Adsorption and desorption of uranium(VI) onto humic acids derived from uranium-enriched lignites

Uranium is a natural radioactive trace element existing in rocks, soils, and coals. Due to human activities such as mining of U-enriched ores, nuclear fuel processing and coal combustion (Jing et al. 2015; Zhang et al. 2016), the occurrence of uranium in soil and groundwater is a serious environmental problem. The mobility and toxicity of uranium in the aquatic environment depend on its speciation. Under oxic conditions, U(VI) is the main oxidation state of uranium, present in natural water systems primarily as the uranyl cation (UO₂²⁺) or as uranyl–hydroxide–carbonate complexes. Under anoxic conditions, U(VI) can be reduced to U(IV) by inorganic reductants. The mobility and biotoxicity of U(VI) is much higher than U(IV). A number of recent studies have focused on understanding the interaction between U(VI) and water (Sun et al. 2016; Wang et al. 2017a), yet an eco-friendly and practical technology for the removal of U(VI) from aqueous solution has not been developed.

Humic acid (HA) is a natural product that is ubiquitous in soils and natural waters, and HA can control trace element cycling and pollutant mobility. HA is a colloid having various functional groups, such as carboxylic, phenolic hydroxyl, alcoholic hydroxyl and carbonyl, which have shown remarkable effect in removing heavy metals. Because HA is an eco-friendly substance, the use of it would not introduce foreign materials into an engineered reaction system; its utilization as an adsorbing material has been a topic of intense research in recent years. Khalili et al. (2017) extracted HA from the Azraq soil and investigated the sorption properties of U(VI). The result of the column experiment showed that HA has good U(VI) uptake properties. HA occurs not only in soils, natural waters, and sediments but also in lignite and bituminous coal. Lignite is a low-rank coal with large amounts of humic and fulvic acids, and due to its low calorific value, high ash content and high content of moisture, lignite is mainly used as a fertilizer or source of HA and fulvic acid due to the high content of organic oxygen (Ciarkowska et al. 2017; Wang et al. 2017b). Many researchers have focused on the adsorption characteristics of HA derived from lignite. Wang et al. (2017b) investigated the characteristic of the HAs from lignite, and they showed that it was rich in hydroxyl and carboxyl groups. Pehlivan & Arslan (2006) appraised the removal of heavy metals with HAs derived from lignites, and HAs proved to be a useful adsorbent for Zn(II) and Cd(II). However, research on the interactions between U(VI) and HA from lignite is limited.

Due to the lignite having abundant organic matter (OM), it can be a biogeochemical trap for uranium. Uranium can co-precipitate with OM and lignite may even contain up to several hundred milligrams of U per kilogram (Hou et al. 2017). The uranium content of lignite in individual regions of China is abnormally high; the main deposits of the uranium-enriched coals in China are in the Yunnan province. The concentration of uranium in the Late Permian coals in the Yunnan province is 200 mg kg⁻¹ (Yang 2007), while the uranium concentration of coals in the Yanshan coalfield of Yunnan Province is 153 mg kg⁻¹ (Dai et al. 2008). The environmental risks of uranium-enriched lignite cannot be overlooked. As lignite is an important source of HA, research into U(VI) adsorption onto HAs derived from lignites, especially the uranium-enriched lignites, is needed.

In our present investigation, we have isolated HAs from three kinds of uranium-enriched lignites in the Yunnan province, China and characterized it by Fourier transform infrared spectroscopy (FTIR), element analysis, determination of HAs acidity and UV–visible (UV-Vis) spectroscopy. The adsorption behavior of uranium on these HAs and a commercial HA were investigated, as well as the effect of initial pH, adsorption kinetics and isotherms to explore whether the high content of uranium present in the HAs can affect the adsorption and reaction system; desorption by two eluting agents (NH₄Ac and EDTA) was studied to understand the nature of adsorption mechanisms.

