Background

Lithium consumption has strong growing demand due to wide lithium application in Li-ion rechargeable batteries, glass, ceramics, grease, nuclear fusion fuel, metallurgical industries, light aircraft alloys, pharmaceuticals (Heidari and Momeni 2017). Solar evaporation, solvent extraction, co-precipitation (using aluminum salt), membrane processes, liquid–liquid extraction—these methods are used for lithium recovery from mines, brines, and salt lakes (Hamzaoui et al. 2003; Epstein et al. 1981; An et al. 2012; Swain 2016; Gao et al. 2016).

Such alternative resources as seawater and secondary waste solutions also should be used. However, the content of this element in seawater is rather small (0.17 mg/l). Among metal ions, Li+ species are characterized by the smallest radius (7.8 × 10−11 m), increased polarization ability and tendency to solvation. This and also high concentration of other inorganic ions (Na+, Ca2+, Mg2+, etc.) complicate Li+ sorption. Thus, development of highly selective materials is very important task.

Adsorption and ion exchange are the most preferable techniques for recovery of inorganic ions from very diluted solutions, inorganic sorbents are very effective for this purpose (Alberti and Massucci 1970). A number of inorganic materials has been proposed, for instance, titanium antimonate (Abe and Chitrakar 1987).

Among inorganic materials, hydrated oxides of multivalent metals are attractive as sorbents of various inorganic ions due to simple synthesis techniques, which requires cheap raw material, high selectivity, particularly towards Li+ (Kim et al. 2008), and possibility to sorb both cations and anions depending on the solution pH (Amphlett 1964; Kudelko et al. 201111). Due to these remarkable properties, oxide materials are used not only as sorbents, but also as modifiers for ceramic (Dzyazko et al. 2006, 2014a, b; Marti-Calatayud et al. 2015; Mora-Gómez et al. 2017) and polymer membranes (Hong et al. 2015; Dzyazko et al. 2017) for electrodialysis, as well as for development of the membranes applied to filtration affected by pressure gradient (Huang et al. 2012; Pang et al. 2014; Myronchuk et al. 2016). Particles of oxides of multivalent metals, which are precipitated in inert polymers (Myronchuk et al. 2016), are smaller than formations of phosphate compounds (Dzyazko et al. 2015, Myronchuk et al. 2016). Calcination of hydrated oxides such as MnO2 (Ooi et al. 1986; Park et al. 2015) or TiO2 (Chitrakar et al. 2014; Shi et al. 2013; Tretyak et al. 2013; Zhang et al. 2010) allows one to obtain Li-selective sorbents.

Spinel-type lithium–manganese oxides (LMO, their composition is LinMn2−xO4, where 1 ≤ n ≤ 1.33, 0 ≤ x ≤ 0.33) were found to be the most attractive materials for preparation of highly selective sorbents, which can be obtained by means of Li+ → H+ exchange (Ooi et al. 1986; Park et al. 2015). With increasing ratio of Li/Mn, higher values of sorption capacity are achieved. However, regeneration with acidic solutions destroys these oxides due to reaction of disproportionation of Mn(III) followed by removal of Mn(II) from the solid.

Other types of lithium-ion sieves were proposed: H2TiO3 with a layered structure and H4Ti5O12 with a spinel structure. These compounds (LTO) are characterized by a higher stability due to large energy of titanium–oxygen bond (Chitrakar et al. 2014; Shi et al. 2013; Zhang et al. 2010). Selectivity is formed by means of saturation of the initial amorphous titanium dioxide with Li+ cations followed by thermal treatment (Tretyak et al. 2013).

In order to provide high sorption rate, small grains of the Li-selective sorbent are attached to the binder (Umeno et al. 2002; Hong et al. 2013; Chung et al. 2014; Xiao et al. 2015; Hong and Chen 2015; Kam et al. 2015), in this case lower sorption capacity is reached (Chung et al. 2017). Other approach is to obtain LMO nanoparticles (Miyamoto et al. 2015). In this case, the limiting stage of sorption is diffusion through grains. Regarding the materials containing large primary particles, sorption rate is determined by chemical reaction (Li+ insertion to spinel structure). Multiple processes of sorption-regeneration result in destruction of the sorbents.

