Materials

The bentonite used as the support was produced in Zhejiang, China, and consists mainly of Ca-montmorillonite which has a unit-cell formula, based on elemental analysis, of Ca²⁺_0.41(Si_7.98Al_0.02)(Al_3.00Mg_0.80Fe³⁺_0.20)O₂₀(OH)₄. The water layer is typically ~3.8–4.1 Å thick when Ca²⁺ is the exchangeable cation, given that the basal spacing is 14.2–14.5 Å (Murray, Reference Murray2006).

The transformer oil for experiments in the laboratory was obtained from Gaoqiao Petrochemical Company of SINOPEC, Shanghai. The water content was ~10–40 ppm.

Preparation of Sodium Polyacrylate (PAA-Na)

Acrylic acid (Shanghai Macklin Biochemical Co., Ltd, Shanghai, China) and 20% NaOH (Shanghai Macklin Biochemical Co., Ltd, Shanghai, China) were used as raw materials, N,N-methylene bisacrylamide (CHEMBEE, Shanghai, China) was used as the crosslinking agent, and potassium persulfate was used as the initiator. PAA-Na was prepared by the aqueous solution polymerization method, as follows: 10 g of acrylic acid was added slowly to 40 mL of 20% NaOH solution in an ice-water bath for neutralization, and then 0.0195 g of initiator (potassium persulfate) and 0.03 g of crosslinking agent (N, N-methylene bisacrylamide) were added and stirred fully until dissolved. The solution was added to a flask equipped with a stirrer, condenser, and thermometer. The hydrogel was obtained after stirring on a magnetic heater in an 80^oC water bath for 3 h. Finally, the hydrogel obtained was dried in a vacuum oven at 80^oC for 24 h to obtain PAA-Na, and then the PAA-Na was ground into powder (>200 mesh) by hand with an agate mortar and pestle.

Preparation of Adsorbent

The bentonite adsorbent utilized in this study was prepared by kneading. To develop a series of modified bentonites, bentonite was mixed with various amounts of PAA-Na powder (10, 20, or 30 wt.%), then kneaded with a dough hook and mixer by adding water (30% of the mass of bentonite), and lastly squeezed into strips using a screw extruder (VanKood, Clay Sculpture Tool, Shenzhen, Guangdong, China).

The adsorbent was dried for 24 h at 80^oC in an oven. Finally, the adsorbent was crushed and sieved. The 20–40 mesh particles were used for the dehydration test. The evaluation of the adsorbent in the laboratory was carried out in a fixed-bed microreactor equipped with a constant-flow pump to control the flow rate. Adsorbent was packed between two layers (each layer is 2 cm) of quartz sand (40–60 mesh) and inserted into the microreactor. The modified bentonite used in all experiments was 20 wt.% PAA-Na unless otherwise specified.

Water Content

The trace-water content of the transformer oil was measured every hour after dehydration using a Karl Fischer Titrator (MKC-710B, KEM, Kyoto, Japan). The dehydration rate, X%, was calculated using the following equation (Eq. 1):

(1) $X=\left(1\text{-}{C}_i/C_0\right)\times 100$

where C ₀ is the water content of the feedstock and C _i is the water content of the products. The adsorption process was performed at 25^oC, atmospheric pressure, and a weight hourly space velocity of 8 h^–1. Weight hourly space velocity is defined as the weight of feed flowing per unit weight of the adsorbent per hour. It can determine the treatment capacity and operating conditions of adsorbents.

Adsorption Isotherm

The adsorption isotherms of water in the unmodified bentonite, 20 wt.% PAA-Na modified bentonite, and PAA-Na were studied at 10 and 25^oC. The Freundlich and Langmuir models were used to describe the equilibrium characteristics as given by the following equations (Eqs 2, 3):

(2) $V=K_F·{\left(P/P_0\right)}^{1/n}$

(3) $V/\left(V_L-V\right)=K_L·\left(P/P_0\right)$

where V is volume adsorbed (cm³/g), P is Presure (kPa), P ₀ is the saturated vapor pressure of water at the corresponding temperature (kPa), K_F and n are Freundlich constants, K_L is the Langmuir constant.

