Jump to the scientific article here

J.Konstapel, Leiden,16-10-2026.

Europa’s batterij-industrie worstelt met een afhankelijkheid van Chinese machines en technici, wat leidt tot faillissementen en vertragingen bij fabrieken zoals Northvolt en ACC.

Nederland kiest daarom voor een nichestrategie: het ontwikkelen van eigen productieapparatuur, componenten en onderzoeksconsortia via BCC-NL en het groeifondsprogramma ‘Material Independence & Circular Batteries’.

De blog analyseert deze inspanningen met een theoretisch ‘vacuum net’-model, dat de fundamentele grens van alle batterijchemie beschrijft: elke energiedichtheidswinst gaat gepaard met een grensvlak dat onder spanning staat en kan breken.

Het model voegt toe dat batterijen geen energie ‘bevatten’, maar een tijdelijke toestand zijn die meetbaar is via een ‘hold time’-ladder, en biedt concrete ontwerpaanwijzingen om grensvlakken te stabiliseren.

based on Blog: The future of Batteries.

On 16 October 2026, Het Financieele Dagblad described Europe’s attempt to make batteries without China. The article’s conclusion was blunt: the difficulty is not the chemistry, it is the machines. European battery factories buy their production equipment from China, and the knowledge needed to run that equipment comes with it — in the form of Chinese technicians who stay on the factory floor.

The record supports this. Northvolt bought about 1,200 machines from the Chinese supplier Wuxi Lead, kept hundreds of Chinese technicians far longer than planned, and went bankrupt in March 2025. ACC still struggles with scrap rates; in September 2026, Stellantis halted two French car plants for lack of ACC cells, and ACC flew in a team of experts from China. PowerCo runs its Salzgitter plant almost entirely on Chinese machines, with roughly 200 supplier specialists on site. Verkor lost its anchor customer after eighteen months of delay.

The numbers behind the headlines are stark:

• China made more than 80 percent of the world’s battery cells in 2025 (IEA).
• In Europe, 98 percent of cell capacity is owned by Asian companies (Deloitte, 2025).
• Battery packs in China were about 35 percent cheaper than in Europe in 2025 (IEA).
• A quarter of the European cell capacity announced since 2022 has been cancelled or shelved (Transport & Environment).

Since 9 October 2025, the problem has had a legal edge as well. China’s MOFCOM Decision No. 58 places under export licence: cells of 300 Wh/kg and above, cathode material and artificial graphite, and the machines for winding, stacking, filling, hot pressing and formation. The licence requirement was suspended until 10 November 2026. Whether it stays suspended is a decision made in Beijing, not in Brussels.

2. What the Netherlands is already doing

The Dutch response was not improvised yesterday. In 2022 the national Action Agenda set a goal: by 2030, Dutch machine builders supply the European gigafactories. The cabinet deliberately chose components, machines and niches — not a domestic cell gigafactory. A letter to parliament of February 2026 repeats that choice. The FD article describes the execution of a decision taken four years ago.

The network behind it:

• BCC-NL in Helmond is the hub, founded in 2021 by DAF, VDL, Damen, ELEO, TNO, TU/e, Brainport Development, RAI and NMT. Its director since March 2025 is Mustafa Amhaouch, formerly twenty years at ASML.
• The growth-fund programme “Material Independence & Circular Batteries” runs from 2024 to 2031 with a budget of at most €296 million, across three lines: raw materials and recycling; materials, components and equipment; and systems for transport and grid. Twenty-two projects have been running since March 2026.
• National research consortia cover the field: BatteryNL (the electrode-electrolyte interface, €9.3 m, TU Delft), ANION (nanoscale electrochemical interfaces, €23.6 m, Leiden), NANEXBAT (sodium-ion, €8 m, Twente), SMART-CBAT (materials, processes and equipment, €22.3 m, TU Delft), SLDBatt (8–100 hour storage, >€30 m, Twente), CIMBATT (heavy transport, €30 m, VDL), CRM-Lion (critical material recovery, €23 m, TNO), and 10X-Factor(y) (domain walls and memristors, €30 m, Groningen).