All the reagents were analytical grade and purchased from Sinopharm Chemical Reagent Co. (Shanghai, China). The U solution was prepared with UO₂(NO₃)₂·6H₂O in deionized water (18.25 M cm⁻¹, total organic carbon ≤ 10 μg L⁻¹) (UPT-I, Ulupure, China). All the glassware, polyethylene tubes, and Teflon beakers were soaked in a 20% (V/V) HNO₃ solution for 48 h and then washed with deionized water before and after each experiment to avoid the binding to metals. The pH was adjusted by adding moderate amounts of hydrochloric acid or sodium hydroxide using a pH meter (PB-10, Sartorius), which was calibrated with different buffers (pH = 1.68, 4.00, 6.86, 9.18).

Three samples of HA used in this study were extracted from three different lignite mines respectively, which were the Mengwang coal mine (MW), the Dazhai coal mine (DZ) and the Huimei coal mine (HM). All of these mines are located in the Lincang city, Yunnan province, China. The uranium content in the lignites was determined by the method from Dai et al. (2011); the uranium concentration of the lignites from above-mentioned mines is 101.22, 60.86 and 16.71 mg kg⁻¹, respectively. A commercial sample of HA (C-HA) was obtained from the Tianjin Guangfu fine chemical research institute. The extraction of the three HA samples (MW-HA, DZ-HA, and HM-HA) was conducted with the alkali extraction method (Cihlář et al. 2016). One hundred grams of the air-dried lignite sample was ground to grain size <0.2 mm and homogenized. Then the lignite sample was mixed with 0.5 mol L⁻¹ NaOH and 0.1 mol L⁻¹ Na₄P₂O₇·10H₂O (1:1) at a volume ratio of 10:1 (mass/V) and shaken for 3 h. Then, the mixture was centrifuged at 5,000 rpm for 20 min, after which 6 mol L⁻¹ HCl was added to the supernatant until obtaining the pH of 1 to precipitate the HAs. Afterwards, 0.5% HCl-HF solution was added to precipitated HAs for 24 h to eliminate the inorganic impurities and then centrifuged at 5,000 rpm for 20 min. Obtained HAs were repeatedly washed until chloride-free and freeze-dried. All the HA samples were finely milled and used in the form of humic powder.

The contents of C, H, N, S, and O in the HAs were measured using an elemental analyzer (Vario MACRO). The ash content was determined by combustion of 200.0 mg of HAs in a muffle furnace at 850 °C for 2 h. HAs were acid digested using a conventional HF-HNO₃-HClO₄ digestion procedure on an electrothermal plate. The concentration of the metallic elements was measured with inductively coupled plasma mass emission spectroscopy (ICP-MS) (ELAN DRC-II, PerkinElmer SCIEX., USA). The internal standard is Rh (10 μg L⁻¹) in 3% (V/V) HNO₃. The accuracy of the duplicate analyses was within ±5%.

The total acidic groups, the carboxyls, and the phenol-hydroxyls in the HAs were measured by conventional methods (Klučáková & Pekař 2008). Absorbance was measured at 465 nm and 665 nm (E4 and E6 respectively), for a 0.2 g L⁻¹ solution of HAs at pH 6.5, and then the E4/E6 ratio was calculated.

The chemical compositions of the four HA samples were characterized by FTIR spectroscopy (Nicolet 6700, Thermo Corp., USA). HA samples were dried at 45 °C for 3 hours in the oven, then prepared by mixing with KBr powder (1%, m/m) and pressing the mixture into a sheer slice. The FTIR spectra were scanned in the transmission mode from 4,000 to 400 cm⁻¹, and the FTIR data were analyzed by OMNIC 8.2.0.

For modeling the kinetic parameters of the U(VI) adsorption by the HAs, the first-order equation (Equation (3)), pseudo-second-order equation (Equation (4)), and intraparticle diffusion model (Equation (5)) have been used.

