Exploring matrix effects and quantifying organic additives in hydraulic fracturing associated fluids using liquid chromatography electrospray ionization mass spectrometry

Marika Nell and Damian E. Helbling *
School of Civil and Environmental Engineering, Cornell University, 220 Hollister Hall, Ithaca, NY 14853, USA. E-mail: damian.helbling@cornell.edu; Fax: +1 607 255 9004; Tel: +1 607 255 5146

Received 24th March 2018 , Accepted 15th May 2018

First published on 17th May 2018

Hydraulic fracturing (HF) operations utilize millions of gallons of water amended with chemical additives including biocides, corrosion inhibitors, and surfactants. Fluids injected into the subsurface return to the surface as wastewaters, which contain a complex mixture of additives, transformation products, and geogenic chemical constituents. Quantitative analytical methods are needed to evaluate wastewater disposal alternatives or to conduct adequate exposure assessments. However, our narrow understanding of how matrix effects change the ionization efficiency of target analytes limits the quantitative analysis of polar to semi-polar HF additives by means of liquid chromatography electrospray ionization mass spectrometry (LC-ESI-MS). To address this limitation, we explored the ways in which matrix chemistry influences the ionization of seventeen priority HF additives with a modified standard addition approach. We then used the data to quantify HF additives in HF-associated fluids. Our results demonstrate that HF additives generally exhibit suppressed ionization in HF-associated fluids, though HF additives that predominantly form sodiated adducts exhibit significantly enhanced ionization in produced water samples, which is largely the result of adduct shifting. In a preliminary screening, we identified glutaraldehyde and 2-butoxyethanol along with homologues of benzalkonium chloride (ADBAC), polyethylene glycol (PEG), and polypropylene glycol (PPG) in HF-associated fluids. We then used matrix recovery factors to provide the first quantitative measurements of individual homologues of ADBAC, PEG, and PPG in HF-associated fluids ranging from mg L−1 levels in hydraulic fracturing fluid to low μg L−1 levels in PW samples. Our approach is generalizable across sample types and shale formations and yields important data to evaluate wastewater disposal alternatives or implement exposure assessments.

Environmental significance

The complex matrix of hydraulic fracturing (HF) associated fluids has limited the applicability of electrospray ionization-based analytical techniques for quantitative analysis of polar to semi-polar chemical additives. Improved understanding of the concentrations of various analytes in HF-associated fluids is an essential prerequisite to evaluate wastewater disposal strategies or assess the environmental risk of contamination events or spills. We systematically evaluated matrix recovery factors for seventeen priority HF additives and applied them to provide the first known quantification of several HF additives in HF-associated fluids. Our approach allows us to overcome the uncertainties associated with complex matrices and can be generalized to other wastewater samples across wells and shale formations.


The use of hydraulic fracturing (HF), coupled with horizontal drilling, has led to a boom in unconventional shale gas production over the course of the past decade. For example, as the United States (US) sought to become a natural gas exporter, hydraulic fracturing played a critical role—already in 2015, hydraulically fractured wells accounted for 67% of all US natural gas production.1,2 However, concerns about the environmental and human health impacts of HF remain.3,4 In the HF process, hydraulic fracturing fluid (HFF), which is a mix of makeup water (MW – i.e., surface water, groundwater, or recycled wastewater),5 a proppant, and up to two percent chemical additives, is injected into a well at high pressure and temperature to increase the permeability of the target formation. When pressure is released from the well, a mix of geogenic brine and HFF returns to the surface as flowback water (FW). Over time, this wastewater will continue to flow from the well as produced water (PW) and its matrix will more closely resemble the geogenic brine sourced from the formation porewater, although there is no clearly defined point at which FW turns to PW.6,7 While relatively few HF additives are used to fracture any single well, over one thousand HF additives have been disclosed including biocides, corrosion inhibitors, and surfactants.7 A review of these additives found that up to 37% could have endocrine disrupting effects and 25% could have mutagenic or carcinogenic affects, highlighting the need for both toxicological studies and the ability to determine the level of exposure in the event of environmental contamination.8 Because contamination can occur at any point in the HF process, it is critical to establish quantitative analytical methods that detect a broad range of contaminants of concern in all HF-associated fluids including MW, HFF, FW, and PW.

To date, most analytical methods developed and applied to characterize the organic composition of HF-associated fluids have utilized gas chromatography mass spectrometry (GC-MS) to quantify hydrophobic organic constituents that are less likely to be persistent and mobile in groundwater or surface water.9–11 Very few studies have focused on quantification of polar to semi-polar HF additives and their transformation products that are likely to be more relevant for water quality. Liquid chromatography electrospray ionization mass spectrometry (LC-ESI-MS) offers sensitive and accurate analysis of polar to semi-polar analytes in water samples and is expected to play an important role in improving our understanding of the chemistry of HF-associated fluids. A few studies have used LC-ESI-MS to detect semi-polar to polar additives in HF wastewater samples,12–17 though all have been qualitative or semi-quantitative in nature. This limits our ability to extrapolate the data for toxicity studies or exposure assessments.

One major limitation of applying existing LC-ESI-MS methods for the quantification of HF additives in environmental samples is the complex and changing matrix of HF-associated fluids, which can lead to complicating matrix effects in ESI-based analyses.16,18 Matrix effects can lead to enhanced or suppressed ionization of target analytes, and there are at least two important ways in which matrix effects can limit chemical analyses in HF-associated fluids. First, inorganic or organic matrix constituents that co-elute with target analytes may enhance or suppress the ionization of the target analytes. For example, surfactants can dominate the surface of droplets formed in the ESI source, enhancing their ionization and detection, but suppressing the ionization of other co-eluting analytes.19–21 Second, complex matrix chemistry can affect the ways in which a target analyte is ionized, including the types of adducts which may be formed during ESI. For example, an analyte that predominantly forms protonated adducts [M + H]+ during ESI when present in a clean water matrix may form a disproportionate amount of sodiated adducts [M + Na]+ during ESI when present in saline FW or PW samples.13 Sodiated adducts pose a problem for typical methods of quantification because they do not fragment as well as protonated adducts.15

Matrix effects have confounded environmental analytical chemistry for decades and a variety of techniques have been developed to account for matrix effects in LC-ESI-MS analyses.19,22–24 The most widely used technique is the addition of isotope labeled internal standards (ILISs).25 If an ILIS can be acquired or synthesized for each target analyte, then any matrix effect experienced by the analyte will also be experienced by the ILIS and appropriate corrections can be made during quantification. Unfortunately, very few ILISs are available for polar to semi-polar HF additives. Another approach is to prepare matrix-matched calibration curves.26,27 Whereas this is an appropriate and effective technique, it is generally only applied across samples with uniform matrices and may require additional preparation to create a matrix blank that does not contain any of the target analytes.28 The matrix of FW and PW varies from well to well and evolves over time, making preparation of matrix-matched calibration curves impractical and cost-prohibitive.6,29,30 Finally, standard addition and the calculation of matrix recovery factors (MRFs) have been used to quantify analytes by relating responses obtained in, for example, calibration curves measured in one matrix to responses obtained in samples with a more complex matrix.16,22,23,31 MRFs calculated for a given analyte in a particular matrix have the added benefit of providing a metric by which the mechanisms of enhanced or suppressed ionization efficiency can be carefully examined.