The Dutch research effort divides into eight lines: silicon and carbon anodes (LeydenJar, E-magy, CarbonX, LionVolt); solid state and electrolytes (Thionics, Twente, Utrecht, Delft, Eindhoven); sodium-ion (NANEXBAT, Alfen with CATL in Almere); long-duration storage (Ore Energy, Elestor, AquaBattery); production equipment (SALD, SparkNano, Kalpana, Powall, TNO Holst Centre); recycling (SK tes, Back to Battery); the interface inside the cell (BatteryNL, ANION); and storage in a wall (Twente, Groningen).

A few people hold the network together — Marnix Wagemaker (Delft), Mark Huijben (Twente) and Moniek Tromp (Groningen) each sit in several consortia; André Schilt connects TNO Holst with BCC-NL, and Holst has spun off LionVolt, LeydenJar and SALD.

The gaps are just as telling. No Dutch party was found working on: inline measurement on foil and coating at production scale (the skill exists in chip metrology — Nearfield raised €330 million in June 2026, but for chips); current without applied voltage in an asymmetric cavity; or logic and storage on three values (+1, 0, −1). Lithium-sulfur and quantum batteries are also absent.

3. The model: the vacuum net in plain terms

This essay reads the battery through a theoretical model — the vacuum net — and asks what follows for how batteries are made, measured and used.

Imagine a fishnet under tension. Strands run through it and cross; where windings hold each other there is a knot; between strands there are meshes — openings with a size and a direction. The vacuum net theory reads the vacuum itself as such a net, built on five axioms:

1. Strand. There is one strand; every distinction is the strand touching itself.
2. Closure. A configuration persists only if it closes on itself in phase.
3. Memory. The net keeps the orientation of what was laid down first.
4. Context. Constants are state functions of the local net.
5. Scale. The same winding appears at every depth.

Three objects follow. A winding is a strand returning on itself in phase: that is energy. A knot is windings holding each other in three directions: that is mass. A mesh is an opening between strands. Power, being energy per unit of time, scales with one over the open size of a mesh.

Two rules do the work in this essay:

• The window rule. Each crossing of strands carries a trit: +1, 0 or −1. Three neighbouring crossings form a window. A window closes when the sum of its values is divisible by three — 9 of the 27 possible windows close; the other 18 stand open under tension. What closes, stays.
• The mesh energy. One mesh of size L holds an energy of π²ℏc/(720 L). At the natural mesh size of 45.588 nm, that comes to 9.51 × 10⁻²¹ J.

What a battery is, in this model

In everyday speech a battery “contains” energy. In the net it does not. Energy is a closed motion, not a substance in a jar. A battery is a state that is held for a time — and a hold time can be measured.

That measurement produces a ladder. Each rung has its own carrier:

Hold time	Carrier	Dutch work
Seconds	Flywheel, supercapacitor, polar wall	Twente (dielectric films); Wang, Delft (MXenes)
Hours	Lithium-ion, sodium-ion	LeydenJar; NANEXBAT; Alfen
Days	Iron-air, salt, flow	Ore Energy; Elestor; AquaBattery; SLDBatt
Months	Molecules	Hydrogen pilots

No rung replaces another. The public grid fuses all phases into one — what is generated is used at the same instant, hold time zero. A cell asked to bridge days does the work of another rung, and that is why it grows heavy, expensive and dependent on scarce material.

The one limit every line shares

The lines of battery research differ in chemistry, but they meet the same limit. Every gain in energy density comes from an electrode that changes shape, and the boundary layer must follow that change without breaking:

• Silicon anode: volume grows by more than 300 percent; the interphase cracks and re-forms every cycle.
• Lithium metal: lithium plates unevenly and grows along cracks.
• Solid state: solid-solid contact is lost; filaments grow along grain boundaries.
• Lithium-sulfur: intermediate products dissolve away from the cathode.
• Magnesium, calcium: the boundary layer seals and stops passing the ion.

Each boundary joins two structures that close differently — crystal and liquid, metal and ceramic, lattice and film. By the window rule, such a boundary is an open window: it stands under tension and does not stay unless something keeps closing it.