Characterization

Fourier-transform infrared (FTIR) spectra were recorded using a Nicolet-6700 (Thermo Scientific, Waltham, Massachusetts, USA) FTIR spectrophotometer by mixing the sample with dried KBr and scanning at a resolution of 4 cm^–1. Thermogravimetric analysis (TGA) of the samples was carried out using a TA SDT Q600 instrument (TA Instruments, New Castle, Delaware, USA) by heating samples in the range 35–800°C at a heating rate of 10°C/min under air flow (30 cm³/min). X-ray diffraction (XRD) of the representative samples (bentonite and PAA-Na-modified bentonites) was performed using a Rigaku-3016 (Rigaku, Akishima, Tokyo, Japan) instrument with CuKα radiation (λ = 0.154 nm), a generator voltage of 45 kV and a current of 25 mA, a step size of 0.02°2θ, and a scanning rate of 1°2θ/min from 3 to 50°2θ. Scanning electron microscopy (SEM) of the sample was carried out using a JEOL-JSM6301-F Electron Microscope (JEOL, Akishima, Tokyo, Japan) with an Oxford INCA/ ENERGY-350 microanalysis system (Oxford Instruments, Abingdon, Oxfordshire, England). Surface area and average pore diameter of the samples were measured by N₂ adsorption-desorption using a JWGB-JQ200C (JEGB, Beijing, China) surface area analyzer. Samples were degassed under vacuum at 120^oC for 2 h prior to the adsorption measurement to evacuate the adsorbed moisture. The surface area was calculated by the Brunauer-Emmett-Teller (BET) equation. Micropore volume-size distribution was calculated using the method of Horvath-Kawazoe (H-K) (Horváth & Kawazoe, Reference Horváth and Kawazoe1983) and mesoporous volume-size distribution curve was calculated using the Barrett-Joyner-Halenda (BJH) method (Seaton et al., Reference Seaton, Walton and Quirke1989) according to the adsorption branch.

FTIR Spectra of Sodium Polyacrylate (PAA-Na)

The (FTIR) spectrum (Fig. 1a) of PAA-Na showed a characteristic peak at 3420 cm^–1, which was attributed to the hydroxyl stretching modes of the adsorbed water. The peaks appearing at 2926 and 2853 cm^–1 were attributed to the methyl and methylene stretching vibrations. The peaks appearing at 1024 and 1411 cm^–1 were attributed to the sulfonic and amide stretching vibrations. The peaks appearing at 1561 cm^–1 were attributed to the asymmetric and symmetric stretching vibrations of the carboxylate anion. Moreover, the peak corresponding to the carbon-carbon double bond between 1620 and 1630 cm^–1 disappeared, suggesting the carbon-carbon double bond was broken and participated in the reaction. The presence of these hydrophilic groups revealed by the infrared spectra proved the hydrophilicity of PAA-Na, and the appearance of the above characteristic peaks illustrated the successful synthesis of PAA-Na (Liu et al., Reference Liu, Huang, Pan, Wang and Li2021)

Fig. 1 a FTIR spectra of PAA-Na, FTIR spectra of bentonite and modified bentonite b before and c after water adsorption

Water adsorption was monitored in situ using FTIR (Fig. 1b, 1c). The intensity of the hydroxyl stretching band of the bentonite and modified bentonite increased after water adsorption. The peak area in the range 3600–2500 cm^–1 increased linearly with the amount of water adsorbed, by 13.97% in the unmodified bentonite and by 31.19% in the modified bentonite. The peak area of modified bentonite increased more because carboxyl creates a hydrogen bond with water in addition to hydroxyl.