The objectives of this study were to (i) evaluate the LC-ESI-MS acquisition parameters for a diverse set of polar and semi-polar HF additives, (ii) explore how matrix chemistry influences the ionization behavior of each of the additives through a modified standard addition approach, (iii) screen for the occurrence of HF additives in MW, HFF, FW, and PW samples, and (iv) apply MRFs to quantify the HF additives in the MW, HFF, FW, and PW samples. We collected field samples from two unconventional shale gas wells in Morgantown, WV. Our approach elucidates the ways in which matrix chemistry influences the ionization behavior of certain types of HF additives and enables the first known quantification of several priority HF additives in field samples by means of LC-ESI-MS.


Standards and reagents

We selected nineteen HF additives or likely transformation products based on their amenability to LC-ESI-MS analysis,12 their identification as additives of concern,4,32 or their inclusion in the FracFocus chemical disclosure for the sampling location. Compound names, compound uses, CAS numbers, and chemical structures are provided in Table ESI1 of the ESI. We acquired twelve of the HF additives as individual compounds of varying purity and seven as homologous mixtures containing varying numbers of individual homologues (e.g., polyethylene glycol, PEG). See Tables ESI2 and ESI3 for details on suppliers and purities of individual compounds and homologous mixtures, respectively. We prepared stock solutions of each individual compound or homologous mixture at a concentration of 1 g L−1 in LC-MS grade methanol (Omnisolv, VWR) or nanopure water (produced by a Milli-Q system, EMD Millipore). We then used these stock solutions to prepare standard solutions and mixtures of all HF additives in nanopure water at 1 mg L−1 and 100 μg L−1. We stored all standard solutions and mixtures at −20 °C.

Compound tuning

We optimized MS acquisition parameters for each individual HF additive by direct infusion into a quadrupole-Orbitrap mass spectrometer (QExactive, ThermoFisher Scientific). We used a syringe pump to deliver 20 to 50 μL min−1 of each standard solution ranging from 100 μg L−1 to 10 mg L−1 to a tee connection receiving 50 μL min−1 to 200 μL min−1 of mobile phase (90% LC-MS grade water and 10% LC-MS grade methanol, both augmented with 0.1% formic acid by volume) and connected directly to the ESI source. We identified the optimal polarity and the exact mass of the dominant ion or adduct (e.g., [M + H]+, [M + Na]+, etc.) in full scan MS acquisitions in positive and negative polarity modes. We identified the optimal collision energy for MS/MS fragmentation and the exact masses of the dominant MS/MS fragments for each additive under a range of normalized collision energies (NCEs). We defined the optimal collision energy as the value that resulted in the highest total intensity of unique MS/MS fragments. Identifying optimal collision energies for homologous mixtures required chromatographic separation. We acquired these values using the LC-ESI-MS gradient method described below and acquired MS/MS data using NCE values ranging from 15 to 90.

Sample collection

In collaboration with the West Virginia Water Research Institute and the Marcellus Shale Energy and Environment Laboratory (MSEEL), we collected fourteen water samples from two unconventional shale gas wells in Morgantown, WV. The wells are known as MIP 5H and MIP 3H33 and both are horizontal wells drilled from the same pad. MIP 5H was hydraulically fractured on 11/6/15 and MIP 3H was hydraulically fractured on 11/9/15. We collected seven water samples related to each well, including a sample of the MW and HFF from the day the well was completed, a sample of FW from the day pressure was released from the well, and four weekly samples of PW from the first month that the wells were producing. The MW was sourced from the Monongahela River, although any water that was stored from drilling may have been mixed with the Monongahela River water. We collected all samples in 1 L TraceClean amber glass bottles (VWR). A complete listing of samples, sample dates, and corresponding sample metadata can be found in Table ESI4. We stored the samples at 4 °C for up to one month before they were shipped to our laboratory, where we then stored them at −20 °C until analysis.

Sample preparation

Prior to all experiments and analyses by LC-ESI-MS, we thawed raw samples, adjusted their pH to 9.8–10 using ammonia, filtered them with 0.45 μm PTFE filters13 (Restek), adjusted them to neutral pH using formic acid, and diluted them using nanopure water. Adjustments to pH and filtration were required to remove particles and precipitates that interfere with LC separations. Dilution was required to reduce the intensities of mass spectral features. MW, FW, and PW were diluted by a factor of 10. HFF was diluted by a factor of 100 for matrix recovery experiments and a factor of 1000 for quantification of benzalkonium chloride (ADBAC).

Matrix recovery experiment

To explore the ways in which the water sample matrices influence the ionization of the HF additives relative to their ionization in nanopure water, we spiked a mixture of the nineteen selected HF additives into each of the fourteen prepared water samples and nanopure at varying concentrations. We selected concentrations so that the resulting chromatographic peaks would have an intensity between 1E7 and 7E7 in nanopure water, ensuring that the compound would be detected in the prepared water samples even if its ionization was suppressed by 90% and would not overfill the mass detector if its ionization was enhanced. It must be noted that we selected the spiked concentration in this way to enable robust estimation of MRFs, though this approach may limit the application of MRFs for quantification if the concentration of the additive in the sample is significantly different than the spiked concentration. The selected concentration for each of the HF additives is provided in Table ESI5. We measured each of the prepared and spiked water samples by means of LC-ESI-MS and calculated MRFs according to eqn (1):
image file: c8em00135a-t1.tif(1)

We conducted matrix recovery experiments in triplicate and we report all MRFs as the average of three measurements.

