Organics and inorganics in flow back and produced water from shale gas operations: treatment and identification of glycols using gas chromatography–mass spectrometry

The shale gas boom in the United States (USA) will likely play an important role in meeting the current energy demand increases (Bellani et al. 2021). The shale oil and gas industry is the fastest-growing energy-related industry in the country owing to horizontal drilling during hydraulic fracturing (Bellani et al. 2021). Hydraulic fracturing requires approximately two to seven million gallons of an aqueous solution for each drilling operation (Lee et al. 2011; Rahm 2011; Stephenson et al. 2011; Nicot & Scanlon 2012; Suarez 2012; Goodwin et al. 2013; Hickenbottom et al. 2013), and this solution consists of 98–99% water along with sand and fracturing additives (Lee et al. 2011; Spellman 2013). After the flow back begins, approximately 20–40% of the initial water input is recovered over 4 weeks, indicating that most of the fracturing additives remain in the geological formation (Bai et al. 2013). It is therefore crucial to assess flow back/produced water quality and ultimately determine adequate treatment or disposal techniques. The U.S. Environmental Protection Agency (EPA) proposed total petroleum hydrocarbons (TPHs) and benzene, toluene, ethylbenzene, and xylene (BTEX) limits of less than 100 μg/L and 5.0 ml/L, respectively, in the effluent (United States Environmental Protection Agency 2005).

Propylene glycols (PGs), ethylene glycols (EGs), and various ethoxylated compounds are considered as commonly used chemical additives for the hydraulic fracturing process (Carter et al. 2013). Glycols act as scale inhibitors, biocides, and friction reducers (Carter et al. 2013; Kim et al. 2020). Carter et al. (2013) reported that PG and EG account for more than 15% of additives used during hydraulic fracturing. Ethylene glycol monobutyl ether (EGME) and EG ranked third and fourth as the most widely used additives for hydraulic fracturing between 2005 and 2009 in the USA, respectively (Waxman et al. 2022). The characteristics of flow back and produced water from shale oil and gas operations vary based on the characteristics of the formation, especially total dissolved solid (TDS) concentrations (Osborn et al. 2012; Vengosh et al. 2014).

Treatment approaches for flow back and produced water may differ depending on the purpose of the effluent (Jenner & Lamadrid 2013; Lutz et al. 2013). Flow back water can be reused for fracturing in other wells; however, the water should meet a certain quality level for successful hydraulic fracturing. Flow back and produced water may be evaporated to decrease the total amount of waste. However, this approach is not always applicable due to unsuitable geological conditions for injection in some regions (Lutz et al. 2013). In other cases, untreated wastewater can be injected into subsurfaces below groundwater formations (Rahm 2011; Jenner & Lamadrid 2013). Electrocoagulation (EC), ultrafiltration (UF), granular activated carbon (GAC), and reverse osmosis (RO) have been used to treat produced water to improve its quality and potential reuse for switch grass and canola irrigation (Pica et al. 2017). High levels of TDS in flow back and produced water are difficult to treat with conventional wastewater treatment processes (Lefebvre et al. 2004; Altaee & Hilal 2014; Lester et al. 2014). Due to the various chemicals used as fracturing additives, such as biocides or toxins, biological treatments are not a suitable option (Stringfellow et al. 2014; Rogers et al. 2015). High levels of total organic carbon (TOC) in flow back and produced water can harm membrane processes and reduce the effectiveness of direct treatment options (Pandey et al. 2012).

