Survey of ²²⁶Ra/ ²²⁸Ra and Inorganic Constituents in Iowa Private Drinking Water Wells

Abstract

Radium is a naturally-occurring radioactive element found in Upper Midwest aquifers, yet exposure risks for unregulated water sources in this region are unknown. A survey of ^226/228Ra and other inorganic constituents in 94 private wells was conducted in 2019 across ten Iowa counties. Within private wells, 53% measured below minimum detectable activity whereas 6.3% have ^226/228Ra levels above the U.S. EPA MCL. ^226/228Ra occurrence was not associated with well-depth or chemical surrogates, but were generally consistent with the lithology and historical water quality trends for aquifers. In 2020, 11 of the ^226/228Ra-impacted wells were re-sampled to assess the efficacy of in-home water treatment. Water softeners reduced ^226/228Ra to non-detectable activities, but other inorganic constituents persisted and their removal required more advanced treatment. This study suggests that ^226/228Ra occurrence in private wells can reasonably be anticipated from public water supply data if reliable information is available to identify the source aquifer.

Introduction

The most abundant radium (Ra) isotopes (^226/228Ra) are found at detectable activities in groundwater across the Upper Midwestern region of the United States. ^226/228Ra occur in these systems because they are daughter products within the ²³⁸U and ²³²Th decay series, respectively, and form soluble divalent cations in water. For public water systems (PWSs) in this region, elevated combined Ra (as the sum of total ²²⁶Ra and ²²⁸Ra activity) in raw groundwater poses a known treatment challenge (Dematatis et al., 2020; Mathews et al., 2018; Stackelberg et al., 2018; Szabo et al., 2012). ^226/228Ra is regulated under the Safe Drinking Water Act (SDWA) with a United States Environmental Protection Agency (U.S. EPA) enforceable Maximum Contaminant Level (MCL) of 5 pCi/L (or 0.185 Bq/L) for combined Ra. Radium is of concern because exposure to drinking water with elevated concentrations has been associated with increased risk of blood, bone, lung, breast, and bladder cancers (Bean, Isacson, Hahne, et al., 1982; Bean, Isacson, Hausler, et al., 1982; Isacson et al., 1985; Lyman et al., 1985).

From 2001 to 2020, combined Ra was among the most frequent MCL violations for PWSs regulated under SDWA in the state of Iowa, where over seventy-five percent of residents rely on groundwater for their drinking water supply (Iowa Department of Natural Resources, 2019). A U.S. Geological Survey study of ^226/228Ra activity in PWS supply wells found a significant cluster of samples measuring above 5 pCi/L in Iowa groundwater from the Cambrian-Ordovician, Dakota, and Devonian aquifer systems (Szabo et al., 2012). Targeted analyses of samples from the Cambrian-Ordovician aquifer in Iowa and Wisconsin found that ²²⁶Ra generally occurred at somewhat higher concentrations than ²²⁸Ra in PWS groundwater sources (although the human health risk associated with samples was most often reflective of ²²⁸Ra levels due to its greater unit cancer risk factor) (Stackelberg et al., 2018). In Wisconsin, Dematatis et al. (2020) discovered increasing levels of combined Ra in unconfined and confined regions of the Cambrian-Ordovician aquifer serving PWSs over the last two decades.

Prevalence of Ra in untreated PWS source water across Iowa and the Upper Midwestern region of the U.S. raises concerns about Ra isotopes in private well water sourced from shared aquifer systems and the ability of household water treatment systems to reduce total Ra (Szabo et al., 2012). Private wells are not regulated under SDWA, and the burden of monitoring and treating contaminants in private drinking water wells falls entirely on the consumer. Point-of-entry (POE) and point-of-use (POU) devices can be installed in residences served by private wells, and many types of common household water treatment (e.g. water softening, reverse osmosis) can effectively remove Ra (Bennett, 1978; Brinck et al., 1978; Fox & Sorg, 1987). National Sanitation Foundation International and the American National Standards Institute test and endorse a list of water softeners for reduction of ²²⁶Ra and ²²⁸Ra (NSF International, 2022), but water treatment systems can be expensive for private well users to install, maintain, and repair (Fox & Sorg, 1987; Orloff et al., 2004).

A major challenge to determining elevated risk for ^226/228Ra contamination and treatment needs in private wells is insufficient data on regional aquifer distribution of total Ra concentration in groundwater (Canu et al., 2011). Frequent testing for private wells is needed to discover threats to water quality in untreated groundwater and check performance of POU and/or POE treatment systems (Fox et al., 2016; MacDonald & Tippett, 2020). However, testing for combined Ra isotopes is burdensome and costly for well users because it requires access to expensive and time-consuming analysis by certified public health lab personnel trained in radiochemical techniques. Further, awareness of ^226/228Ra in groundwater is limited among private well users. For example, the Iowa Private Well Tracking System (PWTS) operated by the Iowa Department of Natural Resources (IDNR) reports a total of 148,504 tests for either coliform, fecal coliform, or nitrate conducted between 1989 and 2021, whereas only 22 wells have been tested for ^226/228Ra over this period (Iowa Department of Natural Resources, 2022a).

