Increasing olivine cathode gravimetric energy density by increasing exchangeable lithium ion content or average discharge voltage

Abstract

The present technology discloses lithium metal polyanion (LMX) compounds. A battery cell including the LMX compound as a cathode may have a weight capacity of more than 170 mAh / g. The present technology utilizes machine learning to provide synthesis conditions and stoichiometry of the LMX compound to maximize the gravimetric energy density of the battery cell.

Description

[0001] Cross-references to related patent applications

[0002] This patent application claims the benefit of U.S. patent application serial number 63 / 357,439, filed on June 30, 2022, entitled “INCREASING GRAVIMETRIC ENERGY DENSITY FOR LITHIUM-METAL POLYANION BY INCREASING EXCHANGEABLE LI-ION CONTENT AND AVERAGE DISCHARGE VOLTAGE,” under 35 U.S.C. §119(e), which is incorporated herein by reference in its entirety. Technical Field

[0003] The technology involves producing higher gravimetric energy density (GED) than lithium iron phosphate (LFP) batteries, and more specifically, involves increasing the exchangeable lithium ion content and / or average discharge voltage through a combination of experiments and machine learning models. Background Art

[0004] Lithium-ion batteries are widely used as the most promising portable energy source in electronic devices due to their high operating voltage, high energy density and good cycle performance. Lithium-ion batteries are used in electric vehicles and hybrid vehicles. Among these lithium-ion batteries, olivine-type cathode materials such as LiMPO4 (M = Fe and Mn) have attracted much attention due to their low cost and high intrinsic safety. However, they exhibit poor electrochemical properties, which is mainly due to their low conductivity.

[0005] Lithium metal phosphates (LMP) with M=Mn have been extensively studied and described in the literature. LMP has been used as a cathode material for Li-ion batteries. This LMP cathode material has a nominal discharge voltage of 4.1 V and therefore has a gravimetric energy density (GED) that is nominally 20% higher than that of lithium iron phosphate (LFP). However, because the orientation of the two-phase interface hinders the path for Li ion (Li+) diffusion, LMP has poor kinetics and lithium (Li) utilization. Composites of LMP and LFP that increase the average nominal voltage have been described, but these composites have not yet been widely used.

[0006] Conventional LFP has limited gravimetric energy density (GED) due to its relatively low discharge voltage (nominal 3.45 V) and modest capacity (eg, theoretical gravimetric capacity of 170 mAh / g and actual capacity ranging from 140 to 165 mAh / g).

[0007] There has been a lot of research on LiMPO4 or LiMSiO4 compounds, where PO4 or SiO4 is the only anion, where M represents a transition metal or two or more transition metals. There is very limited research on mixtures of PO4 anions and SiO4 anions. For example, U.S. Patent No. 5,910,382A discloses a Li-ion cathode including a mixture of PO4 anions and SiO4 anions. U.S. Patent No. 6,136,472 also discloses a mixed Li-ion SiO4 and PO4 composition in a sodium superionic conductor (NASICON) structure, and includes a composition of Li a M ’ (2-b) M ” b Si c P 3-c O 12 Mixtures of PO4 and SiO4 have also been used as solid electrolytes for Na-ion batteries. The most famous example is the crystal of the NASICON structure, with the chemical formula Na3Zr2(PO4)(SiO4)2.

[0008] Multimodal distribution is a common technique to achieve higher green density in ceramics. A common application is 3D printing. In binder jetting, the ability to achieve high green density is limited by the thickness of the powder layer, which can be overcome by mixing powders that include different particle sizes or different particle size distributions.

[0009] Batteries are an important part of many devices, from power tools to home power systems to electric and hybrid vehicles, including many other applications. Lithium iron phosphate (LFP) has been developed for use in power applications such as power tools, starting batteries, and hybrid electric vehicles. Due to the low density of LFP, its use in battery electric vehicles (BEV) is limited. Batteries are a key technological pillar on which many other technologies are built. Given the wide range of applications that use batteries, the design requirements for developing battery cathode materials suitable for their applications are equally wide-ranging. Unfortunately, developing new batteries can be a time-consuming process and expensive.

[0010] Machine learning has shown promising results in a variety of applications. In the field of materials science, machine learning is used to develop new materials, optimize existing materials, and predict the properties of materials. One area of ​​focus in the field of materials science is the synthesis of cathode materials for lithium-ion batteries. Lithium-ion batteries are used in many applications, including portable electronics, electric transportation, and energy storage systems. The performance of these batteries depends in part on the cathode material used.

[0011] Lithium iron phosphate (LFP), nickel cobalt aluminum oxide (NCA), and nickel cobalt manganese oxide (NMC) are commonly used cathode materials for lithium-ion batteries. The synthesis of these cathode materials is a complex process involving various precursors and synthesis processing conditions. Changing the precursors and synthesis processing conditions allows the properties of the cathode materials to be optimized. However, when optimizing the properties of the cathode materials, selecting the precursors and the ratio of the precursors and controlling the synthesis processing conditions are challenging, expensive, and time-consuming.

[0012] Carbon coating is a common technique to improve the conductivity of cathode active materials in lithium-ion batteries. Carbon coating can improve the conductivity of cathode active materials without changing other inherent properties. Uniform coating of carbon on LFP helps to avoid charge accumulation and undesirable chemical reactions. Carbon coating on cathode active materials or compounds (such as LFP, LMP or lithium metal polyanion (LMX) compounds) can affect the cycling performance of battery cells containing carbon-coated cathode powders.

[0013] It is desirable to obtain cathode materials with improved properties at reduced cost. However, the development cycle of cathode materials with improved properties is very long. Therefore, there is still a need to develop methods to accelerate cathode material synthesis and battery cell production. Summary of the invention

[0014] This technology uses machine learning to provide the synthesis conditions and stoichiometry of lithium metal polyanion (LMX) compounds represented by formula (I) and formula (II) to increase the weight energy density of battery cells.

[0015] In one aspect, a powder comprising a lithium metal polyanion (LMX) compound is represented by formula (I):

[0016] Li(Li x TM y TM (1-x-y) )(P,A)O4 formula (I)

[0017] Wherein 0.1≤x, 0≤y<1 and Li / (TM+TM')>1, wherein TM is at least one element selected from Mn, Mg, Zn, Ca, Ni, Co, V, Al, Ti, Zr, Mo, Cr or other transition metals. TM' is a combination of Fe and Mn transition metals.

[0018] In some variations, at least one process variable or at least one stoichiometric variable for producing the compound represented by Formula (I) may be provided by a machine learning algorithm.

[0019] In some variations, TM is Mo and the compound consists of Li[Li] 0.2 Fe 0.2 Mn 0.5 Ti 0.1 PO4 means.

[0020] In some variants, TM is V, and the compound is represented by Li[Li] 0.1 Fe 0.8 V 0.1 PO4.

[0021] In some variants, the compound is represented by Li[Li] 0.1 Mn 0.6 Mg 0.2 V 0.1 PO4.

[0022] On the other hand, a method is provided for designing LFP compounds. The method includes using a machine learning (ML) algorithm-assisted design combined with experimental methods to optimize the composition of the LFP compound to achieve a weight capacity of more than 170 mAh / g.

[0023] In some variants, the method may further include synthesizing the compound to form a powder. The method may further include evaluating the electrochemical performance of the powder and the battery cell. The method may further include using the electrochemical performance and powder information to train a machine learning (ML) model. The method may further include using the energy density of the battery cell as an output to fit a Gaussian process model (constrained by the powder level metrics falling within a set of specifications). The method may further include using an acquisition function to determine N variants that may increase the energy density to be evaluated in the next iteration. The method may further include synthesizing the N variants. The method may further include evaluating the electrochemical performance of the powder and the battery cell, repeating the experiment, and training the ML model until the difference in consecutive iterations is below a threshold.

[0024] In some variants, the cathode active material may include LMX powder.

[0025] In some variants, the cathode may include the cathode active material.

[0026] In some variants, the battery cell may include a cathode, a separator, and an anode, wherein the battery cell has a weight capacity of more than 170 mAh / g.

[0027] In a further aspect, a powder comprising a lithium manganese phosphate compound represented by formula (II):

[0028] Li[Fe 1-x-y Mn x TM y (P,A)O4 formula (II)

[0029] where 0.15 < x < 0.45, 0.20 < y < 0.45, and TM is at least one element selected from Mn, Mg, Zn, Ca, Ni, Co, V, Al, Ti, Zr, Mo, and Cr.

[0030] In some variations, the compound comprises Li[Fe 0.4 Mn 0.3 Mg 0.3 ]PO4.

[0031] In some variations, based on X-ray diffraction (XRD) analysis, the compound Li[Fe 1-x-y Mn x Mg y ]PO4 has the same structure as LiFePO4.

[0032] In some variations, A in formula (I) and formula (II) represents one of V, Si or W.

