Inverse Punicines: Isomers of Punicine and Their Application in LiAlO₂, Melilite and CaSiO₃ Separation

Abstract

The transition to sustainable energy systems demands efficient recycling methods for critical raw materials like lithium. In this study, we present a new class of pH- and light-switchable flotation collectors based on isomeric derivatives of the natural product Punicine, termed inverse Punicines. These amphoteric molecules were synthesized via a straightforward four-step route and structurally tuned for hydrophobization by alkylation. Their performance as collectors was evaluated in microflotation experiments of lithium aluminate (LiAlO₂) and silicate matrix minerals such as melilite and calcium silicate. Characterization techniques including ultraviolet-visible (UV-Vis), nuclear magnetic resonance (NMR) and electron spin resonance (ESR) spectroscopy as well as contact angle, zeta potential (ζ potential) and microflotation experiments revealed strong pH- and structure-dependent interactions with mineral surfaces. Notably, N-alkylated inverse Punicine derivatives showed high flotation yields for LiAlO₂ at pH of 11, with a derivative possessing a dodecyl group attached to the nitrogen as collector achieving up to 86% recovery (collector conc. 0.06 mmol/L). Preliminary separation tests showed Li upgrading from 5.27% to 6.95%. Radical formation and light-response behavior were confirmed by ESR and flotation tests under different illumination conditions. These results demonstrate the potential of inverse Punicines as tunable, sustainable flotation reagents for advanced lithium recycling from complex slag systems.

1. Introduction

Among the many elements that are deemed to be key resources in a transition to renewable energies, lithium is one of high demand but a very minor actual recycling rate [1,2]. However, the annual demand of lithium is estimated to increase by 18% p.a., mostly driven by its application in electromobility and energy storage, which will lead to a depletion of natural reserves by the year 2050 [3,4]. This has recently led to an increased research interest in the recycling of lithium, especially from used Li-ion batteries (LiB). Besides hydrometallurgical treatment, today’s approach to recycling end-of-life LiBs is mainly a pyrometallurgical process, which allows for cobalt, nickel and rare earth element recovery, whereas other elements such as lithium, aluminum and manganese are left in a slag. Current processing technology projects are actively exploring methods to selectively produce lithium-rich artificial minerals within that slag, identify effective separation techniques and develop lithium-rich concentrates. Lithium aluminate (LiAlO₂) is an example of an engineered artificial mineral (EnAM), formed from a blend of lithium oxide, aluminum oxide, silicon dioxide and calcium oxide [5,6]. Matrix-forming minerals, however, are mostly silicates: in several reported slags, minerals like Gehlenite (Ca₂Al [AlSiO₇]), which is part of the melilite mineral group, or Wollastonite (CaSiO₃), made up the majority of the slag matrix [7,8]. Separation of these Li-bearing EnAM phases from their matrix materials is then facilitated by hydrometallurgical separation processes, one of them being froth flotation.

Froth flotation makes use of different surface hydrophilicities and the resulting “de-wetting-ability” of dispersed mineral particles [9,10]. The process involves creating an aqueous suspension and conditioning it with the addition of specific reagents, such as collectors, foaming agents, depressants, activators and regulators. Introducing air into the mixture allows hydrophobized particles to collide with and attach to rising air bubbles, while hydrophilic particles remain in the pulp. These reagents influence the interfacial properties between particles and air bubbles, which, along with hydrodynamic parameters, are crucial for successful flotation. Common collectors in application for oxidic minerals are, for example, sodium oleate (NaOL) or sodium dodecyl sulfate (SDS) [11]. Several other types of cationic collectors have been under current research for oxide minerals, namely, phosphinate collectors [11,12], guanidine collectors [13] or novel Gemini collectors, which feature two separate hydrophobic moieties in one collector molecule [14]. These collectors are either strictly cationic or anionic surfactants. However, the natural product Punicine 1 (1-(2′,5′-dihydroxyphenyl)pyridinium chloride) and its derivatives can change their charge depending on the pH value (Scheme 1) and, as switchable molecules, also change their properties in light (vide infra). Therefore, surfactant-functionalized derivatives of Punicine 1 have been the subject of our ongoing research in the context of lithium recycling [15,16].

