Optimizing Recycling Processes for Mixed LFP/NMC Lithium-Ion Batteries: A Comparative Study of Acid-Excess and Acid-Deficient Leaching

Abstract

This study explores the optimization of hydrometallurgical processes for recycling lithium-ion batteries (LIBs) containing a mixture of lithium iron phosphate (LFP) and nickel–manganese–cobalt (NMC) cathodes. Two approaches were investigated: acid-excess leaching and acid-deficient leaching with residue recirculation. A design of experiments (DoE) framework was applied to assess the impact of key parameters, including sulfuric acid and hydrogen peroxide concentrations, as well as solid-to-liquid (S/L) ratios, on the dissolution yields of target metals (Ni, Mn, Co, and Li). Acid-excess leaching achieved nearly complete dissolution of target metals but required additional purification steps to remove impurities. Acid-deficient leaching with a 60% recirculation of leaching residue improved dissolution yields by up to 12.5%, reduced reagent consumption, and minimized operational complexity. The study also evaluated separation strategies for manganese and cobalt through solvent extraction. Results indicate that while acid-excess leaching offers higher yields, acid-deficient leaching with residue recirculation is more cost-effective and environmentally friendly. These findings provide valuable insights for developing sustainable LIB recycling technologies.

1. Introduction

Since their commercial introduction in 1991, lithium-ion batteries (LIBs) have become indispensable in modern society, powering a wide range of electronic devices. Today, they play a pivotal role in the ecological, energy, and digital transitions [1,2,3]. However, the proliferation of LIBs presents a twofold challenge: they pose a significant environmental hazard if improperly disposed of while simultaneously representing a valuable secondary source of critical metals [4]. The recycling of these batteries is imperative for several reasons: to recover the valuable metals they contain, to mitigate pollution from end-of-life batteries, and to reduce the carbon footprint associated with battery production [5]. Furthermore, recent European regulations have established specific targets for the use of recycled materials in battery production. By 2030, 12% of cobalt (Co), 4% of lithium (Li), and 4% of nickel (Ni) in new batteries must be derived from recycled sources. These targets are set to increase to 20%, 10%, and 12%, respectively, by 2035 [6].

Most of the current processes for treating batteries involve a combination of physical, thermal, and chemical operations [7,8]. Typically, lithium-ion battery (LIB) recycling begins with comminution, a mechanical stage that breaks down the batteries to release their components. This stage is followed by sieving, which separates the coarse steel casing, plastics, and metallic foils from the fine, liberated active particles. Mechanical sorting enhances the liberation of LIB components before further metallurgical processes. The fine fraction (under 1 mm) that remains after sorting, commonly referred to as “black mass” (BM), contains a mix of cathode and anode active materials along with impurities like metallic particles from the foils. Anode material can be either separated from cathode material by flotation or completely removed by thermal treatment. In modern recycling technologies, this black mass serves as the starting point for chemical processes aimed at recovering valuable metals such as nickel (Ni), cobalt (Co), manganese (Mn), and lithium (Li). These processes are typically hydrometallurgical, involving leaching, separation, purification, and either precipitation, extraction, or electrolysis, with the exact process design tailored to the cathode materials present in the battery.

While current recycling efforts largely focus on electric vehicle (EV) batteries, there is an increasing need to address the recycling of smaller batteries, such as those used in electric bicycles and scooters. The streams of these smaller spent batteries, which are becoming more prevalent as the market grows, contain a mix of lithium iron phosphate (LFP) and nickel–manganese–cobalt (NMC) chemistries. The recycling methods for this stream may differ from those used for streams produced from the mineral processing of EV batteries due to the variations in material composition. In electric vehicles (EVs), the composition of batteries is well defined, with cathode materials primarily consisting mainly of either NMC (LiNi_xMn_yCo_zO₂) or LFP (LiFePO₄) technologies. As a result, recycling methods for these materials in separate streams have been extensively studied [9,10,11,12,13] while a few research has addressed the treatment of mixtures of LFP and NMC [14,15]. Additionally, there is a growing trend of combining cathode active particles from different sources. For example, LFP can be blended with NMC to take advantage of its higher energy density. Therefore, it is crucial to develop an agile recycling process that can handle this mixture of cathode materials.

Different options are available to process this type of mixture by selecting a suitable leaching agent, such as organic or inorganic acid or even deep eutectic solvent (DES), to achieve high leaching efficiency while minimizing operation costs and waste production [16]. Inorganic acids offer favorable results at low cost but produce toxic wastes. Conversely, many organic acids and DES are biodegradable though they pose economic challenges. Recent volatility in metal prices has prompted a re-evaluation of LiBs-related projects, with a particular focus on economic considerations [17]. In this study, we focus on acid leaching, which has proven effective for NMC materials. Nearly 100% leaching efficiency is achieved for Co, Mn, and Ni under conditions of 1 mol/L of H₂SO₄, 4.5% (Vol. %) H₂O₂, a pulp density of 50 g/L and ambient temperature [18]. Similarly, sulfuric acid shows promising results for LFP leaching in terms of 96.85% of lithium recovery using 0.3 mol/L of H₂SO₄, an H₂O₂/Li molar ratio of 2.07, and an H₂SO₄/Li molar ratio of 0.57 during 120 min at 60 °C [12]. Special attention is given to reagent consumption to minimize both environmental and economic impacts. Zou et al. demonstrated that a process with excess sulfuric acid followed by impurities removal leads to a good recovery with few impurities [19]. In contrast, in a previous study, we used a stoichiometric quantity of acid to selectively leach target metals but achieve slightly lower dissolution yields [20]. Therefore, two different routes can be employed to process black masses containing a mixture of LFP and NMC: (i) a multi-step process that achieves high dissolution efficiency for both target metals and impurities, requiring subsequent purification stages to remove co-dissolved impurities from the leaching solution (Figure 1a), and (ii) selective leaching of the target metals under suitable conditions, leaving impurities in the leaching residue that can be partially reintroduced into the leaching stage (Figure 1b).