Chemical surrogates or markers for radium that are easier to analyze have been proposed to overcome this testing burden and prioritize test sites, but there are concerns regarding their effectiveness (Schrag, 2017). For example, radon gas (²²²Rn), which is the daughter product of ²²⁶Ra, could be a chemical marker, but ²²²Rn concentrations are not correlated with groundwater ²²⁶Ra measurements because of natural disequilibria (Rosholt Jr, 1959). Other anionic constituents (e.g., SO₄^2-) or alkaline earth metals concentrations may also serve as possible surrogates for ^226/228Ra contamination because the mobility of Ra is controlled by co-precipitation with common anions found in ground water (Molinari & Snodgrass, 1990; Schrag, 2017). For example, Schrag (2017) proposed elevated Mg and Ca concentrations as a possible predictor, and Brandt et al. (2015) identified non-equilibrium Ra entrapment that should be congruent with SO₄^2- or Ba²⁺ release. However, positive correlations between Ra and these chemical predictors have not been thoroughly studied and should be evaluated through screening of raw groundwater at private well hydrants (Schrag, 2017).

In this work, we analyzed water from 94 unique private wells across ten Iowa counties in the fall of 2019 to determine the prevalence of Ra in private well water. The private wells were selected for sampling because of their proximity to public monitoring systems from the same aquifer that had previously measured elevated Ra activity. A secondary goal of this initial survey was to evaluate potential relationships between combined Ra detection and other, more easily measurable water quality parameters (co-occurring inorganic species) that could be used to better predict ^226/228Ra occurrence in private well water. In subsequent testing between November 2020 and February 2021, we selected a subset of 11 homes to re-sample both the outdoor (well hydrant) and indoor (primary drinking water tap) water to evaluate the efficacy of in-home treatment devices for removal of ^226/228Ra and other commonly encountered inorganic groundwater constituents (i.e. Mn, As, Li and Ba). This work improves our understanding of the occurrence of Ra isotopes and their removal by in-home water softeners, while highlighting currently overlooked risks posed to consumers from other inorganic contaminants less frequently monitored in private wells.

Experimental Methods

Site Selection.

For the initial survey in fall 2019, private wells in ten Iowa counties were identified for sampling based upon available water quality monitoring data from the IDNR. Compliance history for the PWS in the selected counties is summarized in Supporting Information (SI), Table S1. We mapped radium occurrence data from public groundwater monitoring wells and MCL violations in public water systems across Iowa from the last 20 years (SI, Figures S1 and S2). From this assessment, we identified ten counties (Allamakee, Carroll, Cherokee, Emmet, Greene, Johnson, Kossuth, Palo Alto, Woodbury, and Wright) with elevated Ra activity and targeted private wells that likely draw from the same aquifers. County public health departments then assisted with identification of well users willing to participate in sample collection and testing, but participation was voluntary for residents. We note that the sampling locations are clustered due to the voluntary nature of the study and can be attributed to locations of residential neighborhoods within the counties where samples were collected.

Following confirmation of participants, kits containing 1 L HDPE bottles were mailed to each county. Water sampling was conducted by local public health sanitarians in the selected Iowa counties from September to November 2019 from a total of 94 private wells. County public health sanitarians collected samples at either well hydrants (greater than 46% of samples, n = 53) or drinking taps (at least 13% of samples, n = 12). Samples were collected concurrently with yearly nitrate and bacteria samples, which are collected at the location of the well user’s preference. For the remaining samples, sample location was not specified by the sanitarian. Approximately 2 L of sample was collected from each outlet in individual 1 L HDPE bottles. These bottles were shipped to the University of Iowa overnight and acidified to 2% nitric acid (TraceMetal Grade, Fisher Chemical) upon receipt, minimizing hold times between sample collection, shipping, and receiving to less than one week for quality control.

During the re-sampling campaign from November 2020 to February 2021, samples were collected from only four (Carroll, Cherokee, Greene, and Woodbury) of the original ten counties. A 2 L sample was collected from the well hydrant or hose bib and a second 2 L sample was collected at point-of-use from the outlet identified by each well user as the primary drinking water tap for household members. Due to the COVID-19 pandemic, indoor primary tap water samples were collected by study participants (with clear instructions to prevent contamination of samples), and well hydrant samples were collected by sanitarians. Acidified samples were stored for a minimum of 16 hours and maximum of 6 months before analysis for Ra isotopes per U.S. EPA guidelines.

Analytical Methods.

Samples were acidified to a final concentration of 2% nitric acid upon receipt. A few water samples were noted to contain visible particulates, which were dissolved during acidification. Acidified samples were analyzed for ²²⁶Ra and ²²⁸Ra using EPA Method 904.0 and 903.0 (U.S. EPA, 1980a; U.S. EPA, 1980b), but used a combined methodology previously employed by Nelson et al. (2014) that meets EPA guidelines for analysis of radionuclides in water. For quality assurance, a lab control blank and lab control spike were prepared using deionized water acidified to a final concentration of 2% nitric acid and analyzed concurrently with every ten samples. In addition, two matrix spikes and a matrix blank were prepared with tap water acidified to a final concentration of 2% nitric acid. The matrix spikes were prepared with NIST ²²⁶RaCl₂ and ²²⁸Ra(NO₃)₂ standard solutions to a combined Ra activity of 15 pCi/L, which is 30 times the targeted minimum detectable activity for each isotope (Caution! ²²⁶Ra and ²²⁸Ra are radioactive and must be handled by trained personnel in a licensed research facility). Radium measurements were conducted using a gas-flow proportional counter (Model LB4200, Mirion) with count times of 60 minutes. ²²⁶Ra was quantitated via alpha measurement of separated Ra (held for two weeks to allow for ²²⁴Ra decay) and ²²⁸Ra was quantitated via beta measurement of separated ²²⁸Ac, the direct decay product of ²²⁸Ra. Samples with less than 70% carrier recovery for ²²⁶Ra or ²²⁸Ra were re-analyzed using the additional water sample remaining from each site.