Double hydrated oxides containing MnO2 possess higher selectivity towards Li+ comparing with oxides based on one metal (Hong et al. 2018). Regarding sorption of heavy metal cations, this is explained by high microporosity of manganese oxide, since micropores are considered as selective sorption centers (Mal’tseva et al. 2011). Calcination of oxides of complex composition containing two or even more multivalent metals allows one to obtain lithium conductors (Bohnke 2008). The materials of this type were applied to preparation of Li-selective membrane for recovery of this element from sea water using reverse electrodialysis (Hoshino 2015). However, mechanical durability of the membranes is expected to be rather low.

Double titanium–manganese oxide (LTMO) possesses selectivity towards Li+ (Jiang 2012). This sorbent was obtained by using a method of solid-state reaction, the grains consisted of coarse particles. The aim of the investigation was to obtain and test the nanomaterial of similar composition. As expected, the nanosized particles would provide high sorption rate.

Experimental

Synthesis of LTMO and LTO

Synthesis of titanium–manganese sorbents was carried out through redox reaction between TiCl4 and KMnO4 solutions. The concentration of each solution was 1 M, their volumes were equal. An excess of HCl, which is formed during TiCl4 hydrolysis, provides Mn(VII) → Mn(II) transformation:

$${\text{TiCl}}_{4} + 2{\text{H}}_{2} {\text{O}} \to {\text{TiO}}_{2} + 4{\text{HCl}}$$
(1)
$$2{\text{KMnO}}_{4} + 16{\text{HCl}} \to 2{\text{MnCl}}_{2} + 2{\text{KCl}} + 8{\text{H}}_{2} {\text{O}} + 5{\text{Cl}}_{2} .$$
(2)

Further disproportionation reaction occurs:

$$4{\text{KMnO}}_{4} + 3{\text{MnCl}}_{2} + 4{\text{HCl}} \to 7{\text{MnO}}_{2} + 4{\text{KCl}} + 2{\text{H}}_{2} {\text{O}} + 3{\text{Cl}}_{2}$$
(3)

The resulting reaction:

$$2{\text{TiCl}}_{4} + 2{\text{KMnO}}_{4} + 2{\text{H}}_{2} {\text{O}} \to 2{\text{TiO}}_{2} { + }2{\text{MnO}}_{2} + 2{\text{KCl}} + 2{\text{H}}_{2} {\text{O}} + 3{\text{Cl}}_{2}$$
(4)

Gel consisting of Ti and Mn hydrated dioxides is obtained by this manner. Poor solubility of KMnO4 in water when added to the TiCl4 solution provides control over the rate of neutralization of excess acidity, slowing down the process of gel formation. Optimal conditions for the formation of the double oxide are achieved by this manner.

The obtained hydrogel was washed with distilled water until MnO4− and Mn2+ were absent in the effluent, then it was dried under room temperature. The inorganic material was in the form of durable black granules of irregular shape. The grains were washed with 12% aqueous ammonia solution and distilled water, then were dried again.

A 1 M TiCl4 solution containing equimolar amount of citric acid was used to synthesize hydrated titanium dioxide. Gel was precipitated with 10% NH4OH solution and dried at room temperature.

The granular sorbents (hydrated oxides of both Ti and Ti–Mn) were loaded with lithium by treatment with a 1 M LiOH solution for 48 h. A volume ratio of sorbent and solution was 1:100. Further the sorbent was separated from the solution, washed with distilled water and dried in air at room temperature down to constant mass. The lithiated form of the samples was calcinated at 500, 600 and 700 °C for 4 h. To remove Li+ ions (provide Li+ → H+ exchange), the samples were washed with 0.5 M HCl and analyzed with an atomic absorption method using a PYE UNICAM SP 9 spectrometer (Philips).

Characterization of the samples

For chemical analysis, the samples of the obtained materials were analyzed with an X-ray fluorescent spectrometer X-Supreme 8000 XRF (Oxford Instruments). Microphotographs were obtained by using a Trek DCM510 camera with a RL05-48 ring diode illuminator attached to a PZO optical microscope (PZO, Poland).

TEM images of granules were obtained by using a transmission microscope JEOL JEM 1230 (JEOL, Japan). Scanning Electron Microscope Tescan Mira 3 LMU (Czech Republic) was also applied to the research.

