Silicon Anode Materials — Why Only Small Amounts Are Still Blended In: The Volume-Expansion Wall and 2026 Commercialization Trends

What Is the Topic: Ten Times the Theoretical Capacity, Single-Digit Blending in Practice

Silicon anode materials have a theoretical capacity more than ten times that of graphite, yet in commercial batteries they remain an additive blended into graphite in small amounts. The reason is that two problems are tangled together: silicon swells by more than 300% when it takes up lithium, and silicon oxide (SiOx), which tempers that swelling, consumes lithium during the first charge. This article works through the two problems in turn and reads the scope of rights of Daejoo Electronic Materials’ registered patent KR101960855B1, which targets the initial-efficiency problem.

Starting with the industry situation, the movement is clear. In a press release dated May 20, 2026, POSCO Future M stated that it had secured mass-production technology for silicon anode materials through product testing and quality verification with major domestic and overseas customers, and that it is pursuing mass-production supply in 2028. The company announced that even in tests raising the blending ratio to 20% or more, it retained over 80% of initial capacity after 1,000 charge-discharge cycles, contrasting this with “the blending ratio of existing batteries, which had stayed in the single digits.” As priority targets it named premium electric vehicles, humanoid robots and urban air mobility (UAM). Earlier, on March 11, 2026, it signed a memorandum of understanding (MOU) on advanced battery-material technology development with U.S. silicon anode company Sila; according to press reports, the two companies are at the stage of starting joint materials research.

The tension is already contained in the announcement itself. POSCO Future M wrote that “volume expansion is the main challenge for commercialization” and described the existing blending ratio as single-digit. In other words, silicon’s capacity advantage has long been known, but the amount that can go into a cell is still bound by process, lifetime and cost constraints. This article answers three questions: why does silicon swell, what do silicon oxide and magnesium doping trade off, and how far do patents in this area draw the scope of rights?

Basic Principle: The More Lithium It Holds, the More It Swells

The theoretical capacity of graphite is 372mAh/g, while silicon reaches 4,200mAh/g. Graphite uses “intercalation” (insertion between layers), slotting one lithium between its layers for every six carbon atoms, whereas silicon accepts up to 4.4 lithium atoms per silicon atom and forms an actual alloy. The larger the storage, the larger the structural change. During alloying the volume grows by more than 300%, and it shrinks back on discharge, every single cycle.

By analogy, graphite is like slipping thin sheets of paper between the shelves of a bookcase, while silicon is like a sponge that swells several times over when it soaks up water. Soak and wring a sponge hundreds of times and it eventually crumbles; in an electrode this is called pulverization (particles breaking into fine pieces). The broken fragments lose their electrical connection to the conductive agent and current collector, and usable capacity falls sharply.

Silicon oxide (SiOx) is easiest to understand as that sponge placed inside a rigid frame. Part of the silicon is dispersed in an oxide matrix (the base material surrounding the particles), which improves expansion and cycle-life behavior. In exchange, capacity drops to about four times that of graphite (about 1,500mAh/g), and during the first charge the oxide component reacts irreversibly with lithium, so the initial coulombic efficiency (the ratio of the first discharge to the first charge) falls to 75% or below. Part of the lithium that came over from the cathode gets trapped in the anode and never returns.

Magnesium doping is an approach to reduce this “trapped lithium.” The KR101960855B1 specification describes a manufacturing method in which a silicon/silicon-dioxide powder mixture and magnesium metal are reacted in the vapor phase under reduced pressure (0.01-2 torr) at 800-1,600°C, forming magnesium silicate crystals evenly within the oxide matrix. The direction is to have magnesium occupy the oxide sites first, before lithium can react there, thereby reducing first-charge losses.