From each acidified private well sample, additional aliquots were filtered with a 0.45 μm syringe filter and reserved for additional analysis. Five mL subsamples were prepared in triplicate for total metals analysis by Inductively Coupled Plasma Mass Spectrometry (ICP-MS; Agilent) using a modified version of EPA Method 200.8 (U.S. EPA, 1994). Hold times for samples collected in 2019 (up to eight months) were longer than six months because of laboratory closures in response to the COVID-19 pandemic. During the re-sampling campaign, samples for ICP analysis were held for a maximum of two months before analysis. Quality assurance and quality control information for all analyses are summarized in Table S2 and Table S3.

Well User Survey.

All 94 residences were contacted by mail for a follow-up survey, with an option to return the completed paper survey or submit responses online using Qualtrics XM. The survey included a series of yes/no questions about household water treatment and consumption behavior (Supporting Information, Figure S3). Before mailing, this study was approved by the University of Iowa Institutional Review Board for use of personal data for well users.

Data Analysis and Sources.

Information about the Iowa private wells surveyed in this study was provided by the IDNR PWTS (Iowa Department of Natural Resources, 2022a), including well age, depth, ownership, and historic water quality survey data (SI, Table S4). PWS compliance data and location information was sourced from the Safe Drinking Water Information System (SDWIS). Shapefiles for maps were sourced from the Iowa Open Geospatial Data database (State of Iowa, 2022) and produced with the open-source mapping software QGIS (Version 3.10).

Well Aquifer Assignment.

The aquifer that supplies water to each well was assigned using the best available data. PWTS (Iowa Department of Natural Resources, 2022a) and GeoSam (Iowa Geological Survey, 2022), another well database operated by the Iowa Geological Survey (IGS), contain construction records for some wells. Information in these records, such as casing lengths and geologic materials encountered, helps to determine precisely where water is supplied to a well. PWTS and GeoSam were queried to identify wells with construction records. For wells with records, the aquifer assignment is the one identified using the records. For wells without records, the aquifer assignment was based on the well depth reported in PWTS. Using IGS ArcGIS coverages of depths to aquifers and referencing wells with reliable construction records in close proximity, the most likely aquifer supplying water to the well was assigned. For a few wells, PWTS records did not contain a well depth. In this case, no aquifer was assigned because of insufficient data.

Results and Discussion

Fall 2019 Private Well Study

Water Quality Analysis for Radium.

Radium concentrations were measured for 94 total private wells and analyzed based upon location, collection point, and well-depth. ²²⁶Ra and/or ²²⁸Ra were detected in 44 private wells (46.8% of total samples) (Figure 1), with six wells (6.4%) measuring above 5 pCi/L. Results for all samples are summarized in Table 1 based on the county in which they were located, Table 2 summarizes these results based on the aquifer assignments for each sampled well, and Table 3 provides results for the subset of samples with known collection points (i.e., well hydrant or hose bib versus primary drinking water tap).

Figure 1.

Map of ²²⁶Ra and ²²⁸Ra activity concentrations in 10 Iowa counties: (a) Emmet, Palo Alto, Kossuth, Cherokee, Wright, Woodbury, Carroll, and Greene, (b) Johnson, and (c) Allamakee.

Table 1.

Summary of ²²⁶Ra and ²²⁸Ra activity concentrations in private well samples from Fall 2019 survey of 94 wells based on their distribution across 10 Iowa counties. Error associated with the mean detected combined Ra is the standard deviation of the samples.

County Name	Number of Private Well Samples	Aquifer Distributiona	Number Above MDA	Number Exceeding MCL	Median Detected Combined Ra (pCi/L)	Mean Detected Combined Ra (pCi/L)	Maximum Detected Combined Ra (pCi/L)
Allamakee	4	2 CO; 2 C	1	0	*	*	1.4
Carroll	9	3 D; 3 B; 2 P; 1 U	5	2	3.7	3.6 ± 2.6	6.9
Cherokee	10	3 B; 2 A; 2D; 3 U	4	1	2.3	2.6± 1.9	5.3
Emmet	10	7 B; 2 D; 1 U	3	2	7.7	7.2± 5.1	12.0
Greene	10	6 B; 1 D 3 P	7	0	3.2	3.1± 1.3	4.8
Johnson	11	5 SD; 4 B; 2 U	6	0	1.9	2.3± 1.5	4.9
Kossuth	10	5 B; 1 SD; 1 D; 2; M; 1 U	8	0	1.1	1.2± 0.3	1.8
Palo Alto	10	4 A; 5 B; 1 D	1	0	*	*	3.4
Woodbury	10	5 B; 5 D	4	1	1.7	2.4± 1.9	5.2
Wright	10	9 M; 1 U	5	0	2.2	2.3± 1.2	4.3
All Samples	94		44	6	1.9	2.7± 2.3	12.0

^aDistribution of aquifers sourced for wells in the county where A = Alluvial; CO = Cambrian-Ordivician; C = Cambrian; D = Dakota; B = Buried Sand & Gravel; P = Pennsylvanian; SD = Silurian/Devonian; M = Mississippian; and U = Unknown

^*Insufficient number of samples above MDA.

Table 2.

Summary of ²²⁶Ra and ²²⁸Ra activity concentrations in private well samples from Fall 2019 survey of 94 wells based on the aquifer from which wells were sourced. Included for comparison are historical water quality data for public supply wells. Error associated with the mean detected combined Ra is the standard deviation of the samples.