Punicine 1 is a natural alkaloid and mesomeric betaine from the Pomegranate tree punica granatum. Depending on the pH value and light scenario, Punicine is able to be switched between different electronic configurations. Thus, Punicines can be cationic, neutral (as two tautomeric mesomeric betaines), anionic or even dianionic, as shown in Scheme 1. Punicine and its derivatives can be easily prepared and functionalized by a facile building block chemistry [17,18]. However, in order to optimize and refine the collectors, also isomeric structures of Punicine are of great interest. Thus, 4-(2′,5′-dihydroxyphenyl)pyridine 2 is an isomer of Punicine and can be referred to as 4-inverse Punicine 4IP, which stands for “inverse punicine with a link via the 4-position of the pyridine”. It has been reported to interact differently with surfaces, such as Pt electrodes, depending on its protonation state. In acidic surroundings, the pyridine nitrogen is protonated; therefore, interaction with surfaces is coordinated by the hydroxy moieties, otherwise coordination is also possible via the free nitrogen electron pair. This also leads to different molecular orientation on the surface [19]. In this study, the synthesis, characterization and application of the 4-inverse Punicine 2 (4IP) and its derivatives is investigated. Advantages of 4-inverse Punicine are the free valence electron pair on the pyridine nitrogen, which offers new donor possibilities compared to Punicine 1 as well as another derivatization anchor. In the following, a straightforward synthesis and characterization of such inverse Punicines is presented. Application of these compounds in separation of lithium aluminate from two matrix minerals is examined.

2. Materials and Methods

2.1. General

All chemicals for this project were purchased from commercial suppliers and used without further purification. Lithium aluminate (>98%) was obtained from SIGMA-ALDRICH (St. Louis, MO, USA). Melilite was obtained from Vata de Sus, Felváca, Romania. As a natural product, it is intergrown with Vesuvianite and Granate in small proportions. Amorphous calcium silicate (>99%) was purchased from SIGMA-ALDRICH and Wollastonite was obtained from the geological collection of Clausthal University of Technology. The Wollastonite provided is a natural product that was intergrown with calcite (CaCO₃) in small proportions. ¹H and ¹³C NMR spectra were measured on the Bruker Avance (400 MHz) and an Avance III (600 MHz) NMR spectrometer at Clausthal University of Technology, Germany. An Alpha-T FT-IR spectrometer from Bruker with a platinum ATR module was used to produce the IR spectra (Clausthal University of Technology, Germany). The spectra cover a range from 400 to 4000 cm⁻¹. The high-resolution mass spectra were measured with an Impact II mass spectrometer from Bruker. UV-Vis spectra were measured on the JASCO V650 (190 nm to 900 nm) spectrometer (Clausthal University of Technology, Germany). Precision cuvettes made of quartz glass from HELLMA (Müllheim, Germany) were used: Type No. 110-QS, light path 10 mm with a volume of 3 mL. The measurements were performed in de-ionized water.

2.2. Syntheses

The synthesis of 4-inverse Punicine 2 (4IP) and its surfactant functionalized derivatives 7a–c followed a straightforward four-step route starting from commercially available 1,4-dimethoxybenzene 3 (Scheme 2). First, selective bromination of the aromatic system with N-bromosuccinimide (NBS) was carried out, which gave 4 in an excellent yield. In a second step, an alkyl chain was introduced to achieve surfactant functionalization via Friedel–Crafts acylation. The respective alkoyl chloride reacted with 4, utilizing AlCl₃ as Lewis acid to form the 1-(2,5-dimethoxy-4-bromophenyl)-1-alkanones 5a–c in excellent yields. Longer reaction times led to formation of side products due to demethylation by AlCl₃, which should be avoided. In order to create the C-C bond between the dimethoxybenzyl- and the pyridyl moiety, the brominated dimethoxybenzenes were reacted with 4-pyridyl boronic acid in a Suzuki–Miyaura cross-coupling reaction, which gave the respective 2,5-dimethoxyphenylpyridines 6a–d in acceptable to good yields. Lastly, demethylation of the methoxy groups took place under mild conditions using BBr₃ [1M in dichloromethane (DCM)], which left us with the desired products 2 and 7a–c in excellent yields.