Dehydration Test of Various Adsorbents

The PAA-Na powder swells several times its volume after water adsorption (Fig. 2), which increases the fixed-bed pressure and, thus, is not conducive to application in industrial oil dehydration. In contrast, after kneading with bentonite, the adsorbent volume did not change significantly after water adsorption. The stable physicochemical properties make PAA-Na-modified bentonite a good candidate for application in industrial oil dehydration. The steady dehydration rates of various adsorbents are shown in Table 1. Zeolite A is one material that is used extensively in industry to dehydrate water by physical adsorption. The channel diameter of the zeolite 5A is 0.5 nm, and the initial developmental intention of this material was that water molecules can enter its pore channel structure freely. Because the molecular size of oil is larger than the pore channel diameter, oil molecules cannot enter, so the pore volume of zeolite 5A can theoretically be occupied only by water molecules. The above characteristics of zeolite 5A can enhance the pore-volume utilization and avoid the competitive adsorption of oil and water. However, the actual dehydration effect for transformer oil does not reflect these advantages. This is because transformer oil has a higher viscosity than common reagents, e.g. benzene, toluene, and xylene, and its lack of fluidity at the orifice of zeolite 5A micropores limits both the internal diffusion of water molecules and the process of water molecules entering the micropores to be adsorbed.

Fig. 2 PAA-Na powder a before and b after water adsorption; PAA-Na modified bentonite c before and d after water adsorption

Table 1 Steady dehydration rate of various adsorbents

Anhydrous calcium chloride, a material commonly used in laboratories to achieve dehydration with the aid of chemisorption, is used commonly in organic synthesis. Although chloride ion loss is a problem when using calcium chloride in fixed-bed experiments, the goal here was to explore the depth of chemisorptive dehydration, in contrast to the zeolite 5A described above. According to the data in Table 1, the dehydration rate of anhydrous calcium chloride can reach 62%, which is greater than the 46% of zeolite 5A. The two main reasons for calcium chloride having a greater dehydration rate than zeolite 5A are: (1) during water adsorption by anhydrous calcium chloride, there is no diffusion resistance for water molecules because the adsorption-active sites of anhydrous calcium chloride are exposed to external surfaces directly; and (2) water adsorption by anhydrous calcium chloride is based on chemisorption, forming CaCl₂·6H₂O to absorb water molecules, and the force between calcium chloride and water molecules is stronger than the van der Waals force of physical adsorption or hydrogen bonding.

PAA-Na and calcium chloride both had a high selectivity for water adsorption in transformer oil. Water adsorption by PAA-Na is lower than calcium chloride because water adsorption by PAA-Na is a physical process, whereas calcium chloride water adsorption is a chemical reaction. The negative aspect of water removal from the oil by calcium chloride is that it introduces chloride ions into the oil, which affect the quality of the oil and corrode the metal equipment.

Bentonite is a good dehydrating agent, and the dehydration rate can reach 68.3%, according to Table 1. Most of the water is adsorbed as hydration water surrounding exchangeable cations in the interlayer region of smectites. Some additional water can be adsorbed by micro- or even mesopores. This additional water has been termed ‘external water’ (Kaufhold et al., Reference Kaufhold, Dohrmann and Klinkenberg2010). These characteristics are responsible for the better water adsorption. As a result, the PAA-Na modified bentonite synthesized in the current work has a greater dehydration rate than other adsorbents because it combines the advantages of both PAA-Na and bentonite.

Dehydration Test of PAA-Na-Modified Bentonites

A series of modified bentonites with various PAA-Na contents were prepared to test the effect of the PAA-Na additive. The water content of unprocessed transformer oil was 28 ppm. A steady dehydration rate of various modified bentonites are shown in Fig. 3a. The optimal amount of PAA-Na to modify the bentonite turned out to be 20 wt.%. When the addition of PAA-Na was <20 wt.%, the dehydration rate increased as the amount of PAA-Na added increased. When the addition was >20 wt.%, the dehydration rate decreased, indicating that the addition of PAA-Na will not increase the dehydration rate indefinitely and that an optimal value exists. At the same time, these observations proved that the water adsorption ability of modified bentonite depends on the joint action of PAA-Na and bentonite. PAA-Na contains abundant hydrophilic groups, so the addition of PAA-Na can improve the dehydration ability of bentonite, and the greater water content of the unprocessed transformer oil, the higher the dehydration rate (Fig. 3b).

