Multivalent Batteries

Every chemistry in a production EV today moves a single-charge ion, Li⁺ or Na⁺, between the electrodes. Multivalent batteries swap this for an ion carrying two or three charges: Zn²⁺, Mg²⁺, Ca²⁺, Fe²⁺ or Al³⁺. On paper that gives more charge per ion and metals that are cheap and abundant. This page is the umbrella for the site’s existing Rechargeable Aqueous Zinc Ion, Aluminium Ion and Calcium Ion pages. It covers what changes when the ion carries more charge, where the headline numbers mislead, and what a cell or pack engineer should expect from the field. Much of it draws on a 2026 review in Materials Today by Anjan, Bharti, Mahato, Oh and Koratkar [1].

What changes with a multivalent ion

The capacity of an electrode depends on how many electrons are exchanged per atom or ion, n, and on the molar mass of the material. A divalent ion delivers twice the charge of a monovalent ion for each ion that moves, and a trivalent ion three times. For a metal anode the theoretical specific capacity is:

where C is the specific capacity (mAh/g), n is the number of electrons transferred per metal atom, F is the Faraday constant (96,485 C/mol) and M is the molar mass (g/mol). The 3.6 converts coulombs per gram into mAh/g. Multiplying by the metal density gives volumetric capacity in mAh/cm³.

The pattern is clear. On a mass basis lithium is unbeaten, and only aluminium comes close. On a volume basis zinc, iron and aluminium are three to four times better than lithium metal. That volumetric advantage is the main reason the research community keeps returning to these metals.

Metal battery or metal-ion battery

The literature uses “multivalent-ion” loosely, and the review makes a useful distinction [1]. In a multivalent metal-ion battery the ion shuttles between two host structures, rocking-chair style, exactly as in a lithium-ion cell with a graphite anode. In a multivalent metal battery the negative electrode is the metal itself (zinc foil, magnesium foil) and works by plating and stripping.

Only the metal battery gets the volumetric capacities in the table above. Put the ion in a host anode and most of the advantage goes, because the host adds mass and volume just as graphite does in a lithium-ion cell. So the multivalent case depends on making a metal anode cycle reliably, which is the same problem the Lithium Metal Battery community has been working on for decades, with different chemistry.

Where the headline numbers mislead

Anode capacity is not cell energy. Two things take back much of the advantage.

Voltage. Zinc and iron sit well above lithium on the potential scale, and many of the practical systems use aqueous electrolytes, which limit the cell to roughly the 1.23 V stability window of water (with some overpotential margin). A quick comparison of energy per unit volume of anode metal shows the effect. Lithium metal at a nominal 3.7 V: 2,062 mAh/cm³ × 3.7 V ≈ 7.6 Wh/cm³. Zinc at the 1.35 V nominal of a zinc-manganese oxide cell: 5,854 mAh/cm³ × 1.35 V ≈ 7.9 Wh/cm³. A fourfold capacity advantage turns into rough parity once voltage is included, and that is before any cathode is added. These are our estimates for illustration, not a cell design.

Cathode. The cathode, not the anode, usually sets the cell energy. Finding a host that accepts a doubly or triply charged ion reversibly, at useful voltage and capacity, is the hard part (see next section). The result can be seen in real numbers. The site’s aqueous zinc-ion page lists 65 to 90 Wh/kg at cell level. Eos Energy’s Z3 zinc-based stationary module is rated at 0.8 kWh for 20.5 kg [2], which is about 39 Wh/kg at module level (our arithmetic). Both are well below the LFP cells they would compete with.

System-level view: equal pack energy and power

Electrode and cell numbers are only half the comparison. The fairer test is to hold usable pack energy, power and bus voltage constant and see how much of the material advantage is left. Lower cell voltage changes the series count, and that carries through to interconnects, sensing and packaging.