This study aims to model and optimize the dissolution yields of target metals for both methods using a design of experiments (DoE) approach. The processes are then compared based on dissolution yields, reagent consumption, impurity removal efficiency, and the potential reintroduction of leaching residue into the leaching stage. The target recovery aims for approximately 90% dissolution yield of the element of interest, with minimal impurities present before the separation step. Furthermore, the separation of manganese and cobalt is examined as both methods produce a clean pregnant leaching solution (PLS), rich in Ni, Mn, Co, and Li, with pH values of 4.5–5, suitable for solvent extraction. To be considered valuable, solvent extraction processes have to produce solutions with a minimum of 98% purity.

2. Materials and Methods

2.1. Materials

Leaching experiments were performed on pristine materials of NMC and LFP (Xiamen Tob New technology Co., Ltd., Xiamen, China) whose chemical compositions were confirmed by elemental analyses with Agilent (Agilent Technologies, Inc., Santa Clara, CA, USA) MP-AES (

Table 1. Chemical compositions of LFP and NMC materials used in this work for leaching experiments, molar ratio vs. Li, and particle size distribution (D10, D50, and D90 in µm).

		Li	Ni	Mn	Co	Fe	P	O	D10	D50	D90
LFP	Wt. %	4.64				34.23	18.01	43.12	0.020	0.074	4.73
	Molar ratio	1.00				0.92	0.87	4.03
NMC	Wt. %	7.71	49.41	4.30	6.78			31.80	0.019	0.094	7.53
	Molar ratio	1.00	0.76	0.07	0.10			1.79

). Table 1 also shows the particle size distributions in NMC and LFP determined by laser diffraction using Malvern Mastersizer 3000 (Malvern Panalytical, Inc., Massy, France).

Elemental analyses confirmed that the composition of the cathode material provided by Xiamen Tob New Technology Co corresponds to NMC811 and LFP even if it appears that Ni and Mn contents are slightly lower than expected (LiNi_0.76Mn_0.07Co_0.10O_1.79 instead of LiNi_0.8Mn_0.1Co_0.1O₂). LFP also contains slightly less iron and phosphorus than expected (LiFe_0.92P_0.87O_4.03 instead of LiFePO₄).

Industrial black masses typically contain aluminum (Al) and copper (Cu) from electrode foils, in addition to Li, Ni, Mn, and Co from NMC, as well as Li, Fe, and P from LFP [7,15,16]. To simulate this composition, 2% (Wt. %) of Al and 1% (Wt. %) of Cu [17] were added to the black mass mixture. The final mixture comprised 48.5% NMC, 48.5% LFP, 2% Al, and 1% Cu by weight.

2.1.1. Leaching

The leaching experiments were performed in 100 mL borosilicate glass flasks shaken in a mechanical stirring apparatus (Gherardt Laboshake, Strasbourg, France) thermostated with a Gherardt Thermoshake (THL 500/1) at 200 rpm and 30 °C for 240 min. The leaching reagent was appropriate amounts of sulfuric acid (95–97%, Fluka, lot #SZBG3220H), hydrogen peroxide (30% weight, Fisher scientific, lot 223487) and deionized water (resistivity = 18 Mohms, VWR Puranity TU 6+). Leaching experiments were conducted by stirring 20 mL of the leaching reagent with an appropriate amount of black mass to reach the desired value of solid-to-liquid ratio (S/L). Leaching solutions, after cooling down to room temperature, were filtered using a 0.45 µm Teflon Digifilter from SCP Science (Villebon, France).

Elemental analyses were used to calculate the dissolution yield of the metal M (%D(M)) as follows:

$$\%\mathrm{D}\left(\mathrm{M}\right)=100\times {\displaystyle \frac{\mathrm{C}\left(\mathrm{M}\right)}{{\mathrm{C}}_{\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}}\left(\mathrm{M}\right)}}$$

(1)

where C(M) is the concentration of the metal M in solution after leaching and C_total(M) is the concentration of the metal M that would be obtained after complete leaching.

2.1.2. Solvent Extraction

Appropriate amounts of Cyanex^®272 (bis-2,4,4-trimethylpentylphosphinic acid, Syensqo, purity = 90%) and DEHPA (bis(2-ethylhexyl) phosphate, Sigma Aldrich, Paris, France, purity = 97%) were diluted in kerosene (low odor, Sigma Aldrich, Paris, France, purity ≤ 100) to prepare extraction solvents. The final concentrations of Cyanex^®272 and DEHPA were 0.9 mol/L and 0.48 mol/L, respectively. During the solvent extraction tests, 5 mL of the extraction solvent was mixed with 5 mL of the PLS with a phase volume ratio O/A = 1 (the phase/volume ratio (O/A) is the quotient of the volume fraction of the organic phase and the volume fraction of the aqueous phase). The pH of the aqueous phase in contact with the organic phase was adjusted to the target values by adding 10 mol/L NaOH (98%, anhydrous, Sigma Aldrich, Paris, France). The biphasic system was then shaken at 200 rpm and at room temperature for 30 min in a controlled shaker (Gherardt Laboshake THL500/1, Strasbourg, France). After settlement, the biphasic system was centrifugated at 3000 rpm for 3 min using a SIGMA 3–16L centrifuge (VWR, Paris, France). Stripping and scrubbing were performed with 0.1 mol/L sulfuric acid and 3 g L⁻¹ MnSO₄, respectively (O/A = 1, room temperature). Aqueous solutions were analyzed before and after extraction. The metal concentration in the organic phase (${\left[M\right]}_{org}$) was deduced by mass balance and the extraction of the metal M$({\%E}_M$) was calculated as follows:

$${\left[M\right]}_{org}=\left({\left[M\right]}_{aq,0}-{\left[M\right]}_{aq}\right)\times {\displaystyle \frac{V_{aq}}{V_{org}}}$$

(2)

$${\%E}_M=100\times {\displaystyle \frac{\frac{{\left[M\right]}_{org}}{{\left[M\right]}_{aq}}\times \frac{V_{org}}{V_{aq}}}{\left(1+\frac{{\left[M\right]}_{org}}{{\left[M\right]}_{aq}}\times \frac{V_{org}}{V_{aq}}\right)}}$$

(3)

where ${\left[M\right]}_{org}$—concentration of the metal M in the organic phase after extraction, ${\left[M\right]}_{aq}$—concentration of the metal M in the aqueous phase after extraction, ${\left[M\right]}_{aq,0}$—concentration of the metal M in the aqueous phase before extraction, $V_{aq}$—volume of the aqueous phase, and $V_{org}$—volume of the organic phase.