Fig. 3 a Steady dehydration rate of feed oil by bentonite modified with various amounts of PAA-Na; b steady dehydration rate of feed oil at various water contents using bentonite modified with 20 wt.% PAA-Na

Adsorption Isotherms

The water adsorption tests above were conducted using real oil, but did not reveal the water-adsorption process in detail. To provide further understanding, experiments were undertaken with water concentrations ranging from 0 to 100%, with the expectation of gaining a broader view of water adsorption by modified bentonite.

Adsorption isotherms were obtained to provide information on adsorbate equilibrium relationships, affinities, and adsorption capacities of the adsorbents. Better recognition of the essential principles of adsorption and the parameters derived from the various equilibrium isotherm models could provide important information on the prediction of adsorption. Data were plotted according to the Freundlich isotherm model (Fig. 4), and the corresponding parameters and determination coefficient (R²) (Tables 2 and 3) indicated that this model predicts well the adsorption of water vapor onto bentonite, PAA-Na, and modified bentonite (20 wt.%) under saturated water vapor conditions at 25 and 10°C. P ₀ in Fig. 4 is the saturated vapor pressure of water at 25 and 10°C. The adsorption capacity and adsorption rate of bentonite improved after modification; PAA-Na modified bentonite exhibited a greater ability to entrap water molecules than bentonite due to the presence of hydrophilic groups, which was corroborated by the FTIR spectra (Fig. 1a). In the low relative pressure region (P/P ₀ < 0.45), the modified bentonite had the highest water adsorption capacity. This finding is consistent with the results achieved by using modified bentonite to remove trace amounts of water from the oil. In the high relative-pressure region (P/P ₀ > 0.45), the adsorption capacity of PAA-Na was greater than either the bentonite or modified bentonite, which is because the greater the external water content, the greater the osmotic pressure in the polymer framework of PAA-Na, thus causing the PAA-Na to adsorb more water.

Fig. 4 Water-adsorption isotherms of bentonite, PAA-Na, and bentonite modified with 20 wt.% PAA-Na

Table 2 Parameters derived from application of the Freundlich adsorption isotherm

Table 3 Parameters derived from application of the Langmuir adsorption isotherm

Adsorption Thermodynamics

The adsorption process can also be described by adsorption thermodynamics. The state functions ΔS ^θ (kJ/mol), ΔH ^θ (kJ/mol), and ΔG ^θ (kJ/mol) can be calculated from the following formulae (Saleem et al., Reference Saleem, Afzal, Qadeer and Hanif1992) (Eqs 4, 5):

(4) $\Delta G^{\theta }=-RTln\left(K_C\right)$

(5) $ln\left(K_C\right)=\left(\mathrm{T\Delta }{S}^{\theta }–\Delta H^{\theta }\right)/\mathrm{RT}$

(6) $K_C=K_F·\rho {\left(10^6/\rho \right)}^{\left(1-1/n\right)}$

where K_C is the thermodynamic equilibrium constant without units (dimensionless) (Eq. 6), where ρ is the density of saturated water vapor at 20°C and 25°C (Ghosal & Gupta, Reference Ghosal and Gupta2015), R is the universal gas constant (8.314 J∙mol^–1∙k^–1); T is system temperature; and ΔH ^θ and ΔS ^θ were obtained from the slope and intercept of the relationship between K_C and 1/T. The calculated results are listed in Table 4. The fact that the ΔG ^θ value decreased as the temperature increased, as shown in Table 4, indicates that the process was thermodynamically favorable. The increased randomness throughout the adsorption process is indicated by the positive value of ΔS ^θ. The positive ΔH ^θ value indicated that the adsorption process was endothermic, which matches the adsorption isotherm result.