Differential thermal analyses were carried out in the range of 20–800 °C at a scanning rate of 5 °C/min («Derivatograph Q - 1500 D», Paulik-Paulik-Erday, Hungary). For X-ray analysis, a diffractometer DRON-3M (LOMO, Russia) with external standards certified by SiO2 (standard 2θ) and Al2O3 (intensity standard) was used (Cu-radiation, Ni-filter) (Nat Inst of Standards and Technology 1991). Isotherms of nitrogen adsorption–desorption were obtained with an automatic gas adsorption analyzer of surface and porosity Quantachrome AS1Win (Quantachrome Instruments, USA). The specific surface area was estimated from the isotherms with the BET method, the total volume of micro- and mesopores was determined with the BJH method (Gregg and Sing 1982).

Testing of the samples

Preliminarily fraction of 0.16–0.25 mm was taken for study of sorption under dynamic conditions in a column. A diameter of the column was 0.6 cm, the height of the sorbent bed was 17 cm. The recovery of Li+ ions was carried out from the solution containing 1 × 10−4 M LiCl. Superficial flow velocity was 0.18 m/h or one volume of the sorbents bed (BV) per hour. Li+ content at the column outlet was analyzed, sorption capacity was calculated as pointed in Helfferich (1995).

Further sorption from the solution containing 0.1 M LiCl, 0.5 M NaCl and also NH4OH-NH4Cl buffer (the initial pH values were 8, 10 and 12) was investigated under batch conditions. The solution containing LiOH (0.1 M) and NaCl (0.5 M) was also used, the initial pH was about 10. The mass ratio of sorbent:solution was 0.01:50, the time of contact was 24 h. The pH of the equilibrium solution was determined using the I-160NI ionomer (Analitprylad, Ukraine). The distribution coefficients (Kd) and selectivity coefficients (Ks) were determined as \(K_{\text{d}} = \bar{C}_{\text{Li}} /C_{\text{Li}}\) and \(K_{\text{s}} = \bar{C}_{\text{Li}} C_{\text{Na}} /(C_{\text{Li}} \bar{C}_{\text{Na}} )\), where C and \(\bar{C}\) are the concentration of ions in the solution and ion exchanger (Helfferich 1995).

Results

Morphology of grains: thermogravimetric analysis

The synthesized titanium–manganese sorbent is in a form of shiny granules with average size of 0.25–0.5 mm (Fig. 1a). Shape of the grains is close to that of titanium dioxide (Fig. 1b).

Fig. 1
figure 1

Images of the sorbents: LTMO (a) and LTO (b)

Full size image

Crystallization of the oxyhydrate matrix occurs during the thermal treatment, which provides mechanical durability of the grains. This procedure is accompanied by shrinkage and compaction of the material. According to the TG–DTA analysis, one or two clear endothermic effects and several exothermic effects are recorded in the high-temperature range (Table 1). Thus, the endothermic peak at 100–250 °C is related to the removal of weakly bonded water, which causes the main mass loss. The second endopeak corresponding to condensation of OH groups is observed only for lithium-substituted titanium oxide (TiO2–Li). In this case, Ti–O–Mn and Ti–O–Ti bonds are formed. In the temperature range of 300–500 °C, overlapping of thermal effects occurs due to removal of water from different structural positions. In the range of growth of the exothermic effect (250–650 °C), the mass loss is insignificant, several small exopeaks are recorded. They are related to crystallization of the amorphous phase and polymorphic transformations. Thus, spinel structure is formed above 500–600 °C. This assumption can be confirmed by the data from an X-ray diffraction analysis.

Table 1 Data of thermogravimetric studies
Full size table

X-ray analysis

In the temperature range of the first and second endoeffects (100–400 °C), the samples are amorphous. In the case of titanium dioxide anatase structure is formed at 400–500 °C, this phase is transformed into rutile above 700 °C (Diebold 2003). The data of X-ray diffraction analysis for the LTO sample show crystalline structure above 500 °C (Fig. 2). The sample, which was calcinated at 500 °C, consists of anatase (76.4%) and lithium titanium spinel LiTi2O4 (23.6%) (Table 2). Regarding spinel phase, its space group is Fd3m, a = 8.25 Å, cell volume is 561.61 Å3. With an increase in the calcination temperature up to 700 °C, the content of anatase reduces (decrease in the intensity of its X-ray reflexes). At the same time, intensity of the reflexes corresponding to rutile and Li4Ti5O12 (the space group is Fd3m, a = 8.35 Å, cell volume is 583 Å3) increases. Therefore, the composition of LTO depends on thermal treatment conditions. Li4Ti5O12 phase is formed at 700 °C. A mixture of crystalline structures of lithium titanate, anatase and rutile is formed in the region of lower temperature.