One point is easy to miss. Fixing the bus voltage and the energy also fixes the pack’s amp-hours, whatever the chemistry. A low-voltage cell does not get to be a bigger cell; the pack simply needs more of them in series, each with the same Ah and less energy. The series count is:

Peak current is also fixed by power and bus voltage, so every series joint carries the full pack current. With one joint per series element, each of resistance R_j, the interconnect loss at peak power is approximately:

where P is pack power, V_bus the nominal bus voltage, V_cell the nominal cell voltage and R_j the resistance of one joint (busbar plus weld). As a worked example, take a 75 kWh usable, 150 kW pack on a 400 V bus. That fixes the pack at about 188 Ah and the peak current at 375 A.

Three things follow.

• Interconnects. Joint count, and with it joint resistance, scales with 1/V_cell. The zinc pack has about 2.7 times the series joints of the NMC pack, and about 2.7 times the I²R loss in them. That is also 2.7 times as many welds to make and inspect. See our Busbars pages for joint design.
• Sensing. Voltage channels, balancing circuits and harness wires scale the same way if every series element is monitored, which is the usual lithium-ion practice (see BMS Sensors).
• Packaging. The fixed overhead of each cell (can or pouch, tabs, terminals, spacing, fixings) is paid about 300 times instead of about 110, on cells that each hold roughly a third of the energy. That pushes cell-to-pack efficiency down on top of the cell-level Wh/kg gap.

The voltage window matters too. The ratio of full-charge to end-of-discharge voltage sets the range the inverter or DC-DC converter has to handle, and it should be compared chemistry by chemistry. We have not quantified it here.

For an EV on a 400 V or 800 V bus, very little of the material advantage survives. Anode energy per unit volume was already at rough parity with lithium once cell voltage was included. The cell-level Wh/kg gap then turns into roughly 3.5 times the cell mass of NMC, and the higher series count adds cost, loss and failure points.

Stationary storage changes two of these terms. Low-voltage modules feeding a DC-DC converter are normal practice there. Eos’s Z3 module, for example, runs at 22 to 48 V DC [2], so much of the bus-voltage penalty moves into power electronics the system often needs anyway. Second, a cell with a water-based electrolyte that gasses rather than burns when overcharged may not need monitoring on every cell. Module-level sensing could remove much of the sensing penalty. That is our interpretation rather than a published BMS architecture. This is why aqueous zinc competes in BESS on cost per kWh, safety and cycle life, not on energy density.

The charge density problem

The same property that gives multivalent ions their capacity also makes them slow. A small ion carrying two or three charges has a high charge density. It holds on tightly to the solvent molecules around it, so desolvation at the electrode surface costs more energy, and it interacts strongly with the anions of the host lattice, so solid-state diffusion is slow. The review ranks the five main metals by charge density with calcium lowest, then zinc and magnesium, iron higher, and aluminium far above the rest [1]. Calcium’s relatively low charge density is the basis of the “calcium paradox” discussed on our Calcium Ion page.

Repeated insertion of a highly charged ion also strains host structures. Most of the cathode work summarised in [1] is about dealing with this: expanding the interlayer spacing, pre-inserting “pillar” species to hold layers apart, introducing defects to lower the migration barrier, and coating particles to stop them dissolving. For an engineer this means rate capability and cycle life at practical loadings remain the main open questions.

The main candidates compared

Zinc has the head start. It is stable enough to use as foil in water-based electrolytes, cells can be built without a dry room, and zinc chemistries already make up a large share of primary cells. The problems are the familiar ones from the nickel-zinc world: dendrites, gassing and shape change. The review points out that several fixes that work in the lab do not scale. Water-in-salt electrolytes suppress hydrogen evolution but are expensive, and some of the zinc alloy anodes are made by routes that cannot be scaled up [1].

