42 L of produced water (PW) samples, 16L of PW-1 and 26L of PW-2, were obtained from an oil extraction operation in South Kuwait. The PW had been submitted to a three-way separator and corrugated plate interceptors. Nevertheless, it still carried appreciable amounts of hydrocarbons. Spent coffee grounds (SCGs) were obtained from a selection of commercially available coffee brands. Pyridine (BDH) was dried over solid KOH and N,N-dimethylacetamide (Sigma Aldrich) was dried over MgSO₄. Octanoyl chloride (Sigma Aldrich), acetyl chloride (Sigma Aldrich), and palmitoyl chloride (Sigma Aldrich) were used without further purification. Potassium hydroxide (KOH pellets, Sigma Aldrich), sodium hydroxide (NaOH pellets, BDH Analr), anhydrous sodium carbonate (soda, eurolab), activated charcoal (Sigma Aldrich, acid-washed with hydrochloric acid), hexane (Panreac) and dichloromethane (CH₂Cl₂, Sigma Aldrich) were used as is.

Torrefaction of SCGs was carried out in a Carbolite oven. Post-reaction, samples were dried in an MMM Ecocell drying cabinet. Reaction mixtures were stirred on WiseStir magnetic stirrers. The ash contents of SCGs, the hydroxide extracted SCGs as well as the SCG-esters were determined by heating respective samples in crucibles (79C-00, Waldenwanger, Berlin) in a Carbolite oven at 600˚C for 3 h, where the mass of the resulting ash is expressed as weight % of the originally weighed-in sample. The initial produced water analysis was carried out at the CORE labs, Abu Dhabi, UAE. The ICP-OES (Inductively Coupled Plasma - Optical Emission spectroscopy) analyses of the salt crystals were carried out by CCIC Middle East FZE – Fujairah Branch. Both laboratories are accredited. For the ICP-OES measurements, an Agilent ICP-OES Model-5110 was used after digestion of the samples with HNO₃ in a microwave digestor. Infrared spectra of the SCG materials, the derived esterified cellulosic materials as well as of the crystallized calcium salts were carried out as KBr pellets with a Perkin Elmer Spectrum 2 FT-IR spectrometer. TOC measurements were performed with a Shimadzu TOC-4200. For the brine surface temperature readings, a Trister^® infrared thermometer TS-1/T1 was used with an initial calibration utilizing a normal immersion glass thermometer

For the adsorptive filtration of PW, two glass columns, A (length, 35 cm, inner diameter 2.25 cm) and B (length 35 cm, inner diameter 1.70 cm) were used. Both columns have an outflow that is controlled with a Teflon^® stop-cock. In experiment A, column A was filled with a well-mixed sorbent mixture of cellulose octanoate (2.40 g) and activated carbon (7.20 g, 1:3 w/w). In experiment B, column A was filled with a well-mixed sorbent mixture of cellulose acetate (2.40 g) and activated carbon (7.20 g, 1:3 w/w). In experiment C, column B was filled with a well-mixed sorbent mixture of cellulose palmitate (1.20 g) and activated carbon (3.60 g, 1:3 w/w). In all the experiments, sand (30 g) was added as a protective layer on top of the sorptive material. The columns were run with 1 atm pressure, initially at 1L per day (column A) and 500 mL per day (column B), respectively. The de-oiled water was collected in plastic jerry-cans, which thereafter were closed.

To the de-oiled PW, solid anhydrous sodium carbonate (Na₂CO₃, 7.0 g per 200 mL de-oiled PW) was added slowly, and the resulting mixture was allowed to stir for 10 min. at rt. The precipitate was filtered off, utilizing a sintered glass filter. The filtrate was evaporated in a two stage process, one outdoors and one indoors. This option was decided upon due to the simplicity of the experimental set-up, where soil temperature as well as solar irradiation contributed to the evaporation of the majority of the water. The last part of the evaporation was carried out under controlled conditions in a drying oven at 37˚C. PW was placed in open plastic containers and set outside. While experiments were performed from July to December 2023, with different temperatures in summer/autumn/winter leading to different evaporation rates, this paper focuses on work carried out in October and November 2023. Daytime temperatures in October ranged from 20˚C to 34.4˚C, in November from 19.4˚C to 30.2˚C, and in December from 18.7˚C to 28.9˚C (Figure 3). Typically, experiments were carried out with 500 mL of de-oiled PW. Crystals were collected directly from the evaporation pond and analyzed by ICP-OES (Agilent ICP-OES Model-5110). Alternatively, the crystals were collected by vacuum filtration using a sintered glass funnel, dissolved in a minimal amount of water, and then set for evaporation in a MMM Ecocell drying cabinet at 37˚C. Crystals were collected using a sintered glass funnel. Filtrates obtained from the respective operations were re-added to the evaporation ponds.