Analytical HPLC-ESI-MS method

We measured standard solutions, standard mixtures, prepared water samples, and spiked water samples by means of high-performance liquid chromatography (HPLC) electrospray ionization (ESI) quadrupole-Orbitrap mass spectrometry (MS) (QExactive, ThermoFisher Scientific) using an analytical method that we adapted from previous work focusing on the broad detection of organic micropollutants (log[thin space (1/6-em)]Kow −3 to +6) in water samples.34,35 Briefly, the analytical method incorporated large-volume injection to retain polar and semi-polar compounds on a C18 trap column (Hypersil Gold aQ, 2.1 × 20 mm, 12 μm particle size, ThermoFisher Scientific) while diverting the majority of inorganic matrix constituents to waste. Samples were then eluted from the trap column and onto a C18 analytical column (XBridge, 2.1 × 50 mm, 3.5 μm particle size, Waters) for chromatographic separation. Details on the mobile phase composition and gradient programs are provided in Tables ESI6 and ESI7 and elsewhere.34,35 We performed MS analysis in rapid polarity switching mode to include positive and negative ESI in the same run. Details of MS and MS/MS acquisition parameters are provided in Table ESI8. For quantification of HF additives, we created a six-point calibration curve by diluting the standard mixture with nanopure water. To explore methods for quantification of individual homologues present in a homologous mixture, we created separate six-point calibration curves with a pure standard of benzyldimethyldodecyl ammonium (ADBAC-C12) and a homologous mixture containing ADBAC-C12, ADBAC-C14 and ADBAC-C16. Finally, we determined retention times (RTs) and limits of detection (LODs) for each of the HF additives included in the standard mixture by inspection of the calibration curves as described in the ESI. Screening for HF additives in water samples was done by matching RTs, accurate masses, and MS/MS fragments with the standards of the individual compounds as previously described.35

Statistical analyses

We used Microsoft Excel 2016 to conduct one-way ANOVA tests to compare MRFs among individual homologues in homologous series. We used IBM SPSS Statistics (version 25) to conduct a Friedman test followed by pairwise comparisons with a Bonferroni correction for multiple comparisons to assess differences in the mean MRFs across water samples. All statistical tests used an alpha value of 0.05 to evaluate significance.

Results & discussion

Analytical response to HF additives

The optimized MS and MS/MS acquisition parameters for each of the nineteen HF additives measured in nanopure water during compound tuning is provided in Table ESI9; representative homologues are described for HF additives acquired as homologous mixtures. Eighteen of the HF additives ionized more efficiently in positive polarity mode, with 2-acrylamido-2-methylpropanesulfonic acid being the only individual compound that ionized more efficiently in negative polarity mode. We noted a number of ions and adducts during compound tuning including protonated adducts [M + H]+, deprotonated compounds [M − H], sodiated adducts [M + Na]+, ammoniated adducts [M + NH4]+, adducts of methanol [M + CH3OH + H]+, and adducts incorporating various conjugates of the HF additives. The three quaternary ammonium compounds (ADBAC, cocamidopropyl hydroxysultaine (CAPHS), and didecyldimethylammonium chloride) carry positive charges and were measured as [M]+. RTs and LODs acquired from the analysis of calibration curves are also provided in Table ESI9. The analytical method simultaneously detects seventeen of the nineteen HF additives with adequate chromatographic resolution and LODs, with the exceptions being ethylenediaminetetraacetic acid (EDTA) and bis(hexamethylene) triamine which were not detected with our analytical method when spiked into the HF fluids. The LODs for each of the HF additives in nanopure water range between 50 ng L−1 to 50 μg L−1 which is more sensitive than previously reported analytical methods for ADBAC, glutaraldehyde, and cocoamidopropyl surfactants,13 likely due to our use of large-volume injection.

Calculation of matrix recovery factors

The matrix chemistry of HFF wastewaters is complex.6,29,36 The TDS and inorganic cation concentrations measured in each of the fourteen water samples from MIP 3H and 5H are provided in Table ESI4 and Fig. ESI1, respectively. Of note among these data are increases in TDS concentrations in FW and PW samples (from ppm levels in MW and HFF to parts per thousand levels in FW and PW) and the concomitant increase in the concentrations of inorganic cations in those water samples, particularly of sodium and calcium cations. Additionally, changes in total organic carbon concentrations resulting from natural organic matter (NOM) and the presence of surfactants and other HF additives have been routinely reported in HFF wastewaters.29,37 We expect these changes in matrix chemistry to alter the ionization behavior of HF additives in a number of possible ways, though we have insufficient knowledge to predict how these complex changes in matrix chemistry will alter the ionization behavior of HF additives. Therefore, we calculated MRFs for each of the seventeen detected HF additives by spiking a mixture of each additive into each of the prepared water samples collected from MIP 3H and 5H. MRFs describe the ways in which the water sample matrices influence the ionization of the HF additives relative to their ionization in nanopure water. MRFs greater than 1.0 indicate enhanced ionization in a matrix relative to nanopure and MRFs less than 1.0 indicate suppressed ionization in a matrix relative to nanopure.

We hypothesized that the overall magnitude of the MRFs will vary for compounds that ionize in different ways (e.g., those that form different types of adducts) and among water samples. To test this hypothesis, we calculated MRFs for each additive and in each water sample and present a portion of these data in Fig. 1 (the full dataset is provided in Table ESI10). We present the average of triplicate MRFs for all individual compounds and a representative homologue for each homologous mixture that generated nonzero MRFs and were either positively charged, deprotonated, or formed protonated or sodiated adducts. Average MRFs greater than 1.0 are shown with shades of blue and average MRFs less than 1.0 are shown with shades of red. The coefficient of variation of all MRFs was typically less than 15%, indicating that our estimates of average MRFs are relatively robust in most instances.

image file: c8em00135a-f1.tif
Fig. 1 MRFs for fourteen of the selected compounds in each of the prepared water samples. MRFs in blue indicate enhancement and MRFs in red indicate suppression. MW is makeup water; HFF is hydraulic fracturing fluid; FW is flowback water; PW is produced water.

The data in Fig. 1 provide important insights on the behavior of each additive in each type of water sample. First, it is clear that most of the additives exhibit suppressed ionization in most types of water samples. This is consistent with the expectations for a complex matrix, where interfering organic compounds may co-elute with target analytes.38,39 Second, we expected that the quaternary ammonium compounds [M]+ would not exhibit significant signal suppression because surfactants dominate the surface of droplets formed during ESI21 and would not experience adduct shifting because their ionization is independent of the matrix or other water chemistry parameters. Therefore, it is notable that the ionization behavior of ADBAC-C12 is rather stable across all matrix types, whereas the ionization of CAPHS-C7H15 and didecyldimethylammonium is generally suppressed across all matrix types. Further, ADBAC-C12 and CAPHS-C7H15 have similar RTs (18.0 and 18.2, respectively) which we expected would contribute to similar changes in ionization behavior if the effects were due to co-elution. These observations complicate our interpretation of these data and suggest that the ionization cannot be generalized from one cationic compound to another, which limits our ability to predict the behavior of untested cationic compounds. Finally, the additives that favor protonated adducts [M + H]+ or are deprotonated [M − H] in nanopure exhibit limited matrix effects in MW and HFF, but their ionization is significantly suppressed in FW and PW. In comparison, the additives that favor sodiated adducts [M + Na]+ in nanopure exhibit suppressed ionization in MW, HFF, and FW, but their ionization is significantly enhanced in PW. We expected this latter observation considering the elevated sodium concentrations in the PW samples (Fig. ESI1). However, we found FW 5H to have the highest sodium concentrations among all of the water samples, yet we did not observe enhanced ionization of additives that favor sodiated adducts in the FW 5H sample. Additionally, we found no statistically significant association between sodium concentration and the ratio of the peak areas of sodiated and protonated adducts among additives that exhibited both types of adducts, suggesting that sodium concentration was not predictive of shifting adduct formation. We therefore conclude that the enhanced ionization of additives that favor sodiated adducts in the PW is not exclusively the result of increasing sodium concentrations, but is rather related to other features of the matrix that remain unidentified.

