratory release rates were similar for all three geologic units. Initially, we expected the rate to be different since the units were deposited at different times over the past 700 ka. Theoretically, release rates decrease with increasing time since erosion. Therefore we expected the release rate to be highest in the Wedron Formation, and lowest in the MA. An explanation for the similarities in release rates is that there is reworking of sediments during subsequent glacial episodes. If so, a diffusion model for calculating He release would not be feasible in this setting because an erosion time would be difficult to estimate. [46] In addition to the laboratory data that appear to show release of stored helium at rates higher than U/Th production, we also see field evidence of elevated release rates. One well screened about 5 m below the water table in the Wedron has 27 mcm 3 STP kg water of radiogenic 4 He. It is unlikely that this groundwater contains 4 He from external bedrock sources because about 100 m of diamicton and aquifer separate the well secreen from the bedrock. Using the in situ produced 4 He and a U/Th accumulation rate of mcm 3 STP kg water yr, the groundwater would have an age of 3000 years. Conceptually, this seems at least an order of magnitude too old. Using the laboratory-derived accumulation rates, the groundwater age is between 30 and 200 years, which is plausible Dissolved He [47] On the basis of the evolution of 3 He/ 4 He ratios along a flow path, the radiogenic 4 He appears to be coming from the same source. This can be demonstrated by plotting the excess air corrected 3 He/ 4 He ratio (( 3 He total 3 He excess air )/( 4 He total 4 He excess air )) versus the fraction of 4 He from atmospheric equilibrium (( 4 He solubility /( 4 He total 4 He excess air )) (Figure 4). At atmospheric equilibrium, the 3 He/ 4 He ratio is and the fraction of atmospheric 4 He is 1.0. At the other end of the graph where the fraction of atmospheric 4 He is 0.0, the 3 He/ 4 He ratio will be that of He generated in the aquifer. The dissolved He data from all three 9of13

10 VAN DER HOVEN ET AL.: HELIUM 4 IN BURIED-VALLEY AQUIFERS Figure 8. Comparison of 4 He ages calculated by release experiments (squares), 4 He ages calculated by U/Th concentrations (triangles), and 14 C ages (diamonds). Error bars are shown for selected samples to reduce visual clutter. The error estimate for 14 C ages are discussed in the text. For 4 He ages the error bars are calculated on the basis of the high and low accumulation rates. Other 4 He ages based on U/Th concentrations are all greater than 10,000 years. geologic units plot along a linear trend (R ) from the atmospheric value to the in situ production ratio of (the y intercept of the linear regression). This 3 He/ 4 He ratio compares well the mean laboratory release ratio for the higher-temperature steps of In comparison, Tolstikhin et al. [1996] report a release ratio of for whole rock crushed samples of shale. [48] Within the Wedron and Glasford Formations, we think the accumulation of radiogenic 4 He is predominantly from in situ sources. Groundwater in these units is separated from potential bedrock sources by >30 m of MA, and the vertical gradient is down into the aquifer. In addition, where paired Glasford-Mahomet wells have been sampled, there is not a strong chemical gradient to drive diffusive transport (Figure 6). [49] Within the MA, there appears to be in situ production as well as an external source from the underlying bedrock. Groundwater flowing southwest from the main recharge area centered on the town of Paxton appears to have primarily in situ produced radiogenic 4 He based on the relatively low linear accumulation with distance (Figure 7). Carbonate bedrock underlies this part of the aquifer. In general, carbonates have low U, Th and radiogenic 4 He concentrations. However, no radiogenic 4 He data are available for these rocks, and a diffusive He flux from the bedrock cannot be ruled out. [50] At about 45 km southwest of Paxton, there is a lithologic change from dominantly carbonate to dominantly shale. This also corresponds with a dramatic increase in the rate of radiogenic 4 He accumulation (Figure 7). Although no bedrock data are currently available, Paleozoic shales have been found to have high radiogenic 4 He concentrations [Tolstikhin et al., 1996; Sheldon et al., 2003]. This may result in a diffusive flux into the aquifer. In the western part of the study area, Panno et al. [1994] infer an advective flux of water from the underlying bedrock based on elevated chloride concentrations. This advective flux correlates with fracture zones associated with the LaSalle Anticlinorium. The very high radiogenic 4 He in the West Deep well (3052 mcm 3 STP kg water ) may indicate that this well is close to a fracture zone. [51] To the north and northeast of Paxton, the rate of radiogenic 4 He accumulation is also high (Figure 7). Panno et al. [1994] infer an advective flux from the bedrock in this area based on sulfate concentrations. Although the underlying bedrock is predominantly carbonate, radiogenic 4 He is likely to be higher in the carbonate than in the overlying aquifer. [52] While the evidence indicates areas where in situ or external sources are dominant, all samples plot along the same line on Figure 4, indicating that both sources have similar isotopic composition. We interpret this to mean that regional bedrock is also the source of the glacial sediments deposited on top of the bedrock A 4 He and 14 C Age Comparison [53] Only a limited number of wells in the MA have been sampled for both 4 He and 14 C. However, both have been sampled throughout the study area. For comparison, ages based on the two techniques are plotted versus distance from the main recharge area centered around Paxton (Figure 8). Both techniques show increasing age away from Paxton, but 4 He ages are uniformly younger than 14 C ages. [54] The 14 C ages were calculated using the interactive geochemical modeling program NETPATH [Plummer et al., 1996]. The program uses geochemical mass balance reactions to estimate the dissolution or precipitation of minerals in order to generate the chemical and isotopic composition of the water from one well to another along a groundwater flow path. Ideally, NETPATH requires that the initial and final wells are positioned on the same groundwater flow path. This is a problem for the MA because it is completely 10 of 13