Magnesium has a good potential (-2.37 V) and high volumetric capacity, and was long described as dendrite-free. The review repeats that view [1], but Davidson and co-workers at Texas A&M and Purdue grew fractal magnesium dendrites several millimetres long in symmetric Mg cells [3]. They used a Grignard electrolyte (0.5 M MeMgCl in THF) at an average current density of 9.2 A/m², which is 0.92 mA/cm² and not an extreme rate by lithium-ion standards. Their model links the morphology to the balance between plating rate and magnesium surface self-diffusion [3]. One electrolyte and one test set-up do not show that magnesium always grows dendrites, but magnesium is better treated as less prone to them than lithium rather than immune. The bigger obstacle is the electrolyte: magnesium passivates in conventional carbonate electrolytes, and the electrolytes that do plate it well tend to be corrosive or have narrow voltage windows. The site does not yet have a dedicated magnesium page.

Calcium and aluminium share the same problem: a surface layer on the metal that does not conduct the working ion. For calcium it is the SEI; for aluminium it is the native oxide. Their dedicated pages cover the detail.

Iron is the cheapest of all. The review quotes about $98 per tonne against around $15,000 per tonne for lithium [1], though prices for lithium in particular have moved a lot and this ratio should be treated as indicative. Iron is heavy (molar mass about eight times lithium’s), so its case is stationary storage, not vehicles.

Emerging candidates: chromium, manganese, copper and cadmium

Beyond the five main metals, the review covers four less familiar anodes [1]. Each one fixes one of zinc’s weaknesses and brings a new problem of its own.

Chromium has the highest volumetric capacity of any multivalent metal, nearly twice zinc’s, because it gives up three electrons and is dense. Pure chromium has always been ruled out because its oxide film stops the ion getting in or out. The review reports a way round this: a chromium-rich high-entropy alloy (Cr-Bi-Cu-Sn-Ni) whose mixed oxide surface lets Cr³⁺ through at some phase boundaries while others stop the oxide thickening. It plated and stripped for over 10,000 hours (about 5,000 cycles) at around 20 mV overpotential [1]. That result comes from the review’s own lead author’s group, so it is worth waiting for independent replication, and it is a symmetric-cell result, not a full cell.

Manganese is the obvious question for anyone who knows it as a cathode material. As an anode it would give about 976 mAh/g and 7,035 mAh/cm³ (our arithmetic), and at -1.18 V it sits between zinc and aluminium, so it could give a higher cell voltage than zinc. The catch is reactivity. In ordinary aqueous electrolytes manganese corrodes, evolves hydrogen and grows an insulating oxide, and the review lists its plating and stripping reversibility at under 50 cycles [1]. Current work uses water-in-salt or non-aqueous electrolytes to suppress hydrogen evolution. For now manganese earns its place on the cathode side, where Mn²⁺ is even added to the electrolyte of zinc-MnO₂ cells to limit cathode dissolution.

Copper goes the other way. Its potential is positive, so it cannot reduce water and avoids hydrogen evolution entirely, and the review reports coulombic efficiencies of about 99.99% [1]. The price is voltage: a copper anode needs a high-voltage cathode, such as a Prussian blue analogue, to make a useful cell. The review sees it as a candidate for long-life stationary storage, where voltage matters less.

Cadmium plates flat rather than forming dendrites and resists corrosion in water. One study cited in the review ran dendrite-free for over 2,000 hours at 99.93% coulombic efficiency [1]. Cadmium’s toxicity is the problem, and anyone who remembers the phase-out of nickel-cadmium will doubt that closed-loop recycling would make it acceptable for new products.

Our reading is that these four will end up in specific applications rather than as general replacements. Chromium is the only one aimed at higher energy density. Copper and cadmium trade voltage or toxicity for very long cycle life, and manganese depends on an electrolyte that does not yet exist.

Sulfur and selenium cathodes

One way round the cathode bottleneck is to use a conversion cathode rather than an intercalation host. Sulfur (1,675 mAh/g theoretical) and selenium (678 mAh/g, but with a volumetric capacity close to sulfur’s because of its density) react with the metal to form sulfides or selenides, with no host lattice to strain [1]. The review covers zinc-, magnesium-, calcium-, iron- and aluminium-sulfur and selenium cells. They inherit the polysulfide shuttle that limits lithium-sulfur, and selenium is scarce and expensive, so it is more useful as a research tool or a dopant in sulfur than as a bulk cathode.