2.2. Methods

2.2.1. Analytical Methods

Prior to analysis, the samples were diluted in 2% (Vol. %) HNO₃ (65%, Sigma Aldrich, lot #STBK3044) for elemental analyses using a microwave plasma-atomic emission spectrometer (Agilent 4210 MP-AES, Santa Clara, CA, USA).

The samples were diluted by a factor of 10 for Cu and Al analyses whereas 200-fold dilutions were performed for Ni, Mn, Co, Li, Fe, and P analyses. Each dilution was performed with 2% (Wt. %) nitric acid (65%, Supelco, lot K53085243 106, Bellefonte, PA, USA). Elemental analyses of Li, Ni, Mn, Co, Fe, P, Cu, and Al were performed at 610.635, 341.476, 279.482, 350.631, 373.486, 213.618, 510.554, and 396.152 nm, respectively.

Solid samples were analyzed by X-ray diffraction (XRD) and scanning electron microscopy (SEM). All samples were ground prior to the XRD analysis. The mineralogy was determined using a D8 Discover Bruker diffractometer (Palaiseau, France) equipped with an X-ray tube emitting Cu-Kα1 radiation (λ = 1.5406 Å) at 40 kV and 40 mA. X-ray diffraction patterns were recorded with a LynxEye detector under ambient conditions from 2θ = 2.5° to 65° (step 2θ = 0.035°, recording time = 3 s/step). Analyses were performed using DIFFRAC. EVA Bruker software (V14.0.0) and diffraction peaks were identified using the powder diffraction file database.

SEM analyses were carried out using the Tescan VEGA3 (Bruker, Palaiseau, France) equipped with an energy-dispersive spectrometer (EDS, Bruker XFlash6 with an area of 15 mm², Palaiseau, France) for semi-quantitative analyses. The imaging voltage was set at 20 kV. Sem images are available in the supporting information (Supplementary Figure S1).

2.2.2. Design of Experiments

One of the objectives of this study was to assay the impact of parameters (solid-to-liquid ratio, and H₂SO₄ and H₂O₂ concentrations) on the dissolution yields of target metals during the leaching process by conducting an orthogonal design of experiments (DoE). This approach allows for drawing reliable conclusions from a reduced number of experiments, thus offering a more efficient alternative compared to other design of experiments methods [21,22]. JMP 14 Pro (V14.0.0) statistical software was used to model the dissolution yields of the metals as a function of the above-mentioned parameters.

The solid-to-liquid ratio was selected between 50 and 90 g/L to reflect typical industrial applications. For acid-excess leaching, the sulfuric acid concentration ranged from 1 and 3 mol/L.

In acid-deficient leaching, the acid concentration served as the limiting factor, with the aim of using the maximum possible concentration while ensuring insufficient acid remained to complete the reaction by the end of the leaching process. To assess this, the acid consumption during the reaction was quantified. Li et al. and Fan et al. identified the following reactions for LFP and NMC leaching with sulfuric acid in the presence of hydrogen peroxide (H₂O₂) [9,12]:

$${2\mathrm{LiFePO}}_{4\left(\mathrm{s}\right)}+{\mathrm{H}}_2{\mathrm{SO}}_4+{\mathrm{H}}_2{\mathrm{O}}_2\leftrightharpoons {\mathrm{Li}}_2{\mathrm{SO}}_4+{2\mathrm{FePO}}_{4\left(\mathrm{s}\right)}+{2\mathrm{H}}_2\mathrm{O}$$

(4)

$${2\mathrm{LiMO}}_{2\left(\mathrm{s}\right)}+{3\mathrm{H}}_2{\mathrm{SO}}_4+{\mathrm{H}}_2{\mathrm{O}}_2\leftrightharpoons {2\mathrm{MSO}}_{4\left(\mathrm{s}\right)}+{\mathrm{Li}}_2{\mathrm{SO}}_4+{4\mathrm{H}}_2\mathrm{O}+{\mathrm{O}}_{2\left(\mathrm{g}\right)}$$

(5)

where M represents Ni, Mn, or Co in the NMC cathode material.

Reactions (4) and (5) defined the following equation for calculating the concentration of sulfuric acid required to complete the leaching of the black mass at a solid/liquid ratio S/L (in g/L):

$$\left[{\mathrm{H}}_2{\mathrm{S}\mathrm{O}}_4\right]={\displaystyle \frac{S}{L}}\left(0.5\times {\displaystyle \frac{{\%}_{LFP}}{M_{LFP}}}+1.5\times {\displaystyle \frac{{\%}_{NMC}}{M_{NMC}}}+1.5\times {\displaystyle \frac{{\%}_{Al}}{M_{Al}}}+{\displaystyle \frac{{\%}_{Cu}}{M_{Cu}}}\right)$$

(6)

where ${\%}_x—$weight percentage of the element x in the black mass and $M_x—$molar mass of x.

Therefore, Equation (6) demonstrates that the acid concentration required for acid-deficient leaching is directly proportional to the S/L ratio if the NMC/LFP ratio remains constant. Using Equation (6), it is determined that a maximum of 0.51 mol/L sulfuric acid is needed at S/L = 50 g/L to achieve complete dissolution of Ni, Mn, Co, and Li from the NMC-LFP black mass under acid-deficient conditions. Similarly, it is expected that the dissolution yields of Ni, Co, Mn, and Li would reach 78% when leaching is performed with 0.4 mol/L sulfuric acid. This calculation was confirmed experimentally: the use of 0.4 mol/L sulfuric acid in the presence of 5% (Vol. %) H₂O₂ under acid-deficient conditions (S/L = 50 g/L) resulted in dissolution yields very close to the expected value of 78%, specifically 77.9% for Ni, 75.7% for Mn, 74.9% for Co and 79.4% for Ni.