NH₄Ac and EDTA were chosen to study the desorption of U(VI) from the HAs. Twenty milligrams of HAs were added into 20 mL of 10 mg L⁻¹ U(VI) solution in 50 mL conical flasks. The desorption experiments were maintained at pH of 5 and temperature of 25 °C, and shaking for 24 h at a shaking rate of 100 rpm to ensure approaching equilibrium. The maximum sorption capacity of the MW-HA, DZ-HA, HM-HA, and C-HA were 3.96, 1.51, 6.02, and 9.50 mg g⁻¹, respectively. After centrifugation and filtration, the HA samples were washed with deionized water and freeze-dried. Two milligrams of solid samples were added into 50 mL of the eluting solutions with varying concentrations and shaken for 8 h. The desorption amount can be determined from the concentration of the U(VI) in the solution.

The elemental compositions of the HAs and the ratios of O/C and H/C are given in Tables 1 and 2. The results show that the C-HA contains more oxygen compared with the MW-HA, DZ-HA, and HM-HA. The O/C ratio reveals the hydrophilicity degree of the HAs. A higher value of this ratio indicates that the HA may have a higher content of oxygen functional groups (e.g. -COOH, -OH) (Cihlář et al. 2014). This is reflected in the determination of HAs acidity as shown in Table 3, in which the total of the acidic groups of C-HA is higher than the other three kinds of HAs, while the total of the acidic groups of DZ-HA is the lowest among all of the HAs. The decrease of the total acidic groups is in the order of C-HA > HM-HA > MW-HA > DZ-HA, which has a close correlation relationship with O/C ratio. E4/E6 ratio is associated with the degree of humification and the molecular weight of the HAs. High E4/E6 ratio indicates less degree of aromaticity and lower molecular weight. The decrease of E4/E6 ratio, as shown in Table 3 in the order of DZ-HA > MW-HA > HM-HA > C-HA, is an indication of increasing humification, aromaticity and molecular weight.

FTIR spectroscopy of the HAs shown in Figure 1 indicates that all the samples show a broad band at 3,420 cm⁻¹ which is assigned to the O-H vibration of the carboxylic and alcoholic groups (Pehlivan & Arslan 2006). The weak absorption around 2,350 cm⁻¹ is assigned to the hydrogen bonded groups (O-H stretching of COOH) (Ghabbour et al. 2004). The weak band at 1,440 cm⁻¹ in the MW-HA, DZ-HA and HM-HA is attributed to the O-H deformation or C = O stretching of the phenolic groups (Khalili & Al-Banna 2015). The band at 1,620 cm⁻¹ is assigned to the aromatic carbonyl and carbonyl motion in the carboxyl groups. The intensity of band 1,620 cm⁻¹ shows that the C-HA and HM-HA contain more carboxylic acids than the other samples. The band at 1,620 cm⁻¹ is also a measure of the humification. The higher intensity of 1,620 cm⁻¹ observed in the C-HA and HM-HA further confirms a higher humification than in the other samples. The band at 1,170 cm⁻¹ is assigned to the C-O vibration of tertiary alcohol. Two bands at 985 and 732 cm⁻¹ in the MW-HA, DZ-HA and HM-HA are due to the bending of aromatic C-H. FTIR spectra of the HA samples indicates that the amount of carboxyl functional groups in the C-HA and HM-HA may be higher than the MW-HA and DZ-HA; the content of functional groups shown in Table 3 also confirms this.

The effect of the initial pH on the U(VI) sorption is shown in Figure 2. The sorption of U(VI) on the four different HAs increases abruptly with the increase of pH when the pH is less than 3, then achieves a peak and remains constant at pH of 5–8. When pH > 8, the removal percentage decreased. The pH of the U(VI) solution affects the species distributions of U(VI), and the surface charge of HAs. These factors affect the adsorption mechanism and the U(VI) removal percentage. The uranium removal percentage of the C-HA is higher than the HAs derived from lignites. This indicates that the functional groups of the HAs play a major role in the adsorption of the U(VI), which may be because the C-HA has a higher content of carboxylic group (2.02 mmol g⁻¹) and phenol-hydroxyl group (1.32 mmol g⁻¹) than the other HAs and therefore binds more U(VI).