[0033] BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The various advantages and features of the present technology will become apparent by referring to the specific embodiments shown in the accompanying drawings. Those skilled in the art will understand that these drawings only show some examples of the present technology and do not limit the scope of the present technology to these examples. In addition, the skilled person will understand the principles of the present technology by using the accompanying drawings, in which the following are described and explained with additional specificity and detail:

[0035] Figure 1 shows a top view of a battery cell according to some aspects of the disclosed technology;

[0036] Figure 2 shows a side view of a set of layers of a battery cell according to some aspects of the disclosed technology;

[0037] Figure 3 is a workflow showing steps for cathode synthesis and characterization at the powder and cell levels according to some aspects of the disclosed technology;

[0038] Figure 4 LiFePO4 and Li[Fe 0.4 Mn 0.3 Mg 0.3 ]X-ray diffraction (XRD) patterns of PO4 compounds;

[0039] Figure 5 The Li[Fe 0.4 Mn 0.3 Mg 0.3 ]Charge and discharge data of PO4 compounds;

[0040] Figure 6 According to some aspects of the disclosed technology, Li[Fe 0.4 Mn 0.3 Mg0.3 ]PO4、LiFe 0.6 Mn 0.4 Comparison of voltage versus capacity during charge or discharge of PO4, LiMnPO4, and LiFePO4 compounds;

[0041] Figure 7 illustrates an example of a deep learning neural network that may be used to implement a perception module and / or one or more verification modules in accordance with some aspects of the disclosed technology; and

[0042] Figure 8 An example processor-based system is shown that may implement some aspects of the disclosed technology.

[0043] Detailed Description

[0044] The detailed description described below is intended to describe various configurations of the subject technology and is not intended to represent the only configuration in which the subject technology can be practiced. The accompanying drawings are incorporated herein and constitute a part of the detailed description. The detailed description includes specific details to provide a more thorough understanding of the subject technology. However, it will be clear and obvious that the subject technology is not limited to the specific details described herein and can be practiced without these details. In some cases, structures and components are shown in block diagram form to avoid blurring the concepts of the subject technology.

[0045] The disclosures of these patents, patent applications, and publications are incorporated herein by reference in their entirety to more fully describe the prior art known to those skilled in the art as of the date of the invention described and claimed herein. If there is any inconsistency between patents, patent applications, and publications and the present disclosure, the present disclosure shall prevail.

[0046] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present disclosure belongs. Unless otherwise indicated, the initial definitions provided herein for a group or term apply to the group or term throughout this specification, either alone or as part of another group.

[0047] In order not to obscure the basic principles of the present invention, well-known features of lithium-ion battery technology known to those skilled in the art have been omitted or simplified. Portions of the following description will be presented using terms commonly used by those skilled in the art.

[0048] i. Definition

[0049] The "capacity" of a battery or battery cell is a measure of the charge that the battery can store and is determined by the mass of active materials contained in the battery. Capacity represents the maximum amount of charge that can be extracted from a battery under certain conditions. A battery has a discharge current in amperes that it can deliver over time. The capacity of a battery is measured in ampere-hours (Ah).

[0050] "Gravimetric capacity" is the capacity per unit mass (mAh / g). Gravimetric capacity is also called specific capacity.

[0051] The “gravimetric energy density” or specific energy of a battery or battery cell is a measure of the energy contained in the battery relative to its weight, usually expressed in Watt-hours per kilogram (W-hr / kg).

[0052] The “volumetric energy density” of a battery or battery cell is a measure of the energy contained in the battery relative to its volume, usually expressed in Watt-hours per litre (W-hr / litre).

[0053] "Tapped density" is a material property of a powder. The tapped density of a powder is determined after a prescribed number of tapping steps on the powder bed. More specifically, the tapped density takes into account the pores and voids between the particles and is not based on a loose powder bed, but rather the bed after a prescribed number of tapping steps. The tapped density of a powder is a measure of the ratio of the mass of powder to the volume occupied by the powder after a prescribed number of tapping steps on the powder bed. The tapped density is different from the bulk density of a powder, which would be determined if the powder were loosely introduced into a graduated cylinder. The bulk density takes into account the pores and voids of the loose powder bed.

[0054] A redox reaction is a chemical reaction involving the transfer of electrons between two substances. A redox reaction is any chemical reaction in which the oxidation number of a molecule, atom, or ion changes by gaining or losing electrons.

[0055] ii. Overview

[0056] The disclosed technology addresses the need in the art to increase the gravimetric energy density (GED) of lithium iron phosphate (LFP) batteries. The GED of LFP is limited by its relatively low discharge voltage (e.g., about 3.4V) and moderate capacity (theoretical 170mAh / g; actual 140 to 165mAh / g). In order to compete with ternary cathode materials, it is useful to increase the GED. The GED of LFP is increased by changing the stoichiometry to add additional exchangeable Li, and by appropriately charge compensating Li by replacing iron (Fe) with metals that allow higher oxidation states and / or partially replacing phosphorus (P) in anionic PO4 (reducing the oxidation state of phosphorus and redoxing phosphorus).

[0057] Machine learning (ML)-assisted design combined with experimental methods (called active learning) can be used to optimize lithium (Li) content, properties, Fe site doping amount, and synthesis conditions. The lithium metal polyanion (LMX) compounds obtained by ML-assisted design can have higher GED than conventional LFP compounds by increasing their capacity and / or average discharge voltage.

[0058] The present technology includes an improved Fe-rich cathode that has a better price / performance ratio than conventional LFP and Ni-rich nickel-manganese-cobalt (NMC) cathodes, as shown in the performance (energy density Wh / liter) versus cost ($ / kWh) curve. The disclosed Fe-rich cathode design can achieve peak performance comparable to NMC622 and offers multiple advantages, including improved tap density, up to 20% improvement in energy density through a combination of cation and polyanion chemistry changes, and utilization of raw materials that are readily available through the U.S. supply chain.

[0059] More specifically, the present technology replaces Fe and phosphate polyanions to improve gravimetric energy density without compromising other metrics. LiFePO4 allows extraction of one Li per formula unit. Orthosilicate (i.e. silicate anion SiO4 4- and any salts and esters thereof) is another class of cathodes. For example, Li2MSiO4 is an exemplary orthosilicate, where M represents one or more transition metals. These orthosilicates often have lower redox voltages but allow up to two Li to be extracted per formula unit because M changes its oxidation state from 2 + becomes 3 + becomes 4 + , which effectively doubles the material's capacity. However, silicate systems have poor cycle life because the crystal structure undergoes various phase changes as lithium is inserted and removed from the system.

[0060] The present technology also involves the simultaneous substitution of Fe with M and PO4 with SiO4, which exploits the higher theoretical energy density of silicates and overcomes the limitation of short cycle life through the stabilization effect of phosphate polyanions.

[0061] iii.Battery cells

[0062] Figure 1 A top view of a battery cell 100 is shown in accordance with some aspects of the disclosed technology. The battery cell 100 may correspond to a lithium-ion battery cell used to power devices used in consumer, medical, aerospace, defense, and / or transportation applications.

[0063] The battery cell 100 includes a stack 102, which includes multiple layers, including a cathode having a cathode active material, a separator, and an anode having an anode active material. More specifically, the stack 102 may include a strip of cathode active material (e.g., aluminum foil coated with a lithium compound) and a strip of anode active material (e.g., copper foil coated with carbon). The stack 102 also includes a strip of separator material (e.g., a conductive polymer electrolyte) located between the strips of cathode active material and the strips of anode active material. The cathode, anode, and separator layers may remain flat, forming a planar configuration.

[0064] The housing may include, but is not limited to, a bag, such as a flexible bag, a rigid container, etc. Back Figure 1During assembly of the battery cell 100, the stack 102 is enclosed in a housing. The stack 102 may be in a planar or wound configuration, although other configurations are possible.

[0065] The stack 102 may also include a set of conductive sheets 106 connected to the cathode and anode. The conductive sheets 106 may extend through a seal in the housing (e.g., formed using sealing tape 104) to provide a terminal for the battery cell 100. The conductive sheets 106 may then be used to electrically connect the battery cell 100 to one or more other battery cells to form a battery pack. The battery cell may be used in a battery electric vehicle. In some variations, the battery cell 100 may be a button cell.

[0066] Batteries can be combined into battery packs in any configuration. For example, a battery pack can be formed by connecting battery cells in series, parallel, or series-parallel configurations. The cells so connected can be enclosed in a hard shell to complete the battery pack, or can be embedded in the housing of a portable electronic device such as a laptop computer, tablet computer, mobile phone, personal digital assistant (PDA), digital camera, and / or portable media player.

[0067] Figure 2 A side view of a set of layers of a battery cell according to some aspects of the disclosed technology is presented. The set of layers may include a cathode current collector 202, a cathode active material 204, a separator 206, an anode active material 208, and an anode current collector 210. The cathode current collector 202 and the cathode active material 204 may form the cathode of the battery cell, and the anode current collector 210 and the anode active material 208 may form the anode of the battery cell. To create a battery cell, the set of layers may be stacked in a planar configuration, or stacked and then wound into a wound configuration.

[0068] As described above, cathode current collector 202 may be aluminum foil, cathode active material 204 may be a lithium compound, anode current collector 210 may be copper foil, anode active material 208 may be carbon, and separator 206 may include a conductive polymer electrolyte.