Mechanisms contributing to MRFs

Enhanced or suppressed ionization of HF additives can be caused by at least two mechanisms: changes in ionization efficiency resulting from co-eluting matrix constituents or the shifting of adduct ratios across varying matrices.16 To further examine the mechanism behind the MRFs reported in Fig. 1, we calculated the ratio of the peak areas of sodiated and protonated adducts among additives that exhibited both types of adducts in spiked nanopure and prepared water samples. These adducts were selected because they were the dominant adducts detected for the majority of additives in the environmental matrices. We normalized the ratio of the peak areas of sodiated and protonated adducts calculated in the prepared water samples to the ratio calculated for the spiked nanopure. This normalization allowed for an examination of the enhancement of adducts even if the sodiated adduct was already dominant in nanopure water. The results of this analysis are provided in Fig. 2. Of the ten additives that exhibited both sodiated and protonated adducts, six exhibited a clear trend in the shift of the ratio of these adducts in the different matrices. The remaining additives either did not exhibit any notable trend or had at least one adduct that was not detected in all samples. Interestingly, all six of the additives that exhibited trends in adduct shifting also exhibited enhanced ionization in the PW samples as shown in Fig. 1. For 2-butoxyethanol, bis(2-ethylhexyl) phthalate, butyl glycidyl ether, and a representative nonylphenol ethoxylate (NPE-EO7), polyethylene glycol (PEG-EO9), and polypropylene glycol (PPG-PO6), the ratio of sodiated to protonated adducts was the same in MW as in nanopure. However, the ratio shifted to favor the protonated adducts in the HFF samples and to favor sodiated adducts in the PW samples. These observations confirm that the enhanced ionization observed in the PW samples is the result of adduct shifting, though it remains unclear why enhanced ionization by means of adduct shifting is only observed in PW samples and not in the similarly saline FW samples.
image file: c8em00135a-f2.tif
Fig. 2 The ratio of sodiated adduct peak area to protonated adduct peak area for each chemical in each sample, normalized to the same ratio in spiked nanopure water. Any values greater than one indicate a shift favoring sodiated adducts, whereas any values lower than one indicate a shift favoring protonated adducts. NP is nanopure water; MW is makeup water; HFF is hydraulic fracturing fluid; FW is flowback water; PW is produced water.

We also sought to explore the mechanisms contributing to the measured MRFs across all homologues in each of the homologous mixtures. In the case of ADBAC, we could only calculate a MRF for ADBAC-C12 because it was the major component of the homologous mixture. For PEG, PPG, and CAPHS, there was no statistically significant difference in the average MRF calculated for each member of the homologous mixture (one-way ANOVA, p = 0.90, p = 0.26, p = 0.49, respectively). For CAPDMA, there were significant differences in the MRFs calculated among the homologues (one-way ANOVA, p = 1.43 × 10−8), but there was no clear trend among the differences, so we cannot currently propose a mechanism which explains these differences. Finally, there were also significant differences in the MRFs calculated among the NPE homologues (one-way ANOVA, p = 4.97 × 10−4), and there was a significant negative association between the magnitude of the MRF and the chain length of the homologue between NPE-EO5 to EO10. Further analysis of the ratio of sodiated to protonated adducts normalized to the ratio in nanopure water indicated that, as the ethoxylate chain length increased, the prevalence of sodium adducts in the FW and PW was less pronounced. While we could not find literature supporting this specific shift, this is in line with the shift towards protonated and ammoniated adducts observed in PEG and PPG with longer ethoxylated chains.14

Generalizing matrix recovery factors across sample types

We finally sought to explore whether the MRFs measured among the HF additives could be generalized across water samples, which would enable extrapolation of MRFs to other water samples of similar types. The data in Fig. 3 summarize the range of MRFs calculated among seventeen of the HF additives (all except EDTA and bis(hexamethylene) triamine and only one representative homologue among the compounds included in homologous mixtures) in each of the water samples. The box plot presents the median MRF among the seventeen HF additives (solid line in box) along with the first and third quartiles and the minimum and maximum values. The data reveal that the MW have median MRFs close to 1.0 with a relatively narrow range and the MRFs for the MW are never greater than 1.0. This is consistent with the TDS and inorganic cation measurements and the calculated MRFs for MW are within the ranges expected for surface waters.40 The calculated MRFs for FW samples also have a relatively narrow range and are generally less than 1.0. In HFF and PW, at least 25% of the HF additives in all samples exhibited either no change in ionization or enhanced ionization relative to nanopure, demonstrating more complex relationships between matrix chemistry and ionization potential in these water samples. When all sample types are considered, there were statistically significant differences in the distributions of the MRFs among the water samples (Friedman test, p = 0.002). Specifically, we found statistically significant differences between FW 5H and PW 3H from W1 (pairwise comparison, p < 0.001), W2 (pairwise comparison, p = 0.001), and W3 (pairwise comparison, p = 0.039). Despite some variability in the MRFs calculated among the PW samples, we did not find statistically significant differences among PWs sampled from the same well at different times (all combinations, pairwise comparison, p = 1.0) or between PW sampled from different wells (all combinations, pairwise comparison, p = 1.0). This suggests that MRFs calculated for an HF additive in any PW sample could be extrapolated to quantify that additive in any other PW sample, though this is only exemplified among our PW samples and further research would be required to see if this observation can be generalized to samples collected from other wells.
image file: c8em00135a-f3.tif
Fig. 3 Matrix recovery factors (MRFs) for HF additives across makeup water (MW), hydraulic fracturing fluid (HFF), flowback water (FW) and produced water (PW). MW, FW, and PW were diluted 1[thin space (1/6-em)]:[thin space (1/6-em)]10 and HFF was diluted 1[thin space (1/6-em)]:[thin space (1/6-em)]100 prior to analysis. Boxes represent the distribution of MRFs for each of the seventeen additives that were successfully detected in each of the fluids. The line in the box represents the median MRF, the edges of the boxes represent the first and third quartile, the ends of the whiskers represent the minimum and maximum MRFs. Samples labeled with the same letter were not found to have statistically significant differences in distribution using a Friedman test followed by pairwise comparisons with Bonferroni corrections for multiple comparisons.