Direct quaternization of the 4-inverse Punicine 2 by alkylation of the pyridine nitrogen was unsuccessful, which we suspect to be due to the formation of two betainic tautomers shown in Scheme 1. The desired product was then obtained in good yields by alkylation of the pyridine ring of 6a prior to demethylation of the two methoxy groups of the resulting pyridinium salts 8a,b, which hinders the possibility to form tautomers. Demethylation gave the target inverse Punicines 9a and 9b in very good yields. N-Dodecyl pyridinium bromide 10 was obtained by reacting pyridine with 1-bromododecane in toluene. Detailed descriptions of the syntheses as well as spectroscopic characterizations can be found in the Supplementary Materials. An abbreviation system based on “4IP” for compound 2 seems logical and convenient for the discussion of the results, as it can be extended to include the substitution pattern and summarizes the substitution pattern, e.g., C12-4IP for 7b, where C12 stands for the total number of carbon atoms in the side chain, N-C1-4IP for the N-methylated derivative 9a and N-C12-4IP for the N-dodecyl-substituted derivative 9b.

2.3. Contact Angle Measurements: Procedures

Contact angles with water were determined using the Washburn method employing the DCAT25 tensiometer by DataPhysics Instruments GmbH, Filderstadt, Germany. Powder samples of the investigated phases (4 g) were conditioned with the surfactants in aqueous dispersion at ~pH 11 (±0.2) for 5 min, filtered off and dried at 75 °C overnight to ensure no thermal degradation of collectors. The pH was beforehand adjusted with 0.1 M NaOH. Calibration of the powder samples (c-value) was performed with n-hexane.

2.4. Froth Flotation

To assess the impact of inverse Punicines as hydrophobizing agents, we conducted microflotation experiments using modified 250 mL Hallimond tubes made of Pyrex or quartz glass. These tubes feature a medium-pore-fritted glass that disperses air into the system. The airflow was set at 32 cm³/min. A magnetic stirrer operating at a constant speed of 500 rpm ensured thorough mixing of the dispersion. In each experiment, 2.00 g of the mineral, or, respectively, a mixture of 1.00 g LiAlO₂ and 1.00 g Melilite/CaSiO₃, underwent flotation, with variations in the amount of added collector. The pH value was regulated using Britton–Robinson Buffer and 0.1 M aqueous NaOH. The samples were conditioned as follows: 1 min mineral and 25 mL of water; 1 min pH adjustment with buffer solution; 1 min after adding the collector. Collector solutions were 50 mmolar solutions of the respective compound in ethanol. Following conditioning, the Hallimond tube was topped with pure water to reach 250 mL, and flotation started, which was then continued for three/five minutes. The concentrate was then separated using a funnel trap. The tailings remained in the fritted tube and were collected for further analysis. Both concentrate and tailings were filtered and dried over 24 h at 80 °C. Recovery yields were determined based on the mass ratios of concentrate and tailings. Each experiment was repeated at least three times, then the 95% confidence interval was calculated. We also varied light conditions during flotation, as prior research has shown that light scenarios affect the recovery rates of LiAlO₂ when using inverse Punicines as froth flotation collectors. The trials alternated between daylight flotation (>5000 lux) and UV light irradiation (390–400 nm, 4500 lux). UV experiments took place in a closed box (60 × 60 × 60 cm) containing the Hallimond tube. Each wall had four LEDs from AVONEC (Wesel, Germany) (Premium 3W LED, 390–400 nm, UV-A color, max power: 750 mA, operating voltage: 3.5–4.5 V), powered by an APC-16-700 unit (input: 100–240 V, output: +24 V, max. 700 mA, class 2 power supply), irradiating the tube. For details of the Hallimond tube and the irradiation experiment setup, see the Supplementary Materials.