Coagulation is usually applied to wastewater because of its high efficiency in reducing suspended and colloidal particles (Wang et al. 2012). Flow back and produced water have similar characteristics, such as high total suspended solids (TSS) and TDS (Kim et al. 2019), and thus are frequently treated by coagulation processes (Esmaeilirad et al. 2015; Rosenblum et al. 2016). Activated carbon is one of the most widely used treatments for industrial and municipal wastewater due to its high ability to absorb organic compounds and relatively reasonable costs (Zhang et al. 2013). Activated carbon is useful for reducing hydrocarbons from saline water (Karapanagioti 2007) and surfactants in flow back and produced water (Wu & Pendleton 2001). GAC is preferred over powdered activated carbon (PAC) for shale oil and gas operations owing to the relatively low carbon usage rate and reasonable costs (Hackney & Wiesner 1996). The adsorption efficiency of activated carbon depends on a broad range of organic compounds as well as contact time, dose, activated carbon quality, size, adsorbate polarity, and salinity. However, activated carbon has significant limitations in removing poly and hydrophilic compounds. Rosenblum et al. (2016) investigated the application of chemical coagulants and PAC in reducing organic compound concentrations, such as TPHs, glycols, and dissolved organic carbon (DOC) from produced water in Colorado.

The guidelines for the possible treatment of produced water from conventional and unconventional oil and gas have been published (Andrew et al. 2005; Fakhru'l-Razi et al. 2009). Since limited investigations for main glycols used as hydraulic fracturing additives have been performed, this study focused not on the treatment but identified PG, EG, and EMGE in flow back and produced water after each treatment process. The objectives of this study were to assess the efficiency of EC, UF, GAC, and RO water treatment processes in reducing PG, EGME, EG, BTEX, TPH, DOC, TOC, TDS, and turbidity in flow back and produced water from shale oil and gas operations in Colorado.

Flow back and produced water were sampled from the Julesburg Basin in Weld County, Denver, between October 27, 2015, and April 8, 2016. Samples (20 mL) were transferred to glass vials for gas chromatography–mass spectrometry (GC–MS) and then stored at 4 °C before use. The samples represented mixed samples because flow back and produced water were collected directly from a tank located in the central processing facility.

Flow back and produced water samples were treated in a series of wastewater treatment processes, as presented in Supporting Information, Figure S1, and detailed below. After each treatment, the samples were transferred into 40 mL amber glass vials to measure the concentration of various water quality parameters.

Jar tests were performed using two metal plates connected to a power supply, and the system included one cathode and one anode. Continuous flow EC used the same power supply as the second-stage test and a SHURflo diaphragm pump, model No. 8000-953-238 (Flow Technologies Group of Pentair, Inc., Minneapolis, MN, USA), and the system was composed of six stacked iron plates and anode and cathode electrolytic cell linkages. Flocculation was conducted using a Phipps and Bird 900 Model Jar Tester (Phipps & Bird, VA, USA) at a speed of 15 rpm. A total of 60 L was processed with the mixer at a paddle speed of 50 rpm. The addition of deionized water with microbubbles generated pressurized microbubbles at 80 psi in the vessel.

UA420 E hollow fiber filters and a MANN + HUMMEL UA420-BT housing (MANN + HUMMEL, Ludwigsburg, Germany) were used during the UF process. The modified hydrophilic membrane material with a pore size of 0.05 μm was used. The water was pumped using a SHURflo diaphragm pump, model No. 8000-953-238 (Flow Technologies Group of Pentair, Inc., Minneapolis, MN, USA) with a flow rate of 0.5 gmp.

GAC testing was performed using three columns composed of 3-inch diameter PVC pipes, caps, nozzles, and plastic tubing. The tubing was used to connect the bottom of each column and was located upward at the entrance for the next column. Acid-washed mesh-activated charcoal coconut shells (12 × 30) were applied as the adsorption medium. The volume of the columns was approximately 24,000 cm³. A Masterflex peristaltic pump (Masterflex, IL, USA) was used, pumping at 50 mL/min. The GAC1230C AW adsorption medium served as isotherms and was combined with 400 mL of UF brine using 10 different masses of GAC (5–50% of mass). The same conditions were applied to the Phipps & Bird Jar Testers (Phipps & Bird, VA, USA) as explained in the EC section. A total of 400 mL of UF brine and coconut charcoal were mixed in a beaker at a speed of 25 rpm for 1 h. The brine–GAC mixture was filtered using a Whatman filter (Cytiva, Bucks, UK) with a 6-μm pore size to remove suspended inorganic carbon particles before analyzing the TOC content.