Hydrogen peroxide is consumed during LFP and NMC leaching because of (i) oxidation of iron(II) into iron(III) in LFP in order to promote lithium dissolution and FePO₄ formation; (ii) reduction of Co(III) into Co(II) in order promote cobalt dissolution as Co(II) is more soluble than Co(III), which destabilizes the NMC structure and favors its dissolution; and (iii) chemical reactions with impurities present in the leach solution. For instance, the presence of Cu and Al impurities can contribute to the decomposition of H₂O₂ according to the following reactions [23]:

$$\mathrm{Cu}+{\mathrm{H}}_2{\mathrm{SO}}_4+{\mathrm{H}}_2{\mathrm{O}}_2\leftrightharpoons {\mathrm{CuSO}}_4+{2\mathrm{H}}_2\mathrm{O}$$

(7)

$$2\mathrm{Al}+{3\mathrm{H}}_2{\mathrm{SO}}_4\leftrightharpoons {\mathrm{A}\mathrm{l}}_2{{(\mathrm{S}\mathrm{O}}_4)}_3+{3\mathrm{H}}_{2\left(\mathrm{g}\right)}$$

(8)

In previous studies, H₂O₂ concentrations between 3 and 6% (Vol. %) H₂O₂ (corresponding to 0.9–1.8 mol/L) were used to leach NMC materials in the presence of sulfuric acid under acid-excess condition [18,24,25]. In order to determine the range of H₂O₂ concentration for the DoE, a series of experiments were conducted to leach LFP/NMC mixtures under acid-excess and acid-deficient conditions at various H₂O₂ concentrations (Figure 2). The highest dissolution yields of Ni, Mn, and Co without significant Li loss were obtained when 3–5% (Vol. %) H₂O₂ were added into 1 mol/L H₂SO₄ under acid-excess or acid-deficient leaching conditions. Therefore, DoE was conducted with H₂O₂ concentration ranging between 2 and 5% (Vol. %).

3. Results and Discussion

3.1. Leaching

3.1.1. Acid-Excess Leaching

Table 2. Experimental parameters for the DoE regarding the acid-excess leaching of a mixture of LFP/NMC/Al/Cu (48.5%/48.5%/2%/1%), and corresponding dissolution yields %D of Ni, Mn, Co, and Li.

H2SO4 (mol/L)	H2O2 % (Vol. %)	S/L (g/L)	%D(Ni)	%D(Mn)	%D(Co)	%D(Li)
2	4	42.9	100.0	97.9	97.6	89.3
1	3	50	92.6	90.1	90.7	89.2
3	3	50	94.5	93.1	93.1	88.2
1	5	50	100.0	99.8	98.9	89.9
3	5	50	100.0	100.0	99.7	89.1
2	2.6	70	80.4	81.1	82.1	88.3
0.6	4	70	92.8	92.4	93.1	89.9
2	4	70	89.9	92.0	93.1	92.4
2	4	70	92.0	95.2	95.6	92.7
2	4	70	89.7	91.9	92.7	91.9
3.4	4	70	90.5	90.7	91.1	92.3
2	5.4	70	97.7	100.0	100.0	90.4
1	3	90	74.4	73.5	74.8	88.9
3	3	90	79.1	79.7	80.3	89.1
1	5	90	96.9	100.0	100.0	93.7
3	5	90	91.7	98.1	96.8	98.9
2	4	97.1	78.2	78.6	79.2	86.0

presents the dissolution yields of Ni, Mn, Co, and Li under various experimental conditions. The dissolution yield of lithium is not significantly influenced by any of the parameters listed in Table 2 (H₂SO₄ and H₂O₂ concentrations and solid-to-liquid S/L). Its mean value is (90.6 ± 2.9)%. Billy et al. [26] explained that NMC dissolution begins with lithium deintercalation, which depends solely on proton concentration. They demonstrated that maximum lithium extraction during leaching can be achieved when the proton concentration is in excess. In the present study, proton is indeed in excess as lithium concentrations ranged from 0.40 to 0.72 mol/L and sulfuric acid concentrations ranged from 1 to 3 mol/L (S/L = 50–90 g/L). Meshram et al. [27] observed that lithium dissolution was influenced by the S/L ratio. However, dissolution yields of lithium only decreased by 4%, which is comparable to our standard deviation (2.9%). Table 2 also shows that the amount of proton did not influence metal dissolution under acid-excess conditions, similar to the findings by Nadimi and Jalalian Karazmoudeh [18].

The following equation was used to fit the dissolution yields of cobalt, nickel, and manganese as a function of S/L and H₂O₂ content given that the H₂SO₄ concentration does not influence the dissolution yields under acid-excess conditions as discussed previously:

$$\begin{array}{cc}\mathrm{\%}\mathrm{D}\left(\mathrm{M}\right)={\mathrm{\mu }}^{exp,M}& +{\mathsf{\alpha }}_{H_2{O}_{2,M}}\left(\mathrm{\%}{\mathrm{H}}_2{\mathrm{O}}_2-\mathrm{\%}{\mathrm{H}}_2{\mathrm{O}}_{2\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{n}}\right)-{\mathsf{\alpha }}_{{\displaystyle \frac{S}{L}},M}{\displaystyle \frac{\left(\mathrm{S}/\mathrm{L}-{\mathrm{S}/\mathrm{L}}_{mean}\right)}{20}}\\ & +{\mathsf{\beta }}_{H_2{O}_2.{\displaystyle \frac{S}{L}},M}\left(\mathrm{\%}{\mathrm{H}}_2{\mathrm{O}}_2-\mathrm{\%}{\mathrm{H}}_2{\mathrm{O}}_{2\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{n}}\right){\displaystyle \frac{\left(\mathrm{S}/\mathrm{L}-{\mathrm{S}/\mathrm{L}}_{mean}\right)}{20}}\end{array}$$

(9)

where µ^exp—the experimental average of the dissolution yield reported in Table 2, ${\mathsf{\alpha }}_{H_2{O}_2}$—coefficient related to the H₂O₂ concentration, ${\mathsf{\alpha }}_{S/L}$—coefficient related to the SL/ratio, ${\mathsf{\beta }}_{H_2{O}_2.S/L}$—interaction term between the H₂O₂ concentration and S/L ratio. The coefficients $\mathrm{\%}{\mathrm{H}}_2{\mathrm{O}}_{2\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{n}}$ (Vol. %) and ${\mathrm{S}/\mathrm{L}}_{mean}$ (in g/L) represent the mean values of the H₂O₂ content and S/L ratio in the DoE.