2.5. Particle Size Measurements

The particle size distribution of mixed samples and single mineral samples of LiAlO₂, Wollastonite and Melilite was determined with dry measurement via the HELOS laser diffractometer from SYMPATEC (Clausthal-Zellerfeld, Germany) at an air pressure of 4 bar.

2.6. X-Ray Diffraction Analysis (XRD)

Chemical analysis, separation and comminution of LiAlO₂, Melilite and Wollastonite were performed at the Department of Geoscience, Institute of Geotechnology and Mineral Resources. Powder X-ray diffraction (PXRD) was performed using a PANALYTICAL X-Pert Pro Diffractometer with a Co X-ray tube (MALVERN PANALYTICAL GmbH, Kassel, Germany, λ = 1.7902 Å), measured at 40 kV and 40 mA. The measurement was conducted over an angular range of 4 to 90°. A step size of 0.026° was chosen, with a measurement time of 150.45 s per step. The analysis was carried out using the HighScore v5.3a program, and phase identification was performed using the PDF-2 ICCD XRD database as well as the American Mineralogist Crystal Structure Database. Semi-quantitative phase identification was conducted using the reference intensity ratio (RIR) method without Rietveld refinement [20]. For XRD spectra of the minerals and mineral mixtures, see the Supplementary Materials.

2.7. Electron Spin Resonance Spectroscopy (ESR): Procedures

Continuous-wave ESR spectra of compounds 2 and 9a were recorded on a Bruker Elexsys 580 spectrometer at X-band frequencies of 9.66–9.72 GHz at ambient temperature (2 mW microwave power) and at a temperature of 80 K (2 µW). Data were acquired with a time constant of 2.56 ms and a conversion time of 10.24 ms, using a modulation amplitude of 0.1 mT at ambient temperature and 0.2 mT at 80 K. A BRUKER MD4EN dielectric ring resonator was employed. The g tensor and hyperfine tensors for semiquinone and anion radicals of both compounds were computed with ORCA by density functional theory (DFT) [21]. First, geometry was optimized at B3LYP/def2-SVP level. Second, ESR parameters were computed with the B3LYP functional, the EPR-III basis set for protons and the EPR-II basis set for all other atoms. Third, the list of hyperfine coupled nuclei was reduced to ¹⁴N and ¹H nuclei for which at least one principal value of the hyperfine tensor exceeded 5.6 MHz (twice the modulation amplitude at ambient temperature). Spectra were then simulated with EasySpin [22] while accounting for neglected nuclei by Voigtian line broadening with 0.1 mT broadening parameters for both the Lorentzian and Gaussian components.

2.8. Zeta Potential Measurements

The measurements were conducted with the Zetasizer Nano from MALVERN at Clausthal University of Technology, Germany. A suspension of 1 g of the mineral was prepared in de-ionized water and left to settle so that large particles sedimented. pH values were adjusted by the addition of Britton–Robinson buffer. 1 mL of the supernatant dispersion was then taken and mixed with 7.2 µL of the respective collector solutions (50 mmol/L). Measurements were repeated three times.

3. Results

First, a thorough characterization of the Punicine isomers by means of UV-Vis, NMR- and ESR-spectroscopy was performed. To achieve this, 4-inverse Punicine 2 as well as N-C1-4IP 9a were examined as reference substances for the molecule class.

3.1. UV-Vis

The 0.1 mmolar solutions of 2 and 9a, respectively, were prepared in a 1:1 mixture of EtOH and H₂O. The spectra are shown in Figure 1.

4-Inverse Punicine 2 in neutral solution shows an absorption band at 260 nm as well as a shoulder at 220 nm, which resembles the neutral charge form. Decreasing the pH value then leads to protonation of the molecule on the pyridine; hence, absorption bands shift, and at 260, 290 and 360–390 nm absorption now occurs. Under slightly basic conditions, the shoulder at 220 nm disappears; however, no new absorption bands show, it is still present in neutral state. Raising the pH further (pH = 13) now leads to deprotonation, resulting in a broad absorption between 290–320 nm.