RO was performed using a Sterlitech SEPA CF cross-flow cell filter unit (Sterlitech, Auburn, WA, USA), and a GE Osmonics flat plate RO membrane (GE, Osmonics, TX, USA) was used to evaluate various materials and operating pressures. Detailed information on the AG and SE series is provided in Supporting Information in Table S1.

EG (CAS No. 106-21-1) was obtained from Fisher Scientific. PG (CAS No. 57-55-6), EG butyl ether (CAS No. 111-76-2), 1,3-propanedial (CAS No. 504-63-2), and ENVI-Carb Plus were purchased from Sigma Aldrich. Solid phase extraction (SPE) was performed based on a previously reported method with minor modifications (Sidisky et al. 2010). First, the cartridge was conditioned after rinsing with 1 mL of methylene chloride, 2 mL of methanol, and 3 mL of LC-grade water. The solvents were transferred to the cartridge, and the plunger of a glass syringe placed on top of the cartridge was pressed. A sample (5 mL) spiked with 100 μL of 500 μg/mL 1,3-propanediol (internal standard (IS)) in water was transferred to the cartridge and dried under a vacuum for 10 min. Glycols were eluted with 2 mL of methanol and methylene chloride in an 8:2 (volume/volume) ratio after 1 min of soaking. The elute was preserved at 4 °C until analysis and injected without adjusting the volume. The vacuum used for drying over 10 min was provided from a fume hood vacuum outlet and a syringe was used to push the liquid to the cartridge for sample loading and elution. Two samples from each treatment process were randomly selected to generate sample blanks without adding IS and extracted via SPE. Samples for the calibration curves were set by diluting authentic standards in methanol and spiking them with an equivalent amount of IS. The samples for the calibration curve were directly injected without performing SPE extraction.

Trace 1310 GC coupled with a Thermo ISQ-LT MS system (Thermo Fisher Scientific, MA, USA) was used to analyze the samples after SPE extraction and calibration curve generation. Samples (2 μL) were injected at a ratio of 5:1, the inlet was maintained at 250 °C, and the transfer line at 240 °C. Short chain fatty acids (SCFA) separation was performed on a 30 mm DB-WAX Ultra Inert (DBWAXUI) column (J&W, 0.25 mm inner diameter, 0.25 μm film thickness) (Agilent, CA, USA). The oven temperature was maintained at 110 °C for 1 min, increased at 10 °C/min to 160 °C, and increased further at 30 °C/min to 240 °C. Helium was used as the carrier gas and the flow was maintained at 1.2 mL/min. Selected ion monitoring (SIM) mode was applied for mass scanning at values of 31, 33, 41, 43, 45, 57, 58, 61, 62, 75, 87, and 100 at a rate of 10 scans per second. Methanol blanks were injected three times between sample runs to remove any carryover signals. The total ion chromatograms of the samples and standard are presented in Supporting Information, Figure S2. The mass spectra for PG, EMGE, EG, and IS are presented in Supporting Information, Figure S3.

Initially, samples after SPE for the calibration curve were prepared using water in equivalent amounts. Concentrations of 5, 10, 15, and 30 μg/mL were measured for the first set of calibration curves as suggested in a previous study (Sidisky et al. 2010). However, the results indicated that most samples had concentrations of approximately 0.1–5 μg/mL. A second set of calibration curves was then run with concentrations of 0.2, 1, 2, 10, and 20 μg/mL (see Supporting Information, Figure S4). All calibration curves had high coefficients of determination. The final calibration curves were generated based on the second calibration curves by applying four concentration points (0.2, 1, 2, and 10 μg/mL).

The coefficient of variation for the entire analysis, including sample preparation and data acquisition, was computed based on the three replicates. The coefficients of variations for EG and PG were 8 and 12%, respectively. Raw-produced water samples were rerun after prefiltration to reduce turbidity and remove suspended solids by using a 0.45 μm pore size filter.