Aquifer	Number of Private Well Samples	Number (Percent) Above MDA	Number (Percent) Exceeding MCL	Median Detected Combined Ra (pCi/L)	Mean Detected Combined Ra (pCi/L)	Maximum Detected Combined Ra (pCi/L)	Number of Wells	Ra Detections (%)	Exceedances of Ra MCL (%)
Data from this study							Public water supply data (1974–2019)
Alluvial	6	2 (33%)	0	2.2	2.2 ± 1.7	3.4	215	61%	1%
Buried Sand & Gravel	38	13 (34%)	1 (3%)	1.9	2.5 ± 1.8	7.7	109	81%	7%
Dakota	15	10 (67%)	4 (27%)	2.9	4.0 ± 3.5	12.0	110	96%	30%
Pennsylvanian	5	5 (100%)	1 (20%)	4.0	3.4 ± 1.9	5.4	16	100%	19%
Mississippian	11	7 (64%)	0	1.9	2.1 ± 1.0	4.3	62	95%	15%
Silurian/Devonian	6	4 (67%)	0	1.0	1.2 ± 0.4	1.8	202	77%	11%
Cambrian-Ordivician	2	0	0	*	*	*	250	98%	62%
Cambrian	2	1 (50%)	0	*	*	1.4	13	100%	46%
Unknown	9	2 (22%)	0	3.1	3.1 ± 2.6	4.9

^*Insufficient number of samples above MDA.

Table 3.

Summary of ²²⁶Ra and ²²⁸Ra activity concentrations in private well samples from Fall 2019 survey compared by known sampling location. Error associated with the mean detected combined Ra is the standard deviation of the samples.

Sample Location	Known Number of Samples	Samples Above MDA	Number of Samples Exceeding MCL	Median Detected Combined Radium	Mean Detected Combined Radium	Range, Above MDA
Collected at Well Hydrant or Hose Bib	53	41	5	1.9	2.8 ± 2.3	0.7 – 12.0
Collected at Primary Drinking Water Tap	12	3	1	3.4	*	1.4 – 5.2

^*Insufficient number of samples above MDA.

Forty-one of the 44 samples with detectable Ra²⁺ were collected at the well hydrant, which precedes any household water treatment that may be used by the homeowner. Of the six samples measuring above 5 pCi/L, only one sample was collected at a primary drinking water faucet. Household water treatment systems were not identified during sample collection; thus, it is unknown if this sample was untreated or treated water. Well depth was recorded in the PWTS for 86 of the wells included in this study and ranged from 12 feet to 560 feet (SI, Table S2). Seventeen samples contained both ²²⁶Ra and ²²⁸Ra, but there was no apparent correlation in their relative abundance (SI, Figure S4, linear regression R² = 0.14). Forty-one samples above the MDA (minimum detectable activity) for ²²⁶Ra and/or ²²⁸Ra were matched to a recorded well depth, but no apparent association between well depth and combined Ra was observed in these samples (SI, Figure S5, linear regression R² = 0.11 for entire dataset), even when limiting analysis to specific aquifer units.

All wells exceeding the MCL for Ra²⁺ in this sampling were located in northwest Iowa (Figure S6), drawing from the Dakota (4 exceedances out of 15 wells), Pennsylvanian (1 exceedance out of 5 wells), and buried sand and gravel (1 exceedance out of 38 wells) aquifers (Table 2). Radium was detected in all aquifer units tested except the Cambrian-Ordovician, which may be due to the limited number of sampled wells that draw from that aquifer (n = 2).

The relative abundance (based on exceedances) and occurrence of Ra²⁺ in the sampled private wells are generally consistent with expectations from the lithology and historical water quality data associated with the aquifers being sourced (see Table 2). We note that in Table 2, historical radium data was obtained from the “SourceWaterQuality” file publicly available at the State of Iowa Open Geospatial Data database (State of Iowa, 2022), which contains radium results from public water supply wells from 1974–2019. For example, radium is derived from the radioactive decay of naturally occurring uranium and thorium, which are common in granitic and metamorphic crystalline rocks and in associated weathered sedimentary deposits in the central US (Szabo et al., 2012). These types of rock fragments are commonly found in the sands and gravels that constitute alluvial and buried sand and gravel aquifers in Iowa. Notably, the percentage of wells with radium detections in alluvial and buried sand and gravel in this study was significantly lower than that found in the historical data. We speculate that the buried sand and gravel aquifers in this study may be shallower than those in the historic data and, therefore, have younger water that is lower in radium. In contrast, older groundwater from deeper wells that has been in contact with the geologic parent materials longer will have more time to accumulate radioactive decay products (Rowden, 2008).

The Dakota aquifer produces water generally from sandstones, and the Pennsylvanian produces from sandstones and limestones. Both aquifers, however, also contain layers of shale, a sedimentary rock that can be enriched in uranium (Rowden, 2008). All but one well with a ^226/228Ra MCL exceedance were open to either the Dakota or Pennsylvanian aquifer. In fact, the percentage of wells in the Dakota and Pennsylvanian aquifers that exceed the MCL in this study (27% and 20%, respectively) correspond closely to the percentage of exceedances in the historic wells (30% and 19%, respectively).

The Mississippian and Silurian/Devonian aquifers consist mostly of limestones and dolomites, which are generally considered to contain little uranium (Bell, 1963). However, radium was detected frequently in this work (64% and 67%, respectively) and historically (95% and 77%, respectively) in these aquifers, but not typically at levels that exceed the ^226/228Ra MCL. It is known that radium can be detected in aquifers that do not have a concentration of radioactive elements in the rocks. For example, chemical processes in groundwater and on rock surfaces, such as mineral dissolution, desorption, and ion-exchange reactions, can influence the release of radium from aquifers (Szabo et al., 2012). Oxygen-poor water has also been associated with high radium concentrations (Szabo et al., 2012), and many wells in the Mississippian and Silurian/Devonian aquifers are oxygen poor, with dissolved oxygen measurements less than 1 mg/L.