[0069] iv. Lithium Metal Polyanion (LMX) Compounds

[0070] This technique helps determine the stoichiometry of LMX compounds that increase gravimetric energy density over conventional LFP. Increasing lithium content increases capacity, improves stability, and increases gravimetric energy density in various aspects.

[0071] In some aspects, the stoichiometry of the LMX compound includes compounds with Li / (Fe+TM)>1 (i.e., the ratio of Li to the sum of Fe and TM is higher than 1), and the charge is compensated for the additional Li by appropriate selection of transition metals (TM) and / or partial substitution of P in the anion PO4 (lowering its oxidation state and making it redox active), where TM represents one or more substitutional cation dopants. This is different from the compounds in U.S. Pat. No. 9,178,215, which describes compounds with excess Li / (Fe+Mn+D) (which is a maximum of 1.05), and 0.35≤Mn≤0.60, and the dopant D is 0.001≤D≤0.1. In other words, U.S. Pat. No. 9,178,215 describes a compound in which the ratio of Li to the sum of Fe, Mn and D is at most 1.05, while the amount of Mn is in the range of 35-60% metal and the amount of dopant is in the range of 0.1%-1.0%.

[0072] Multiple redox reactions of Mn (e.g. Mn 2+ / 3+ and Mn 3+ / 4+ ) can lead to higher GED. More specifically, obtaining a wider range of Mn oxidation states can provide additional electrons, thereby leading to higher GED. The present technology shows that doping with about 0.3 mol of Mg can help Mn in olivine-type cathode materials. 3+ / Mn 4+ Redox Utilization. The compound does not need to have excess lithium.

[0073] In some aspects, by placing Li + Add to Fe 2+ sites to increase the exchangeable Li + This is done by replacing some of the Fe with transition metals in higher oxidation states (which can also have multiple higher oxidation states). 2+ to compensate the charge.

[0074] Li + Add to the Fe site and by reducing the average oxidation state of the anion, PO4 3- To (P,A)O4 (-3+x) To compensate the charge, a compound represented by the following formula (A) is generated:

[0075] Li(Li x TM' (1-x) )(P,A)O4 formula (A)

[0076] In addition to Li, x amount of more Li is added to the transition metal site + , so that Li, Li x and TM' 1-xThe total charge of is balanced by the anionic charge, where TM' represents the combination of Fe and Mn. The anion (P,A)O4 is redox active because the presence of A allows the exchange of additional Li + , where A can be silicon (Si), vanadium (V) and tungsten (W) etc.

[0077] In some aspects, the compounds can be summarized as including substitution of Fe with other transition metals according to formula (B):

[0078] Li(Li x TM y TM' (1-x-y) )(P,A)O4 formula (B)

[0079] Among them, Li x TM y TM' (1-x-y) Fe in LiFePO4 2+ The element of the site, y represents the fraction of TM, x represents the fraction of excess Li, 0.1≤x, 0≤y<1, and Li / (TM'+TM)>1, x, y, z represent atomic percentages or molar fractions. Wherein TM is one or more elements selected from Mn, Mg, Zn, Co, V, Al, Ti, Zr, Mo, Cr or other transition metals, and TM' represents a combination of Fe and Mn. Transition metals balance the charge during delithiation by oxidizing to a more positive oxidation state.

[0080] In one embodiment, x = 0.2, y = 0.1, M = Ti, and formula (B) becomes Li[Li] 0.2 Fe 0.2 Mn 0.5 Ti 0.1 PO4, where [Li] represents the Fe of LiFePO4 2+ Li in the site + Here, the titanium (Ti) is at +4 due to the charge neutrality of the original material.

[0081] In another embodiment, x=0.1, y=0.1, M=Mo, and formula (B) becomes Li[Li] 0.1 Fe 0.8 Mo 0.1 PO4, where [Li] represents Fe 2+ Li in the site + Here, Mo is in the +3 oxidation state and swings to +6 to utilize all the Li + . Because of Mo 3+ to the end 6+ With higher voltages, the average voltage also increases.

[0082] In another embodiment, x = 0.1, y = 0.1, M = V, and formula (B) becomes Li[Li] 0.1 Fe 0.8 V 0.1 PO4, where [Li] represents Li at the Fe 2+ site, vanadium (V) is in the +3 oxidation state and can swing to +5 to utilize all Li + . Due to the higher voltage of V + to V 3+ , the average voltage of the compound also increases. 5+

[0083] In some aspects, the compound composition is represented by formula (C) as follows:

[0084] Li[Li 1-x-y-z Fe x Mn y TM z (P,A)O4 Formula (C)

[0085] where x represents the fraction of Fe, y represents the fraction of Mn, z represents the fraction of TM to Li, 0 ≤ x ≤ 1, 0 ≤ y ≤ 0.8, 0 < z ≤ 0.6, and the transition metal (TM) includes one or more of Mg, Zn, Co, V, Al, Ti, Zr, Mo, or Cr. In one embodiment, x = 0, y = 0.8, z = 0.2, and formula (C) becomes Li[Mn 0.8 Mg 0.2 PO4. 0.8 of Mn can balance the charge by oxidizing Mn 2+ to Mn 3+ during the delithiation process. During the delithiation process, Mn 3+ is further (partially) oxidized to Mn 4+ to utilize the remaining 0.2 Li.

[0086] In some variants, 0 ≤ x ≤ 1.0. In some variants, 0 ≤ x ≤ 0.9. In some variants, 0 ≤ x ≤ 0.8. In some variants, 0 ≤ x ≤ 0.7. In some variants, 0 ≤ x ≤ 0.6. In some variants, 0 ≤ x ≤ 0.5. In some variants, 0 ≤ x ≤ 0.4. In some variants, 0 ≤ x ≤ 0.3. In some variants, 0 ≤ x ≤ 0.2. In some variants, 0 ≤ x ≤ 0.1.

[0087] In some variants, 0 ≤ y ≤ 0.8. In some variants, 0 ≤ y ≤ 0.7. In some variants, 0 ≤ y ≤ 0.6. In some variants, 0 ≤ y ≤ 0.5. In some variants, 0 ≤ y ≤ 0.4. In some variants, 0 ≤ y ≤ 0.3. In some variants, 0 ≤ y ≤ 0.2. In some variants, 0 ≤ y ≤ 0.1.​

[0088] In some variations, 0 < z ≤ 0.6. In some variations, 0 < z ≤ 0.5. In some variations, 0 < z ≤ 0.4. In some variations, 0 < z ≤ 0.3. In some variations, 0 < z ≤ 0.2. In some variations, 0 < z ≤ 0.1.

[0089] By combining the above Li[Mn 0.8 Mg 0.2 PO4 compound with additional Li and V, a cathode material with the formula Li[Li 0.1 Mn 0.6 Mg 0.2 V 0.1 PO4 can be synthesized. In other words, with respect to formula (C), x = 0, y = 0.6, TM = Mg, V and z = 0.3 (i.e., the total percentage of transition metals Mg and V, 0.2 + 0.1).

[0090] v. Olivine-type cathode material without excessive lithium

[0091] The present technology also provides GED improvements without introducing excessive lithium. Excessive lithium is often used to provide additional capacity. Therefore, removing excessive lithium results in a lower energy capacity. However, the present technology solves the limitation of removing excessive lithium by utilizing additional transition metals. More specifically, the olivine-type cathode material can include LiMPO4 (M = Fe and Mn). Using Mn as the transition metal can provide additional capacity by simultaneously using both the Mn 2+ / 3+ and Mn 3+ / 4+ redox reactions.

[0092] In some aspects, multiple redox reactions of Mn can be utilized in the olivine-type cathode material, such as the Mn 2+ / 3+ and Mn 3+ / 4+ redox reactions. The olivine-type cathode material is represented by the following formula (D):

[0093] Li[Fe 1-x-y Mn x TM y (P,A)O4 Formula (D)

[0094] where 0.15 < x < 0.45, 0.20 < y < 0.45, where TM is at least one element selected from Mn, Mg, Zn, Ca, Ni, Co, V, Al, Ti, Zr, Mo, and Cr, etc., x represents the fraction of Mn, and y represents the fraction of TM.

[0095] In some variants, 0.15 < x < 0.45. In some variants, 0.20 < x < 0.45. In some variants, 0.25 < x < 0.45. In some variants, 0.30 < x < 0.45. In some variants, 0.35 < x < 0.45. In some variants, 0.40 < x < 0.45. In some variants, 0.15 < x < 0.40. In some variants, 0.15 < x < 0.35. In some variants, 0.15 < x < 0.30. In some variants, 0.15 < x < 0.25. In some variants, 0.15 < x < 0.20.

[0096] In some variants, 0.25 < x < 0.35. In some variants, 0.20 < x < 0.35. In some variants, 0.20 < x < 0.30. In some variants, 0.15 < x < 0.30. In some variants, 0.15 < x < 0.25. In some variants, 0.10 < x < 0.35. In some variants, 0.10 < x < 0.30. In some variants, 0.10 < x < 0.25. In some variants, 0.10 < x < 0.20. In some variants, 0.10 < x < 0.15.