The signal intensity for 1,3-propanediol in samples ranged from 1.5 × 10⁴ to 4.2 × 10⁴, which was less than 6% of the signal intensity for the spiked IS level. Variations were observed after each treatment; hence, the background was not subtracted. A summary of the spiked IS signals is provided in Supporting Information, Table S2.

A Shimadzu TOC-VCSH analyzer (Shimadzu, Kyoto, Japan) was used to measure the TOC. The standard method 2540 and EPA method 180.1 were used to measure TDS and turbidity, respectively. BTEX, TPH, Mg, Ca, Na, HCO₃, Cl, and B were measured at the ALS Global Environmental Laboratory (Fort Collins, CO, USA) using the USEPA Method 310.

Water quality parameters (TOC, DOC, TDS, EG, and PG) were determined by Tukey's test for multiple pairwise comparisons and the analysis of variance. R version 3.20 (R Core Team 2019) was used to perform the statistical analyses. Every water quality parameter after each treatment was considered significant when P-values were less than 0.05.

Flow back and produced water samples used in the water treatment processes were obtained from the central processing facility. The water quality characteristics varied depending on the sampling date because of temporal variability in flow back and produced water quality from this facility.

Table 1 presents the characteristics of the flow back and produced water from the central processing facility.

The pH value of produced water was approximately 6.6. A high TOC level indicated higher concentrations of dissolved cross-linkers remaining in the fracturing fluid. The relatively older produced water had a higher TDS level compared to the newer produced water. The relatively high fluctuations in TDS concentration may have been due to temporal variations. Bicarbonate, calcium, chloride, and sodium were considered the main ions in the produced water. Kim et al. (2016) showed that flow back and produced water from two different wells that used different types of fracturing fluids had similar TDS and TOC values and further assessed the temporal variations in Northeast Colorado for 200 days after flow back water production began. Magnesium and boron ions were also observed in the present study.

Significant differences were not observed between the raw flow back and produced water and the water samples after UF or EC treatment (P-values were higher than 0.99; see Supporting Information, Table S3). In general, activated carbon reduced the concentration of organic compounds in wastewater. However, the TDS concentration was reduced by approximately 15% after GAC (from 21,500 to 18,470 mg/L after GAC). Esmaeilirad et al. (2015) showed that divalent ions like Ca²⁺ or Mg²⁺, the main ions in flow back and produced water, got removed after EC treatment. However, EC removed approximately 1% of inorganic compounds in the present study. Some organic ions can be adsorbed on the GAC sorbent through accumulation based on the water quality. The most frequently adsorbed ion was Ca²⁺ (Lambert et al. 2002), which is considered a dominant ion in flow back and produced water (Barbot et al. 2013; Kim et al. 2016). Mg, Al, and Fe can also be adsorbed on the GAC sorbent (Lambert et al. 2002). The removal of approximately 15% TDS was probably due to the accumulation of these inorganic ions in the GAC bed. RO can effectively reduce more than 95% of the TDS concentration depending on factors that include membrane quality, water quality, and RO equipment condition (Funston et al. 2002; ALL Consulting 2003; Barrufet et al. 2005). Although RO reduced the TDS by approximately 90% in this study (18,470–2,550 mg/L), the TDS level was still high for reuse because recycled water with a TDS value higher than 2,000 mg/L can only be reused for the irrigation of tolerant plants and permeable soils (United States Environment Protection Agency 2004).

The BETX concentration from the raw flow back and produced water reached approximately 9,300 μg/L and mostly consisted of benzene (5,200 μg/L) and toluene (3,200 μg/L). GAC removed most of the BTEX by more than 95% (from 3,482 to 166 μg/L), which was similar to a previous finding in which approximately 95% of BTEX was removed after GAC treatment (Caetano et al. 2017). The GAC treatment process is usually applied to reduce organic compounds (Caetano et al. 2017), and the BTEX concentration complied with the discharge policy requirements except for benzene (discharge permitted: benzene (5 μg/L), toluene (100 μg/L), ethylbenzene (100 μg/L), and xylene (200 μg/L)) (Nevada Division of Environmental Protection Fact Sheet). The drinking water standards were also met, except for benzene (drinking water standard: benzene (5 μg/L), toluene (1,000 μg/L), ethylbenzene (700 μg/L), and xylene (10,000 μg/L)) (Water Research Center Database). Benzene has strict discharge and drinking water limits because it is considered carcinogenic (Water Research Center Database).