Table 4 Thermodynamic parameters for bentonite, PAA-Na, and modified bentonite

Thermal Stability Analysis

Thermogravimetric curves (Fig. 5) of bentonite and modified bentonite (20 wt.% PAA-Na) revealed that the weight loss due to adsorbed water occurred between 35 and 150°C for bentonite (Fig. 5a, DTG curve) and between 35 and 200°C for modified bentonite (Fig. 5b, DTG curve). Another weight loss was observed at ~200–500°C due to thermal decomposition of the PAA-Na polymer. The residual amount of modified bentonite was 76%, while that of bentonite was 85%, and the reduction of the residue was due to the decomposition of PAA-Na at high temperature.

Fig. 5 TGA curves of a bentonite and b bentonite modified with 20 wt.% PAA-Na

Water molecules were removed between 35 and 200°C (Fig. 5), so drying temperatures of 80, 120, and 160°C were selected for pretreatment (Fig. 6). After 24 h of drying, the modified bentonite was used for transformer oil dehydration at 25^oC and atmospheric pressure with a weight hourly space velocity of 8 h^–1. The dehydration curves of modified bentonite pretreated at 80 and 160^oC were similar. The modified bentonite pretreated at 120^oC showed the best dehydration effect. The pretreatment temperature of the modified bentonite, therefore, was selected as 120^oC.

Fig. 6 Steady dehydration rate at various drying temperatures of bentonite modified with 20 wt.% PAA-Na

Influence of Weight Hourly Space Velocity

The space velocity is an important parameter in the adsorption experiment for modified bentonite (20 wt.% PAA-Na); under various space velocities (ranging from 8 to 60 h^–1), the extent of water adsorption was largely consistent, and ultimately reached ~80% (Fig. 7). Operating conditions, therefore, can be changed flexibly in industrial applications. Modified bentonite can maintain a stable dehydration rate even when the space velocity fluctuates significantly. At a space velocity of 30 h^–1, each gram of adsorbent can treat 4750 g of transformer oil and each gram of adsorbent can adsorb 139.3 mg of water from oil.

Fig. 7 Steady dehydration rate of feed oil containing 28 ppm water at various space velocities

Transformer Oil Quality Index after Dehydration

PAA-Na modified bentonite was developed as an adsorbent specifically for dehydration, and the trace water in transformer oil should be removed selectively without reducing the quality of the transformer oil. In order to assess changes in quality indicators of transformer oil before and after treatment with modified bentonite, a series of tests was carried out (Table 5). After examination and comparison of oil quality indicators before and after dehydration, modified bentonite was shown to be effective at trace-water removal from transformer oil and the breakdown voltage was also improved. Modified bentonite also reduced kinematic viscosity and dielectric loss factor; it may be that  the modified bentonite also adsorbs other components from the oil, thereby further improving its performance.

Table 5 Transformer oil quality index before and after dehydration

Analysis of Morphology

X-ray diffraction was used to analyze the mineral composition and crystal structure (Park et al., Reference Park, Ayoko and Frost2011). Figure 8 shows the XRD patterns of bentonite and PAA-Na-modified bentonite before and after water adsorption. The patterns are very similar, which suggests that the PAA-Na did not intercalate into the interlayer spaces of the bentonite. The interlayer distance should increase and the d ₀₀₁ spacing should decrease if PAA-Na was intercalated into the interlayer (Natkański et al., Reference Natkahski, Kuśtrowski, Białas, Piwowarska and Michalik2012), but such changes were not observed, as reported in the literature (Guerra et al., Reference Guerra, Merat, San Gil and Dieguez2008). The several sharp diffraction peaks of the five samples were due to a large amount of silica (Kawi & Yao, Reference Kawi and Yao1999) in the bentonite. Comparing the micro morphologies of bentonite and modified bentonite (20 wt.%) in Fig. 9, showed that the layers of both bentonites remained relatively intact, the edges appeared curled, and the crystal form was not destroyed, consistent with the conclusions from XRD. After PAA-Na modification, the bentonite layers became more compact, and the granular protrusion was significantly reduced.