The Equation (9) does not include the quadratic effect due to the F-values lower than unity (see Supplementary Table S1 in the Supplementary Materials). The parameters of Equation (9) are gathered in

Table 3. DoE parameters of Equation (9) for the dissolution of Ni, Co, and Mn under acid-excess condition and quality indicators of the fit (F-value, p-value, R², $R_{adj}^2$ and RMSE).

	Ni	Mn	Co
${\mathrm{\mu }}^{exp,M}$	90.6	91.4	91.7
${\mathsf{\alpha }}_{H_2{O}_{2,M}}$	6.1	7.5	6.9
${\mathsf{\alpha }}_{{\displaystyle \frac{S}{L}},M}$	6.4	5.0	4.7
${\mathsf{\beta }}_{H_2{O}_2.{\displaystyle \frac{S}{L}},M}$	2.8	3.5	3.4
$\mathrm{\%}{\mathrm{H}}_2{\mathrm{O}}_{2\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{n}}$	4	4	4
${\mathrm{S}/\mathrm{L}}_{mean}$	70	70	70
F-value	68.8	49.5	44.4
p-value	<0.0001	<0.0001	<0.0001
R2	0.94	0.92	0.91
$R_{adj}^2$	0.93	0.90	0.89
RMSE	2.17	2.64	2.62

.

The independence of the dissolution yields from acid concentration is confirmed by the F-values reported in Table 3, which exceed the threshold F-values from the Fisher–Snedecor table (see more details in the Supplementary Materials, Table S2). Instead, it is solely a function of the solid-to-liquid ratio (S/L in g/L) and the H₂O₂ content (Wt. %).

The quality of the fit used to describe the dissolution yields of Ni, Co, and Mn as a function of S/L and H₂O₂ content was evaluated using the indicators reported in Table 3 derived from the least-squares method implemented in the JMP software. The mathematical expressions of these indicators, i.e., F-value, p-value, coefficient of determination (R²), adjusted coefficient of determination (R²_adj), and the deviation of the residuals (RMSE), are provided in the SI section. The F-value and p-value indicate the agreement between the model and the experimental values. R² measures the influence of the model parameters on the discrepancy between experimental data and predictions, while RMSE reflects the dispersion of the results relative to the model.

The F-values obtained for modeling the dissolution yields of nickel are higher than those for manganese and cobalt. However, all F-values are sufficiently high to indicate that the model fits the data well for the dissolution of nickel, manganese, and cobalt. Additionally, the low p-values support this conclusion. The coefficient of determination (R²) ranges from 0.91 to 0.94, indicating that 91–94% of the observed variation is explained by the model parameters, with only 6–9% remaining unexplained. However, Karazhiyan et al. [28] emphasized that a high R² value alone does not guarantee a good model. The close agreement between R² and R²_adj confirms the good quality of the model. Similarly, the RMSE values indicate that the experimental data closely align with the model for the dissolution yields of Ni, Co, and Mn.

Figure 3 shows that the dissolution yields increase with the amount of H₂O₂ as H₂O₂ facilitates the reduction of Co(III) to Co(II), which is more soluble.

Conversely, the dissolution yields decrease as the solid-to-liquid (S/L) ratio increases due to the reduced availability of lixiviant to dissolve the black mass [27].

3.1.2. Acid-Deficient Leaching

Equation (9) shows that the acid concentration is directly related to the S/L ratio under acid-deficient leaching conditions provided the LFP/NMC ratio remains constant. Therefore, leaching efficiency depends only on the S/L ratio and H₂O₂ concentration.

Table 4. Dissolution yields of Ni, Mn, Co, and Li at various H₂O₂ contents and solid-to-liquid ratios (S/L) during acid-deficient leaching of LFP/NMC/Al/Cu (48.5%/48.5%/2%/1%).

H2O2 in % (Vol. %)	S/L (g/L)	%D(Ni)	%D(Mn)	%D(Co)	%D(Li)
4	48.4	78.6	78.7	76.6	81.2
3	50	77.9	75.7	74.9	79.4
5	50	79.0	76.2	76.3	78.8
2.9	70	76.3	74.5	76.1	80.0
4	70	78.5	76.4	77.7	80.3
4	70	77.7	79.2	79.4	80.3
5.1	70	75.4	73.1	75.1	79.1
3	90	71.5	72.3	72.8	80.2
5	90	76.5	76.7	79.2	81.2
4	91.6	75.5	77.0	78.1	80.3
Average values (%)		76.7	76.0	76.6	80.1
Standard deviations (%)		2.2	2.2	2.1	0.8
RSD (%)		2.9	2.9	2.7	1.0

presents the dissolution yields of Ni, Co, Mn, and Li under acid-deficient leaching and their average values. The S/L ratio (and, then, the H₂SO₄ concentration according to Equation (6) and the H₂O₂ content do not significantly influence the dissolution yields of these metals. This is confirmed by the very low F-values reported in the Supplementary Table S3 (1.35, 1.41, 1.17, and 0.83 for Ni, Mn, Co, and Li, respectively) and the low relative standard deviation (RSD) of the dissolution yields of Ni, Mn, Co, and Li (RSD = 2.9%, 2.9%, 2.7% and 1.0% for Ni, Mn, Co, and Li, respectively).