Adsorption filtration as a tertiary treatment for PW has been reviewed extensively [40] [41] [42] [43] . Biomass waste has been recognized as adequate sorption material [26] [44] [45] that is often cheaply sourced, if not always easy to be disposed of. Biomass waste can be transformed into activated carbon [46] [47] , but can also be used largely unprocessed or as a scaffold for chemical transformations leading to materials with a tailored surface. Ground nut-shells, specifically walnut shells and pecan nut shells [48] [49] [50] are standardly used as sorption material in the treatment of PW, also from oil and gas. Other biomass waste materials that have been investigated include corn cobs [51] , kapok (Ceiba pentandra) [52] , sago fronds (Cycas circinalis) [53] , biomass from the giant false agave (Agave gigantea) [54] , and empty fruit bunches of the oil palm (Elaeis oleifera) [55] . In all of these cases, the biomass was processed to obtain the cellulose content. In recent years, the possibility of using spent coffee grounds (SCGs) as biowaste derived sorption material for treating oily water was communicated [56] [57] [58] . This approach separates the cellulose from the lignin by aq. hydroxide extraction, and then optionally derivatises the so-obtained cellulose by either esterification or silylation. As SCGs are a major accumulating biowaste that is readily available, with globally 6 million tons produced annually

[59] , we opted to adapt the protocol described in ref. 56 and 57. An additional advantage of SCGs is that it is already in powdered form. After drying at 100˚C, CGS was subjected to extraction with 3w% aq. NaOH solution. It is here that much of the lignin which contains phenolic groups is separated from the remaining material, which is mostly cellulosic, by entering the aqueous phase as lignin salt. In this step, hemicellulose is partially also extracted as hemicellulose can be hydrolyzed by aqueous base more easily than cellulose. The cellulosic material was filtered off and dried. During the extraction process, 37.3% - 44.7% of the mass of the material was lost. The water used in the extraction process can be recycled by distillation. We obviated a bleaching process of the material with sodium hypochlorite (NaOCl).

Analysis of the infrared spectra of the original, dried SCGs and the extracted, dried SCGs reveal a number of differences, one of the main being the disappearance of the IR band at 1744 cm⁻¹, which is attributed to ester functions present in the lignin and potentially to polysaccharide pectins, the latter of which are less likely to be present in spent coffee grounds (Figure 4). The absorption of the ester carboxyl group has been studied both in lignin isolated from natural sources [60] , in synthetically modified, esterified lignins [61] as well as in esterified lignin components [62] and the absorption ranges between 1717 cm⁻¹ and 1760 cm⁻¹, with 1744 cm⁻¹ signaling an aliphatic ester function. The infrared spectrum shown in Figure 4(b) matches infrared spectra of cellulose from the literature [63] . Next, the cellulosic material was subjected to esterification reactions to produce a hydrophobic surface of an appreciable thickness. A larger number of procedures are known for the preparation of cellulosic esters [64] , some of which use N,N-dimethylacetamide (DMAc) as solvent. The authors opted to use acyl chlorides as reagents in DMAc as solvent and pyridine as catalyst and HCl scavenger (Figure 5), obviating the use of the known DMAc-LiCl system [65] . The resulting cellulosic esters were filtered, washed with CH₂Cl₂ - hexane and dried. IR spectra of the esters (Figure 5) showed the characteristic ester carboxyl band at 1748 cm⁻¹. A comparison of the IR spectrum of the cellulose acetate (Figure 6(a)) obtained showed that it exactly matches the IR spectrum of commercially available cellulose acetate [66] . The ash (mostly inorganic) content of the sieved and dried SCGs was found to be

1.11 ± 0.06 w%. After the SCGs were extracted with 3w% aq. NaOH to separate out the lignin, the ash content of the remaining cellulosic material was measured at an appreciable 10.96 ± 1.23 w%. This value was seen to be significantly decreased upon acylation of CGS, with 4.86 w% ash content for cellulose palmitate and 0.26 w% ash content for cellulose acetate. It is believed that salts including remaining sodium ions were still attached to the extracted CGS, which were then released upon reaction of the hydroxyl groups with the acyl chlorides.