[0097] In some variants, 0.20 < y < 0.45. In some variants, 0.20 < y < 0.40. In some variants, 0.20 < y < 0.35. In some variants, 0.20 < y < 0.30. In some variants, 0.20 < y < 0.25. In some variants, 0.25 < y < 0.45. In some variants, 0.30 < y < 0.45. In some variants, 0.35 < y < 0.45. In some variants, 0.40 < y < 0.45.

[0098] In some variants, 0.25 < y < 0.35. In some variants, 0.20 < y < 0.35. In some variants, 0.20 < y < 0.30. In some variants, 0.15 < y < 0.30. In some variants, 0.15 < y < 0.25. In some variants, 0.10 < y < 0.35. In some variants, 0.10 < y < 0.30. In some variants, 0.10 < y < 0.25. In some variants, 0.10 < y < 0.20. In some variants, 0.10 < y < 0.15.

[0099] Experiments have demonstrated the successful synthesis of the high weight energy density (GED) olivine compound represented by formula (D).

[0100] In one embodiment, x = 0.3 and y = 0.3, TM = Mg, and formula (D) becomes Li[Fe 0.4 Mn 0.3 Mg0.3 ]PO4. The fraction of Mg is outside the conventional range of 0.2>y>0.

[0101] The Li[Fe 0.4 Mn 0.3 Mg 0.3 ]PO4 compounds. The synthesis utilizes the Mn 3+ / 4+ Oxidation reduction. Mn 3+ / 4+ Has a higher redox voltage than Mn 2+ / 3+ , which has a redox voltage of about 4.1 V. Mn 3 + / 4+ It also has a higher redox voltage than Fe 2+ / 3+ , which has a redox voltage of about 3.45 V. Therefore, using Mn 3+ / 4+ Redox increases GED and offsets the capacity reduction caused by the removal of excess lithium.

[0102] In addition, Li[Fe 0.4 Mn 0.3 Mg 0.3 Doping the PO4 compound with about 0.3 mol of Mg can help utilize Mn in olivine-type cathode materials. 3+ / 4+ In addition, from the X-ray diffraction (XRD) analysis, Li[Fe 0.4 Mn 0.3 Mg 0.3 ]PO4 compound exhibits an XRD pattern similar to that of LiFePO4.

[0103] vi. Machine Learning Assisted Optimization of LMX Cathode Active Materials

[0104] To achieve the optimal cathode, thousands to tens of thousands of possible variations need to be tested, and the synthesis and testing of each variation is resource-intensive, expensive, and time-consuming. For example, such variables to be adjusted in the design space include at least: (1) Li excess (x), (2) the properties of the substitutional cation dopant (M), (3) the amount of M (y), (4) the properties of the substitutional polyanion dopant (A), (5) the amount of A (z), (6) the synthesis temperature T1, (7) the synthesis residence time at temperature T1, (8) the synthesis temperature T2, and (9) the synthesis residence time at temperature T2.

[0105] In some variations, one or more doping elements (eg, Mg or Zn, etc.) may be added to the compound in relatively large amounts to obtain a high GED.

[0106] There is no analytical relationship between the variables and the metric of interest (such as cathode electrochemical performance, particle size, tap density, phase purity, etc.), and the goal is to determine the possible results measured by the metric of interest for a set of parameters for the variables.

[0107] Active learning refers to a class of machine learning models that guides efficient and minimalistic data collection to build a model that maps inputs of variables (design variables) to outputs quantified by metrics of interest. Specific implementations involve Bayesian optimization to balance exploration and exploitation strategies. The two components of Bayesian optimization are 1) the model function and 2) the acquisition function.

[0108] For the model function, Gaussian processes are used due to their probabilistic basis and ability to encode physically based kernels for the covariance function.

[0109] The goal of Bayesian optimization is to use a set of observations and suggest where to evaluate the experiment next. The acquisition function is usually a cheap function that can be evaluated at a given point that is commensurate with the expected degree of evaluation of f at x to be used in the minimization problem. The acquisition function can be optimized to select the location of the next observation. In the context of optimization problems, it can also be interpreted as a loss function. Typical choices for acquisition functions include probability of improvement, expected improvement, upper confidence limit, etc. Some acquisition functions, such as expected improvement, are more suitable for research settings, where the goal of the experiment is to "explore" the design space, while "upper confidence limit" acquisition functions are more suitable for global maximization (or minimization), such as in a more development environment.

[0110] The gain function can be one of an upper confidence limit, expected improvement, or information gain. Based on the intent of the experimental activity (research vs. development), there is a trade-off between exploration and exploitation. In addition, the optimization can be performed in a batch setting, meaning that at each iteration, multiple data points can be collected in parallel (limited by available resources).

[0111] For example, the workflow for an experimental campaign to increase gravimetric energy density could be as follows:

[0112] Step 1: Synthesize N variants from the variables in the design space solved above and evaluate the powder specifications and button cell electrochemical data. Define the variant as a vector of values ​​for the variables in the design space above. This can serve as seed data for training the model.

[0113] Step 2: Using the coin cell energy density as the output, fit a Gaussian process model, constrained by the set specification that the powder level indicator should meet.

[0114] Step 3: Using the acquisition function, determine N variations [N may vary] to evaluate in the next iteration that may or are predicted to increase energy density.

[0115] Step 4: Synthesize N variants from step 3 and evaluate powder specifications and coin cell electrochemical data.

[0116] Step 5: Repeat steps 2-4 until the experimental budget is exhausted, the difference between consecutive iterations is below a threshold, or the iterations meet a set of specifications for the target gravimetric energy density (GED).

[0117] Cathode development involves trade-offs. The algorithm can provide a Pareto-optimal selection of design variables that improves gravimetric energy density without severely compromising rate capability, electrical resistance, tap density, and other quantities. The algorithm can also work in the presence of noisy data and categorical variables.

[0118] Cathode development is used to investigate a specific excess Li range to achieve a target gravimetric energy density (GED) by combining experiments with machine learning.

[0119] When excess Li is used to increase GED, other properties may become worse or may be compromised. Machine learning can help find the appropriate trade-off. For example, machine learning predictions can help find how life cycle, capacity, voltage, energy retention, stability, etc. will be affected. The optimization is multi-objective, including increasing GED and trying not to compromise factors such as transport properties (conductivity, surface reaction kinetics, Li+ diffusion rate, etc.) and cycle life. The optimization will be Pareto optimized and find trade-offs. All other metrics can also be measured and used to guide experiments. For example, to maximize GED, constraints may include keeping the voltage less than 4.1V, utilizing elements that are still abundant (e.g., to reduce material costs), while also aiming to obtain the same or improved transport properties.

[0120] Machine learning can help reduce the number of experiments to run to reach a target GED. There is an efficiency gain in the number of experiments to run to achieve a set trade-off. Each set of experiments will be informed by existing experiments and data. In contrast to traditional experimental design, where the design is static (i.e., the experiments to be run are predetermined), Bayesian optimization and / or active learning methods dynamically decide on data collection and build a response surface model at each incremental data collection. This allows one to use the most updated information to decide on the course of further experiments.

[0121] vii. Cell Development Workflow

[0122] Figure 3 A workflow is shown for steps of cathode synthesis and qualification at the powder and cell levels according to an embodiment of the disclosure. As an example, a workflow 300 is provided for forming a battery cell. The workflow or process 300 includes (1) synthesis, (2) powder metering, (3) cell prototyping, and (4) cell testing.

[0123] Synthesis is the process of forming cathode powder. Figure 3 As shown, the synthesis includes mixing precursors, which involves Li:Fe or Li:Mn or dopant stoichiometry. The synthesis also includes grinding under wet or dry conditions. The synthesis also includes calcination at various temperatures and times. The synthesis also includes surface treatment, which also involves chemistry.

[0124] Precursor materials (e.g., lithium salts, phosphates, silicon dioxide) can be chemically reacted in a wet lab to synthesize powders (e.g., LMX). One method of synthesizing powders includes solid-state synthesis. Solid-state synthesis provides a continuous process that can be easily scaled up to increase production. For solid-state synthesis, the precursor materials are not reacted during the grinding stage. Therefore, the powders need to be mixed after grinding.

[0125] The result of the grinding process is a slurry in which the precursor can be ground into a small size (e.g., submicron). Several different mills can be used to grind the powder into a slurry. For example, a horizontal disc mill can be used to grind the powder into a submicron size. As another example, a planetary ball mill can be used to grind the powder into a slurry. In some cases, a planetary ball mill may be preferred because the planetary ball mill can be configured to process a variety of different compositions or powders in a separate tank. In other words, a planetary ball mill can improve output by grinding a variety of different compositions at the same time. One disadvantage of a planetary ball mill is that the planetary ball mill may require additional monitoring of temperature and gas due to the generation of unwanted gases during grinding.

[0126] In some cases, water can be used as the grinding solvent. In other cases, when the material is incompatible with water, alcohol or other grinding solvents can be used.