Under acid-excess or acid-deficient conditions, H₂O₂ is essential for achieving high dissolution yields (Figure 2). However, during acid-deficient leaching, the acid becomes the limiting factor for metal dissolution, and the yield cannot exceed 70–80%, even with 5% (Vol. %) H₂O₂. Additionally, H₂O₂ concentrations between 3% and 5% (Vol. %) do not significantly affect the dissolution yields. A similar conclusion applies to the S/L ratio: while a low S/L ratio is required for high yields, no significant influence on the dissolution yields was observed at a high S/L ratio between 50 and 90 g/L.

3.1.3. Residue Reintroduction Under Acid-Deficient Leaching: A Strategy to Increase Leaching Yields

Reintroduction of the leaching residue in the leaching reactor was investigated to improve the recovery of target metals for acid-deficient leaching. It was considered acid-deficient leaching of 1000 g of black mass by H₂SO₄ (H₂SO₄ concentration = 0.4 mol/L calculated by Equation (6)) in the presence of 3% (Vol. %) H₂O₂ at S/L = 50 g/L. An iterative mathematical model based on mass balance calculations was developed to calculate the dissolution yields of metals under acid-deficient conditions in the presence of a reintroduction loop (Figure 1b).

The quantity of the metal M in the leaching residue at the iteration i + 1 ($m_{M,res,i+1}$) was calculated by mass balance as follows:

$$m_{M,res,i+1}=(1-{\displaystyle \frac{\%D\left(M\right)}{100}})\times m_{M,intro,i}$$

(10)

where $m_{M,intro,i}$—quantity of the metal M in the solid introduced in the reactor at the iteration i, and $\%D\left(M\right)$—dissolution yield of the metal M.

Under acid-deficient conditions, the dissolution yield of metal M is constant and equal to 76.7% for Ni, 76.0% for Mn, 76.6% for Mn, and 80.1% for Li (average values in Table 4), regardless of the value of S/L, sulfuric acid concentration, or H₂O₂ content.

The quantity of the metal M in the solid at the iteration i + 1 can be calculated using the following equation:

$$m_{M,intro,i+1}={\displaystyle \frac{\%R}{100}}\times m_{M,res,i+1}+m_{M,intro,0}$$

(11)

where %R—a percentage of leaching residue reintroduced in the leaching reactor and $m_{M,intro,0}$—amount of the metal M contained in the black mass initially introduced in the reactor (i = 0).

Then, it is possible to calculate the dissolution yields of the metals at iteration i + 1 in the presence of a reintroduction loop using the following equation:

$${\%D\left(M\right)}_{recirculated,i+1}={\displaystyle \frac{\%D\left(M\right)\times m_{M,intro,i+1}}{m_{M,intro,0}}}$$

(12)

It was considered that the model converged when $\left|{\%D\left(M\right)}_{recirculated,i+1}-{\%D\left(M\right)}_{recirculated,i}\right|$ < 0.0001 for all target metals.

The reintroduction of 60% (Wt. %) of the leaching residue to increase the solid-to-liquid (S/L) ratio was adopted, which increased the S/L ratio from 50 g/L to 88 g/L while maintaining the parameters within the design of experiments (DoE) range. Under acid-deficient conditions, the dissolution yields of Ni, Mn, Co, and Li increased significantly when 60% (Wt. %) recirculation was applied (

Table 5. Dissolution yields of Ni, Mn, Co, and Li under acid-deficient leaching of LFP/NMC/Al/Cu (48.5%/48.5%/2%/1%) with and without reintroduction of the leaching residue. Experimental conditions: 0.4 mol/L H₂SO₄ + 3% (Vol. %) H₂O₂; temperature = 30 °C; time = 240 min. * The recirculation is responsible for an increase in the S/L ratio from 50 g/L to 66 g/L without the reintroduction loop, and 88 g/L after the reintroduction loop (these values of S/L still remain within the range of validity of the DoE model).

	%D(Ni)	%D(Mn)	%D(Co)	%D(Li)
Experimental test for acid-deficient leaching without recirculation (S/L = 50 g/L)	76.7	76.0	76.6	80.1
Experimental test for acid-deficient with 60%(Wt. %) reintroduction(S/L = 66 g/L) *	87.1	87.1	87.4	89.4
Acid-deficient leaching with Calculation with 60%(Wt. %) recirculation after 1 reintroduction loop (S/L = 66 g/L) *	87.4	86.9	87.4	89.7
Calculation with acid-deficient leaching with 60%(Wt. %) recirculation after 7 reintroduction loops (S/L = 88 g/L) *	89.2	88.8	89.1	91.0

).

To validate these calculations, a mixture of 16 g of leaching residue corresponding to a reintroduction percentage of 60% (Wt. %) of the leaching residue and 50 g of fresh LFP/NMC/Al/Cu black mass (48.5%/48.5%/2%/1%) was leached under acid-deficient conditions (H₂SO₄ concentration = 0.45 mol/L, H₂O₂ content = 3% (Vol. %), S/L = 50 g/L, 30 °C, leaching duration = 240 min). The XRD pattern of the resulting leaching residue showed that it primarily consisted of unleached NMC and FePO₄ (Supplementary Figure S2). Since only 76.7%, 76.0%, and 76.6% of Ni, Mn, and Co, respectively, were leached, it was inferred that approximately 25% (Wt. %) of NMC remained unleached in the residue. Therefore, the mixture of the black mass and the 60% (Wt. %) of leaching residue contained 15% of unleached NMC in addition to FePO₄.