For the adsorptive filtration (Figure 7), two glass column were used, both 35 cm in length, but one with an inner diameter of 2.25 cm (column A) and the other with an inner diameter of 1.70 cm (column B). Column A was filled with cellulosic ester (octanoate or acetate, 2.40 g) and commercially available activated carbon (7.20 g), column B was filled with cellulose palmitate (1.20 g) and activated carbon (3.60 g). To the sorbent layer was added a layer of sand (30 g/15 g) as protection. Thereafter, PW was subjected to adsorptive filtration on these mixed sorbents, with 15.2 L PW-1 run over the cellulose-octanoate/AC column, 17.1 L PW-2 run over the cellulose-acetate/AC column, and 7.7 L PW-2 run over the cellulose-palmitate/PW column. One reason that a mixture of cellulose-ester and AC was used rather than cellulose-ester alone was the worry that due to the small particle size sole use of cellulose ester might lead to clogging once enough oil had been adsorbed. The small particle size of the sorbent and the use of sand as protective layer made back-washing unsuitable for this set-up, in contrast to using walnut shell granules as sorbent material [50] [67] [68] . Thus, clogging, even for the mixed sorbent, limits the amount of PW that can be treated per g sorbent material rather than a perceived concentration of effluent oil at the column outflow in form of an oil effluent breakthrough. The TOC of the effluent water was measured to be below 50 ppm. In the case of using cellulose-octanoate/AC as sorbent for the filtration of 15.2 L of PW 14.2 mL oil could be retained. This oil could be recovered, partly by squeezing out the oil from the sorbent material. The oil showed a density of 0.9586 g/mL corresponding to an API gravity of 16.02. The sulfur content of the recovered oil was found to be 5.42%. While PW is now de-oiled, the salt content has remained the same. This was true for PW-1 filtered over CGS-octanoate and for PW-2 filtered over CGS-acetate or CGS-palmitate.

After de-oiling the PWs, the salt inherent in PWs was to be precipitated as industrial salt. As the adsorptive filtration of PW does not change its inorganic ion content, Table 1 shows the approximate ion concentration in the PW under study, after being de-oiled by adsorptive fitration over CGS-octanoate and Table 2 shows the approximate ion concentration of PW-2 after being de-oiled either by CGS-acetate or CGS-palmitate. The density of PW-2 at this stage was ca. 1.113 g/mL. After de-oiling, PWs maintain high Ca [7670 ppm for PW-1], Mg [1730 ppm for PW-1] and Sr [255 ppm for PW-1] concentrations. All of these divalent earth-alkaline metals form salts that are scalants, and their presence in industrial salt (10 pound brine) is unwanted. Thus, their concentration needs to be reduced significantly. The overall idea was to remove the earth-alkaline ions as carbonates under necessarily basic conditions, acidifiy the filtrate and subject the filtrate to evaporation, first in an outdoor evaporation pond. Figure 8 shows the overall process schematically.

To achieve a carbonate concentration in solution in order to remove the earth-alkaline metal species as carbonates can be achieved principally in two different ways – 1) by saturation of the aq. solution with CO₂ which could be obtained as part of flue gases from a separate process or 2) by the addition of solid, but water-soluble alkali metal carbonates. In both cases, it must be noted that in an equilbrium the carbonate concentration in the aq. solution is pH dependent as shown in Figure 9, with the equlibria shown in Figure 10 playing a role. In the case of adding solid alkali metal carbonates to the solution it can be expected that the reaction to the little soluble earth alkaline metal carbonates such as calcium carbonate (CaCO₃) is faster than the carbonic acid/bicarbonate/carbonate equilibrium can establish itself.

When adding aq.1N KOH to PW, both in presence and in absence of CO₂, the pH of the analyte moves to about pH 10, where it then remains upon further addition of aq. KOH for a time (Figure 11). It is here that reactions take place with the earth-alkaline metal ions, where metal hydroxides are formed in the absence of CO₂ and a mixture of metal carbonates and metal hydroxides are formed in the presence of CO₂ (see below). The low solubility of both carbonates and hydroxides mean that all added hydroxide ions are bound up in the precipitated metal salts. Only upon reaction completion does the pH increase upon further addition of KOH. It should be noted that the addition of aq. KOH to synthetic aq. sodium chloride solution of a concentration similar to that in PW shows no such buffering zone in respect to change of pH.

Indeed, adding aq. 1N KOH or aq. 1N NaOH solutions dropwise to the PW in the presence of CO₂ leads to a rapid crystallization of a colorless solid which becomes more pronounced with increased pH. Filtration yields a colorless solid, that is hygroscopic, which indicated that a certain amount of Ca(OH)₂ would be found in the mixture. Indeed, the IR spectrum shows a strong band for the OH stretching vibration of Ca(OH)₂ at 3700 cm⁻¹. The CO stretching vibration of CaCO₃ at 1445 cm⁻¹ is less in evidence (Figure 12(a)). If the mixture is stirred under CO₂ for a further 6 h, Ca(OH)₂ converts to Ca(CO)₃, driven also by the different solubilities of the salts [solubility product K_sp(25˚C) CaCO₃ = 4.96 × 10⁻⁹; K_sp(25˚C) Ca(OH)₂ = 4.68 × 10⁻⁶] [72] .