For N-C1-4IP 9a, also distinct absorption patterns are observed at various pH levels, indicating structural and electronic changes in the molecule. At pH 4, the sample exhibits a prominent band at 270 nm alongside a shoulder at 210 nm, a pattern consistent in slightly acidic conditions. This reflects the deprotonated, betainic structure. However, in more acidic environments (pH 2 and 1), the molecule’s absorption shifts, revealing bands at 260 nm and an additional band at 325 nm, which indicate reprotonation to the salt form. At neutral and basic pH levels, the data show bands at 260 nm and 380 nm, which are indicative of the molecule’s anionic form. As the pH reaches a highly alkaline level of 13, the appearance of absorbance between 280 to 320 nm highlights the presence of the dianion, reflecting further deprotonation.

3.2. Electron Spin Resonance Spectroscopy (ESR)

Previous research has shown pyridinium compounds, such as Punicines, or pyridinium ylides, can exist as pyridinyl radical species, some of them stable enough to be isolated [18,23,24]. Such radical species allow Punicines to act as light- and pH-switchable collectors in lithium recycling [15,16]. To determine if the isomers of Punicine also possess a radical character, they were subjected to ESR measurements. Samples were measured as a powdered solid as well as in ethanolic solution. Deprotonation was carried out with NaOH. For Punicine, in the simplest case, a disproportionation for radical formation can be assumed, which can also be adapted for inverse Punicine (Scheme 3) [25]. Tautomers of the radicals can also be formulated. For 4-inverse Punicine 2, the ESR spectra suggest a stable semiquinone radical (Figure 2). Further ESR measurements were also conducted for 9a, where the alkylation of the pyridine nitrogen hinders the tautomerism shown in Scheme 3. For this compound, the ESR spectra are inconsistent with a radical localized on a single inverse punicine unit. The observed spectra are narrower than expected for such species, indicating smaller hyperfine couplings. Such behavior would be expected upon oligomerization where the unpaired electron is distributed over several inverse Punicine units. A complete overview about ESR measurements and simulations is given in the Supplementary Materials.

3.3. Zeta Potentials

For the applicability of surfactants as collectors in mineral processing procedures such as froth flotation, insights about particle–collector interaction are of great interest. As previous research in our group has shown, a chemisorption of Punicines via substitution reaction on the hydrolyzed lithium aluminate surface takes place [14]. For zeta potential measurements, we chose inverse Punicine collectors 7b (C12-4IP) and 9b (N-C12-4IP) functionalized with a C12-alkyl chain as their surfactant functionalization. To further understand the role of OH-moieties in Punicines and inverse Punicines as collector molecules, zeta potentials of Punicine 8b (Dimethoxy-N-C12-4IP) were studied additionally (Figure 3). For the measurements, a dispersion of lithium aluminate was adjusted to the respective pH values using Britton–Robinson buffer and left standing so larger particles could settle. Afterwards, 2 mL of the supernatant dispersion were treated with a 50 mmolar ethanolic solution of the collector so that a collector concentration of 0.06 mM was achieved. At pH 2, all collectors exist as cationic species, interaction with the LiAlO₂ surface therefore only results in a slight increase of zeta potential, which might be due to interaction with the surface via the oxygen atoms of the hydroxy/methoxy moieties, among other factors. As the dimethoxy collector 8b (Dimethoxy-N-C12-4IP) is not deprotonable in higher pHs, it generally leads, as a cation, to an increase in lithium aluminate zeta potential, as it can interact with the otherwise negatively charged surface of LiAlO₂ due to physisorption. At weakly acidic pH, neutral/betainic collectors 7b and 9b do not have a significant effect on zeta potential. At strongly basic pH, though, the anionic collectors can interact with the mineral via their respective olate moieties, thus decreasing the zeta potential further.

Also, the zeta potential of pure calcium silicate was determined in pH dependency (Figure 4). Notably, calcium silicate is also close to isoelectric at every pH besides pH = 2, where enrichment of protons in the shear layer of the particle surface leads to a positive zeta potential. A zeta potential close to the isoelectric point might destabilize the dispersion and facilitate agglomeration of calcium silicate particles, which consequently would hinder floatability. However, since calcium silicate is a gangue material and to be separated, this is beneficial.