Using the same method to calculate the acid requirement under deficiency, it was determined that 0.49 mol/L of H₂SO₄ was needed. The experimental dissolution yields were then compared with those calculated using i = 0 (no recirculation loop). Table 5 shows good agreement between the experimental yields and the values predicted by the mathematical model for Ni, Mn, Co, and Li. The difference in dissolution yields for Ni, Mn, Co, and Li was 0.3%, 0.2%, 0.0%, and 0.3%, respectively. In conclusion, the implementation of partial recirculation in the acid-deficient leaching of the LFP-NMC black mass increases the dissolution yields of Ni, Mn, Co, and Li by 12.5%, 12.8%, 12.5%, and 10.9%, respectively. The only additional costs are the implementation of a solid residue recirculation system and an increase in the H₂SO₄ concentration from 0.40 mol/L to 0.49 mol/L. Details about the mass balance calculations are available in Supplementary Table S4.

3.1.4. Comparison of Acid-Excess Leaching and Acid-Deficient Leaching with Leaching Residue Reintroduction

Acid-excess leaching produces a PLS with high recovery rates for target metals, achieving nearly 100% dissolution yields for Ni, Mn, and Co. However, this method also dissolves impurities, such as 17% of Cu, 100% of Al, and, notably, 100% of Fe from LFP materials. These impurities must be removed at some stage of the process. While impurities can be removed in a single step by increasing the pH, this approach causes co-precipitation [23]. Co-precipitation is exacerbated by the presence of LFP as transition metals precipitate with phosphate ions. Zou et al. reported losses of 12.7% Ni, 26.4% Mn, 17.3% Co, and 9.2% Li during single-step impurity removal [13]. To minimize losses, gentler processes are preferred. For instance, Peng et al. adjusted the pH to 3.5 to remove Fe and Al, followed by a two-step operation for Cu removal: electrodeposition (recovering 90% of Cu) and solvent extraction, achieving a metal loss rate of only 0.4% [24]. Similarly, Virolainen et al. used ion exchange for impurity removal; however, this also removed Mn, requiring an additional separation step [25]. Wang and Friedrich used cementation with Fe powder to remove Cu, followed by pH adjustment to 3.5 to remove Fe and Al, resulting in a metal loss rate of 3–5% [26]. Thus, while acid-excess leaching achieves high dissolution yields, impurity management increases operational costs due to additional steps, higher reagent consumption, and waste generation.

Conversely, Fe and Al are entirely absent from the PLS produced under acid-deficient leaching, while Cu is present (

Table 6. Dissolution yields of Ni, Mn, Co, Li, Fe, Al, and Cu after acid-excess and acid-deficient leaching without reintroduction of the leaching residue and after acid-defect leaching with 60% (Wt. %) reintroduction of the leaching residue (calculated using the mass balance model based on Equations (10)–(12)). Experimental conditions: black mass = LFP/NMC/Al/Cu (48.5%/48.5%/2%/1%), leaching reagent = 1 mol L⁻¹ H₂SO₄ for acid-excess leaching and 0.4 mol L⁻¹ H₂SO₄ for acid-deficient leaching, 3% (Vol. %) H₂O₂, S/L = 50 g/L, temperature = 30 °C, leaching duration = 240 min.

	%D(Ni)	%D(Mn)	%D(Co)	%D(Li)	%D(Fe)	%D(Al)	%D(Cu)
Experimental test of acid-excess leaching without reintroduction	100	100	99.9	88.0	100	17	100
Experimental test of acid-deficient leaching without reintroduction	76.7	76.0	76.6	80.1	0	0	5.8
Calculation under acid-deficient leaching with reintroduction	89.2	88.8	89.1	91.0	0	0	13.3

). Moreover, reintroducing the leaching residue into the reactor increases the dissolution yield of Cu, from 5.8% to 13.3%, as Cu is partially reintroduced into the solution. Thus, acid-deficient leaching with partial residue reintroduction significantly enhances the recovery of target metals, offering fewer drawbacks compared to acid-excess leaching.

To comprehensively compare the processes, it is essential to evaluate not only the dissolution yields of the elements but also the reagent consumption and the number of operational steps as these factors significantly influence both the economic (OPEX and CAPEX) and environmental impacts.

Table 7. Dissolution yields of Ni, Mn, Co, and Li along with the number of operations and reagent consumption depending on the leaching process: (1): acid-excess leaching and impurities removal; (2): acid-deficient leaching with 60% (Wt. %) recirculation. Black mass = LFP/NMC/Al/Cu (48.5%/48.5%/2%/1%), S/L = 50 g/L).

	%D(Ni)	%D(Mn)	%D(Co)	%D(Li)	Number of Unit Operations	mol H2SO4 (per Liter of PLS)	mol H2O2 (per Liter of PLS)	mol NaOH (per Liter of PLS)
(1)	95.5	95.5	95.4	84	3	1	2.1	0.6
(2)	89.2	87	89	90.7	1	0.4	1.3	0

highlights the dissolution yields of the target metals, the number of operations, and reagent consumption. The acid-excess process achieves higher dissolution yields for Ni, Mn, and Co, with respective increases of 6.3%, 6.7%, and 6.3% compared to the acid-deficient process. However, lithium recovery is 7% higher in the acid-deficient process due to residue recirculation, a difference that could also be achieved by implementing recirculation in the acid-excess process.

The main advantage of the acid-deficient process is its lower CAPEX as it involves only one step compared to three for the acid-excess process (leaching and the removal of copper, iron, and aluminum). Additionally, the acid-deficient process has lower OPEX, requiring less H₂SO₄ and H₂O₂ and eliminating the need for NaOH. Last but not least, the composition of the PLS after acid-deficient leaching with a reintroduction loop facilitates its treatment by solvent extraction with classical extractants.

3.2. Solvent Extraction

After leaching, the solution contains both the target metals and various impurities. Several methods are employed to concentrate and purify the solution such as precipitation, cementation, solvent extraction, ion exchange, etc. Each of these methods is selected based on the specific metal being processed, the composition of the leach solution, and economic considerations. The goal is to efficiently and selectively recover valuable metals while minimizing impurities and environmental impact. In the present case, the most appropriate method is solvent extraction as precipitation cannot separate cobalt and nickel and demands a high amount of reagent to neutralize the acidity of the leach solution; cementation is not relevant and ion exchange is not relevant as the metal concentration is too high in the leach solution. Therefore, we decided to work on the production of a leaching solution in which the quality (presence of impurities and metal concentrations) is adapted for solvent extraction, which is a mature technology in hydrometallurgy.