In this study, only 4 wells were tested from the Cambrian-Ordovician and the Cambrian aquifers (2 wells each), and the percentages of radium detections in these samples was far lower than those in the historical data. Indeed, wells in the Cambrian-Ordovician and Cambrian aquifers are the most likely historically to exceed the ^226/228Ra MCL (see Table 2). A recent study of the Cambrian-Ordovician aquifer in the north-central US found that water that was recharged into the aquifer long ago, contains greater amounts of dissolved minerals, and is low in dissolved oxygen is more likely to leach radium from its surrounding rock (Stackelberg et al., 2018). The Cambrian aquifer is the deepest aquifer in the state, and relatively few wells penetrate the aquifer. Because of the depth, the Cambrian aquifer across much of Iowa is associated with higher dissolved minerals, low dissolved oxygen, and water that was recharged into the aquifer long ago. We note that the Cambrian-Ordovician and Cambrian aquifers samples in the study were collected in Allamakee County in northeast Iowa, where both aquifers are closer to the surface. Consequently, recharge to the aquifer in this location occurs faster, which may explain the fewer detections and exceedances of the ^226/228Ra MCL.

Water Quality Analysis of other Constituents.

Other water quality parameters, for example high total dissolved solids (which were not measured in the present study) within the Dakota Aquifer (Burkart, 1984), have been suggested as a concurrent metric for elevated Ra²⁺ concentrations. To explore possible chemical surrogates that could be screened more easily to assess Ra contamination, all well samples from the fall 2019 sampling were also screened for total metals. We initially evaluated concentrations of alkaline earth cations as possible predictors for Ra contamination and results for Mg, Ca, Sr, and Ba analysis in well samples are summarized in SI, Table S5 (based on county) (Schrag, 2017). Mg and/or Ca were detected in every private well sample, excluding two private wells in Allamakee County taken at point-of-use outlets, possibly after unknown treatment processes. Among known raw groundwater samples collected in northwest Iowa counties (n = 53), Mg and Ca concentrations range from below detection to up to 60 mg/L Mg and 180 mg/L Ca. Concentrations of Sr and Ba were most frequently detected at levels below 1 mg/L, although some wells measuring between 1–2 mg/L for both elements were identified across sites in northwestern Iowa counties. These samples are all identified as well hydrant samples taken before any in-home water treatment processes. There were weak to no associations between combined Ra and alkaline earth metal concentrations (SI, Figure S7, R² values for linear regressions < 0.20).

The distributions of Mg, Ca, Sr, and Ba were also examined as a function of the aquifer source for each well (Figure S8). For example, by far the highest levels of Sr were found in wells drawing from the Pennsylvanian aquifer, followed by buried sand and gravel wells whose distribution of Sr was more variable. Higher levels of barium, on the other hand, were found in wells drawing from the Mississippian, Silurian-Devonian, and those wells for which we were unable to confidently assign an aquifer. Hardness causing ions were relatively consistent across all aquifers, with Ca levels typically exceeding Mg. We caution, however, that the number of sampled wells sourced from certain aquifers (e.g., Cambrian and Cambrian-Ordivician) is small.

Concentrations of ²³⁸U (²²⁶Ra progenitor) were also evaluated but found to be non-detectable at most sites (n = 70) (SI, Table S6). A cluster of private wells sampled in Cherokee County approached or exceeded the EPA MCL of 30 μg/L for total U, with values ranging from 30 to 70 μg/L (n = 4). One sample above 30 μg/L was collected from a well hydrant, whereas no information was provided for the remaining three private wells. ²²⁶Ra and/or ²²⁸Ra were not detected at any of these sites, but this finding may simply reflect well-established differences in the geochemical environments that favor Ra and U mobility (Mathews et al., 2018; Porcelli & Swarzenski, 2003; Stackelberg et al., 2018; Szabo et al., 2012; Vengosh et al., 2022). Although elevated U concentrations in groundwater are generally not of concern in major aquifer systems throughout Iowa (Smith & Field, 2007), our analysis found U levels to be greatest in alluvial and buried sand and gravel wells, in addition to those wells for which we had insufficient information to assign an aquifer (Figure S9).

Occurrence of Ra²⁺ relative to As and Mn was also assessed due to concern for private well water quality in the region (Schnoebelen et al., 2017). There is high awareness of As issues among private well users in Cerro Gordo county, which neighbors Kossuth and Wright counties, because of past work and outreach conducted by Cerro Gordo County Public Health (Miao & Fry, 2018; Schnoebelen et al., 2017). Other major inorganic groundwater constituents (Cr, Fe, Ni, Co, Cu, Zn, Cd, Pb) were also measured by EPA Method 200.8 in fall 2019, but none were detected at concentrations above recognized health limits (i.e., U.S. EPA MCL or Health Advisories).