Comparison: Graphite, Silicon Oxide, Mg-Doped Composite and Pure Silicon

Material	Theoretical / representative capacity (mAh/g)	Initial coulombic efficiency (%)	Main challenge
Graphite	372	90 or higher	Capacity ceiling
Silicon oxide (SiOx)	About 1,500	75 or lower	Initial irreversible reaction
Mg-doped SiOx composite	About 1,500 (theoretical); 640 discharge capacity in the example	91 (Example 1)	Process cost
Pure silicon	4,200	–	Volume expansion of 300% or more

What this table says: As capacity rises, initial efficiency and volumetric stability waver; the Mg-doped composite raises initial efficiency to 91% at the price of a heavier process burden. The example discharge capacity of 640mAh/g is a value for the whole composite including the carbon coating, so it should not be compared directly with theoretical material capacities.

Placing the patent specification’s examples and comparative examples side by side reveals that performance is decided less by adding magnesium than by which crystal phase forms.

Case	Crystal phases	Discharge capacity (mAh/g)	Initial efficiency (%)	Capacity retention after 50 cycles (%)
Example 1	Si + MgSiO₃	640	91	92
Example 2	Si + MgSiO₃ + Mg₂SiO₄	630	89	87
Comparative Example 1 (no Mg)	Si + SiO₂	620	85	80
Comparative Example 2	Si + Mg₂SiO₄	597	86	53

What this table says: Even with the same magnesium doping, 50-cycle retention was 92% when enstatite (MgSiO₃) was the main crystal phase, while Comparative Example 2, which formed only forsterite (Mg₂SiO₄), reached just 53%. These figures show that what the patent seeks to protect is not “adding magnesium” in general but “which magnesium silicate crystal phase is formed.”

Patent Perspective: The Scope Staked Out by KR101960855B1

Bibliographic data: KR101960855B1 (title: “Silicon oxide composite for anode material of secondary battery and method for preparing the same”) was filed by Daejoo Electronic Materials on March 20, 2017, published on October 1, 2018 as KR20180106485A, and registered on March 21, 2019. According to Google Patents its legal status is Active and its anticipated expiration date is March 20, 2037. The inventors are listed as Oh Seong-min, Kim Jang-han and Kim Da-eun.

Gist of claim 1 (restated): The original article summarized claim 1 as silicon particles, MgSiO₃ crystals, magnesium content and specific gravity, but the actual independent claim also includes a surface carbon coating as an element. Broken into elements, it reads as follows.

• Silicon particles with a crystal size of 1-25nm are present in an SiOx (0<x<2) matrix
• MgSiO₃ (enstatite) crystals are contained in the same matrix
• A carbon coating is present on the surface
• Magnesium is 2-20 parts by weight per 100 parts by weight in total, and specific gravity is 2.3-3.2

The dependent claims narrow the scope further: a form that also contains Mg₂SiO₄ (forsterite) crystals (claim 2), X-ray diffraction peaks at 28-29° for silicon and 30.5-31.5° for MgSiO₃ (claim 3), carbon content of 1-20 wt% (claim 5), carbon coating thickness of 5-100nm (claim 6), average particle diameter of 0.1-15μm (claim 9), and specific surface area of 1-40m²/g (claim 10).

Filing history: Family members sharing the same priority date (March 20, 2017) are registered as U.S. US11043668B2 (granted June 22, 2021, anticipated expiration January 20, 2038), Japan JP6306767B1, China CN108630887B and CN113410446B, and Europe EP3379611B1. Notably, U.S. claim 1 adds a condition that forsterite is excluded or kept smaller than enstatite. Whereas Korean claim 2 keeps coexisting forsterite inside the scope of rights, in the U.S. the crystal-phase ratio becomes a more direct limiting element.

Insight: the center of gravity of this patent is not “doping” but “phase selection.” The 53% figure for Comparative Example 2 is the patentee itself recording in the specification the inferiority of a forsterite-dominant structure. This can be read in two directions. If a competitor designs around forsterite, there may be room to keep literal distance, but that design follows a path the patentee has already disclosed as lower-performing. Conversely, the more enstatite is increased in pursuit of performance, the closer one gets to both the Korean and U.S. claims. In the end, FTO (freedom-to-operate review) for Mg-doped SiOx is more likely to be decided by XRD crystal-phase analysis than by the composition sheet.