Fig. 8 XRD patterns of bentonite and bentonite modified with 10, 20, and 30 wt.% PAA-Na, a before and b after water adsorption for 12 h

Fig. 9 SEM images of a-I, II bentonite and b-I, II bentonite modified with 20 wt.% PAA-Na

BET Analysis

The pore structures of the bentonite and the modified bentonites were investigated by N₂ adsorption-desorption isotherms (Fig. 10). Isotherms of both bentonite and modified bentonites are type IV according to the classification of IUPAC. The hysteresis loop indicated the presence of mesopores with the capillary condensation of liquid nitrogen on its surface (Gautier et al., Reference Gautier, Muller, Le Forestier, Beny and Guegan2010). The mesopores were stacked layer by layer (Korichi et al., Reference Korichi, Elias and Mefti2009). Table 6 presents the results of surface area, total pore volume, average pore diameter, and most frequent pore width for the bentonite and the modified bentonites. After modification, the surface area of the bentonites decreased from 168.2 to 5.0 m²/g, the average pore diameter increased from 6.2 to 30.5 nm, and the pore volume decreased from 0.3 to 0.04 cm³/g. The surface area and pore volume values for the modified bentonites were smaller than for bentonite; this may be due to the blockage of some of the finer pore junctions in the bentonite by the intercalation of the PAA-Na (Mahboub et al., Reference Mahboub, Mouzdahir, Elmchaouri, Carvalho, Pinto and Pires2006). These PAA-Na particles dispersed on the bentonite surface and in the pores. According to the results of dehydration and surface-area tests, the best dehydration effect of modified bentonite (20 wt.% PAA-Na) occurs because of the combination of pore adsorption in the bentonite and expansion of PAA-Na. When the amount of PAA-Na added was >20 wt.%, PAA-Na blocked the pore severely and reduced the water adsorption. The behavior of water adsorption changed with the addition of PAA-Na. The water-adsorption capacity of modified bentonite was increased with the addition of PAA-Na when it was <20 wt.%, due to the hydrophilicity of PAA-Na. Further addition of PAA-Na, on the other hand, blocked the pores and reduced the surface area of the bentonite, reducing its dehydration ability. The optimum addition of PAA-Na was found to be 20 wt.% (Fig. 3a), which balanced the benefits and drawbacks of the PAA-Na. The surface area and pore volume were reduced significantly, and the pore channels with smaller pore sizes were blocked by PAA-Na (Fig. 11). Only a few pores with relatively larger pore sizes remained (Table 7).

Fig. 10 Bentonite and bentonite modified with 10, 20, and 30 wt.% of PAA-Na: a N₂ adsorption-desorption isotherms and b pore-size distributions

Table 6 Textural properties of the investigated samples

Fig. 11 Bentonite and bentonite modified with 20 wt% of PAA-Na: A N₂ adsorption-desorption isotherms (a) before and (b) after water adsorption; and B pore-size distributions a before and b after water adsorption

Table 7 Textural properties of bentonite modified with 20 wt.% PAA-Na before (a) and after (b) water adsorption

Broad Applicability of PAA-Na Modified Bentonite

The adsorption performance of various adsorbents in different oils varied, and the static adsorption capacity in different oils of several common adsorbents was demonstrated in Fig. 12. Different adsorbents were tested for 24 h under the conditions of 25^oC, atmospheric pressure, and adsorbent:oil ratio of 1:20. PAA-Na modified bentonite had a high dehydration rate in xylene, naphtha, n-heptane, and diesel, indicating that its applicability is extremely wide.

Fig. 12 Static adsorption capacities of various adsorbents and bentonite modified with 20 wt.% PAA-Na for various oils