Elevated total As and Mn concentrations (i.e., above the As MCL and Mn Health Advisory) were detected in several samples during fall 2019 (see a summary of Mn and As results in SI Table S6). For 50 wells with detectable As, 16 exceeded the EPA MCL of 10 μg/L, but none of these samples were collected at a known drinking water outlet. For a subset of eight sites measuring above the MCL, information on sample collection location was not provided following sample shipment. Levels of As were greatest in buried sand and gravel wells, as well as those sourced from the Dakota and Mississipian aquifers (Figure S9). Mn was detected in 80 private wells, with 57 samples exceeding the EPA secondary MCL (SMCL) of 50 μg/L. Further, this subset of SMCL exceedances included 21 wells with Mn concentrations above the EPA Health Advisory of 300 μg/L (U.S. EPA, 2004). Nine of these samples with exceedances were taken at the well hydrant, but the collection location is unknown for the other 12 exceedances. Moderate to high levels of Mn were found across alluvial, buried sand and gravel, Dakota, Mississippian, and Pennsylvanian wells (Figure S9), with the highest levels in water drawn from alluvial, buried sand and gravel and the Pennsylvanian aquifers. We note that our As and Mn results agree well with analyses conducted on samples taken concurrently at the same outlets during the Iowa Well Survey conducted by the State Hygienic Laboratory at the University of Iowa in 2019 (see SI Table S7, which compares results to available data in the Iowa PWTS).

Collectively, the incidence of As, Mn, and U in raw groundwater is notable because naturally occurring inorganic contaminants associated with bedrock geology are undertested within private well monitoring in Iowa. Iowa’s Grants to Counties Water Well program has supported testing for nitrate, total coliform, and fecal coliform, because of the acute health risks in drinking water. In 2016, the program began to cover costs for As testing in private wells, either at the user’s hydrant or primary drinking water tap. Based on fall 2019 survey results presented herein, point-of-use concentrations of both naturally occurring and anthropogenic inorganic and radiological contaminants may require additional yearly monitoring to better understand actual exposure risks from these species for private well users.

Water Use and Behavior Survey.

Concurrent to water sample collection and analysis in fall 2019, a survey was developed and distributed to study participants to better assess their home drinking water source and water use behavior. Thirty well users (out of the 94 residences; 31.9%) responded to the questionnaire, which was disseminated with results mailings for the fall 2019 study (survey responses are summarized in SI, Table S8). All respondents stated they and other occupants of their household drink, cook, and bathe with water from the private well on their property that was sampled in the fall 2019 study. No respondents reported behavior such as purchasing bottled water to avoid using private well water. Of the 30 respondents, 28 (93%) users indicated they use a water softener to treat drinking water, without specification if the softener was connected to only a hot water line or both hot and cold plumbing on their primary drinking water tap. In addition to water softening, a subgroup of 13 well users (43%) used a second treatment device, including specific whole-home filter models and reverse osmosis systems. Of this subset, five well users employed point-of-use or pitcher filters for drinking water.

The questionnaire results indicate that, among survey respondents, consumers widely rely on their private well water as their primary drinking water source, and more than half of these consumers use the water without any additional treatment beyond softening. For in-home treatment, potential risks from drinking water exposure will depend on the efficacy of these units for contaminant removal. These survey results reinforce the need for more comprehensive characterization of private well water quality, including radium, arsenic or manganese, and emerging chemical threats at the well water source and primary tap used for drinking and cooking. To explore the relationships between specific contaminant reduction and in-home treatment in more detail, we conducted a re-sampling campaign based upon initial Ra detection within the private well samples.

Winter 2020–2021 Private Well Study

Site selection for the 2020 re-sampling study was voluntary and twelve households from the fall 2019 study consented to be resampled for Ra isotopes approximately one year after the original campaign. These 12 households are located in Carroll, Cherokee, Greene, and Woodbury counties, and radium results from the Winter 2020–2021 campaign generally were consistent with data from the Fall 2019 sampling (comparison data provided in SI, Table S9).

Exact water treatment systems employed by these well users were unknown until self-reported by 2020–2021 study participants (information requested shown in SI, Figure S10). Self-reporting revealed that knowledge of water softener characteristics, salt type, and treatment system operation was limited among households sampled at a primary drinking water outlet (n = 11). For example, only two well users identified that sodium chloride was used in their water softeners. One well user also specified use of a commercially available reverse osmosis system (Culligan Water^®) in addition to a water softener, with both systems used to treat cold and hot lines before the primary drinking tap. No participants in this study provided information identifying resin type or commercial brand name for their water softening systems. Access to a primary drinking water outlet was not available for one resampled well (labeled as Site X); thus, sample collection occurred only at the well hydrant.

Figure 2 compares the concentrations of Ra measured at the well hydrant and at the primary drinking water tap for each re-sampled site. Combined Ra measured above the MCL for three of the 12 raw groundwater samples collected, including at Site X. In two sites (F and I), Ra was reduced from levels exceeding the MCL at the hydrant to below detectable activity at the tap after water softening. Two other primary point-of-use taps registered activity above MDA for Ra (Site B with no in-home water treatment and Site H with hot and cold water softening), but these tap water concentrations measured <50% of the MCL for combined Ra. In locations where there were detectable levels of Ra at the hydrant, all but one (Site B, which used no in-home treatment) showed statistically meaningful reduction in Ra levels at the primary drinking tap. In total, seven sites measuring detectable Ra at the hydrant demonstrated removal to non-detectable activity following treatment. No sites measured above the combined Ra MCL at the primary tap.

Figure 2.

Summary of ²²⁶Ra and ²²⁸Ra activity concentrations with one standard deviation of the mean of duplicate samples shown as error in paired private well hydrant and primary drinking tap samples during Winter 2020 to 2021.