Reading multivalent papers as an engineer

Most multivalent results are reported per gram of cathode active material, at low loading and with flooded electrolyte. The review is candid about this. It recommends that sulfur and selenium cathodes be tested at more than 4 mg/cm² with lean electrolyte of 5 to 7 µL per mg of active material, and notes that most published work uses less than 2 mg/cm² [1]. It makes the same point for zinc anodes, where many studies use current densities and areal capacities far below practical levels.

Rule of thumb: before taking a multivalent result seriously, check the areal loading (mg/cm² or mAh/cm²), the electrolyte-to-active-material ratio, the anode excess (thick metal foil hides poor coulombic efficiency) and whether the energy is quoted per gram of cathode or per cell. A cycle-life claim at 0.1 mAh/cm² says very little about a 3 mAh/cm² cell.

The review also states that multivalent ions are less likely to drive thermal runaway than lithium [1]. That is fair for aqueous zinc, where the electrolyte is water, but it is a property of the electrolyte as much as the ion. A magnesium or calcium cell with a flammable ether electrolyte needs its own abuse testing.

Medical and implantable devices

An unusual part of the review is its focus on healthcare [1]. Zinc, magnesium and iron are biocompatible, and magnesium and zinc can be made to dissolve in the body at a controlled rate, which opens up implants that power themselves for a fixed period and then resorb. Here energy density matters less than safety, form factor and biocompatibility, which plays to the strengths of aqueous multivalent cells. It is a niche, but it may be one of the first places where rechargeable multivalent cells find real use outside stationary storage.

What to take from this

Multivalent chemistry is a materials-research field, not a pack-design option, for now. The volumetric capacities are real, but once voltage and cathode are included the advantage over lithium shrinks to roughly parity at best, and the main candidates still have unsolved anode or electrolyte problems at practical loadings. The near-term engineering relevance is aqueous zinc in stationary storage, where low cost, non-flammable electrolyte and tolerance of abuse matter more than Wh/kg. Our Grid Battery Energy Storage Systems page covers where that fits. Magnesium and calcium are the ones to watch for higher-voltage cells, but both need an electrolyte breakthrough first. The review’s call to test at commercially relevant loadings is the right one, and it is a useful filter for anyone tracking the field.

The review also opens by saying lithium-ion will soon hit a ceiling of about 300 Wh/kg [1]. The cell data does not support that. Our Cell Energy Density Roadmap to 2035 uses cell data up to 2026 and shows graphite/silicon-blend NMC and NCA production cells at around 350 Wh/kg. Next-generation EV-format cells reach about 380 Wh/kg, and silicon-anode aviation cells about 450 Wh/kg. So the target multivalent chemistries have to beat keeps rising, which makes their path to vehicles harder still.

References

1. A. Anjan, V. K. Bharti, M. Mahato, I.-K. Oh and N. Koratkar, “Multivalent metal-ion, sulfur, and selenium batteries for energy and healthcare applications: Current status, challenges, and future outlook,” Materials Today, 95 (2026) 103286. doi:10.1016/j.mattod.2026.103286 (open access, CC BY-NC)
2. Eos Energy Enterprises, Eos Z3 product sheet, October 2023.
3. R. Davidson, A. Verma, D. Santos, F. Hao, C. Fincher, S. Xiang, J. Van Buskirk, K. Xie, M. Pharr, P. P. Mukherjee and S. Banerjee, “Formation of Magnesium Dendrites during Electrodeposition,” ACS Energy Letters, 4(2) (2019) 375–376, letter with Supporting Information. ACS Energy Letters

See also: Rechargeable Aqueous Zinc Ion, Aluminium Ion, Calcium Ion, Sodium-Ion Battery, Potassium Ion Battery, Lithium Metal Battery, Lithium Sulfur, Battery Chemistry Roadmap, Cell Energy Density Roadmap to 2035