The rule does not say why a seam breaks. It says which configurations persist. That is enough to turn the seam into three design questions: Where is the seam? Can it be laid so that fewer windows stay open? And can the motion take place in a mesh — an open space — rather than in the seam itself?

4. The model against the measurements

The rules were set against numbers that Dutch and international groups have published.

Closure orders crystal orientations. Crystal directions are written with three Miller indices; for the low-index directions these are trit triples, so the window rule applies directly. Of the ⟨100⟩ family (6 members), none close; of ⟨111⟩ (8 members), 2 close; of ⟨110⟩ (12 members), 6 close. An open direction carries passage; a closed direction holds. The rule therefore predicts transport through a film in the order ⟨100⟩ > ⟨111⟩ > ⟨110⟩.

The measurements agree:

• Lithium titanate (Cunha et al., 2019), at 3C: (100) about 313 mAh/g, (111) about 283, (110) about 277 — exactly the order of the open fraction.
• Lithium manganese oxide (Huijben et al., Twente, 2022): after 1,000 cycles at 5C, (100) keeps 55 percent and (110) keeps 33 percent, with more surface-held charge in the closed direction. An earlier Twente series (Hendriks et al., 2018) also found (100) best.

The coupling at a wall is a quantity of that wall. The theory’s coupling law reads Δg/g = γδ, where δ is the relative change in density and γ is a state function of the wall, not one universal number. Palasantzas’s group in Groningen measures Casimir forces on phase-change films:

• GST: when the film crystallises, the force rises by about 40 percent below 100 nm, while density goes from 5.87 to 6.27 g/cm³ (δ = 0.068). Measured γ: 5.9.
• AIST: force contrast of 17–25 percent against densification of 5.2 percent. Measured γ: 3.3 to 4.8.

Passage scales with one over the gap. Garret Moddel’s group in Colorado measured current without applied voltage across metal-insulator-metal junctions carrying optical cavities from 33 to 1,100 nm. The current rose as the cavity grew thinner. Maclay’s fit (2024) shows conductance proportional to 1/L — the exponent of the passage rule.

A polar wall holds about one mesh energy per lattice cell. The Twente multilayer of Nguyen, Koster and Rijnders stores 128.4 J/cm³ at 81 percent efficiency. With a lattice parameter of 4.15 Å, that is 9.18 × 10⁻²¹ J per cell — 0.97 of one mesh energy. The best films reported up to October 2026 hold between one and 1.65 mesh energies per cell.

Room instead of grip. LeydenJar’s silicon columns, E-magy’s pores and LionVolt’s 3D pillars all leave room for the swelling — in the fishnet image, the motion goes into the mesh, not into the knot. Single-crystal cathodes, which have no internal seams, outlast polycrystalline ones by a wide margin (Dalhousie–Tesla, 2022).

Where the readings stand:

Reading	Status
9 of 27 windows close; 0/6, 2/8, 6/12 per family	counted, exact
Orientation order (100) > (111) > (110)	agrees in two spinels
γ at a wall is a quantity of that wall	GST 5.9; AIST 3.3–4.8
Conductance ∝ 1/L	agrees with Moddel’s series
Polar walls hold about one mesh energy per cell	Twente 0.97; best films up to 1.65
A battery boundary is an open window	reading
Life follows the count of open windows	hypothesis, not yet measured

5. What the net adds: six contributions

The net places the limit of the battery at the seam — and the Netherlands is strongest exactly there. BatteryNL is devoted entirely to the electrode-electrolyte interface; Twente grows crystal films one atomic layer at a time and measures them while they grow; Delft watches lithium move through the interface with neutrons; Eindhoven has the densest cluster of thin-layer equipment builders in Europe; and the chip supply chain has made the country a world leader in measuring and controlling thin layers. That is a different strength from the “paint robot” in the FD headline — and it is the strength the net speaks to.

1. Ranking the seam before it is grown. The window rule ranks crystal orientations by their open fraction before a single film is grown, and it got the order right in two spinels. One experiment would turn this into a design tool: grow all three orientations of one electrode at equal surface area and measure life and rate. Huijben’s group has the films and the pilot line; Wagemaker and Bannenberg in Delft have the instruments to watch the seam while the cell runs. The same count extends to any interface whose two sides can be written as trits — every new coating, interlayer or epitaxial stack gets a ranking at no cost.