The PLS produced under acid-deficient leaching with a recirculation loop contains only 0.4% Cu as an impurity, whereas the PLS obtained by acid-excess leaching contains 0.7% Al, 1.9% Cu, and 32% Fe, which must be removed before performing solvent extraction (Table S5). Therefore, The PLS produced by acid-deficient leaching of LFP/NMC/Al/Cu (48.5%/48.5%/2%/1%) with the reintroduction of the leaching residue in the reactor is of a better quality for solvent extraction, which is widely used to recover target metals (extraction), to remove metals (purification), and to separate metals from PLS [29,30,31].

The PLS prepared by leaching LFP/NMC/Al/Cu (48.5%/48.5%/2%/1%) under acid-deficient conditions was used to investigate the liquid–liquid extraction and the separation of cobalt, manganese, and nickel. DEHPA and Cyanex^®272 are two extractants usually used to recover cobalt, nickel, and manganese from acidic sulfate solution [32,33,34,35].

Figure 4 shows that DEHPA is more selective than Cyanex^®272 to extract Mn selectively toward Co, Ni, and Li since 64% of Mn and only 1.0% of Co, 0.8% of Ni, and no Li is extracted by 0.48 mol/L DEHPA in kerosene at pH 1.6 (Figure 4a), whereas no separation between Mn and Co can be expected with Cyanex^®272 (Figure 4b).

After Mn extraction by DEHPA at pH 1.6 (O/A = 1) and stripping of the extraction solvent with 0.1 mol/L sulfuric acid at O/A = 1, the stripping solution contained 7 mg L⁻¹ Ni, 374 mg L⁻¹ Mn, 4 mg L⁻¹ Co, and no lithium (the initial concentrations in the PLS were 9.2 g L⁻¹ Ni, 0.735 g L⁻¹ Mn, 1.29 g L⁻¹ Co and 2.35 g L⁻¹ Li). Therefore, it was necessary to add a scrubbing step before stripping. The extraction solvent was then scrubbed with an aqueous solution at pH 5 containing 3 g L⁻¹ Mn at O/A = 1. After stripping the scrubbed extraction solvent with 0.1 mol/L sulfuric acid, the resulting stripping solution contained 2.16 g L⁻¹ Mn, with only 0.2 g L⁻¹ of Cu left and no other impurities.

As mentioned previously, Cyanex^®272 is a relevant extractant to separate Co and Ni. After Mn removal, Figure 4b shows 0.9 mol/L Cyanex^®272 in kerosene can extract 49% of Co at pH 3.9, but only 2.2% of Ni is co-extracted. Ni co-extraction could be reduced by performing solvent extraction in a counter-current with several mixer-settlers or by implementing a scrubbing stage with diluted sulfuric acid solution.

Since these extractants are selective for Mn and/or Co, this study suggests two alternative solvent extraction processes to separate Mn and Co from the PLS.

The first process (Figure 5a) starts with the selective extraction of Mn using 0.48 mol/L DEHPA in kerosene at pH 1.6, adjusted with H₂SO₄. The McCabe–Thiele diagram (Figure 6a) shows that this step can successfully remove 99.41% of Mn when using 4 mixer-settlers at O/A = 2, leaving only 0.005 g L⁻¹ of Mn in the initial solution. The loaded organic phase is then stripped with 0.1 mol/L sulfuric acid at O/A = 1.

After Mn removal, Co is extracted from the Mn-free PLS by 0.9 mol/L Cyanex^®272 in kerosene at pH 3.9, adjusted with sodium hydroxide. The McCabe–Thiele diagram (Figure 6b) indicates that this step can successfully remove 99.99% of Co with 4 mixer-settlers at O/A = 2, leaving less than 0.001 g L⁻¹ of Mn in the raffinate. The loaded organic phase is then stripped with 0.1 mol/L sulfuric acid at O/A = 1.

The second flowsheet (Figure 5b) follows a different sequence. It begins with the co-extraction of Mn and Co using 0.9 mol/L Cyanex^®272 in kerosene at a pH of 3.9, adjusted with sulfuric acid. The McCabe–Thiele diagrams (Figure 6b,c) show that this step achieves Mn and Co removal efficiencies of 99.88% and 99.99%, respectively, with four mixer-settlers at O/A = 2. The loaded organic phase is stripped with 0.1 mol/L sulfuric acid at O/A = 1, producing a stripping solution in which the pH is optimal for Mn extraction with DEHPA. The next step involves the selective extraction of Mn from the stripped solution using 0.48 mol/L DEHPA in kerosene at a pH of 1.6. As before, this process can extract 99.41% of Mn using 4 mixer-settlers at O/A = 2. This flowsheet is designed to minimize the consumption of sulfuric acid and sodium hydroxide for pH adjustment as extraction is performed at a pH close to that of the solution.

4. Conclusions

This study presents a novel approach to the hydrometallurgical recycling of mixed LFP/NMC cathode materials from lithium-ion batteries. Departing from traditional acid-excess leaching methods, this research explores acid-deficient leaching combined with residue recirculation. This innovative strategy significantly improves metal dissolution yields while reducing reagent consumption and operational complexity. Notably, the rarely studied residue recirculation under acid-deficient conditions achieves a 12.5% increase in critical metal (Ni, Mn, Co, and Li) dissolution yields, enhancing selectivity and process efficiency without additional purification steps.

Using a design of experiments (DoE) methodology, the study demonstrates that acid-deficient leaching matches the effectiveness of conventional methods while drastically reducing sulfuric acid and hydrogen peroxide consumption. This reduction lowers operational costs (OPEX) and minimizes residue production, a key factor for sustainable battery recycling.

Additionally, the selective leaching of target metals under these conditions generates a cleaner pregnant leach solution (PLS), facilitating the subsequent separation of manganese and cobalt via solvent extraction, as demonstrated in this work.