3.4. Contact Angle Measurements

Contact angles of lithium aluminate, melilite and commercial calcium silicate were determined (Washburn method) for the pure mineral samples as well as after being conditioned with the collectors. The results are presented in

Table 1. Contact angles of the respective minerals, pure and after conditioning with collectors.

Collector	Lithium Aluminate	Melilite	Calcium Silicate
pure mineral	26°	61°	65°
7a	35°	56°	69°
7b	50°	67°	68°
7c	>90°	65°	72°
8b	59°	60°	74°
9b	>90°	70°	78°
10	54°	47°	75°

. A collector concentration of 0.06 mM in the conditioning dispersion (which corresponds to 7.5 µmol collector per gram mineral) was applied. Contact angles exceeding 90° cannot be measured with this method, therefore no exact values are given.

Contact angles deliver information on how well a particular collector is able to perform in froth flotation. However, because flotation involves de-wetting a particle, and contact angle measurements involve wetting a previously dry surface, discrepancies between the two methods may arise. LiAlO₂ by itself is a hydrophilic mineral with a contact angle of 26°, as opposed to the gangue materials melilite and calcium silicate with 61° and 65°, respectively. After treatment with collectors, though, mainly the hydrophobicity of lithium aluminate can largely be influenced, increasing the contact angle especially with collectors 7c (C16-4IP) and 9b (N-C12-4IP) to over 90°, which results in a very hydrophobic particle suited for flotation. Most collectors also have a hydrophobizing effect on the gangue minerals, but not at an immense rate. The significant difference in contact angle on LiAlO₂ between 8b and 9b, which differentiate in methoxy and hydroxy moieties, suggests that the free OH-groups play a significant role in adsorption to the surface and therefore hydrophobized the mineral.

3.5. Froth Flotation of LiAlO₂

To examine inverse Punicines as collectors for the recycling of the engineered artificial mineral (EnAM) LiAlO₂, microflotation experiments were performed (see Section 2.3). First, 4-inverse Punicines with varying alkyl chain lengths C8, C12 and C16 (7a–c) were used as collectors in daylight experiments at pH values of approximately 2 and 11, since at these pH values, the LiAlO₂ dispersion is stable and interaction with the different protonation states of the collectors occurs. By this experiment, the amphoteric character of the inverse Punicines and its effect on collector properties can be gauged. Flotation time was 5 min, respectively.

Best flotation results were observed when flotating lithium aluminate with 7b as collector at the pH of 11, improving the flotation yield from 40.7% without any collector to 85.6% under otherwise same conditions (Figure 5). Even though a longer alkyl chain at the C16-4IP 7c bears a higher hydrophobic parameter logP than C8 and C12, flotation under these conditions at pH 11 results in less lithium aluminate in concentrate.

Table 2. Calculated logP values for collectors in their different protonation states.

	7a	7b	7c
Anionic logP	3.60	5.43	7.25
Neutral logP	3.65	5.31	6.98

shows estimated logP for the synthesized collectors calculated by ChemDraw 23.0.1. using Broto’s method [26,27]. For cationic compounds, no logP can be calculated with the present increment data.

However, in pH 2 it performs better than C8-4IP 7a and C12-4IP 7b. The reason for this phenomenon might be due to the assumed interaction mechanism with the mineral. At acidic pH, inverse Punicines are present in their cationic form and therefore are able to interact with the surface through Coulomb interaction. At pH 11, though, a substitution reaction with OH-moieties on the mineral is suggested [15]. For this, the long C16-alkyl chain in proximity to the olate moieties in the punicine might be of steric hinderance, and the hydrophobization of the mineral is then kinetically hindered.

Furthermore, larger-chain molecules tend to form micelles, which can impair their ability to arrange effectively on the particle surface. This can lead to reduced coating efficiency and consequently to lower floatability [28].