2. The coupling of a wall as a design quantity. With γ known per wall, the change in coupling follows from a density measurement alone. That makes γ measurable for any wall that changes density: a phase-change film, an interphase that densifies during formation, a coating that compacts. Palasantzas has the force set-up; Wagemaker’s neutron depth profiles give the density of a plated layer during operation. Together they would give γ for the interphases that decide a cell’s life.

3. Making without the heat step, outside the licence list. A conventional electrode is a slurry coated on foil and dried in long ovens. The drying step alone costs 11.02 kWh per kWh of cell capacity and 27–48 percent of all factory energy; counting yield, each good kWh carries up to 58 kWh of factory energy — a cell must be fully discharged 23 to 58 times before it has returned the energy of its own production.

The net describes making as a three-step cycle: soften (bring the local state near its tipping point), write (impose a mesh, a layering, an asymmetry), harden (return the state so that only closing windings remain). Hardening selects; it does not place. Processes that work layer by layer, at low temperature, follow this cycle directly.

The machines China has licensed belong to one process order: mix, coat, dry, press, wind or stack, fill, form. Four process families do not appear on the list — plasma deposition (LeydenJar), spatial atomic layer deposition (SALD, SparkNano, Kalpana, TNO Holst Centre), thin-film cells (Huijben, Energy01), and electrodeposition. Each is already Dutch. The FD plan rebuilds the existing order from Dutch coaters and laminators, but that order still contains the licensed steps. A second route chooses the processes that do not need them. It is less mature, it touches the Chinese list less, and it plays directly to Dutch strength.

4. The loop instead of the hand. The FD article names the real difficulty: a machine does not run by itself; the knowledge sits in the people who tune it. Fraunhofer FFB puts first-years gigafactory scrap at 15–30 percent, and each percentage point costs about €30,000 per day — €10.95 million per point per year. At 20 percent scrap that is €219 million a year.

In the net, a state that holds itself is a closed loop. A loop that moves one setting until the output is at its maximum, then holds it there, needs no operator. Where the state of each layer is measured inline and the setting follows, the knowledge sits in the loop — not in the hands of technicians flown in with the machine. This is the thin spot of the Dutch network: the skill exists in chip metrology, the battery application is missing. Joining the two would address the cause of Northvolt’s and ACC’s troubles directly.

5. The wall that holds in three states. A ferroelectric wall has three polarisation states — P+, P0, P−: a trit. The Twente films already store about one mesh energy per lattice cell, and no ion moves house, so there is no seam that changes with every cycle. Twente and Groningen (10X-Factor(y), €30 million) hold the materials. What the net adds is the register: a wall with three states can hold and pass on +1, 0 and −1 directly. Storage and logic on three values are not yet worked on in the Netherlands.

6. Passage instead of stock. One mesh at the natural size holds 9.51 × 10⁻²¹ J; one passage per natural time unit gives 6.25 × 10⁻⁵ W per mesh. Moddel’s patent reports 1.4 pW from an active junction of 0.02 µm² — about 70 W/m², or a coupling η of 2.3 × 10⁻⁹ of the full mesh rate. That η, the share of what the mesh closes that leaves as current, is the one parameter that decides how far this route goes. No Dutch group works on it; Palasantzas, AMOLF and Schall have the cavities, the walls and the tipping point as a knob.

6. The future if the readings hold

Two conditions carry the boldest scenario: Moddel’s 70 W/m² must be reproduced, and stacked layers must add. The rest rests on what is already measured.

Less is asked of the cell. When storage follows the ladder of hold times, each rung does its own work. The cell no longer bridges days; it becomes smaller, lighter and less dependent on cobalt, nickel and graphite. A good part of the demand that drives gigafactories falls away, and storage moves from a remote park to the street, the house and the district.

The seam is designed, not suffered. A cell is built from the seam outward. Orientations, interlayers and coatings are ranked before they are grown; the coupling of each wall is known from its density. Life becomes a design target instead of a test outcome after years. Single crystals, epitaxial films and columns with room for motion become the rule rather than special cases.