11 VAN DER HOVEN ET AL.: HELIUM 4 IN BURIED-VALLEY AQUIFERS buried by diamicton deposits and does not crop out at the surface. Much of the diamicton above the MA is clay rich and low permeability, and shallower aquifers within the Glasford are often disconnected and isolated. However, because the shallower Glasford sands are upgradient of the MA, wells sampled from the Glasford aquifer were selected as the initial groundwater composition for determining the 14 C age of the groundwater in the central region of the MA. Thus the selection of upgradient well chemistry can have a large impact on the calculated age. The error bars for 14 C ages on Figure 8 represent the variability associated with the selection of different (but plausible) upgradient wells. Another problem with calculating the age of the groundwater in the MA using any technique is that there is evidence of significant mixing that occurs due to advection from bedrock. No bedrock wells in the vicinity of the MA area were available for sampling at the time of the 14 C study. The rationale for selection of wells and reactive minerals is detailed by Hackley [2002]. [55] The 4 He ages were calculated by dividing the measured radiogenic 4 He concentration in a well by the laboratory-derived accumulation rate. This assumes that all radiogenic 4 He comes from in situ sources. This assumption may only be valid for wells in the main recharge area and up to about 45 km southwest of Paxton. In the other parts of the study area, there is geochemical evidence for an external bedrock flux. Nevertheless, 4 He ages were calculated for all wells because they provide constraint on the maximum 4 He age. [56] In the central part of the study area (Paxton to 45 km southwest on Figure 8), there is only one well for comparative ages. The 4 He age is 50 years while the 14 C age is 4900 years, a large discrepancy. The 14 C age seems anomalous given that it is within the main recharge area and its age is greater than most other 14 C ages in the study area. On the other hand, a 4 He age of 50 years conceptually seems on the high end of vertical transport to the depth of the MA. The 4 He ages in the area increase to about 230 years at a distance of 44 km from Paxton. Much of the aquifer in this area is overlain by extensive sand and gravel deposits in the Glasford (Figure 2). Recharge from the surface may be semicontinuous throughout this area, resulting in addition of young groundwater over a wide area in this part of the aquifer. Also shown on Figure 8 are 4 He ages calculated using the U/Th concentrations for this part of the aquifer. These ages appear unrealistically old, starting at over 3000 years and rising to almost 10,000, which is older than any age in this study. [57] In the western part of the study area, there is better concordance between 4 He and 14 C ages. In this area it is likely that both methods are impacted by mixing with water from the shale bedrock which adds 4 He and dilutes the 14 C. To the north and northeast of Paxton, 14 C ages increase much faster than 4 He ages. Again, a bedrock flux may explain this discrepancy. In this area, the bedrock is dominantly carbonate, which would decrease the 14 C content while adding lesser amounts of 4 He than would come from shale bedrock Loss of Atmospheric Gases [58] In the MA, most wells in the southwest part of the study area have negative DNe values, indicating loss of atmospheric gases (Figure 7). Subsurface formation of CH 4 gas pockets is suspected to be involved in stripping atmospheric gases from the groundwater. Although CH 4 drift gas pockets are known to exist in the Glasford Formation in this region, no gas pockets are known to exist in the MA. CH 4 is also known to be associated with Paleozoic coal and petroleum deposits in the Illinois Basin. The d 13 C of dissolved CH 4 in the MA and of the drift gas ( 70 to 90%) indicates the CH 4 is of biogenic origin by reduction of CO 2 [Hackley, 2002]. CH 4 from the Paleozoic rock is likely to be of thermogenic origin (d 13 C of 30 to 50%). The isotopic data indicate that a loss of atmospheric gases occurs when groundwater comes in contact with the drift gas. [59] With respect to 4 He, elevated chloride concentrations are interpreted to be due to an advective flux from the shale bedrock. Since shale usually has high He concentrations, the accumulation of radiogenic 4 He in this part of the aquifer appears to be from a flux from the bedrock underlying the MA. In contrast, the loss of atmospheric gases appears to occur in the sediments overlying the MA. Since we are not certain whether the loss of atmospheric gases would result in a loss or gain of 4 He from the groundwater as it passes through the overlying sediments of the Glasford Formation, it would not be appropriate to apply a correction due to loss of atmospheric gases [e.g., Lippmann et al., 2000]. [60] The loss or gain of 4 He from groundwater passing through the sediments overlying the MA would have an impact on the accumulation of radiogenic 4 He and 4 He age of groundwater in the MA. A loss of 4 He due to stripping by drift gas as groundwater flows through the Glasford would result in an decrease of the amount of accumulated radiogenic 4 He and underestimate the 4 He age in the MA. A gain of 4 He from the drift gas would increase the amount of accumulated radiogenic 4 He and overestimate the 4 He age in the MA. [61] Although atmospheric gases have been lost from groundwater in the southwestern part of the study area, examination of Figure 4 appears to indicate that the gas loss has not significantly affected the accumulation of radiogenic 4 He. Figure 4 indicates that there is a linear evolution of the helium isotopic composition from 100% atmospheric helium to 100% subsurface helium. Large losses of helium would cause points to fall well above the line, while large gains would cause points to fall well below the line. We interpret the relatively small scatter about the line to mean that the amount of helium lost or gained when the other atmospheric gases were stripped is small compared to the accumulated radiogenic 4 He from in situ production and the flux from the bedrock underlying the MA. [62] We also cannot rule out gas loss during sampling. However, the use of diffusion samplers is designed to minimize this possibility. Diffusion samplers are lowered to the screened interval of the well where gases dissolved in the groundwater come into equilibrium with the sampler. The column of water above the screen is usually >30 m, imparting enough hydrostatic pressure to keep the gases in solution. After equilibration, the sampler is rapidly removed from the well and sealed. No water is brought to the surface 11 of 13