The U(VI) species are present in the forms of UO₂²⁺, UO₂(OH)⁺, (UO₂)₃(OH)₅⁺, UO₂(OH)₃⁻ and UO₂(OH)₄²⁻ at different pH values, and the U(VI) species change from positively charged to negatively charged with pH ranging from 1 to 14, as simulated by Visual MINTEQ 3.1 in Figure 3. When pH < 5, positively charged UO₂²⁺ exists as the predominant species (≥70%), while UO₂(OH)⁺ is less than 29%. At pH of 6, UO₂(OH)⁺ is the predominant species (≥62%), and the amount of (UO₂)₃(OH)₅⁺ increases. At pH of 6–8, a mixture of species of UO₂(OH)⁺ and (UO₂)₃(OH)₅⁺ is observed. Negatively charged species of UO₂(OH)₃⁻ and UO₂(OH)₄²⁻ become dominant species at pH ≥ 9. Yue et al. (2009) found that slight dissolution of the HA raised the negative charges on the surface in neutral solution, the zeta potential of HA was about −35 mV, and the uranium cations could easily adsorb on the surface of HA. However, with the increasing of pH (pH > 8), it is difficult for the negatively charged species of uranium to adsorb onto the negatively charged surface of HA, which is the major cause for U(VI) removal percentage decreasing at higher pH values.

Figure 6 plots the q_t variation versus time for the U(VI) onto the HAs for various C₀ (50, 100 and 200 μg/L) at 25 °C, pH 5. We conclude from this figure that the U(VI) sorption by the HAs from the lignites needs about 6 h to reach equilibrium; the U(VI) sorption by the C-HA needs about 4 h to reach equilibrium. The value of q_t increased rapidly during the beginning of the adsorption, then the increasing rate slows down, and the external mass transfer diffusion is not the control mechanism of the adsorption process.

To investigate the adsorption mechanism and reaction order, the kinetics data were fitted to three kinetic models. The kinetic rate parameters, the correlation coefficients, and calculated and experimental q_e values are listed in Table 4. According to the intraparticle diffusion model, if the value of C is 0, the intraparticle diffusion is the rate limiting step in the adsorption; if not, the adsorption process is controlled by other adsorption stages. Li et al. (2010) suggested that the heavy metals adsorbed by the HA may include three stages; the first stage is the external sorption, the second stage is attributed to intraparticle diffusion, the last one is equilibrium stage. In this study, intraparticle diffusion is not the main mechanism of the adsorption. As seen from Table 4, the pseudo-second-order model could well simulate the data, among which the r² of the pseudo-second-order model is higher than that for the first-order model and the intraparticle diffusion model. This illustrates that the pseudo-second-order model could describe the adsorption kinetics of U(VI) onto the HAs.

The sorption capacity calculated from the pseudo-second-order equation (q_e2(cal)) is much more close to the experimental value, q_e(exp), than is the first-order equation. It also indicates that the pseudo-second-order model is more suitable for the adsorption system. We find that q_e(exp) increases with the increase of C₀. The increase of U(VI) concentration could provide more contacting with the HA. The values of the rate constant k₂ of HAs were found to decrease with the increase in C₀ from 50 to 200 μg L⁻¹. It indicates that the adsorption of higher concentration of U(VI) to a specific removal percentage would take a longer time.

The adsorption capacity q_e(exp) of the MW-HA, DZ-HA, HM-HA, and C-HA at C₀ = 100 μg/L are 0.87, 0.28, 1.18 and 1.63 mg g⁻¹, respectively. However, the uranium concentration of the MW-HA and DZ-HA are 66.91 and 19.92 mg kg⁻¹, respectively. It shows that the higher concentration of uranium present in the HA does not affect its adsorption capacity. In the pseudo-second-order model, the chemical reaction is crucial in the rate-limiting step. Differences in adsorption capacity and equilibrium time in different HAs may be caused by the number of carboxylic acid and phenolic hydroxyl functional groups.