Factories without fire, near the user. Making by soften-write-harden needs no ovens tens of metres long and no dry rooms. A line that deposits layer by layer at low temperature is small; it fits in a city rather than on the edge of a country. The gigafactory, built around the oven and the formation hall, is no longer the natural form of making a cell. The scale of production follows the scale of use.

The knowledge sits in the loop. Where every layer is measured as it forms and the setting follows, the line holds itself. The knowledge no longer travels with foreign technicians. Scrap falls — and with it the €10.95 million per percentage point per year. A licence decision in Beijing then reaches the spare parts of a machine, not the know-how of a line.

The wall holds and computes. A wall with three states holds energy without moving ions, and holds a trit. Storage and register become one structure, and the fastest rung of the ladder is served by walls that do not wear.

Passage instead of stock. If the passage holds, the change is largest. A house needing 1,450 W continuously needs 20.7 m² of one layer at 70 W/m² — or a plate of about 45 by 45 cm with 100 layers. Such a plate delivers 34.8 kWh per day, where a battery holding one such day at 250 Wh/kg weighs 139 kg. A drone needing 1 kW needs 0.14 m². The battery then stops bridging energy; it only catches peaks — a small buffer in the meter cupboard, the drone, the tool.

What it means for people. In Eindhoven, laboratory staff at LeydenJar assemble test cells by hand all day, at a 95 percent pass rate; in a line that holds itself, their work moves from tuning by feel to reading a state. In Skellefteå, a town of 75,000 built its future on Northvolt and lost 1,000 jobs in one round of cuts, beside technicians from Wuxi who lived in their own camp; a factory without fire, near the user, is less likely to tie a whole town to one plant. In Debrecen, residents raised concerns about CATL’s plant, and in early 2026 the authorities closed parts of it after nine employees were exposed to nickel; a cell asked to do less carries less nickel. In Ukraine, the drones soldiers carry run mostly on Chinese cells, and a licence decision reaches their front line; a plate that delivers continuously would end that dependence. In the household, the base load would no longer depend on a cable, a substation or a battery park — what disappears is the long chain of mine, refinery, gigafactory, ship and grid; what remains is the house and its plate.

---

Annotated references

The occasion

Van Wijnen, J. F. (2026). “Europa maakt batterijen desnoods met een verfrobot om los te komen van China.” Het Financieele Dagblad, 16 October. https://specials.fd.nl/europa-maakt-batterijen-desnoods-met-een-verfrobot-om-los-te-komen-van-china
Why read? LeydenJar, the gigafactories and the Dutch machine plan in one story — the occasion of this essay.

Bits&Chips (2 October 2025). “LeydenJar partners with Chinese firm to scale silicon-anode batteries.” https://bits-chips.com/article/leydenjar-partners-with-chinese-firm-to-scale-silicon-anode-batteries/
Why read? LeydenJar’s partnership with Highpower, which the FD does not mention.

The Local Sweden (2 February 2026). “Northvolt lost ten months due to gamble with untested Chinese machines.” https://www.thelocal.se/20260202/northvolt-lost-ten-months-due-to-gamble-with-untested-chinese-machines
Why read? What it costs when the knowledge sits in someone else’s machines. Read with the next item.

The Local Sweden (27 September 2024). “‘It was terribly mismanaged’: Former Northvolt workers call for accountability.” https://www.thelocal.se/20240927/it-was-terribly-mismanaged-former-northvolt-workers-call-for-accountability
Why read? The counter-voice: former engineers who place the cause in preparation and management.

electrive (30 September 2026). “Battery shortage: Stellantis pauses EV production in France.” https://www.electrive.com/2026/09/30/battery-shortage-stellantis-pauses-ev-production-in-france/
Why read? The ACC shortage and the Chinese experts.

Industriemagazin (14 October 2026). “Volkswagen: Die chinesische Abhängigkeit hinter der großen Batterie-Wette.” https://industriemagazin.at/automobil/volkswagen-die-chinesische-abhaengigkeit-hinter-der-grossen-batterie-wette/
Why read? Gotion’s role in PowerCo from 2021 to now, with the 35 percent price gap.