Fig. 2
figure 2

X-ray pattern of TiO2-Li (a) and TiO2–MnO2–Li (b) obtained at 500 and 700 °C

Full size image
Table 2 Phase composition
Full size table

A size of regions of coherent scattering was calculated from the half-width of reflexes using Scherrer formula (Scherrer 1918):

$$D = \frac{K\lambda }{{(\Delta^{2} - \Delta_{0}^{2} )^{0.5} \cos \theta }},$$
(5)

where \(K = 2\sqrt {\frac{\ln 2}{\pi } \approx 0.9}\). Here D is the crystallite size, λ is the wave length, Δ is the width of reflex at its half-height, Δ0 is the similar parameter, which is caused by the apparatus (this value is determined using LaB6 standard), and θ is the angle corresponding to the reflex maximum. The data are given in Table 2.

In the case of LTMO, Li0.75 Mn0.25Ti2O4 spinel is formed at 500 °C (its space group is Fd3m, a = 8.46 Å, cell volume is 654 Å3) (Lambert et al. 1988). This phase coexists with LiMn2O4 and rutile. With increasing temperature of calcination up to 700 °C, the largest reflexes correspond to the rutile and titanium–manganese spinel enriched with manganese Li(Mn0.8Ti0.2)2O4 (space group Fd3m, a = 8.28 Ǻ, cell volume is 569 Ǻ3) (Cho et al. 1997).

Visualization of nanoparticles and their aggregates

For example, TEM images for the LTMO sample calcinated at 700 °C are given in Fig. 3. The materials are composites, since the image contains no contrast spots, which is confirmed by the data of X-ray analysis. Sizes of primary particles are about 15–40 nm and they form aggregates with size of 50–400 nm, as seen from SEM image (Fig. 4). Smaller aggregates are formed at lower temperature (up to 250 nm); and in this case, the aggregates are more compact.

Fig. 3
figure 3

TEM picture of the LTMO sample calcinated at 600 °C

Full size image
Fig. 4
figure 4

SEM images of the LTMO sorbent calcined at 500 °C (a) and 700 °C (b)

Full size image

Pores and specific surface area

Packing of the sorbent structure with increasing calcination temperature is also confirmed by data of porosity measurements (Table 3). Fraction of micropores in pore volume, which includes micro- and mesopores (the method of capillary condensation, i.e., nitrogen adsorption technique allows us to determine only these types of pores), decreases under elevated temperature. The size of dominant pores shifts toward larger radii indicating merger of the primary particles. The rise in processing temperature over 500 °C is accompanied by a sharp decrease in the pore volume due to shrinkage and compaction of the particles. The method of capillary condensation gives an estimate of the general structural characteristics of the sorbent, but does not reflect the data of lithium-selective sorption centers, a size of which corresponds to Li+ ions. This information could be obtained only from sorption experiments.

Table 3 Structural characteristics of the synthesized materials
Full size table

Li+ sorption under batch and dynamic conditions

From the calculated selectivity coefficients KLi/Na and distribution coefficients \(K_{{{\text{d}}_{\text{Li}} }}\) (Table 4), it can be seen that the lithium distribution coefficients for sorption materials containing manganese are higher than for titanium dioxide; and binary materials are generally more selective to lithium. In addition, the decrease of \(K_{{{\text{d}}_{\text{Li}} }}\) values with transition to higher temperatures of synthesis is worth noting, which is explained by structural transformations in the material at 700 °C. As for the dependence of the selectivity factors on the pH of the adsorbate solution, it can be noted that there is no clear correlation for binary materials. However, there is a difference in the selectivity between sorption from buffer and non-buffer solutions that is explained by the competition between sodium and ammonium ions, as a result of which the sodium sorption is reduced and KLi/Na, respectively, increases.