Figure 12(b) shows the IR spectrum of the filtered solid collected afterwards, clearly showing that CaCO₃ is by now the main constituent. The solid was also analyzed by completely dissolving it in HCl and subsequently determining the total hardness of the solution by standard titration with ethylenediaminetetraacetic acid (EDTA) [73] with the idea that the w% of Ca in Ca(OH)₂ is quite different from that in CaCO₃.

Well realizing that EDTA also complexes Sr²⁺ and Mg²⁺, the results nevertheless clearly show that Ca(OH)₂ comprises less than 5% of the material, although a OH-band is still visible at 3700 cm⁻¹ in the infrared spectrum. Upon addition of solid Na₂CO₃ to PW (7 g for every 200 mL PW), precipitation is very rapid. In the case of an addition of Na₂CO₃ all at once to PW, the filtered solid shows up to 10.8 w% Na in the sample, in addition, 3.5 w% Mg can be found in the solid as well as Sr (8620 ppm) and K (3050 ppm). Upon a portion-wise addition of Na₂CO₃ to PW, lasting a minute, 2.28 - 2.32 w% of Na were found in the filtered solid, Here, also Mg (3.41 w%) and Sr (1.06 w%) were observed. When the filter cake was washed with a minimum amount of water (20 mL per 50 g precipitate), the amount of Na in the CaCO₃ can be reduced to 1.27 w%. It can be expected that all earth alkaline metals precipitate predominately as carbonates. If this is the case, more than 94% of the precipitated calcium is in form of calcium carbonate. At this point, the spectrum of the solid also lacks the tell-tale absorption band in the infrared at 3700 cm⁻¹ for Ca(OH)₂, signaling its absence (Figure 13). In addition, most of the heavy metal content can be removed from PW in this way. In various experiments of adding Na₂CO₃ to de-oiled PW, heavy metals could be noted in the precipitate such as Cu (3.8 - 8.9 ppm), Mn (up to 4.5 ppm), and Ni (up to 6.7 ppm), while the concentrations of these metals in the NaCl crystals gained from the filtered PW were all found to be <0.1 ppm (see below).

Crystallization of salt by evaporation of the filtered and slightly acidified PW gave sodium chloride (NaCl) crystals in their typical cubic shape (Figure 14). The first harvesting was performed after evaporation of 37.5 vol% of the PW-1, the second harvesting after evaporation of overall 67.5 vol% of PW-1, and the final harvesting was carried out after evaporation of 38.7 vol% of the remaining PW-1, leaving 20 vol% PW-1 at the end. Direct evaporation of PW without removal of calcium ions via the carbonate process gave calcium concentrations in the salt crystals of 200 - 400 ppm and magnesium concentrations of 130 - 150 ppm in the first harvest. The last harvest gave salt crystals with a calcium (Ca) concentration of 5100 pm, a magnesium (Mg) concentration of 1490 ppm and a strontium (Sr) concentration of 185 - 190 ppm. In addition, heavy metals such as copper (Cu, 0.56 ppm) and manganese (Mn, 0.36 ppm) could be found.

The first harvest of PW-1 where calcium had been removed by the carbonate method showed Ca concentration of 32 - 46 ppm, with potassium (K) at 39 - 105 ppm, Mg at 54 - 95 ppm and Sr at 2.4 ppm and with lead (Pb), Cu, Mn and nickel (Ni) at <0.1 ppm. The purity of NaCl salt obtained was calculated to be 99.2 w%. The final harvest of the decalcified PW-1 gave a Ca concentration of 220 ppm. Also in evidence were Mg (190 ppm), K (123 ppm) and Sr (11.8 ppm). All heavy metal conc. (Cu, Pb, Mn, Ni) were found to be below 0.1 ppm.

Salt crystals from PW treated with aq. NaOH or aq. KOH in the presence of CO₂, where the solid precipitate formed was immediately filtered after addition of the base, still showed relatively high values of Ca (up to 420 ppm) even in the first harvest due to the fact that in this case there is still an equilibrium that establishes itself among Ca(OH)₂ and CaCO₃ via the species CO₂, HCO 3 − and CO 3 2 − .

NaCl production was also carried out with PW-2, de-oiled by adsorptive filtration on CGS-palmitate. Here, de-oiled PW was evaporated in 500 mL batches, Here, the first harvest was taken after the brine was reduced to 50 vol%, the second harvest when it was reduced to 26.0 vol% and the third harvest when it was reduced to 16.0 vol%. Table 3 shows the content of the obtained salt, where the composition and NaCl purity was calculated upon subtraction of the moisture content of the samples.