The environmental impact of shale oil and gas production by hydraulic fracturing (fracking) is of increasing concern. The biggest potential source of environmental contamination is flowback and produced water, which is highly contaminated with hydrocarbons, bacteria and particulates, meaning that traditional membranes are readily fouled. We show the chemical functionalisation of alumina ceramic microfiltration membranes (0.22 μm pore size) with cysteic acid creates a superhydrophilic surface, allowing for separation of hydrocarbons from frac and produced waters without fouling. The single pass rejection coefficients was >90% for all samples. The separation of hydrocarbons from water when the former have hydrodynamic diameters smaller than the pore size of the membrane is due to the zwitter ionically charged superhydrophilic pore surface.
Samuel J.Maguire-Boyle1, Joseph E. Huseman1, Thomas J.Ainscough2, Darren L. Oatley-Radcliffe2, Abdullah A. Alabdulkarem3, Sattam Fahad Al-Mojil4 & Andrew R. Barron1,2,5
Membrane fouling is essentially eliminated, while a specific flux is obtained at a lower pressure (<2 bar) than that required achieving the same flux for the untreated membrane (4–8 bar).
Although the long-term solution to global energy needs must be based on renewable sources, the present demand for oil and gas shows no sign of a reduction. Unfortunately this comes at a price in terms of resource consumption and possible environmental hazards1,2,3,4,5. The introduction of horizontal drilling coupled with hydraulic fracturing or ‘fracking’ has increased the cost of effective access to oil and gas resources, but water usage and wastewater remains controversial issues6,7. Hydraulic fracturing uses on average 20 million litres of water per well, of which only 10–15% is typically recovered during the flowback stage8. Flowback water as well as post-well completion water (production or produced water) are contaminated with hydrocarbons9, many of which are classified as hazardous, which along with significant bacteriological content means that this water cannot be reused without significant treatment10. The recovery of clean water for recycling and reuse has been viewed in the past by industry as being economically untenable as produced water is notoriously difficult to purify. As a consequence in most cases the waste water is either evaporated or disposed of through deep injection into abandoned gas or oil wells11. In either case storage and transportation of enormous volumes is required, with the potential risk of spillage. Since fresh water is a valuable commodity, wastage is not acceptable, especially in regions of continued drought. Thus, the recyclability of frac (and produced) water is a desirable goal from the environmental and economic viewpoint12.
Recyclability of many industrial wastes has been undertaken using ceramic filtration membranes, primarily for their robust nature and the ability to have select pore sizes with narrow distributions. Unfortunately, their capacity to purify or otherwise separate material has many drawbacks, such as membrane fouling leading to low permeate flux13. These drawbacks have precluded treatment of flowback and produced water because of submicrion particles, colloids, and hydrophobic organics14. Oil emulsions are generally removed by particle filtration (10–1000 μm pore size) and certainly by microfiltration (0.1–10 μm pore size). Shale gas produced water contains 1,000–50,000 mg/L of hydrocarbons, which are divided into the saturate, aromatic, resin and asphaltene (SARA) groupings9. While the emulsions represent a challenge with regard to fouling14, the solubilized SARA chemicals pose an issue for separation since their molecular weights are much lower than the cut-off for microfiltration. Use of a microfiltration membrane is FDA approved for bacteria removal from water, but hydrocarbons (even high molecular weights) pass through such membranes. Natural organic matter is generally in the 0.001–0.1 μm size, which requires ultrafiltration (0.005–1.0 μm pore size) or nanofiltration (0.0005–0.005 μm pore size). Produced water contains natural organic matter, as well as sugars derived from guar gum (used in frac fluids)15,16. If cost and energy consumption were not an issue, then the removal of these impurities may be possible using a multi-stage process to remove each component. It has been suggested that membrane distillation (MD) is an attractive treatment option for shale gas produced water because of its ability to handle high salinity as well as the inherent geothermal heat available to this process17,18.
What is needed is a process that removes hydrocarbons and resins at the same time as bacteria and particulates. Due to fouling ultrafiltration and nanofiltration are not suitable14. The interaction between membrane surfaces and solutes plays an important role in determining the extent of membrane fouling, explained by the mechanisms of pore blocking, cake formation or hydrophobic interaction19,20. Hydrophobic interaction between solutes and membrane material is frequently accepted as one of the predominant mechanisms. Membrane fouling is expected to be less severe with hydrophilic than hydrophobic membranes21,22,23. In addition, the high permeate flux for aqueous eluants is superior to hydrophobic membranes24.
There have been various approaches to alter ceramic membranes to be more hydrophilic with antifouling properties: coatings25, graft polymerization26, and metal substitution27,28. However, the chemistry of alumina microfiltration membranes offers the ability to create direct functionalization of the surface (using carboxylic acids) without changing pore size or membrane stability29. Furthermore, the use of different functional groups on alumina surfaces allows for changes in the wettability of the surface30,31. Cysteic acid (HO2CCH(NH2)CH2SO3H) functionalization (Fig. 1a) forms a super hydrophilic surface (contact angle of 5°, see Fig. 1b,c)32.
Figure 1(a) Schematic representation of the cysteic acid functionalized alumina surface. Photographic image of a water droplet on (b) unfunctionalized alumina and (c) cysteic acid functionalized alumina surface taken immediately upon dropping on the surface since within a few seconds the droplet completely wets the surface. EDS of (d) as received alumina membrane and cysteic acid functionalized alumina membrane magnified on (e) the carbon nitrogen and oxygen peaks and (f) on the aluminium and sulfur peaks. SEM image (g) of cysteic acid functionalized membrane, with associated EDS maps of cysteic acid functionalized membrane of (h) aluminium, (i) oxygen, (j) sulfur, (k) nitrogen, and (l) carbon.