Greatest reductions in total radium concentrations from hydrant to tap were observed when water softening or reverse osmosis was used for treatment relative to cases with no treatment. Water softening, particularly with zeolite-based (Clinoptilolite) systems, was shown by Fan et al. (2016) to be effective at removal of Ra over other alkali-earth cations. Zeolitic systems are more selective over ion exchanges resins, with Ba identified as the major competitor for absorption sites. Reverse osmosis is documented as a best available technology for Ra reduction in household treatment systems but has not been as widely studied as water softening (Association, 2013b; Sorg et al., 1980). We note that while these treatment technologies are likely responsible for most Ra removal, we also cannot rule out some incidental loss due to pipe corrosion chemistry. Radium absorbs to iron and manganese oxides and will co-precipitate in calcium carbonate or sulfate pipe scales (Valentine & Stearns, 1994). Field et al. (1995) demonstrated that pipe scale taken from a well in Delta, Iowa contained 10 Bq/g ²²⁶Ra, whereas concentrations in the water averaged 9.3 × 10⁻⁵ Bq/mL. However, Fisher et al. (2000) noted that the stability of these absorbed or co-precipitated solids can also be an important consideration, as dissolution or dislodgement can lead to higher concentrations of Ra at the drinking water source, which could explain the higher total radium concentration at the Site H tap relative to the hydrant.

As in the fall 2019 survey, additional constituents measured by EPA Method 200.8 on ICP-MS were evaluated to assess their occurrence and the efficacy of in-home water treatment for their reduction at the tap. For corrosion derived metals (e.g., Pb, Cu, Fe), it also allowed us to explore their possible addition to drinking water from premise plumbing components because private wells fall outside of regulations under SDWA and the EPA Lead and Copper Rule (U.S. EPA, 2022), which helps to assess distribution and premise plumbing stability for consumers served by community water systems. For these purposes, 14 distinct metals and metalloids (Li, Na, Mg, Ca, Sr, Ba, As, Se, Mn, Fe, Cu, Zn, Pb, U) were compared in paired well hydrant or hose bib samples with samples collected at a primary drinking water faucet (Figure 3). A summary for 11 of the 14 elements tested including their associated health advisories or regulatory limits for drinking water and any measured exceedances is presented in SI, Table S10. Table S10 excludes results for Mg, Ca, and Sr because no health-based standards, aesthetic guidelines, or recommended intake levels are associated with these constituents.

Figure 3.

Summary of major inorganic contaminant concentrations in paired private well hydrant or hose bib samples and primary drinking tap samples with one standard deviation of the mean of duplicate samples shown as error during winter 2020 to 2021.

Alkali-earth cations were of interest due to their specific removal by water softening systems. From Figure 3, Ca and Mg removal is evident for sites with a water softener supplying cold and hot water, with reductions from 1290–5330 μg/L Mg and 3590–1200 μg/L Ca in source water to 35–620 μg/L Mg and 120–1180 μg/L Ca in treated water. Sr reduction was also observed following water softening for these sites. Although Ba levels were not measured above the MCL (2 mg/L), it was not consistently removed compared to other alkaline earth metals after softening. Notably, any reductions in these metals were paired with demonstrable increases in sodium concentrations at the primary drinking water tap at sites where a water softener is the sole treatment (Sites D, F, G, and I). Among these samples, seven homes exceeded the 20 mg/L recommendation from U.S. EPA for consumers following low-sodium diets (U.S. EPA, 2003).

Persistence of Ba in softened water is notable as it is expected to be removed during water softening (Water Quality Association, 2013a). The affinity series for water softening can generally predict the removal of cations during treatment, but the actual performance in home depends upon factors including the type of resin and salt used in the water softening system (Wachinski, 2006).⁵⁰ Within strong acid cation exchangers, such as sulfonic acid resins, the affinity for the alkali earth cations is Ba²⁺ > Sr²⁺ > Mg²⁺> Ca²⁺ (Hubicki & Kołodyńska, 2012). For these resins, binding increases with higher charge (Ca²⁺ > Na⁺) and increasing atomic number for cations of the same charge (Fan et al., 2016). Weak acid cation resins utilize carboxylate ligands where the affinity trend is reversed from the strong acid resins; smaller alkali earth cations (e.g., Mg²⁺ and Ca²⁺) would be removed preferentially over the larger ones, such as Ba²⁺ or Ra²⁺ (Sorg et al., 1980). Some zeolitic systems, such as clinoptilolite, have been specifically designed for selective capture of the larger alkali-earth cations (Dyer et al., 2018; Fan et al., 2016). Because Ba²⁺ was found to persist through treatment and some raw groundwater and primary tap samples measured above 1 mg/L (the MCL for Ba²⁺ is 2 mg/L), Ba²⁺ concentrations may require additional monitoring in these private wells going forward. Further, additional research may be warranted to better understand why Ba²⁺ is not removed, as expected, by these water softening units.

In-home water treatment also resulted in other notable differences in elemental concentrations between the sample taken at the well hydrant versus primary tap. For most sites with Mn concentrations exceeding the SMCL (50 μg/L) and/or HA (300 μg/L) in raw groundwater, levels were reduced to below the SMCL at the primary drinking water tap, suggesting the formation of insoluble Mn during water treatment. A decrease in Mn concentration was not observed at Site B (no additional water treatment system). Site A also lacked household water treatment system, but a decrease in Mn from 560 μg/L to 25 μg/L was measured from the hydrant to the primary tap. Site A also exhibited Cu concentrations increases in the primary tap compared to the well hydrant, but Zn and Pb changes were negligible. The shifts in Mn and Cu may be attributed to deposition and scale formation in premise plumbing (Gerke et al., 2016), which was the only site to have evidence of corrosion and exchange with plumbing components. At other sites, high Fe, Cu, Zn, and Pb observed at the well hydrant in some samples may be attributable to outdoor pipe components instead of groundwater sources.