Herbert Smith Freehills Kramer (November 2025). “China suspends export controls on lithium-batteries and artificial graphite anode materials.” https://www.hsfkramer.com/notes/mining/2025-posts/china-suspends-export-controls-on-lithium-batteries-and-artificial-graphite-anode-materials
Why read? The legal reading of MOFCOM Decision No. 58 and its suspension.

Fred Gao, Inside China (9 October 2025). “China Implements New Export Controls on … Lithium Batteries, and Artificial Graphite Anode Materials.” https://www.fredgao.com/p/china-implements-new-export-controls
Why read? The list of controlled machines, step by step. Basis for section 5, contribution 3.

IEA (2026). Global EV Outlook 2026 — Electric vehicle batteries. https://www.iea.org/reports/global-ev-outlook-2026/electric-vehicle-batteries
Why read? Who makes which share of the world’s cells. Reading advice: the figure on capacity by headquarters.

The network

BCC-NL, team and partners. https://batterycompetencecluster.nl/team
Why read? Who carries the cluster and who leads which pillar.

Nationaal Groeifonds. “Material Independence & Circular Batteries.” https://www.nationaalgroeifonds.nl/overzicht-lopende-projecten/thema-energie-en-duurzame-ontwikkeling/material-independence—circular-batteries
Why read? The programme lines, goals and money.

Nationale Actieagenda Batterijsystemen 2026. https://batterycompetencecluster.nl/fileadmin/Battery_Competence_Center/NAB_magazine_2026_Digitaal.pdf
Why read? Themes, moonshots and the 37 contributing organisations.

BCC-NL. Feasibility study on cell-production equipment. https://batterycompetencecluster.nl/fileadmin/Battery_Competence_Center/Battery_Cell_Production_Technologies.pdf
Why read? The Dutch side of the “paint robot” plan.

Expertgroep Batterijen (September 2022). Management summary, Actieagenda Batterijsystemen. https://oostnl.nl/sites/default/files/attachments/MANAGEMENTSAMENVATTING-ACTIEAGENDA-BATTERIJSYSTEMEN.pdf
Why read? The 2030 goal for Dutch machine builders, on one page.

TU Delft. BatteryNL. https://www.tudelft.nl/en/2022/tnw/in-search-of-the-coveted-safe-better-longer-lasting-battery-batterynl
Why read? The consortium devoted entirely to the seam. Basis for contribution 1.

Holst Centre. “Next-gen Equipment for Batteries and Battery Materials.” https://holstcentre.com/insights/news/next-gen-equipment-for-batteries-and-battery-materials/
Why read? Who builds the Dutch equipment for the new process families.

Holland High Tech. D&M Vacuumsystemen and TNO, dry coating without a dry room. https://www.hollandhightech.nl/en/programmes-and-projects/projects/novel-dry-coating-equipment-with-controlled-environment-for-cost-and-energy-efficient-high-performance-battery-electrode-manufacturing
Why read? A Dutch machine that removes the dry room.

The model

Konstapel, J. (2026). “The Vacuum.Net Theory: Complete Derivation.” constable.blog, 28 September. https://constable.blog/wp-content/uploads/vdefThe-Vacuum.Net-Theory.pdf
Why read? Axioms, definitions, natural units and the mesh energy. Read the status table at the end first.

Konstapel, J. (2026). “The Grid That Cannot Hold.” constable.blog, 31 August. https://constable.blog/2026/08/31/the-grid-that-cannot-hold-what-is-wrong-with-the-electricity-system-the-thinking-errors-behind-it-and-the-simple-net-alternative/
Why read? The four phases, the hold time τ and the ladder. Basis for sections 3 and 6.

Konstapel, J. (2026). “Applied Vacuum Theory, Parts Two to Four.” constable.blog, 25–27 September. https://constable.blog/2026/09/25/applied-vacuum-theory/
Why read? Soften-write-harden, the self-tuning loop and power = 1/S. Reading advice: Part Four, on energy without a reservoir.

Konstapel, J. (2026). “A Wall with Three States.” constable.blog, 12 October. https://constable.blog/wp-content/uploads/v2A-Wall-with-Three-States-Perovskites-the-Trit-Cube-and-the-Open-Side-of-the-Mesh.pdf
Why read? The ferroelectric wall as a trit. Basis for contribution 5.