Functionalization of the surface of a ceramic membrane also controls the flux rate through the membrane: generally, the more hydrophilic a surface the higher the flux32. Our goal is to show that functionalization of an alumina ceramic membrane with cysteic acid should increase the flux through the membrane for a particular pore size and that cysteic acid-functionalized alumina membranes can separate oil emulsions from frac and produced water.
Results and Discussion
A membrane average pore size of 0.22 μm was chosen since this is in the high end of the ultrafiltration range and low end of the microfiltration range. The membrane should reject oil emulsion, bacteria and particulates by size exclusion; however, sugars and other soluble organic compounds should not be rejected. The pore size is sufficiently small that fouling occurs rapidly with types of components in produced water.
The membranes were functionalized by reacting at 85 °C with cysteic acid in DI water. Reactions were initially carried out in a static reaction vessel; however, it was found that for more than one membrane it is easier to perform the reaction by flowing the cysteic acid solution through the membrane. The treated membranes are very hygroscopic and will attain a wet appearance, even after being dried, even if they are stored in a low humidity environment. The surface area of the membrane (0.358 m2/g) is unchanged upon surface functionalization.
SEM analysis of the surface of the functionalized membranes is indistinguishable from the untreated membrane, but the difference can be seen from EDS analysis. In the EDS spectrum both sulphur and nitrogen peaks are observed for the functionalized membrane (Fig. 1e,f), which are absent for the un-functionalized membrane (Fig. 1d). Mapping of the functionalized membrane for nitrogen and sulfur showed even coverage across the entire cross section (Fig. 1h–l). Nitrogen and sulphur signals were not diminished after the membrane was washed repeatedly or after treatment of multiple batches of frac or produced water, confirming that the cysteic acid is covalently bound to the alumina surface33.
Produced water analysis
The total carbon (TC), non-purgeable organic carbon (NPOC, also known as total organic content or TOC), and total inorganic carbon (TIC) for each produced water sample were measured (Table 1). In our previous report of the quantitative analysis of produced water samples, we found that the TOC of a shale produced water from wells recently drilled varies between 3 and 58 g/L9. Some of the samples studied herein are from dry gas wells and thus represent examples of water with lower initial carbon content.
Table 1: TC, NPOC, and IC content (mg/L) of feed, concentrate, and permeate water for various frack flowback and produced water purified using a cysteic acid functionalized 0.22 μm α-alumina membrane.
The NPOC is significantly higher than the TIC for most of the samples; however, both Roosevelt (Utah) and Scott Sugg (Texas) samples show a TIC:NPOC ratio >1, i.e., the carbonate content is higher than the organic content. There is no relationship between TIC and NPOC, which is consistent with the variation of geology between various shale reservoirs. The important point from this observation is that the source waters vary considerably and any process for treatment will have to deal with a range of feed compositions.
Qualitative and quantitative analyses were undertaken using the setup described in the Fig. 2. Initial analysis by inspection clearly demonstrated the ability of the membrane to screen colloidal hydrocarbons. The clarity of the permeate sample compared to the feed and the concentrated samples indicated that the membrane successfully removes large amounts of water contaminate (Fig. 3a).
Figure 2 Schematic representation of membrane filtration unit with (a) feed inlet, (b) recirculation tank, (c) pump suction line (d) pump, (e) pump discharge valve, (f) pressure gauge pre-filter, (g) gasket holding membrane in housing, (h) membrane housing, (i) ceramic membrane, (j) gasket holding membrane in housing, (k) pressure gauge post-filter, (l) concentrate return line valve used to vary trans-membrane pressure, (m) concentrate return line, (n) permeate pressure gauge, (o) permeate control valve, (p) permeate line.
Analysis of the carbon content of both permeate and concentrate samples in comparison to the feed sample showed that there was significant removal of carbon material in the production water. The carbon of the permeate sample was in the low ppm range (Table 1). The same extremely low carbon content was evident for all types of water analysed. It is worth noting that while only the Roosevelt produced water comes close to being within EPA limits (0.01–1.0 mg/L) after purification, all of the samples are cleaned sufficiently for re-use in a hydraulic fracture or a water flood, allowing for re-use.
Figure 3 (a) Visual inspection of Marcellus shale produced water sample before and after filtration showing the feed (left), concentrate (center), and permeate (right) samples. Representative GC traces for Marcellus produced water using CHCl3 extraction for (b) feed, (c) retentate, and (d) permeate. The large peak in the permeate trace is due to the CHCl3 reference.
The percentage retention of carbon content by the membrane (a measure of the ability of the membrane to remove carbon from the permeate) was calculated using R = 1−(Cp/Cr), where R = rejection coefficient, Cp = permeate concentration and Cr = retentate concentration. For TC and NPOC the rejection coefficient was generally in the range 0.95–0.99 for produced water samples (Table 2). The Scott Sugg flowback sample shows the lowest rejection coefficients.
Table 2: Rejection coefficient (r) for total carbon (TC), non-purgeable organic carbon (NPOC), and total inorganic carbon (TIC) for various frack flowback and produced waters purified using a cysteic acid functionalized 0.22 μm α-alumina membrane.
The carbonate concentration (IC) in the feed is higher than the organic content (NPOC); however, NPOC in the permeate is reduced compared to the feed. The sample taken from the evaporation pit also showed a slightly lower rejection coefficient for a single pass; however, analysis of several samples collected from different locations in the pit showed a variation from 0.78–0.92. The rejection coefficient for TIC (>90%) for the samples, except the Scott Sugg, which is one of the highest TIC:NPOC ratio materials, and the low rejection is possibly associated with the low level of organic present.