Screening for HF additives in water samples

The FracFocus disclosures for MIP 3H and 5H are provided in Tables ESI11–ESI16. The disclosures include 40 unique chemical additives, with 33 of them being organic chemicals and five of those being amenable to LC-ESI-MS analysis (glutaraldehyde, ADBAC, PEG, PPG, and guar gum).12 We first leveraged the high-resolution mass spectral acquisitions obtained for each water sample to screen for the disclosed HF additives, the HF additives included in our compound tuning and MRF experiments, and other priority HF additives and expected transformation products using previously described techniques.35 We detected four of the disclosed additives in at least one of the water samples collected from MIP 3H or 5H: glutaraldehyde, and homologues of ADBAC, PEG, and PPG. We did not detect guar gum, likely due to its ready biodegradability.41 We also did not detect 2-acrylamido-2-methylpropanesulfonic acid, but it was not explicitly included in the HFF mixture; we included it in the analysis because it was a suspected transformation product of disclosed polymers used in the HFF.

Glutaraldehyde, one of the most commonly used biocides,32,42 was disclosed in both wells, but we detected it only in HFF from well 3H ([M + H]+, RT 7.7 min, 1.82 ppm error). The absence of glutaraldehyde in the wastewaters is consistent with previous findings and attributable to its tendency to biodegrade and auto-polymerize.13,43 Similar to what was reported in a previous study, we observed a hydrated and sodiated adduct of a dimerized molecule of glutaraldehyde ([2M + H2O + Na]+, RT 10.4 min, 0.15 ppm error) in HFF 3H at an accurate mass (m/z) of 241.1048. The higher retention time indicates that this polymer is a product of an aldol condensation reaction, and not an artifact formed in the source.13 We also observed a trimer ([3M + H2O + Na]+, RT 11.9 min, 0.28 ppm error), further complicating our subsequent attempts of quantification. To properly quantify the total concentration of glutaraldehyde polymers, it has been suggested that derivatization with 2,4-dinitrophenylhydrazine is necessary.16 Analytical data supporting these observations is provided in Fig. ESI2.

We detected ADBAC, a commonly used antimicrobial surfactant,32,42 as ADBAC-C12 ([M]+, RT 18.0 min, −0.35 ppm error), ADBAC-C14 ([M]+, RT 19.9 min, −0.26 ppm error), and ADBAC-C16 ([M]+, RT 21.3 min, −0.49 ppm error) in both HFF samples with accurate masses (m/z) of 304.2998, 332.3311, and 360.3623, respectively. We also found ADBAC-C12 in all FW and PW samples, which will be discussed later with respect to quantification in detail. Trace levels of other ADBAC homologues were also detected sporadically among the water samples including the C6, C8, C10, and C18 homologues. A Kendrick mass analysis14,15 supporting these observations is provided in Table ESI17.

We detected PEGs ranging from PEG-EO5 ([M + Na]+, RT 8.4 min, 0.58 ppm error) to PEG-EO14 ([M + NH4]+, RT 10.2 min, 0.77 ppm error). In this analysis, PEG-EO5 was the only PEG homologue where the sodiated adduct was dominant. For PEG-EO6 and the PEG homologues with longer ethoxylated chains, the ammoniated adduct was dominant. This shift is consistent with previous studies which detected PEG and PPG in FW and PW from other formations and noted a shift in adducts related to the length of the ethoxylated and propylated chains.14,15,37 Specifically, it was previously noted that PEG homologues with an ethoxylate chain length of seven or fewer form predominantly sodiated adducts and PEG homologues with an ethoxylate chain length of nine or greater form predominantly ammoniated adducts.14 A similar phenomenon was described for PPG, with sodiated adducts being greater in intensity for PPG-PO6 and homologues with shorter propylated chains and ammoniated adducts being greater in intensity for PPG-PO8 and homologues with greater propylated chain length.15 With respect to PPG, this study found PPG homologues ranging from PPG-PO4 ([M + Na]+, RT 11.7 min, −0.02 ppm error) to PPG-PO11 ([M + H]+, RT 18.3 min, 1.13 ppm error). However, we found the sodiated adducts for PPG homologues with greater intensity than the ammonium adduct, even for homologues with chain lengths greater than that of PPG-PO8. Instead, protonated adducts were dominant for PPG-PO8 and other homologues with greater propylated chain lengths. This phenomenon is similar to what was previously reported, however, it suggests that the adducts formed in previous studies may not be entirely due to structural changes promoting stronger complexation of ammonium as previously hypothesized,14,15 but may be at least partially due to the ammonium concentrations in the samples or an instrument-specific phenomenon. Kendrick mass analyses supporting these observations are provided in Tables ESI18 and ESI19.

Interestingly, despite the fact that it was not disclosed in the FracFocus report, we also measured 2-butoxyethanol (2-BE, [M + Na]+, RT 10.7 min, 0.773 ppm error) in all FW and PW samples. We measured 2-BE with an accurate mass (m/z) of 141.0887, which matches the MS spectra acquired for 2-BE using the analytical grade standard (see Table ESI9). We also confirmed MS/MS fragments with accurate masses (m/z) of 57.0707 and 105.0035. Due to its absence from both the FracFocus disclosure and the HFF samples, 2-BE was likely not used as a chemical additive in the process of well completion and may be derived from another source, which has yet to be determined. 2-BE is commonly used as a surfactant, corrosion inhibitor, and nonemulsifier and was found to be in 22.8% of FracFocus disclosures in a 2015 review and may have been used as an additive (e.g., maintenance chemical) subsequent to the fracturing of the wells.32 It has previously been identified in FW and PW samples using 2D gas chromatography coupled to time-of-flight mass spectrometry in FW and PW samples and groundwater samples near drilling operations.11 This is the first known measurement of 2-BE in HF-associated fluids using LC-ESI-MS, although targeted LC-ESI-MS methods have been developed for detection of 2-BE in seawater because it was a component of dispersant formulations used to remediate the effects of the Macondo well blowout.44 Analytical data supporting these qualitative observations is provided in Fig. ESI3.

Quantification of ADBACs

We used the MRFs calculated for each additive to provide the first quantitative estimates of multiple HF additives using LC-ESI-MS. We performed quantification of ADBAC-C12 using calibration curves prepared from a pure standard and from a homologous mixture containing ADBAC-C12, ADBAC-C14, and ADBAC-C16. In the latter approach, we assumed that each individual homologue was ionized with the same efficiency and assigned concentrations to each homologue in each sample of the calibration curve based on the fraction of the total peak area each homologue contained. The results of the quantification of ADBAC-C12 using both approaches is provided in Fig. 4, where error bars represent the uncertainty in the linearity of the respective calibration curves. We quantified ADBAC-C12 in all samples, with concentrations found in the low mg L−1 level in the HFF and in the low μg L−1 level in the MW, FW, and PW. The results also indicate that the quantitative estimates for ADBAC-C12 were not significantly different when using calibration curves prepared from a pure standard or from a homologous mixture. Despite greater uncertainty in the estimates made from the calibration curve prepared from a homologous mixture, both approaches result in concentrations within a factor of two (even when considering uncertainty) which is appropriate for most exposure assessments.
image file: c8em00135a-f4.tif
Fig. 4 Concentrations of ADBAC-C12 calculated using a calibration curve prepared using a pure analytical grade standard and a standard mix of three ADBAC homologues. The left panel has a concentration range of 0–5000 μg L−1 for the y-axis and the right panel has a concentration of 0–10 μg L−1 for the y-axis. *The MRFs for both HFF samples were determined using aliquots that were diluted by a factor of 100, which is 10 times more than the aliquots used for this quantification. This is a conservative estimate because we would expect the matrix effects to be less pronounced in more dilute aliquots. MW is makeup water; HFF is hydraulic fracturing fluid; FW is flowback water; PW is produced water.