Paired outdoor and indoor sampling also identified other recognized health risks to private well consumers regulated under SDWA. Elevated As and Se above the respective MCL values co-occurred at Sites E, G, and I. These sites utilized hot and cold water softening and we observed no change in the As or Se concentrations between the hydrant and tap. This is consistent with both As and Se occurring predominantly as anions at pH values typical of groundwater, thus their recalcitrance to treatment by cation exchange. Uranium concentrations above the MCL were also detected in paired samples at site F and, notably, this well is among three sites in this case study with combined Ra measuring above 5 pCi/L. For all sites with U above detection at the well hydrant, little to no reduction was observed following water softening. We note that the persistence of As and U between the hydrant and tap in homes with only water softening devices is consistent with prior work showing that cation-exchange does not remediate uranium (Lee & Bondietti, 1983) or arsenic (Spayd et al., 2015; Yang et al., 2012).

While Li was not initially analyzed in the 2019 survey, it was measured in re-sampled homes because of growing concerns of its presence in U.S. groundwater supplies (Lindsey et al., 2021). Lithium was detected at the hydrant at all sites and no removal was observed by water softening, excluding Site K that incorporates both water softening and RO as in-home water treatment. Measurements of Li were approximately equivalent within one standard deviation between well hydrant and tap. Six sites exceeded 10 μg/L Li, a health-based screening level based on provisional toxicity values, at the tap (Lindsey et al., 2021; U.S. EPA, 2008). Levels of Li in these samples suggest that additional monitoring is necessary for private wells to ensure that emerging public health concerns are addressed for homeowners.

Public Health Implications and Insights for Private Well Water Monitoring

From this work, it may be possible to better identify private wells vulnerable to Ra contamination if sufficient information is available to allow identification of the primary aquifer from which the well draws water. We found that Ra occurrence and exceedances, for the most part, follow expectations from the lithology and screened well zone. These findings support recent regional studies in the Midwestern U.S that have also noted the importance of local geochemistry on Ra occurrence in groundwater (Dematatis et al., 2020; Stackelberg et al., 2018).

In the absence of a clear chemical surrogate for Ra, the best course of action for local public health officials may be to prioritize Ra testing for those private wells known to draw water from aquifers with historical Ra occurrence. A major barrier to this approach, however, will be the availability of reliable information about private well construction to allow an accurate aquifer assignment. In Iowa, available databases contain some data, but data is often not available for all wells, and there can be concerns about data quality. Aquifer assignments herein required extensive work by trained IGS professionals to make such aquifer assignments with reasonable confidence. This study, therefore, highlights the need for better data collection and sharing for private wells to enable effective public health interventions (e.g., targeted testing campaigns). For existing private wells that lack such data, there may be an opportunity for public health officials to engage local well drillers, who are a likely source of reliable information for wells in their area.

In addition to targeted monitoring guided by private well construction and public water supply data, another approach to limit Ra exposure would be to prioritize effective and well-maintained in-home treatment systems. With the additional sampling campaign in winter 2020–2021, we noted that Ra levels decreased from the well-hydrant to the primary tap because of water softening and/or reverse osmosis. With these devices, homes that exceeded the EPA MCL for total radium at the hydrant were found to be below the detectable Ra threshold at the tap.

We also evaluated other contaminants of concern for public health. Both arsenic and manganese contamination is elevated for some private well users in northwest Iowa, and more frequent testing should be considered to evaluate the scale of contamination in private wells. In Iowa, As has been included in the standard battery of GTC well water tests since 2016 (Shotwell, 2016), and the most recent statewide survey in 2019 sampled every participating well for Mn (Iowa Department of Natural Resources, 2022b). In addition, Se, U, and Li should be considered emerging contaminants of concern for well users because they will require specialized treatment beyond water softening to adequately lower concentrations to acceptable levels. Finally, although softeners play an important role in managing Ra levels at the tap, this comes at the expense of elevated Na concentrations, which were frequently measured at levels well above those recommended for consumers following low-sodium diets.

Our hope is that this work will provide valuable lessons to help to strengthen the quality of future studies on private well water quality. For example, this work illustrates some of the challenges encountered when working with private well users, including enrolling participants, access to in-home sampling (especially during COVID-19), and limited reliable information about source water characteristics, well properties, and in-home treatment systems. We found that county public health sanitarians were effective at identifying participants and collecting samples, but their involvement limited data collection (e.g., sample pH and DO measurements) in the field due to lack of appropriate equipment and time while unforeseen communication barriers resulted in incomplete information about sample collection location (e.g., well hydrant versus primary tap in the fall 2019 campaign).

This work should also motivate more high-resolution studies of naturally occurring radiological and inorganic contaminants in private well water both at the hydrant and primary drinking tap. Given the highly localized contamination patterns observed herein, it is important that more comprehensive testing for private wells be performed. In turn, increased water sampling will require more investment in testing and technical assistance to well users to better understand patterns of contamination. It is also important to evaluate the impact of the premise plumbing and the treatment strategy, including performance across the lifetime of the treatment unit, to fully assess the public health risks of such contaminants in drinking water. Overall, it is our recommendation that state and federal authorities should address increased testing of private wells via a holistic approach that: prioritizes sampling based upon knowledge of localized groundwater quality issues; acknowledges how premise plumbing issues can affect the quality of well water; and emphasizes appropriate operation and maintenance of in-home treatment devices, including widely used water softeners that are not typically considered by consumers for their ability to improve water quality beyond removal of hardness causing ions.

Research Impact Statement:

This study highlights importance of associating primary aquifer data with private wells to identify vulnerability to radium contamination and use of in-home water treatment for effective remediation.

Data Availability Statement:

The data that support the findings of this study are available from the corresponding author upon reasonable request.