UF decreased BTEX by more than 60% (from 9,380 to 3,482 μg/L), consistent with a previous study that reported a 50% removal of BTEX from produced water in the North Sea after UF treatment (Bilstad & Espedal 1996).

Raw flow back and produced water contained 180 mg/L of TPH, which was mostly composed of gasoline range organics ((GRO); 150 mg/L; Figure 2).

The TPH concentration of raw flow back and produced water was higher than that reported by Rosenblum et al. (2016), who observed a TPH concentration of 8–20 mg/L from a well in Colorado and illustrated more than 90% TPH removal using a PAC dose of 1,000 mg/L.

EC removed approximately 80% TPH and 90% GRO, whereas only approximately 50% of diesel range organics (DRO) and oil range organics (ORO) were removed (DRO: from 16.0 to 7.3 mg/L; ORO: from 13 to 6.30 mg/L), indicating that smaller carbon chain TPHs (GRO (C5–C10)) were easier to remove than larger carbon chain ones (DRO (C10–C28) and ORO (C28–C36)). GRO, ORO, and DRO were decreased by more than 95% following the UF process, with ORO decreased to undetected concentration. TPH concentration following the UF process still required further treatment to comply with the discharge limits for ground remediation, which are less than 1 mg/L (Nevada Division of Environmental Protection Fact Sheet). TPH in flow back and produced water was removed by GAC. Therefore, RO was not applied for THP removal in flow back and produced water in this study.

EC removes up to 80% of TPH under ideal working conditions (Pérez et al. 2016). Activated carbon is the most commonly used method for treating municipal and organic wastewater because of its ability to remove various organic compounds. Ayotamuno et al. (2006) showed that more than 95% TPH removal following PAC or GAC (from 9,300 to 360 and 12 mg/L, respectively) using 5–25 g of activated carbon. However, since flow back and produced water usually have high TDS and TOC levels, these waters may reduce the efficiency of the wastewater treatment process. Previously, flow back and produced water TOC concentrations were measured in water samples from unconventional oil and gas wells, and the TOC level was stabilized at approximately 2,000 mg/L based on temporal variations and fracturing fluid types (Kim et al. 2016).

GAC removed more than 95% of TOC and DOC, confirming that activated carbon is the best method for reducing organic carbon. The P-values of TOC and DOC between the UF and GAC processes were lower than 0.001, meaning that TOC and DOC concentrations obtained by the UF process were statistically lower than those obtained by the GAC process (Supporting Information, Tables S4 and S5). Recent studies have reported higher TOC and DOC removal rates (95%) following GAC treatment compared with those presented by Rosenblum et al. (2016), who obtained reductions between 9.5 and 23.3% by PAC using a 1,000 mg/L dose from both vertical and horizontal oil and gas operations. Significant differences in the DOC (P-value = 0.94) and TOC (P-value = 0.98) values were not observed between the GAC and RO processes; however, the TOC and DOC levels decreased by 99% following the RO process; thus, the TOC removal efficiency obtained was higher than that observed in previous studies, in which TOC decreased by approximately 90% using a surfactant-modified zeolite membrane bioreactor RO system (Brookes 2005; Kwon et al. 2008). The EC treatment decreased TPH concentration by more than 8% and the DOC level by only 5%. Therefore, DOC accounted for a small fraction of TPH.

EGME was not detected in any of the samples, indicating that it might have not been used as an additive in the studied well or was degraded by high temperatures and pressure during hydraulic fracturing. The PG concentration was slightly higher than the EG concentration, varying from 0.07 to 5.39 and 0.07 to 5.52 μg/mL, respectively. The highest PG and EG concentrations were found in the fourth sample set (Supporting Information, Table S6).