Konstapel, J. (2026). “What the Path Gives and What Its Description Gives.” Version 1, 16 October.
Why read? Why γ is a state function and not one universal number. Basis for the reading of the Groningen films in section 4.

The measurements

Hassanzadeh, Atul, Wang, Kooi, Tajik, Palasantzas (2025). Physical Review B 112, 155403. https://doi.org/10.1103/zlvn-r3jy
Why read? The GST force contrast of about 40 percent.

Torricelli, van Zwol, Palasantzas et al. (2010). Physical Review A 82, 010101(R). https://arxiv.org/abs/1006.4065
Why read? The first switchable Casimir force, with AIST.

Njoroge, Wöltgens, Wuttig (2002). Journal of Vacuum Science and Technology A 20, 230.
Why read? The densities of amorphous and crystalline GST that give δ.

Hendriks, Huijben et al. (2018). ACS Applied Energy Materials. https://doi.org/10.1021/acsaem.8b01477
Why read? LiMn₂O₄ in all three orientations; (100) best.

Huijben et al. (2022). “Self-assembled epitaxial cathode-electrolyte nanocomposites.” ACS Applied Materials & Interfaces. https://research.utwente.nl/en/publications/self-assembled-epitaxial-cathode-electrolyte-nanocomposites-for-3/
Why read? Retention and surface share per orientation. The closest existing data to the ranking of contribution 1.

Cunha et al. (2019). Orientation-dependent Li₄Ti₅O₁₂ thin films.
Why read? The three-orientation capacity series that follows the closure order. Reading advice: compare at the same C-rate.

Nguyen, Birkhölzer, Houwman, Koster, Rijnders (2022). Advanced Energy Materials 12, 2200517. https://doi.org/10.1002/aenm.202200517
Why read? 128.4 J/cm³ in a Twente multilayer: one mesh energy per lattice cell.

Lv, Wagemaker et al. (2018). Nature Communications.
Why read? Plated lithium seen with neutrons during operation. The instrument for the density of a seam in contribution 2.

Zheng, Zhao, Archer et al. (2019). “Reversible epitaxial electrodeposition of metals in battery anodes.” Science 366, 645. https://news.cornell.edu/node/315386
Why read? An underlayer that sets the direction of growth; cycles rise tenfold.

Dalhousie University / Tesla (2022). Single-crystal NMC532 cells. Journal of the Electrochemical Society. https://electrek.co/2022/05/24/tesla-battery-research-paper-high-energy-density-battery-last-100-years/
Why read? A crystal without internal seams and its very long life.

Schütte, Degen, Walter (2024). Batteries 10(2), 64. https://www.mdpi.com/2313-0105/10/2/64
Why read? Energy per production step per kWh of cell. Source of the 11.02 kWh in contribution 3.

Fraunhofer FFB. “The ramp-up of a gigafactory in battery cell production.” https://www.ffb.fraunhofer.de/en/press/news/the-ramp-up-of-a-gigafactory-in-battery-cell-production.html
Why read? Scrap rates and their cost per day. Source for contribution 4.

Moddel, Weerakkody, Doroski, Bartusiak (2021). “Optical-Cavity-Induced Current.” Symmetry 13, 517. https://arxiv.org/abs/2101.03085
Why read? The cavity series from 33 to 1,100 nm, with eight controls for artefacts.

Moddel, G. US Patent 12,166,434, “Enhanced quantum vacuum energy devices.” https://patents.google.com/patent/US12166434B2/en
Why read? The 1.4 pW and the active area of 0.02 µm² behind the 70 W/m².

Maclay, G. J. (2024). “Modeling Electronic Devices with a Casimir Cavity.” Physics 6(3), 1124–1131. https://doaj.org/article/31210466294848b0b30d3380b0056c2c
Why read? The fit of conductance against one over thickness. Maclay questions the proposed mechanism and leaves the measurement standing.

Hertlein, Helden, Gambassi, Dietrich, Bechinger (2008). Nature 451, 172.
Why read? Critical Casimir forces measured directly. The tipping point as a knob, which Schall uses in Amsterdam.

Scientific Article