[0127] Other synthesis methods (such as hydrothermal synthesis, solvent thermal synthesis, microwave hydrothermal synthesis, etc.) can complement solid-state synthesis. These different synthesis methods can reduce the need for grinding due to the dissolution of materials in solvents during the synthesis process. For example, hydrothermal synthesis can provide more uniform powders. In hydrothermal synthesis, the precursor material is dissolved in a solvent (such as water or alcohol, etc.) to form a solution placed in an autoclave. The chamber is then sealed, heated to a high temperature (such as 200°C), and pressurized under high pressure (such as 300Ps i). Therefore, the reaction of the precursor material is carried out in a solution under high heat and pressure in a small chamber. After the reaction (for example, after 24 hours), small crystals or small particles of powder (such as LMX) are left. Hydrothermal synthesis is a slow batch process and is more difficult to scale up. For example, a typical hydrothermal synthesis can take up to a day to heat the material and complete the synthesis.

[0128] As another example, microwave hydrothermal synthesis and / or microwave solvothermal synthesis can utilize microwaves to quickly heat materials and complete the synthesis (eg, 20 minutes). Microwave-assisted synthesis produces small batch sizes and is difficult to scale up to increase production or yield.

[0129] In some cases, these methods may include "one-pot" synthesis. One-pot synthesis can provide a convenient synthesis method in which all raw materials are combined in one pot and the reaction occurs therein. Thus, "one-pot" synthesis can provide a simplified process without additional precursor reactions, mixing, and other steps. One disadvantage of one-pot synthesis techniques is that these techniques are more difficult to control because, without appropriate controls or precautions, it is possible that undesirable reactions can occur.

[0130] For wet grinding, after the powder is ground into a slurry, the slurry is dried by, for example, spray drying. In some cases, the drying method can result in different characteristics of the resulting powder. For example, changing the nozzle, pressure, temperature, production chamber, etc. can result in different properties of the powder, such as shape, sphere size, etc. In some cases, nitrogen can be used to spray materials that may be sensitive to moisture. Another method of drying the material uses a vacuum oven and / or a microwave oven.

[0131] After drying the cathode powder, the cathode powder is calcined by heating to a high temperature to remove volatile materials. Box furnaces and / or tube furnaces can be used to calcine the cathode powder. Calcination can include various configurable parameters, including temperature, duration, material layers, stack height, gases used in the furnace, heating profile, pressure, etc.

[0132] In some cases, the cathode powder can be treated to improve conductivity. Carbon coating is a common technique for improving the conductivity of cathode active materials in lithium-ion batteries. Carbon coating can improve the conductivity of cathode active materials without changing other inherent properties. Uniformly coating carbon on cathode active materials or compounds helps avoid charge accumulation and undesirable chemical reactions. Carbon coating on cathode active materials or compounds (e.g., LMX) can affect the cycle performance of battery cells produced from carbon-coated cathode powders.

[0133] In some variations, the particles may be coated with a carbon coating. One method of forming the carbon coating is to place a soluble sugar (e.g., glucose) into a water-based slurry. After spray drying, the sugar is present in the spray-dried powder. During calcination, the sugar decomposes and forms a carbon coating around each particle.

[0134] Powder metering includes performing material characterization and analysis of the resulting synthesized powder to determine whether the cathode powder is suitable for the next step (e.g., using the cathode powder for cell prototyping or building a battery cell). Material characterization and analysis of the cathode powder are performed to determine one or more characteristics and / or properties of the cathode powder, such as phase purity, crystallinity, particle size, surface area and tap density of the cathode particles, etc. In some embodiments, powder metering can be performed automatically, and the results of powder metering can be fed back into a machine learning model for identifying precursor materials and process parameters for making powders.

[0135] The phase purity, crystallinity, particle size, surface area of ​​the cathode particles can be determined through material analysis procedures (e.g., facilitated by various analytical equipment), including X-ray diffraction analysis (XRD), scanning electron microscopy (SEM), energy dispersive X-ray spectroscopy (EDS), etc.

[0136] For example, a tap density is a material property of interest. Tap density takes into account the pores and voids between particles and is not based on a loose powder bed, but rather the bed after a specified number of tapping steps. The tap density of the cathode powder is determined after a specified number of powder bed tapping steps. The tap density is different from the bulk density of the powder, which takes into account the pores and voids of the loose powder bed. The bulk density can be determined if the powder is loosely introduced into a graduated cylinder.

[0137] like Figure 3 As further shown, the workflow 300 also includes cell prototyping, such as building a battery cell. Specifically, the cell prototyping includes electrode preparation, cell assembly, and formation of a battery cell (e.g., a button cell). After forming the cell prototype, the cell can be tested (i.e., cell testing) to determine whether the battery cell meets the target cell properties of the battery cell. Some target cell properties include internal resistance, voltage, capacity, and cycle life, etc.

[0138] For example, one target cell property is the capacity of a battery or battery cell, which is a measure of the charge a battery can store and is determined by the mass of active materials contained in the battery. Capacity represents the maximum amount of charge that can be extracted from a battery under certain conditions. A battery has a discharge current in amperes that it can deliver over time. The capacity of a battery is measured in ampere-hours (Ah).

[0139] viii. XRD results

[0140] X-ray diffraction analysis (XRD) is an analytical technique used in materials science to determine some properties of materials, such as crystal structure, chemical composition and other physical properties. XRD is based on the constructive interference of monochromatic X-rays and crystalline samples. X-rays are electromagnetic radiations of shorter wavelengths, which are produced when charged particles with sufficient energy are decelerated. In XRD, the generated X-rays are collimated (i.e. made parallel) and directed to a material sample, where the interaction of the incident rays with the sample produces diffracted rays, which are then detected, processed and counted. The intensity of the diffracted rays scattered by the material at different angles is plotted to display the diffraction pattern.

[0141] Figure 4 The LiFePO4 compound and Li[Fe 0.4 Mn 0.3 Mg 0.3 ]PO4 compound X-ray diffraction (XRD) pattern. The horizontal axis or x-axis represents the 2θ angle, while the vertical axis or y-axis represents the arbitrary intensity or counts. The 2θ angle is used to represent the angle between the detector position and the incident X-ray beam. Figure 4 As shown, XRD pattern 402 represents the XRD result of LiFePO4 compound, while XRD pattern 404 represents Li[Fe 0.4 Mn 0.3 Mg 0.3 ]XRD results of PO4 compounds.

[0142] Figure 4 Further explanation of Li[Fe 0.4 Mn 0.3 Mg 0.3 ]PO4 compound and the XRD pattern 402 of the LiFePO4 compound are superimposed or overlapped, which indicates that Li[Fe 0.4 Mn 0.3 Mg 0.3 ]PO4 compound has similar peaks to LiFePO4 compound. Therefore, XRD analysis shows that Li[Fe 0.4 Mn 0.3 Mg 0.3 ]PO4 compound has the same or substantially similar structure as LiFePO4. XRD analysis also shows that based on XRD analysis, Li[Fe 0.4 Mn 0.3 Mg 0.3 ]PO4 compound has a single phase without any by-products or impurities.

[0143] ix. Battery cell testing and experimental results

[0144] The cathode disk is made of synthetic powder. The density of the cathode disk depends on the size of the powder. More specifically, reducing the size of the powder can increase the density of the cathode disk. For such purpose, a mill is often used to grind the powder into a finer powder.

[0145] Porosity is a measure of the void space in a material. The porosity of a cathode can affect the performance of an electrochemical cell. For example, the porosity of a cathode can be increased by compressing a material containing Li[Fe 0.4 Mn 0.3 Mg 0.3 ]PO4 compound powder to form the cathode disk.

[0146] The cathode disk is assembled with the anode disk and the electrolyte into a button cell battery. 0.4 Mn 0.3 Mg 0.3 ]PO4 cathode disks are used to make button cells. Coin cells are evaluated to determine various properties of the cathode material, including capacity, average voltage, volumetric energy density, and discharge energy retention. Cell experiments may include cycling tests on button cells. Constant current charge / discharge cycles of button cells may be performed.

[0147] The button cell can be loaded into a temperature-controlled chamber that can be connected to a battery tester (e.g., manufactured by Neware or Arbin) and tested according to a customized test protocol. The test temperature can vary from about 10°C to about 45°C. Various test protocols can be designed to evaluate the performance of the button cell, including different test currents (e.g., from C / 100 to 1C), different voltage ranges (e.g., between about 1.5V to about 5V), different numbers of cycles (from 1 cycle to 50 cycles), and various combinations of all the above parameters. For example, a constant current charge / discharge cycle of a button cell can be performed at a rate of C / 10 at about 30°C using an operating voltage ranging from 2.5V to 5.2V.

[0148] Electrochemical testers offer users a variety of battery testing options. Multiple channels can be plugged into an electrochemical tester to allow multiple batteries to be tested simultaneously. These tests allow the user to fully understand the effectiveness of the electrochemical cell being tested by measuring the battery's parameters such as voltage, current, impedance, and capacity. The tester can be connected to a computer to obtain digital test values.

[0149] The following examples are for illustrative purposes only. It will be apparent to those skilled in the art that many modifications, both materials and methods, may be practiced without departing from the scope of the disclosure.

[0150] By Li[Fe 0.4 Mn 0.3 Mg 0.3The experimental results of electrochemical analysis of button cell batteries made of cathode materials formed by ]PO4 compounds demonstrated the high GED potential.