Because the quantitative estimates of ADBAC-C12 were reasonable when using the calibration curve prepared with a homologous mixture, we could extend our analysis to quantify ADBAC-C14 and ADBAC-C16 in the HFF with the total peak area approach.45 We found both of these homologues at the low mg L−1 level in both HFF samples, resulting in total ADBAC concentrations of 3.59 mg L−1 for HFF 3H and 8.01 mg L−1 in HFF 5H. These findings are consistent with the FracFocus report, which indicated that approximately 7 mg L−1 level of total ADBAC-C12 to C16 was present in the HFF. It is notable that we measured significant levels of ADBAC-C14 and ADBAC-C16 in the HFF but not in any of the wastewater samples. This suggests that ADBAC may be transformed in the subsurface, likely through mechanisms of chain shortening to form ADBAC homologues of shorter chain length or by cleavage of the alkyl group to form benzyl dimethyl amine.46 We found qualitative evidence of ADBAC homologues of shorter chain length in the FW and PW samples, though our homologous mixture did not contain those homologues so we could not quantify them.

Quantification of PEGs and PPGs

We used the same total peak area approach that we validated with ADBAC to quantify PEGs and PPGs in the prepared water samples. For quantification, we only considered homologues that made up greater than 5% of the homologous mixture and that were detected with peak intensities at least twice as high as those found in filtered nanopure blanks. These selection criteria limited quantification to PEG-EO10 through PEG-EO14 and to PPG-PO5 through PPG-P08. The results of the quantifications are provided in Fig. 5. Notably, while both PEG and PPG were detected in FWs and PWs, PPG was not detected in MWs or HFFs and PEG was not detected in MWs or HFF 3H. PEG-EO10 to PEG-EO14 homologues were detected in HFF 5H with a total concentration of 60.2 μg L−1. We note two important trends for both groups of homologues. First, total PEG and PPG concentrations (within the range of homologues quantified) are greatest in FW samples and decrease in PW samples. Total PEG concentrations were on the order of 100 μg L−1 in FW samples and ranged down to 4.2 μg L−1 in week 2 PW samples before dropping below our LODs. We found total PPG concentrations in a similar range, and PPG-PO5 remained above our LOD in all PW samples. These measured concentrations are much lower than previous semi-quantitative estimates of total PEG and PPG concentrations in HFF wastewaters.15 The discrepancy can be explained at least in part due to the enhanced ionization of PEGs and PPGs we observed in PW samples resulting in MRFs greater than 1.0; this results in a downward adjustment in estimated concentrations that has not been accounted for in previous studies. Further, our quantification includes only a limited range of PEG and PPG homologues that were present in our homologous mixtures. Like ADBAC, we expect that greater concentrations of shorter chain homologues may be present which would result in higher total concentrations of PEG and PPG. Second, there is a negative association between concentration and chain length of each homologue. In other words, shorter chain homologues are present at greater concentrations, suggesting degradation of longer chain PEGs and PPGs or other alcohol ethoxylates into shorter chain homologues.47 Further work that quantifies additional short chain homologues and evaluates changes in the concentrations of homologues of varying chain length to better assess transformation is warranted.
image file: c8em00135a-f5.tif
Fig. 5 Concentrations of PEG and PPG quantified using homologous mixtures. PPG was not detected in MWs or HFFs. PEG was not detected in MWs or HFF 3H, but PEG-EO10 to PEG-EO14 homologues were detected in HFF 5H with a total concentration of 60.2 μg L−1. FW is flowback water; PW is produced water.


This research addresses two important knowledge gaps related to the measurement of polar to semi-polar chemical additives in HF-associated fluids. First, we report MS and MS/MS acquisition parameters for nineteen priority HF additives and quantify the ways in which seventeen of these additives' ionization is enhanced or suppressed in water samples collected over the life cycle of a HF gas well with MRFs. These data clearly demonstrate that the changing matrix of water samples from MW through PW can have changing effects on the ionization of chemical additives. It is notable that most compounds have suppressed ionization in most types of samples, though additives that form sodiated adducts in nanopure water have enhanced ionization in PW samples. These include some of the most widely reported HF additives including glutaraldehyde, NPEs, PEGs, PPGs, whose concentrations in PW can be overestimated if enhanced ionization is not considered in their analysis. Through further analysis, we demonstrate that enhanced ionization is most likely the result of adduct shifting and we expect this to be generalized across PW samples, perhaps even from wells in different formations. Second, we provide the first quantitative estimates of individual homologues of ADBAC, PEG, and PPG in HF-associated fluids. Notably, our estimates of ADBAC concentrations agree with what was reported in the FracFocus disclosure, adding some additional validation to our approach and providing some evidence of the accuracy of data in FracFocus disclosures. In the case of ethoxylated homologous mixtures, we found increasing concentrations of shorter chain homologues in later wastewater samples, also providing evidence of transformation, likely by the mechanism of chain shortening. Together, these data provide the first quantitative measurements that can be used to inform future toxicity studies or exposure assessments.

Conflicts of interest

There are no conflicts to declare.


The Environmental Defense Fund and Cornell University's David R. Atkinson Center for a Sustainable Future (ACSF) supported this work. M. N. acknowledges NSF GRFP grant no. DGE-1650441. The West Virginia University Marcellus Shale Energy and Environment Laboratory, which is funded by the National Energy Technology Laboratory of the US Department of Energy and the West Virginia Water Research Institute, provided access to the environmental samples referenced in this article. We would like to thank Dr Shree Giri for performing the ICP-OES analysis of these samples.