As ESR-spectroscopy (Section 3.3) shows, inverse Punicines are able to form radicals, which can lead to changeable interaction with the mineral regarding the lighting scenario. As it performed best, 7b was then subjected to microflotation experiments in daylight, UV light (390 nm) and total darkness (Figure 6). Overall, yields are lower than shown in Figure 2, as we used LiAlO₂ of larger particle size (60–100 nm, see the Supplementary Materials) to avoid reaching yield limits and entrainment effects. At pH 11, yields in darkness and UV irradiation increase by around 10%, though no significant difference between UV light and darkness is seen. At pH 2, though, only an increased yield in absence of any light is observed. Taking the UV-Vis experiments into consideration, 4-inverse Punicines only have an absorption band at 390 nm at pH 2; therefore, the reason for this observation might be that at pH 11, C12-4IP 7b does not absorb 390 nm UV irradiation. This results in no radical formation, which is also the case in darkness, and hereby higher flotation yields. However, as collector–mineral interactions are very complex dynamic systems, this might only be a part of the reason for the reported behavior.

Likewise, N-alkylated 4-inverse Punicines were subjected to microflotation experiments (Figure 7). To gain deeper understanding of the structural increments of the collectors and their effect, three collectors have been compared: N-dodecyl pyridinium bromide 10 as a common cationic collector [29] as well as structural increment of N-alkylated inverse Punicines, Dimethoxy-N-C12-4IP 8b and N-C12-4IP 9b. N-Dodecyl-pyridinium bromide 10 is outperformed by the inverse Punicine collectors in regards of lithium aluminate yield in the single mineral microflotation. Collector 8b shows minimally higher yields at pH 11 because it is slightly more hydrophobizing. It is a cationic collector, which possibly works well with LiAlO₂ due to its negative zeta potential. Collector 9b, however, is anionic at pH 11. Its interaction with the lithium aluminate surface is possible by nucleophilic substitution to the surface as we observed previously with other Punicine derivatives [14].

Between UV irradiation and daylight flotation, for these collectors, no significant difference in the LiAlO₂ yield in the concentrate is seen. This is consistent with ESR results, as in N-alkyl-4IPs the unpaired electron appears to be distributed over several inverse Punicine units in an oligomer, making it much less reactive.

Lastly, comparison of the aqueous filtrates obtained from the Hallimond tube flotation experiments further indicates that collector 9b exhibits the strongest affinity for the mineral surfaces. The collectors impart a distinct yellow to orange coloration both in solution and on the surface of the treated minerals. Following filtration of the concentrate and tailings, both the filtrates and the corresponding solids retained this coloration. In contrast, when collector 9b was employed, the resulting filtrate appeared colorless, suggesting a complete and robust adsorption of the collector onto lithium aluminate surfaces.

Comparing to other collectors for the separation of lithium aluminate and melilite, inverse Punicine collectors show slightly higher recovery rates than comparable flotation experiments reported in the literature. Qiu et al. report LiAlO₂ recovery maxima of 56.03% (3 min flotation) using decyl dihydrogen phosphate at pH 11 [30]. Besides the EnM concept, recuperation of lithium from LiBs in the form of LiCoO₂ has been investigated by LIU et al. [31]. They report a recovery of 89.83% of LiCoO₂ by flotation after cryogenic grinding; however, a flotation collector is not given. The EnAM concept is advantageous to this, though, as valuable Co is separated beforehand and can be recycled before Li-rich slags are formed.

3.6. Froth Flotation of Gangue Material: Melilite and Calcium Silicate

Melilites are silicate minerals that serve as a model for the gangue phase in the EnAM slag which has to be separated from the Li-bearing phase. To enrich the precious LiAlO₂ mineral in the concentrate, collectors should prefer hydrophobization of this mineral as opposed to the gangue material. In a screening experiment, comparison of pure melilite flotation against 7b (C12-4IP) as well as commercially available collector sodium oleate was performed (Figure 8). It can be seen that the inverse Punicine collector improves also the gangue material flotation; however, less so than sodium oleate.