Table 4 Selectivity coefficients KLi/Na and distribution coefficients \(K_{{{\text{d}}_{\text{Li}} }}\) obtained under batch conditions
Full size table

Figure 5 illustrates concentration of Li+ ions at the column outlet relatively to the initial concentration as a function of bed volume (a ratio of volumes of the passed solution and sorbent). The data are given for the LTMO sample calcinated at 700 °C. It is shown that when passing about 150 bed volumes of the solution, the degree of lithium recovery reached 99%. The ion exchanger retention time was 78 h. The break-through capacity, at which rapid decrease of the curve is observed, is about 0.27 mmol/g, i.e., ≈ 10% of the maximum value under the given experimental conditions. Micropore filling is fast and does not significantly affect sorption rate. The synthesized composites are mainly mesoporous, diffusion is proportional to the concentration gradient. Under these conditions, the rate of sorption is determined by availability of selective sorption centers. Before the break-through capacity, sorption rate is 3.71 × 10−10 mol/(g s), its lower by 1–2 order of magnitude comparing with sorption of transition metals on ion exchange resins (Dzyazko et al. 2006).

Fig. 5
figure 5

Concentration of Li+ ions at the column outlet relatively to the initial concentration (dark symbols) and sorption capacity (white symbols) as functions of bed volume

Full size image

Regarding other samples, no Li+ removal under dynamical conditions was found despite rather high distribution coefficients obtained under batch conditions (Table 4). It means very slow sorption by the samples, which are free from manganese. This is also valid for the LTMO sample calcinated at 500 °C. As shown for inorganic ion-exchangers (zirconium hydrophosphate), increase of sorption capacity slows down sorption (Dzyazko et al. 2006). Indeed, among LTMO sorbents, the lowest distribution coefficients were found for the sample calcinated at 700 °C.

It was found that the LTMO sorbents show high selectivity towards Li+ ions during sorption from the solution containing also Na+ and NH4+. The best selectivity is reached at pH 10. Decrease of the pH results in deterioration of sorption evidently due to decrease of surface charge density, since oxide materials are amphoteric (Amphlett 1964). Indeed, ion exchange mechanism makes sufficient contribution to sorption (the values of equilibrium pH are ≈ 4).

Cation exchange ability of the samples enhances in alkaline solutions. However, lower distribution and selectivity coefficients were obtained at pH 12. This is caused by competition between Li+, Na+, and NH4+ ions. This assumption is confirmed by the data obtained for the solution containing no ammonia buffer. In this case, higher distribution and selectivity coefficients were found than those for the solution containing the buffer.

Discussion

The LTMO sample containing nanosized particles of rutile and nanosized Li0.75Mn0.25Ti2O4 spinel provides sufficient Li+ uptake probably due to formation of selectivity centers. These centers are evidently formed during synthesis, when Li-loaded forms of hydrated double oxide being calcinated followed by partial Li+ removal. Nanosized particles provide sufficient sorption rate, this allows us to remove Li+ ions from aqueous solutions under dynamic conditions. The material demonstrates preferable sorption of Li+ ions over Na+ and NH4+. Since sorption is rather slow (very low feed velocity is needed), it is necessary to use large volume of the sorbent to achieve the most complete lithium recovery. Other way is grinding the sorbent and use combining method of sorption and microfiltration similar to Zhang et al. (2009). This problem will be investigated in future.

Conclusions

LTO and LTMO nanomaterials were obtained by loading of hydrated oxides with Li+ ions, calcination followed by removal of lithium from the solid. The sorbents are related to nanomaterials; this evidently facilitates sorption. Increase of calcination degree up to 700 °C results in insufficient enlargement of the nanoparticles. At the same time, the grains become larger and their packing are looser. In the case of LTMO containing rutile, anatase and Li0.75Mn0.25Ti2O4 spinel, nanodimensional particles make it possible to remove Li+ ions from aqueous solutions under dynamic conditions. These sorbents show the highest Li+ sorption, when these ions are recovered from multicomponent solutions. This is evidently due to formation of selective sorption centers. The highest sorption capacity was found to be reached at pH 10, and ion exchange mechanism makes its contribution to sorption. Due to rather low sorption rate, the material could be recommended as a modifier to prepare Li+-selective membranes. In order to accelerate sorption, it is recommended to obtain smaller grains and attach them to porous support.