[0151] Figure 5 The Li[Fe 0.4 Mn 0.3 Mg 0.3 ]PO4 compound charge and discharge data. The horizontal axis or x-axis represents the specific capacity of the battery, and the vertical axis or y-axis represents the voltage of the battery relative to Li metal. The operating voltage range is 2.5V to 5.2V relative to Li. The voltage relative to Li metal refers to the reference voltage of the cathode voltage to the voltage of Li metal.

[0152] like Figure 5 As shown, curve 502 represents discharge, while curve 504 represents charge. Curve 502 includes three different regions 506A, 506B, and 506C during the discharge process. The first, second, and third regions 506A, 506B, and 506C represent three different reduction processes, including Mn 4+ Reduction to Mn 3+ , Mn 3+ Reduction to Mn 2+ and Fe 3+ Reduction to Fe 2+ .

[0153] At the beginning of discharge as shown in the first region 506A (e.g., the 0 to 10 mAh / g region), the potential energy drops from about 5 V to about 4.2 V, indicating that Li[Fe 0.4 Mn 0.3 Mg 0.3 ]Mn in PO4 compounds 4+ Reduction to Mn 3+ , which is also called Mn 3+ / 4+ Reduction. The reduction potential is about 4.8 V relative to Li.

[0154] In Mn 3+ / 4+ After reduction, the second region 506B (e.g., 10 to 60 mAh / g region) indicates that Mn 3+ Reduction to Mn 2+ , which is also called Mn 2+ / 3+ The third region 506C (e.g., 60 to 130 mAh / g region) indicates that Fe 3+ Reduction to Fe 2+ , which is also called Fe 2+ / 3+ reduction.

[0155] Figure 6 According to some aspects of the disclosed technology, Li[Fe 0.4 Mn 0.3 Mg0.3 ]PO4、LiFe 0.6 Mn 0.4 Comparison of voltage versus capacity of LiPO4, LiMnPO4 and LiFePO4 compounds during charge or discharge. As shown, charge or discharge curves 602, 604, 606 and 608 represent the charge or discharge of compounds Li z [Fe 0.4 Mn 0.3 Mg 0.3 ]PO4、Li z Fe 0.6 Mn 0.4 PO4、Li z MnPO4 and Li z The relationship between discharge voltage and capacity during charging of FePO4, where the subscript z is a variable during the charging process. At the beginning of charging, when the capacity is zero, z = 1, and at the end of charging, when the capacity reaches the maximum value, such as 170 mAh / g, z = 0, and Li[Fe 0.4 Mn 0.3 Mg 0.3 ]PO4 reaches the highest voltage of more than 4.5V at the end of charging among the four compounds. At the end of charging, LiFe 0.6 Mn 0.4 PO4 and LiMnPO4 compounds may have a voltage of about 4.1 V, which is higher than LiFePO4 which has a voltage of about 3.5 V. GED can be calculated from the area under the charge or discharge curve, e.g. Figure 6 The curves 602, 604, 606 and 608 in FIG. 1 are voltage times capacity. Therefore, the area under the curves 602, 604, 606 and 608 represents the GED of each compound. Therefore, Li[Fe 0.4 Mn 0.3 Mg 0.3 ]PO4 has the largest area under each of the curves 602 among the four compounds, or has the highest GED among the four compounds.

[0156] x. Neural Network Architecture and Machine Learning

[0157] Figure 7An example neural network architecture is shown in accordance with some aspects of the present technology. Architecture 700 includes a neural network 710 defined by an example neural network description 701 in a rendering engine model (neural controller) 330. Neural network 710 may represent a neural network implementation of a rendering engine for rendering media data. Neural network description 701 may include a complete specification of neural network 710, including neural network architecture 700. For example, neural network description 701 may include a description or specification of architecture 700 of neural network 710 (e.g., layers, layer interconnections, number of nodes in each layer, etc.); input and output descriptions indicating how inputs and outputs are formed or processed; indications of activation functions in the neural network, operations or filters in the neural network, etc.; neural network parameters, such as weights, biases, etc.; and the like.

[0158] Neural network 710 reflects architecture 700 defined in neural network description 701. In this example, neural network 710 includes input layer 702, which includes input data, such as powder information and button cell electrochemical data. In one illustrative example, input layer 702 may include seed data, which includes button cell electrochemical data.

[0159] The neural network 710 includes hidden layers 704A to 704N (hereinafter collectively referred to as "704"). The hidden layers 704 may include n number of hidden layers, where n is an integer greater than or equal to one. The number of hidden layers may include as many layers as required for the desired processing results and / or rendering intent. The neural network 710 also includes an output layer 706, which provides the output generated by the processing performed by the hidden layer 704 (e.g., variables in the design space, resulting in a button cell with a button cell energy density or a weight energy density (GED), a tap density, or a volumetric energy density (VED), etc.). In an illustrative example, the output layer 706 can provide parameters for the variables in the design space, which can increase the button cell energy density, GED, tap density, or VED.

[0160] Neural network 710 is, in this example, a multi-layer neural network of interconnected nodes. Each node may represent a piece of information. Information associated with a node is shared between different layers, and each layer retains the information as it is processed. In some cases, neural network 710 may include a feedforward neural network, in which case there are no feedback connections (the output of the neural network feeds back into itself). In other cases, neural network 710 may include a recurrent neural network, which may have loops that allow information to be transferred across nodes as inputs are read.

[0161] Information can be exchanged between nodes by node-to-node interconnection between each layer. The nodes of input layer 702 can activate a group of nodes in the first hidden layer 704A. For example, as shown, each input node of input layer 702 is connected to each node of the first hidden layer 704A. The nodes of hidden layer 704A can convert the information of each input node by applying an activation function to the information. Then, the information derived from the conversion can be passed to the nodes of the next hidden layer (e.g., 704B) and activated, and these nodes can perform their specified functions. Example functions include convolution, upsampling, data conversion, merging and / or any other suitable functions. Then, the output of the hidden layer (e.g., 704B) can activate the nodes of the next hidden layer (e.g., 704N), and so on. The output of the last hidden layer can activate one or more nodes of the output layer 706, providing output at this point. In some cases, although a node in neural network 710 (e.g., nodes 708A, 708B, 708C) is shown as having multiple output lines, the node has a single output and all lines shown as output from the node represent the same output value.

[0162] In some cases, each node or interconnection between nodes may have a weight, which is a set of parameters derived from training neural network 710. Once neural network 700 is trained, it may be referred to as a trained neural network, or a trained machine learning algorithm, which may be used to classify one or more activities. For example, an interconnection between nodes may represent a piece of information learned about the interconnected nodes. The interconnections may have adjustable numerical weights (e.g., based on a training data set) so that neural network 710 adapts to the input and is able to learn as more data is processed.

[0163] The neural network 710 may be pre-trained to process features of data from the input layer 702 using different hidden layers 704 to provide outputs through the output layer 706. In an example where the neural network 710 is used to identify an object collision path from a trained object path prediction algorithm, the neural network 710 may be trained using training data that includes seed data obtained from an experiment, such as button cell electrochemical data or powder information, where the powder is synthesized experimentally. For example, the training seed data may be input into the neural network 710, and the neural network 710 may process the seed data to generate an output that may be used to adjust one or more aspects of the neural network 710, such as weights, biases, etc.

[0164] In some cases, the neural network 710 may use a training process called back propagation to adjust the weights of the nodes. Back propagation may include forward transfer, loss function, backward transfer and weight update. Forward transfer, loss function, backward transfer and parameter update are performed in one training iteration. This process may be repeated for a certain number of iterations for each set of training media data until the weights of each layer are accurately adjusted. For the first training iteration of the neural network 710, since the weights are randomly selected at initialization, the output may include values ​​that are not biased towards any particular category. For example, if the output is a vector with the possibility that the object includes different products and / or different users, the possibility values ​​of each different product and / or user may be equal or at least very similar (e.g., for ten possible products or users, each category may have a possibility value of 0.1). Using the initial weights, the neural network 710 cannot determine low-level features and therefore cannot accurately determine the classification of the object. A loss function may be used to analyze errors in the output. Any suitable loss function definition may be used. Any suitable loss function definition may be used, such as cross entropy loss. Another example of a loss function includes mean square error (MSE), defined as E_total = ∑(1 / 2(target - output)^2). The loss may be set equal to the value of E_total.

[0165] For the first training data set (e.g., images), the loss (or error) may be high because the actual value will be different from the predicted output. The goal of training is to minimize the amount of loss so that the predicted output matches the target or ideal output. The neural network 710 can perform a back pass by determining which inputs (weights) contribute most to the loss of the neural network 710, and the weights can be adjusted so that the loss is reduced and ultimately minimized.

[0166] The derivative of the loss with respect to the weight (expressed as dL / dW, where W is the weight of a particular layer) is used to determine the weight that contributes most to the network loss. After the derivative is calculated, the weight update can be performed by updating all the weights of the filter. For example, the weights can be updated so that they change in the opposite direction of the gradient. The weight update can be expressed as w=w_i-ηdL / dW, where w represents the weight, wi represents the initial weight, and η represents the learning rate. The learning rate can be set to any suitable value, with a high learning rate including larger weight updates and a lower value representing smaller weight updates.