As a second gangue phase model, calcium silicate (Wollastonite) was chosen. Since slags, depending on the cooling rate, can differ between amorphous (rapid cooling) and crystalline silicates (slow cooling) in the matrix, a comparison between amorphous CaSiO₃ and crystalline Wollastonite was investigated (Figure 9). Crystalline Wollastonite shows a slightly higher flotation yield utilizing collector 9b and sodium oleate, while blank flotation almost gives the same result for both minerals. These two collectors produce a more stable foam, which entrains more of the fine particles of the Wollastonite (cf. particle sizes in the Supplementary Materials) and therefore leads to slightly higher yields. As at pH 11 CaSiO₃ bears a zeta potential around zero, no physisorption due to Coulomb interactions with the collector is expected. The low recovery rates as well as the contact angles also indicate that no inherently strong chemisorption of the punicine collectors takes place. Therefore, the low recoveries of this matrix material, regardless of crystallinity, are observed.

Finally, a 1.00 g:1.00 g mixture of LiAlO₂ and Melilite with 8b (Dimethoxy N-C12-4IP), 9b (N-C12-4IP) and N-dodecyl-pyridinium bromide 10 as collectors in a Hallimond tube was flotated to estimate separation capabilities (Figure 10). Lithium grade in this feed composition is 5.27%.

The concentrate was then analyzed via PXRD to give a semiquantitative analysis of the concentrate composition. PXRD results showed a concentrate composition of LiAlO₂ and Melilite as well as LiAl₂(OH)₇(H₂O)₂ and traces of CaCO₃. XRD spectra are given in the Supplementary Materials.

The highest enrichment of lithium aluminate as well as lowest quantity of Melilite were observed with flotation with 9b (LiAlO₂ 64.7%, melilite 29.7%, 4.3% LiAl₂(OH)₇(H₂O)₂) (

Table 3. Concentrate composition as given via XRD analysis without Rietveld refinement.

	LiAlO2 %	Melilite %	LiAl2(OH)7(H2O)2 %	CaCO3 %	Li Grade %
N-C12-4IP 9b
	64.7	29.7	4.3	1.7	6.95
Dimethoxy-N-C12-4IP 8b
	61.7	33.0	4.0	1.0	6.63
N-dodecyl pyridinium bromide 10
	61.3	34.3	3.3	1.0	6.57

). This corresponds to a lithium grade of 6.95% in the concentrate.

These results are only semiquantitative; calculated masses of LiAlO₂ slightly exceed 1.00 g. Partially, this might be due to increased weight of lithium species due to formation of LiAl₂(OH)₇(H₂O)₂. However, a tendency on the separation abilities can be deduced. At pH 11, 9b with its capabilities to chemically absorb on the LiAlO₂ surface, as opposed to the cationic collectors, seems to favor selectivity towards the lithium aluminate in the mineral mixture.

4. Conclusions

This study introduced a new class of flotation collectors, the inverse Punicines, derived from the natural product Punicine, and demonstrated their effectiveness in the selective separation of lithium aluminate from complex silicate matrices. The synthesized compounds exhibit strong pH- and light-switchable behavior, with structural modifications—particularly alkylation—significantly influencing their collector performance. Among the tested variants, the N-C12-4IP (9b) displayed the highest selectivity and adsorption efficiency toward LiAlO₂, achieving flotation recoveries up to 86% under alkaline conditions. The selectivity is further backed by contact angle measurements, as 9b along with 7c show strong hydrophobization for LiAlO₂, but only weak effects on gangue minerals.

Spectroscopic analyses confirmed the amphoteric nature of these molecules and their ability to form radicals, which contributed to their tunable surface activity. Furthermore, the flotation behavior varied with lighting conditions, reflecting the dynamic electronic configurations of these switchable collectors.

Overall, inverse Punicines represent a promising and sustainable alternative to conventional collectors in lithium recycling technologies. Their pH- and light-responsive nature opens new avenues for selective, environmentally adaptive mineral separation processes. Future research may further explore structural refinements and scale-up potential for industrial applications.