The adsorption isotherms of U(VI) onto the HAs are shown in Table 5. The correlation coefficient indicates that the Langmuir and Freundlich model can be used to describe the experimental data. Jaynes (1991) revealed that the adsorption conforms to the Langmuir and Freundlich models when the value of r² is larger than 0.89.

The Langmuir isotherm was used to estimate the sorption capacity and K_L. According to the hypothesis of the Langmuir model, the U(VI) adsorbed on some sites of the HAs, whose binding energies are the same, and the process is single molecular layer adsorption. The sorption capacity, Q_m, of each HA increases with the uplifting of temperature, which indicates that raising temperature is beneficial to the adsorption process. The experimental data also fit well with the Freundlich model, which reveals that some other sorption sites are non-uniform, and the value of n > 1 indicates the adsorption is a favorable process (Khalili & Al-Banna 2015).

The isotherm experiment data fitting the Langmuir and Freundlich models well demonstrate that both single molecular layer adsorption and heterogeneous surface conditions exist at the same time. The adsorption of U(VI) by the HAs extracted from the lignites involve more than one mechanism.

The adsorption and desorption are reversible processes, and the adsorbed metal ions can be desorbed under certain conditions. The desorption efficiency can be used to illustrate the binding degree of HA and metal ions. NH₄Ac was used as the desorption agent to represent the cations adsorbed onto the HA by ion-exchange effect. The desorption amount of the metal ions by EDTA suggests that metal ions adsorbed onto HA is determined by complexation effect and ion-exchange effect.

Different concentrations of NH₄Ac and EDTA were used as eluting agents for the removal of uranium from the HAs; the results are presented in Figure 7. It is clear that when the concentration of NH₄Ac and EDTA increased, the desorption efficiency of uranium increased. The maximum desorption percentages of uranium from the MW-HA, DZ-HA and HM-HA in EDTA solution are 51.32%, 66.21% and 54.78%, and in NH₄Ac solution are 12.22%, 5.24% and 9.74%. It indicates that the uranium adsorption onto the MW-HA, DZ-HA and HM-HA is controlled by the complexation with the organic ligands, and the effect of ion exchange was not the main adsorption mechanism. The results of uranium desorption from the C-HA are not identical to the other three kinds of HAs. The maximum desorption percentage of uranium from C-HA in NH₄Ac and EDTA solution are 19.17% and 44.29%, which may suggest that both the complexation and ion-exchange effect are the control factors for the adsorption of uranium onto the C-HA.

In this work, the characteristics of three different HAs extracted from uranium-enriched lignites and one commercial HA were confirmed by elemental analysis, FTIR, determination of HAs acidity and UV-Vis spectroscopy. The adsorption of U(VI) onto the HAs was examined under various experimental conditions. The effective desorption with two different eluting agents for the U(VI) was also studied.

The results demonstrate that the U(VI) sorption onto the HAs is highly dependent on pH, and the optimum pH level at which all the HAs adsorbed uranium(VI) ranged from 5 to 8. The high uranium content of the HAs (e.g. MW-HA) could release into the solution at the pH value between 1 and 3, which shows that the HAs derived from uranium-enriched lignites present a potential environmental risk when used in acidic conditions. The adsorption kinetics data were best modeled using a pseudo-second-order equation and indicate that chemisorption is the rate-controlling mechanism. The adsorption isotherms fit the Langmuir and Freundlich models well, and prove that both single molecular layer adsorption and heterogeneous surface conditions exist in the adsorption process. The desorption percentage of the MW-HA, DZ-HA, and HM-HA in EDTA is much larger than that in NH₄Ac, which illustrates the adsorption process is controlled by the complexation effect in these HAs. The uranium existing in the HAs may not affect the adsorption capacity of uranium(VI). The results of the chemical characterization of the HAs demonstrate that the carboxylic and phenolic hydroxyl groups or other functional groups play a significant role in controlling the adsorption capacity.

This work was partially supported by the National Natural Science Foundation of China (NSFC, No. 41572233). We thank the anonymous reviewers and editor for their constructive comments and suggestions, which helped us to improve the manuscript.