  1. J. Perrin, T. Cook and US Energy Information Administration, Hydraulically fractured wells provide two-thirds of U.S. natural gas production, https://www.eia.gov/todayinenergy/detail.php?id=26112, accessed 8 June 2017.
  2. K. Dyl, V. Zaretskaya and US Energy Information Administration, In new trend, U.S. natural gas exports exceeded imports in 3 of the first 5 months of 2017, https://www.eia.gov/todayinenergy/detail.php?id=32392, accessed 8 August 2017.
  3. M. Bamberger and R. E. Oswald, Unconventional oil and gas extraction and animal health, Environ. Sci.: Processes Impacts, 2014, 16, 1860–1865 RSC.
  4. E. G. Elliott, A. S. Ettinger, B. P. Leaderer, M. B. Bracken and N. C. Deziel, A systematic evaluation of chemicals in hydraulic-fracturing fluids and wastewater for reproductive and developmental toxicity, J. Exposure Sci. Environ. Epidemiol., 2017, 27, 90–99 CrossRef CAS PubMed.
  5. B. G. Rahm, J. T. Bates, L. R. Bertoia, A. E. Galford, D. A. Yoxtheimer and S. J. Riha, Wastewater management and Marcellus Shale gas development: Trends, drivers, and planning implications, J. Environ. Manage., 2013, 120, 105–113 CrossRef PubMed.
  6. L. O. Haluszczak, A. W. Rose and L. R. Kump, Geochemical evaluation of flowback brine from Marcellus gas wells in Pennsylvania, USA, Appl. Geochem., 2013, 28, 55–61 CrossRef CAS.
  7. US Environmental Protection Agency – US EPA, Hydraulic Fracturing for Oil and Gas: Impacts from the Hydraulic Fracturing Water Cycle on Drinking Water Resources in the United States (Main Report – EPA/600/R-16/236fa), 2016.
  8. T. Colborn, C. Kwiatkowski, K. Schultz, M. Bachran, C. Taylor, T. Colborn, C. Kwiatkowski, K. Schultz and M. Bachran, Natural Gas Operations from a Public Health Perspective, Hum. Ecol. Risk Assess., 2011, 17, 1–25 CrossRef.
  9. B. D. Drollette, K. Hoelzer, N. R. Warner, T. H. Darrah, O. Karatum, M. P. O'Connor, R. K. Nelson, L. A. Fernandez, C. M. Reddy, A. Vengosh, R. B. Jackson, M. Elsner and D. L. Plata, Elevated levels of diesel range organic compounds in groundwater near Marcellus gas operations are derived from surface activities, Proc. Natl. Acad. Sci. U. S. A., 2015, 112, 13184–13189 CrossRef CAS PubMed.
  10. W. Orem, C. Tatu, M. Varonka, H. Lerch, A. Bates, M. Engle, L. Crosby and J. McIntosh, Organic substances in produced and formation water from unconventional natural gas extraction in coal and shale, Int. J. Coal Geol., 2014, 126, 20–31 CrossRef CAS.
  11. G. T. Llewellyn, F. Dorman, J. L. Westland, D. Yoxtheimer, P. Grieve, T. Sowers, E. Humston-Fulmer and S. L. Brantley, Evaluating a groundwater supply contamination incident attributed to Marcellus Shale gas development, Proc. Natl. Acad. Sci. U. S. A., 2015, 112, 6325–6330 CrossRef CAS PubMed.
  12. G. J. Getzinger, M. P. O'Connor, K. Hoelzer, B. D. Drollette, O. Karatum, M. A. Deshusses, P. L. Ferguson, M. Elsner and D. L. Plata, Natural Gas Residual Fluids: Sources, Endpoints, and Organic Chemical Composition after Centralized Waste Treatment in Pennsylvania, Environ. Sci. Technol., 2015, 49, 8347–8355 CrossRef CAS PubMed.
  13. I. Ferrer and E. M. Thurman, Analysis of hydraulic fracturing additives by LC/Q-TOF-MS, Anal. Bioanal. Chem., 2015, 407, 6417–6428 CrossRef CAS PubMed.
  14. E. M. Thurman, I. Ferrer, J. Blotevogel and T. Borch, Analysis of Hydraulic Fracturing Flowback and Produced Waters Using Accurate Mass: Identification of Ethoxylated Surfactants, Anal. Chem., 2014, 86, 9653–9661 CrossRef CAS PubMed.
  15. E. M. Thurman, I. Ferrer, J. Rosenblum, K. Linden and J. N. Ryan, Identification of Polypropylene Glycols and Polyethylene Glycol Carboxylates in Flowback and Produced Water from Hydraulic Fracturing, J. Hazard. Mater., 2017, 323, 11–17 CrossRef CAS PubMed.
  16. K. Oetjen, C. G. S. Giddings, M. McLaughlin, M. Nell, J. Blotevogel, D. E. Helbling, D. Mueller and C. P. Higgins, Emerging analytical methods for the characterization and quantification of organic contaminants in flowback and produced water, Trends Environ. Anal. Chem., 2017, 15, 12–23 CrossRef CAS.
  17. Y. He, S. L. Flynn, E. J. Folkerts, Y. Zhang, D. Ruan, D. S. Alessi, J. W. Martin and G. G. Goss, Chemical and toxicological characterizations of hydraulic fracturing flowback and produced water, Water Res., 2017, 114, 78–87 CrossRef CAS PubMed.
  18. P. F. Ziemkiewicz and Y. Thomas He, Evolution of water chemistry during Marcellus Shale gas development: A case study in West Virginia,, Chemosphere, 2015, 134, 224–231 CrossRef CAS PubMed.
  19. K. L. Rundlett and D. W. Armstrong, Mechanism of signal suppression by anionic surfactants in capillary electrophoresis – electrospray ionization mass spectrometry, Anal. Chem., 1996, 68, 3493–3497 CrossRef CAS PubMed.
  20. N. B. Cech and C. G. Enke, Practical implications of some recent studies in electrospray ionization fundamentals, Mass Spectrom. Rev., 2001, 20, 362–387 CrossRef CAS PubMed.
  21. K. Tang and R. D. Smith, Physical/chemical separations in the break-up of highly charged droplets from electrosprays, J. Am. Soc. Mass Spectrom., 2001, 12, 343–347 CrossRef CAS PubMed.
  22. M. Thompson and S. L. R. Ellison, A review of interference effects and their correction in chemical analysis with special reference to uncertainty, Accredit. Qual. Assur., 2005, 10, 82–97 CrossRef CAS.
  23. M. Stüber and T. Reemtsma, Evaluation of three calibration methods to compensate matrix effects in environmental analysis with LC-ESI-MS, Anal. Bioanal. Chem., 2004, 378, 910–916 CrossRef PubMed.
  24. C. R. Mallet, Z. Lu and J. R. Mazzeo, A study of ion suppression effects in electrospray ionization from mobile phase additives and solid-phase extracts, Rapid Commun. Mass Spectrom., 2004, 18, 49–58 CrossRef CAS PubMed.
  25. E. Stokvis, H. Rosing and J. H. Beijnen, Stable isotopically labeled internal standards in quantitative bioanalysis using liquid chromatography/mass spectrometry: necessity or not?, Rapid Commun. Mass Spectrom., 2005, 19, 401–407 CrossRef CAS PubMed.