[0167] Neural network 710 may include any suitable neural network or deep learning network. One example includes a convolutional neural network (CNN), which includes an input layer and an output layer with multiple hidden layers between the input layer and the output layer. The hidden layers of the CNN include a series of convolutional layers, nonlinear layers, merging layers (for downsampling), and fully connected layers. In other examples, neural network 710 may represent any other neural network or deep learning network, such as an autoencoder, a deep belief network (DBN), a recurrent neural network (RNN), etc.

[0168] As will be appreciated by those skilled in the art, machine learning-based classification techniques may vary depending on the desired implementation. For example, machine learning classification schemes may use one or more of the following alone or in combination: hidden Markov models; recurrent neural networks; convolutional neural networks (CNNs); deep learning; Bayesian symbolic methods; generative adversarial networks (GANs); support vector machines; image registration methods; applicable rule-based systems. When regression algorithms are used, they may include, but are not limited to, stochastic gradient descent regressors and / or passive-active regressors, etc.

[0169] The machine learning classification model may also be based on a clustering algorithm (e.g., a mini-batch K-means clustering algorithm), a recommendation algorithm (e.g., a Miniwise hashing algorithm or a Euclidean locality sensitive hashing (LSH) algorithm), and / or an anomaly detection algorithm (e.g., a local outlier factor). In addition, the machine learning model may use a dimensionality reduction method, such as one or more of the following: a mini-batch dictionary learning algorithm, an incremental principal component analysis (PCA) algorithm, a latent Dir ichlet allocation algorithm, and / or a mini-batch K-means algorithm, etc.

[0170] Figure 8 An example processor-based system in which some aspects of the subject technology may be implemented is shown. For example, processor-based system 800 may be any computing device or any component thereof that makes up the system, wherein the components of the system communicate with each other using connection 805. Connection 805 may be a physical connection via a bus, or directly connected to processor 810, such as in a chipset architecture. Connection 805 may also be a virtual, network connection, or a logical connection.

[0171] In some embodiments, computing system 800 is a distributed system, wherein the functionality described in the present disclosure may be distributed across a data center, multiple data centers, a peer-to-peer network, etc. In some embodiments, one or more of the system components represent a number of such components, each performing some or all of the functionality of the component. In some embodiments, a component may be a physical or virtual device.

[0172] The example system 800 includes at least one processing unit (central processing unit (CPU) or processor) 810 and connections 805 that connect various system components (including system memory 815, such as read-only memory (ROM) 820 and random access memory (RAM) 825) to the processor 810. The computing system 800 may include a cache of high-speed memory 812 directly connected to the processor 810, close to the processor 810, or integrated as part of the processor 810.

[0173] Processor 810 may include any general purpose processor and hardware services or software services, such as services 832, 834, and 836 stored in storage device 830, configured to control processor 810, as well as special purpose processors where software instructions are incorporated into the actual processor design. Processor 810 may be an independent computing system containing multiple cores or processors, buses, memory controllers, caches, etc. Multi-core processors may be symmetric or asymmetric.

[0174] To enable user interaction, the computing system 800 includes an input device 845, which may represent any number of input mechanisms, such as a microphone for voice, a touch-sensitive screen for gesture or graphical input, a keyboard, a mouse, motion input, voice, and the like. The computing system 800 may also include an output device 835, which may be one or more of a variety of output mechanisms known to those skilled in the art. In some cases, a multimodal system may enable a user to provide multiple types of input / output to communicate with the computing system 800. The computing system 800 may include a communication interface 840, which may generally control and manage user input and system output. The communication interface may perform or facilitate receiving and / or transmitting wired or wireless communications via a wired and / or wireless transceiver, including utilizing an audio jack / plug, a microphone jack / plug, a universal serial bus (USB) port / plug, Ports / plugs, Ethernet ports / plugs, Fiber optic ports / plugs, Proprietary wired ports / plugs, Wireless signal transmission, Low energy (BLE) wireless signal transmission, Wireless signal transmission, radio frequency identification (RFID) wireless signal transmission, near field communication (NFC) wireless signal transmission, dedicated short range communication (DSRC) wireless signal transmission, 802.11 Wireless signal transmission, wireless local area network (WLAN) signal transmission, visible light communication (VLC) signal transmission, Worldwide Interoperability for Microwave Access (WiMAX), infrared (IR) communication wireless signal transmission, public switched telephone network (PSTN) signal transmission, integrated services digital network (ISDN) signal transmission, 3G / 4G / 5G / LTE cellular data network wireless signal transmission, ad-hoc network signal transmission, radio wave signal transmission, microwave signal transmission, infrared signal transmission, visible light signal transmission, ultraviolet light signal transmission, wireless signal transmission along the electromagnetic spectrum, or some combination thereof.

[0175] The communication interface 840 may also include one or more global navigation satellite system (GNSS) receivers or transceivers for determining the location of the computing system 800 based on one or more signals received from one or more satellites associated with one or more GNSS systems. GNSS systems include, but are not limited to, the Global Positioning System (GPS) of the United States, the Global Navigation Satellite System (GLONASS) of Russia, the Beidou Navigation Satellite System (BDS) of China, and the Galileo GNSS of Europe. There is no limitation on the operation of any specific hardware device, and therefore the basic features here can be easily replaced with improved hardware or firmware devices when they are developed.

[0176] The storage device 830 may be a non-volatile and / or non-transitory and / or computer-readable memory device, and may be a hard disk or other type of computer-readable medium that can store computer-accessible data, such as a magnetic tape, a flash memory card, a solid-state memory device, a digital versatile disk, a cassette, a floppy disk, a flexible disk, a hard disk, a magnetic tape, a magnetic stripe / strip, any other magnetic storage medium, a flash memory, a memory resistor memory, any other solid-state memory, a compact disc (CD) read-only memory (CD-ROM) optical disc, a rewritable CD optical disc, a digital video disc (DVD) optical disc, a Blu-ray disc (BD) optical disc, a holographic optical disc, other optical media, a secure digital (SD) card, a micro SD (microSD) card, card, a smart card chip, an EMV chip, a subscriber identification module (SIM) card, a mini / micro / nano / pico SIM card, another integrated circuit (IC) chip / card, a random access memory (RAM), a static RAM (SRAM), a dynamic RAM (DRAM), a read-only memory (ROM), a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), a flash EPROM (FLASHEPROM), a cache memory (L1 / L2 / L3 / L4 / L5 / L#), a resistive RAM (RRAM / ReRAM), a phase change memory (PCM), a spin transfer torque RAM (STT-RAM), another memory chip or cartridge, and / or a combination thereof.

[0177] Storage device 830 may include software services, servers, services, etc., and when code defining such software is executed by processor 810, it enables system 800 to perform a function. In some embodiments, a hardware service that performs a particular function may include a software component stored in a computer-readable medium connected to a hardware component (e.g., processor 810, connection 805, output device 835, etc.) to perform the function.

[0178] Embodiments within the scope of the present disclosure may also include tangible and / or non-transient computer-readable storage media or devices for carrying or storing computer-executable instructions or data structures thereon. Such real computer-readable storage devices may be any available device accessible by a general or special-purpose computer, including the functional design of any special-purpose processor as described above. As an example and not limitation, such tangible computer-readable devices may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, disk storage or other magnetic storage devices, or any other device that can be used to carry or store required program code (which is in the form of computer-executable instructions, data structures or processor chip designs). When information or instructions are provided to a computer via a network or another communication connection (hardwired, wireless or a combination thereof), the computer correctly regards the connection as a computer-readable medium. Therefore, any such connection is correctly referred to as a computer-readable medium. The above combination should also be included in the scope of computer-readable storage devices.

[0179] Computer executable instructions include, for example, instructions and data that cause a general-purpose computer, a special-purpose computer, or a special-purpose processing device to perform some function or group of functions. Computer executable instructions also include program modules that are executed by computers in a stand-alone or network environment. Typically, program modules include processes, programs, components, data structures, objects, and functions in the design of a special-purpose processor that performs a task or implements an abstract data type. Computer executable instructions, related data structures, and program modules represent examples of program code means for performing the steps of the methods disclosed herein. A specific sequence of such executable instructions or related data structures represents an example of corresponding actions for implementing the functions described in such steps.

[0180] Other embodiments of the present disclosure can be implemented in a network computing environment with various types of computer system configurations, including personal computers, handheld devices, multiprocessor systems, microprocessor-based or programmable consumer electronics, network personal computers (PCs), minicomputers, mainframe computers, etc. The embodiments can also be implemented in a distributed computing environment, where tasks are performed by local and remote processing devices connected (by hardwired links, wireless links, or a combination thereof) through a communication network. In a distributed computing environment, program modules can be located in both local and remote memory storage devices.

[0181] It will be understood that the cathode materials or cathode active materials described herein can be used in conjunction with any battery cell or component thereof known in the art. For example, in addition to wound battery cells, the layers can be stacked and / or used to form other types of battery cell structures, such as a dual-cell structure. All such battery cell structures are known in the art.

