Graphene Oxide Membranes for High Salinity, Produced Water Separation by Pervaporation

Abstract

Oil and gas industries produce a huge amount of wastewater known as produced water which contains diverse contaminants including salts, dissolved organics, dispersed oils, and solids making separation and purification challenging. The chemical and thermal stability of graphene oxide (GO) membranes make them promising for use in membrane pervaporation, which may provide a more economical route to purifying this water for disposal or re-use compared to other membrane-based separation techniques. In this study, we investigate the performance and stability of GO membranes cast onto polyethersulfone (PES) supports in the separation of simulated produced water containing high salinity brackish water (30 g/L NaCl) contaminated with phenol, cresol, naphthenic acid, and an oil-in-water emulsion. The GO/PES membranes achieve water flux as high as 47.8 L m^-2 h^-1 for NaCl solutions for membranes operated at 60 °C, while being able to reject 99.9% of the salt and upwards of 56% of the soluble organic components. The flux for membranes tested in pure water, salt, and simulated produced water was found to decrease over 72 h of testing but only to 50-60% of the initial flux in the worst-case scenario. This drop was concurrent with an increase in contact angle and C/O ratio indicating that the GO may become partially reduced during the separation process. Additionally, a closer look at the membrane crosslinker (Zn²⁺) was investigated and found to hydrolyze over time to Zn(OH)₂ with much of it being washed away during the long-term pervaporation.

Keywords: 2D materials; desalination; graphene oxide; membranes; oil/water separation; pervaporation; produced water.

Figure 1

(a) Illustration of rejection of chemicals and purification of water via the use of GO membrane pervaporation; (b) method of production of GO membranes; (c) schematic diagram of the pervaporation separation apparatus; (d) schematic of the membrane test cell module showing the flow directions and permeate outlet.

Figure 2

(a) AFM images of graphene oxide flakes spun coat onto Si wafer, (b) histogram of lateral sheet size distribution. The inset shows a photo of a 50 µg cm⁻² GO/PES membrane prepared by vacuum filtration. (c) FTIR spectrum for GO sheets; and (d) cross-sectional SEM image showing surface topology and thickness of a 50 µg cm⁻² GO membrane and the underlying PES support.

Figure 3

Membrane stability enhancement by Zn²⁺ crosslinking: (a) vacuum filtered GO membrane immersed in water for 30 min; (b,c) squares of membrane exposed to high speed mechanical stirring with no zinc treatment: squares in (b) have no Zn²⁺ treatment while the membrane in (c) was soaked in the Zn²⁺ solution for 24 h.

Figure 4

Average water flux and solute rejection for: (a) NaCl solution and pure water at different GO loading; (b) NaCl solution tested by 75 µg cm⁻² GO loading; (c) single organic and PWM solutions; (d) single organic (phenol solution). For (c,d), a 50 µg cm⁻² GO loading was used. The error bars were estimated as ± one standard deviation of three independent measurements.

Figure 5

Example UV/vis spectra for one of the collected samples, and its normalized curve matching the initial PWM curve.

Figure 6

(a) flux/flux₀ (J/J₀); (b) and rejection vs. time for 50 µg cm⁻² GO membrane for the tested solutions.

Figure 7

(a) XPS survey data; (b) C/O ratio (red bars) and Zn atomic % (black bars), high resolution XPS for (c) C 1s and (d) Zn 2p for the unused and used membranes for different solutions.

Figure 8

Average contact angle for unused and used GO/PES membranes tested for different solutions. The error bars were estimated as ± one standard deviation of three independent measurements.