  26. G. Fedorova, T. Randak, R. H. Lindberg and R. Grabic, Comparison of the quantitative performance of a Q-Exactive high-resolution mass spectrometer with that of a triple quadrupole tandem mass spectrometer for the analysis of illicit drugs in wastewater, Rapid Commun. Mass Spectrom., 2013, 27, 1751–1762 CrossRef CAS PubMed.
  27. D. E. Helbling, D. R. Johnson, M. Honti and K. Fenner, Micropollutant Biotransformation Kinetics Associate with WWTP Process Parameters and Microbial Community Characteristics, Environ. Sci. Technol., 2012, 46, 10579–10588 CrossRef CAS PubMed.
  28. H. Trufelli, P. Palma, G. Famiglini and A. Cappiello, An overview of matrix effects in liquid chromatography-mass spectrometry, Mass Spectrom. Rev., 2011, 30, 491–509 CrossRef CAS PubMed.
  29. E. Barbot, N. S. Vidic, K. B. Gregory and R. D. Vidic, Spatial and temporal correlation of water quality parameters of produced waters from Devonian-age shale following hydraulic fracturing, Environ. Sci. Technol., 2013, 47, 2562–2569 CrossRef CAS PubMed.
  30. K. Oetjen, K. E. Chan, K. Gulmark, J. H. Christensen, J. Blotevogel, T. Borch, J. R. Spear, T. Y. Cath and C. P. Higgins, Temporal characterization and statistical analysis of flowback and produced waters and their potential for reuse, Sci. Total Environ., 2018, 619–620, 654–664 CrossRef CAS PubMed.
  31. C. Díez, M. Feinberg, A. S. Spörri, E. Cognard, D. Ortelli, P. Edder and S. Rudaz, Evaluation of Quantification Methods to Compensate for Matrix Effects in the Analysis of Benzalkonium Chloride and Didecyldimethylammonium Chloride in Fruits and Vegetables by LC-ESI-MS/MS, Food Analytical Methods, 2016, 9, 485–499 CrossRef.
  32. J. D. Rogers, T. L. Burke, S. G. Osborn and J. N. Ryan, A Framework for Identifying Organic Compounds of Concern in Hydraulic Fracturing Fluids Based on Their Mobility and Persistence in Groundwater,, Environ. Sci. Technol. Lett., 2015, 2, 158–164 CrossRef CAS.
  33. J. L. Luek, M. Harir, P. Schmitt-Kopplin, P. J. Mouser and M. Gonsior, Temporal dynamics of halogenated organic compounds in Marcellus Shale flowback, Water Res., 2018, 136, 200–206 CrossRef CAS PubMed.
  34. C. M. G. Carpenter and D. E. Helbling, Removal of micropollutants in biofilters: Hydrodynamic effects on biofilm assembly and functioning, Water Res., 2017, 120, 211–221 CrossRef CAS PubMed.
  35. A. L. Pochodylo and D. E. Helbling, Emerging investigators series: prioritization of suspect hits in a sensitive suspect screening workflow for comprehensive micropollutant characterization in environmental samples, Environ. Sci.: Water Res. Technol., 2017, 3, 54–65 RSC.
  36. S. J. Maguire-Boyle and A. R. Barron, Organic compounds in produced waters from shale gas wells, Environ. Sci.: Processes Impacts, 2014, 16, 2237–2248 RSC.
  37. J. Rosenblum, E. M. Thurman, I. Ferrer, G. Aiken and K. G. Linden, Organic Chemical Characterization and Mass Balance of a Hydraulically Fractured Well: From Fracturing Fluid to Produced Water over 405 Days, Environ. Sci. Technol., 2017, 51, 14006–14015 CrossRef CAS PubMed.
  38. P. L. Ferguson, C. R. Iden and B. J. Brownawell, Analysis of Alkylphenol Ethoxylate Metabolites in the Aquatic Environment Using Liquid Chromatography-Electrospray Mass Spectrometry, Anal. Chem., 2000, 72, 4322–4330 CrossRef CAS PubMed.
  39. P. A. Lara-Martín, E. González-Mazo and B. J. Brownawell, Multi-residue method for the analysis of synthetic surfactants and their degradation metabolites in aquatic systems by liquid chromatography-time-of-flight-mass spectrometry, J. Chromatogr. A, 2011, 1218, 4799–4807 CrossRef PubMed.
  40. V. Leendert, H. Van Langenhove and K. Demeestere, Trends in liquid chromatography coupled to high-resolution mass spectrometry for multi-residue analysis of organic micropollutants in aquatic environments, TrAC, Trends Anal. Chem., 2015, 67, 192–208 CrossRef CAS.
  41. Y. Lester, T. Yacob, I. Morrissey and K. G. Linden, Can We Treat Hydraulic Fracturing Flowback with a Conventional Biological Process? The Case of Guar Gum,, Environ. Sci. Technol. Lett., 2014, 1, 133–136 CrossRef CAS.
  42. M. Elsner and K. Hoelzer, Quantitative Survey and Structural Classification of Hydraulic Fracturing Chemicals Reported in Unconventional Gas Production, Environ. Sci. Technol., 2016, 50, 3290–3314 CrossRef CAS PubMed.
  43. G. A. Kahrilas, J. Blotevogel, E. R. Corrin and T. Borch, Downhole Transformation of the Hydraulic Fracturing Fluid Biocide Glutaraldehyde: Implications for Flowback and Produced Water Quality, Environ. Sci. Technol., 2016, 50, 11414–11423 CrossRef CAS PubMed.
  44. C. E. Ramirez, S. R. Batchu and P. R. Gardinali, High sensitivity liquid chromatography tandem mass spectrometric methods for the analysis of dioctyl sulfosuccinate in different stages of an oil spill response monitoring effort, Anal. Bioanal. Chem., 2013, 405, 4167–4175 CrossRef CAS PubMed.
  45. I. Ferrer and E. T. Furlong, Identification of Alkyl Dimethylbenzylammonium Surfactants in Water Samples by Solid-Phase Extraction Followed by Ion Trap LC/MS and LC/MS/MS, Environ. Sci. Technol., 2001, 35, 2583–2588 CrossRef CAS PubMed.
  46. G. A. Kahrilas, J. Blotevogel, P. S. Stewart and T. Borch, Biocides in Hydraulic Fracturing Fluids: A Critical Review of Their Usage, Mobility, Degradation, and Toxicity,, Environ. Sci. Technol., 2015, 49, 16–32 CrossRef CAS PubMed.
  47. K. M. Heyob, J. Blotevogel, M. Brooker, M. V. Evans, J. J. Lenhart, J. Wright, R. Lamendella, T. Borch and P. J. Mouser, Natural Attenuation of Nonionic Surfactants Used in Hydraulic Fracturing Fluids: Degradation Rates, Pathways, and Mechanisms, Environ. Sci. Technol., 2017, 51, 13985–13994 CrossRef CAS PubMed.


Electronic supplementary information (ESI) available. See DOI: 10.1039/c8em00135a

This journal is © The Royal Society of Chemistry 2019