[0182] The various embodiments described above are provided by way of illustration only and should not be construed as limiting the scope of the disclosure. For example, the principles herein are equally applicable to optimization and general improvements. Various modifications and changes may be made to the principles described herein without following the example embodiments and applications shown and described herein, and without departing from the spirit and scope of the disclosure. "At least one" in a group of claim language records means that one member of the group or multiple members of the group satisfy the claim.

[0183] Item 1. A powder comprising Li(Li x TM y TM' (1-x-y) A lithium metal polyanion (LMX) compound represented by )(P,A)O4, wherein 0.1≤x, 0≤y<1 and Li / (TM'+TM)>1, wherein TM is one or more elements selected from Mn, Mg, Zn, Ca, Ni, Co, V, Al, Ti, Zr, Mo, Cr or other transition metals, and wherein TM' is a combination of Fe and Mn transition metals.

[0184] Clause 2. The powder of Clause 1, wherein at least one process variable or at least one stoichiometric variable required for producing the compound represented by formula (I) is provided by a machine learning algorithm.

[0185] Clause 3. The powder of any one of the foregoing clauses, wherein TM is Mo and the compound is represented by Li[Li] 0.2 Fe 0.2 Mn 0.5 Ti 0.1 PO4.

[0186] Clause 4. The powder of any one of the foregoing clauses, wherein TM is V and the compound is represented by Li[Li] 0.1 Fe 0.8 V 0.1 PO4.

[0187] Clause 5. The powder of any one of the foregoing clauses, wherein the compound is represented by Li[Li] 0.1 Mn 0.6 Mg 0.2 V 0.1 PO4.

[0188] Clause 6. A cathode active material comprising the powder of any one of the foregoing clauses.

[0189] Clause 7. A cathode comprising the cathode active material of Clause 6.

[0190] Clause 8. A battery cell comprising the cathode of Clause 7, a separator, and an anode, wherein the battery cell has a weight capacity of more than 170 mAh / g.

[0191] Clause 9. A powder comprising a lithium manganese phosphate compound represented by formula (II): Li[Fe 1-x-y Mn x TM y (P,A)O4, where 0.15 < x < 0.45, 0.20 < y < 0.45, and TM is at least one element selected from Mn, Mg, Zn, Ca, Ni, Co, V, Al, Ti, Zr, Mo, and Cr.

[0192] Clause 10. The powder of Clause 9, wherein x = 0.3, y = 0.3, and TM = Mg, and the compound is represented by Li[Fe 0.4 Mn 0.3 Mg 0.3 PO4.

[0193] Clause 11. The powder of any one of the foregoing Clauses 9 - 10, wherein based on X-ray diffraction (XRD) analysis, Li[Fe 1-x-y Mn x Mgy ]PO4 has the same structure as LiFePO4.

[0194] Item 12. The powder of any one of the preceding items 9 to 11, wherein A represents one of V, Si or W.

[0195] Item 13. A cathode active material comprising the powder of any one of the preceding items.

[0196] Item 14. A cathode comprising the cathode active material of Item 13.

[0197] Item 15. A battery cell comprising the cathode of Item 14, a separator, and an anode, wherein the battery cell comprises a gravimetric capacity of more than 170 mAh / g.

[0198] Item 16. A method for designing the LMX compound of any one of the preceding items 1-5, the method comprising optimizing the composition of the LMX compound to achieve a gravimetric capacity of more than 170 mAh / g using machine learning (ML)-assisted design combined with an experimental method.

[0199] Clause 17. The method of clause 16, further comprising: synthesizing a compound to form a powder of any of the preceding clauses 1-5; evaluating the electrochemical properties of the powder and the battery cell of claim 8; using the electrochemical properties and powder information to train a machine learning model; using the energy density of the battery cell as output to fit a Gaussian process model, subject to the constraint that the powder level indicator falls within a set of specifications; using an acquisition function to determine N variants to be evaluated in the next iteration that may maximize the energy density; synthesizing the N variants; evaluating the electrochemical properties of the powder and the battery cell; and repeating the experiment and training the ML model until the difference in consecutive iterations is below a threshold.

[0200] Item 18. A method for designing the LMX compound of any one of the preceding items 9-12, the method comprising optimizing the composition of the LMX compound to achieve a gravimetric capacity of more than 170 mAh / g using machine learning (ML)-assisted design combined with an experimental approach.

[0201] Clause 19. The method of clause 18, further comprising: synthesizing a compound to form a powder of any of clauses 9 to 12 above; evaluating the electrochemical properties of the powder and the battery cell of claim 15; using the electrochemical properties and powder information to train a machine learning model; using the energy density of the battery cell as output to fit a Gaussian process model, subject to the constraint that the powder level indicator falls within a set of specifications; using an acquisition function to determine N variants to be evaluated in the next iteration that may maximize the energy density; synthesizing the N variants; evaluating the electrochemical properties of the powder and the battery cell; and repeating the experiment and training the ML model until the difference in consecutive iterations is below a threshold.

Claims

1. A powder comprising a lithium metal polyanion (LMX) compound represented by formula (I): Li(Li x TM y TM’ (1-x-y) )(P,A)O4, formula (I) where 0.1 ≤ x, 0 ≤ y < 1 and Li / (TM + TM’) > 1, where TM is one or more elements selected from Mn, Mg, Zn, Ca, Ni, Co, V, Al, Ti, Zr, Mo, Cr or other transition metals, and where TM’ is a combination of Fe and Mn transition metals.

2. The powder according to claim 1, wherein at least one process variable or at least one stoichiometric variable required for producing the compound represented by formula (I) is provided by a machine learning algorithm.

3. A powder according to any one of the preceding claims, wherein TM is Mo and the compound consists of Li[Li] 0.2 Fe 0.2 Mn 0.5 Ti 0.1 PO4 means.

4. A powder according to any one of the preceding claims, wherein TM is V and the compound consists of Li[Li] 0.1 Fe 0.8 V 0.1 PO4 means.

5. A powder according to any one of the preceding claims, wherein the compound consists of Li[Li] 0.1 Mn 0.6 Mg 0.2 V 0.1 PO4 means.

6. A cathode active material comprising the powder according to any one of the preceding claims.

7. A cathode comprising the cathode active material according to claim 6.

8. A battery cell comprising the cathode according to claim 7; a separator; and an anode, wherein the battery cell has a weight capacity of more than 170 mAh / g.

9. A powder comprising a lithium manganese phosphate compound represented by formula (II): Li[Fe 1-x-y Mn x TM y (P,A)O4 formula (II) where 0.15 < x < 0.45, 0.20 < y < 0.45, and where TM is at least one element selected from Mn, Mg, Zn, Ca, Ni, Co, V, Al, Ti, Zr, Mo and Cr.

10. The powder according to claim 9, wherein x=0.3, y=0.3, TM=Mg, and the compound is Li[Fe 0.4 Mn 0.3 Mg 0.3 ]PO4 indicates.

11. A powder according to any one of the preceding claims 9-10, wherein based on X-ray diffraction (XRD) analysis, Li[Fe 1-x-y Mn x Mg y ]PO4 has the same structure as LiFePO4.

12. The powder according to any one of claims 9 - 11, wherein A represents one of V, Si or W.

13. A cathode active material comprising the powder according to any one of the preceding claims.

14. A cathode comprising the cathode active material according to claim 13.

15. A battery cell comprising the cathode according to claim 14; a separator; and an anode, wherein the battery cell has a weight capacity of more than 170 mAh / g.

16. A method of designing an LMX compound according to any one of claims 1 - 5, the method comprising using machine learning (ML) - assisted design in combination with experimental methods to optimize the composition of the LMX compound to achieve a weight capacity of more than 170 mAh / g.

17. The method according to claim 16, the method further comprising: synthesizing the compound to form a powder according to any one of claims 1 - 5; evaluating the electrochemical performance of the powder and the battery cell according to claim 8; using the electrochemical performance and powder information to train a machine learning model; using the energy density of the battery cell as an output to fit a Gaussian process model, which is constrained by the powder - level metrics falling within a set of specifications; using an acquisition function to determine N variants that are likely to maximize the energy density to be evaluated in the next iteration; synthesizing the N variants; evaluating the electrochemical performance of the powder and the battery cell; and repeating the experiments and training the ML model until the difference in consecutive iterations is below a threshold.

18. A method of designing an LMX compound according to any one of claims 9 - 12, the method comprising using machine learning (ML) - assisted design in combination with experimental methods to optimize the composition of the LMX compound to achieve a weight capacity of more than 170 mAh / g.

19. The method according to claim 18, further comprising: synthesizing said compound to form a powder according to any one of the preceding claims 9-12; evaluating the electrochemical properties of the powder and the battery cell according to claim 15; training a machine learning model using the electrochemical properties and powder information; fitting a Gaussian process model using the energy density of the battery cell as output, subject to the constraint that a powder level indicator falls within a set of specifications; Using the acquisition function to determine N variations that are evaluated in the next iteration that are likely to maximize energy density; synthesizing the N variants; evaluating the electrochemical properties of the powder and the battery cell; and The experiment is repeated and the ML model is trained until the difference in consecutive iterations is below a threshold.

Images