The average EG and PG concentrations presented a decreasing trend based on the treatment processes. The GAC process removed a considerable amount of EG and PG that complied with drinking water standards (less than 1 μg/mL). Significant differences were not observed in the EG and PG values between the GAC and RO processes, as the P-values were higher than 0.9 (Supporting Information, Tables S7 and S8).

The PG and EG concentrations in the raw flow back and produced water were lower than those after the H₂O₂–EC process. As the raw flow back and produced water were collected from hydraulic fracturing, interference from the geological formation or fracturing additives may have occurred that negatively impacted the chromatogram despite the application of prefiltration and the SPE method. Figure 1 shows that the TDS level was insufficiently removed by any of the processes except for RO. Hence, the salts in flow back and produced water may not significantly influence the recovery of EG and PE in the chromatogram. Moreover, DOC and TOC may not have been affected by EG and PG in the flow back and produced water due to limited decreases in these parameters after EC and UF treatments.

Previous studies confirmed that the EC process can effectively remove turbidity (Kobya et al. 2003; Bukhari 2008; Merzouk et al. 2009). Turbidity can be removed more effectively by applying a softening process prior to the EC process (Esmaeilirad et al. 2015). In the present study, EC reduced turbidity by 85% (from 172 to 24 NTU) and UF by approximately 99% (from 24 to 1.1 NTU).

The EC process significantly removed the TPH (Figure 3) and turbidity (Figure 5), indicating that TPH and turbidity may interfere with the raw flow back and produced water samples. The recovery of EG and PG using the GC–MS method after applying ENVI-Carb Plus cartridges for SPE was not affected by the presence of salt and oil (Sidisky et al. 2010). Suspended solids, which have a strong positive correlation with turbidity (Hannouche et al. 2011), can block SPE cartridges (Tadeo 2008), indicating that low EG and PG recovery levels may be due to high turbidity levels in the raw flow back and produced water. Prefiltration was expected to decrease the turbidity and increase the recovery of EG and PG in raw flow back and produced water. Prefiltration was applied to five sets of raw flow back and produced water samples using a 10 mL NORM-JECT with a 0.45 μm pore size; however, prefiltration did not improve the recovery of PG and EG from raw-produced water in the current study. The PG and EG concentrations in the raw flow back and produced water may be higher than those after the H₂O₂/EC treatment. Thus, other pre-treatment techniques are essential for PG and EG recovery in raw water samples. Moreover, the use of peroxide during H₂O₂/EC might result in increased PG and EG concentrations due to the formation of glycols, such as polyglycols. Previous studies have shown the presence of polypropylene glycols and polyethylene glycols in produced water in Weld County, Denver, CO, USA (Thurman et al. 2017).

The study investigated specific organic compounds and water quality parameters in flow back and produced water and determined the effects of different treatment processes, including EC, UF, GAC, and RO. GC–MS with SPE was used to measure PG, EG, and EGME levels. EGME was not recovered in any of the raw flow back and produced water samples. However, PG and EG were recovered at values ranging from 0.07 to 5.52 μg/mL after the treatment process. The GAC treatment reduced both the EG and PG levels to values that meet acceptable levels and also reduced the levels of other organic compounds (TPH, BTEX, TOC, and DOC). TPH and turbidity were significantly removed by the EC process, and most of the TPH and turbidity were removed by the UF process. TDS values were effectively decreased by approximately 90% after the RO process. However, the TDS levels remained too high for water reuse in certain contexts.

Although some water quality parameters remained above the reusability threshold for most purposes, the treated water reached a level where it could be reused on tolerant plants and permeable soils. In-depth knowledge and understanding of flow back/produced water quality characteristics, prior and post water treatment processes, can improve water treatment process strategies, reduce wastewater discharge, and improve the consequent reusability of treated water.

This work was funded by MWH Global and Halliburton Inc. We thank Drew Caschette for his support in this project. This research was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (No. 2021R1C1C101040411).

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.