Design-around points: Claim 1 is structured so that literal infringement is established only when all six conditions are met. A design that grows silicon crystals beyond 25nm, a design outside the specific-gravity range, a design without a carbon coating, or a design that pre-occupies the oxide sites with an element other than magnesium may each lack one element.

Design-around scenario (counterargument): However, most of these avoidance routes must be traded against performance loss. Larger silicon crystals raise expansion stress, and removing the carbon coating lowers conductivity. Moreover, if a design near the boundary of a numerical range (for example a crystal size of 26-30nm) achieves substantially the same effect, it may become the target of an infringement-by-equivalents claim. From the patentee’s side too, numerical limits such as specific gravity and crystal size can vary with the measurement method, so in a dispute the measurement conditions themselves may become an issue.

Limits and Outlook

Magnesium doping raised initial efficiency, but it does not eliminate volume expansion itself. That is why current commercial cells generally blend silicon-based materials into graphite, and POSCO Future M’s description of the existing blending ratio as single-digit reflects that reality. Raising the content increases capacity, but the design and process costs needed to absorb the expansion rise with it.

The optimistic outlook is that the announced 80% retention after 1,000 cycles at 20%+ blending leads to mass production in 2028 and lifts the silicon share a step higher. There is also a counter-scenario. POSCO Future M described 2028 mass-production supply as something it is “pursuing,” a plan that can be adjusted according to market demand and the business environment. If EV demand slows or the fast-charging performance of graphite anodes improves, the customers willing to bear the cost of high-content silicon may be limited to small-volume, high-value markets such as robots and UAM.

The strategic alternative is to track IP by treating silicon anode materials not as a “content race” but as a “crystal-phase and structural-design race.” As the Daejoo Electronic Materials patent shows, the scope of rights hinges more on structural parameters such as crystal phase, particle size and coating than on elemental composition. When reviewing supplier diversification, requiring crystal-phase data alongside composition comparisons helps reduce FTO gaps.

Practical action: Planning and IP staff who use or are considering Mg-doped SiOx should obtain XRD crystal-phase (MgSiO₃/Mg₂SiO₄ ratio), silicon crystal size and specific-gravity data from suppliers, and prepare separate element-by-element comparison charts by country against claim 1 of KR101960855B1 and claim 1 of US11043668B2.

Items requiring verification

• The primary source for Sila’s explanation, cited in the original article, that “range and charging speed both improve as silicon content rises”
• The primary source for the silicon content (around 10%) of products applied at Porsche, Audi, Tesla and Hyundai Motor Group, as cited in the original article
• The material type of POSCO Future M’s mass-production product (SiOx-based, Si-C composite, etc.) and the production capacity of the Pohang demo plant
• The reason Google Patents shows a financial institution in the current-assignee field of KR101960855B1 (whether it reflects a registered pledge or similar), to be confirmed in the patent register
• How the registered claims of the JP, CN and EP family members differ from the Korean and U.S. claims

Sources

1. Google Patents, KR101960855B1 – Silicon oxide composite for anode material of secondary battery and method for preparing the same (Daejoo Electronic Materials)
2. Google Patents, US11043668B2 – Silicon oxide composite for lithium secondary battery negative electrode material and method for manufacturing the same (Dae Joo Electronic Materials)
3. POSCO Newsroom, press release on POSCO Future M securing silicon anode mass-production technology (2026-05-20)
4. The Elec, POSCO Future M and U.S. Sila begin joint silicon anode research (2026-03)
5. Money Today, POSCO Future M joins hands with U.S. Sila to develop advanced battery materials (2026-03-13)

Related Reading — LFP vs. NCM: Not Better or Worse, Just Different Optimization Axes · High-Nickel Cathodes: What Is the Price of Energy Density? · Electrolyte, From Liquid to Solid — The Safety-vs-Interfacial-Resistance Tradeoff