FAQ — Salt, Water, Power

Section 0

How to read this FAQ

Three short framing notes so the cards land as intended.

Framing

Every card has three layers — use the one that matches your question.

In one sentence

Top line is the headline; middle is the plain-English explanation; bottom is the engineering case with numbers and citations — read whichever depth your question deserves.

The plain-English version

Each card has a one-sentence TL;DR up top, a plain-English paragraph in the middle, and an engineering-case block at the bottom with numbers, citations, and trade-offs. You can stop at any layer. The plain-English layer is written for a curious non-specialist; the engineering layer is written for a reviewer with a chemistry or finance background. If a card spans more than one screen, that’s on purpose — the depth is there if you want it.

The full engineering case

Phase vocabulary used throughout: Stage 0 = pre-Phase-1 due diligence (~$80–120K; FTO opinion, contract-lab replication of Chen et al. at 1 A/g, hiring pipeline, signed term sheets). Phase 1 = 13-month chemistry validation in Shakopee (~$1.4–1.8M; 520-cell campaign against the canonical kill-gate table). Phase 1.5 = Plantweb integration, BoS engineering, first hyperscaler letter of intent. Phase 2 = first commercial prototype skid. Phase 3 = build-vs-license-vs-partner decision. Phase 4 = commercial scale-up under Rodolphe El Khoury (NA Operations). Kill-gate vocabulary: kill floor = minimum acceptable performance to proceed; target = expected performance at Phase 1 success; stretch = consistent with the projected 30-year envelope. The canonical Phase 1 success table lives in “What does ‘Phase 1 succeeded’ actually look like?” under Section B and overrides any other card.

Framing

What this FAQ does not cover.

In one sentence

Detailed cell schematics, MOFCOM/CFIUS legal opinions, customer NDAs, and the full 30-year Operations Manual live elsewhere on saltwaterpower.org — this FAQ is the public-facing decision document, not the engineering archive.

The plain-English version

This FAQ is the public-facing decision document. It does not cover: detailed cell-level schematics (those live in the Operations Manual at 11_operations.html); MOFCOM/CFIUS legal opinions on Chinese collaboration (handled by Stage 0 trade counsel); customer NDAs and named pilot agreements (under confidentiality until contracts close); the full 30-year service schedule (Operations Manual); the proposal document itself (01_proposal.html). If you’re looking for a specific topic and don’t find it here, check the rest of the hub at saltwaterpower.org.

The full engineering case

Hub layout: 01_proposal.html = the proposal document itself (Executive Summary, nine revenue surfaces, Storage Landscape, Risk Register); 03_protocol.html = canonical Phase 1 success criteria + cell-build SOPs + measurement protocols; 11_operations.html = the 30-year Operations Manual (6 roles, 12 emergency procedures, 30-year service schedule, OSHA-compliant standalone programs for Respiratory / BBP / DOT / EAP / FPP); 10_faq.html = this document. Internal evidence trail at github.com/cscanlan/emerson-proposal (intellectual provenance + revision history). Excluded by design: anything subject to attorney-client privilege, customer-disclosure NDAs, or BIS export-control review.

Framing

Where the open questions still are.

In one sentence

Three honest theses remain in front of us — skid-scale geometry, 30-year calendar aging, customer willingness to pay LCOE pricing — and Stage 0 + Phase 1 are designed to answer each.

The plain-English version

Reviewers should know what this FAQ is honestly uncertain about. Three theses remain open: (1) whether Chen’s coin-cell chemistry transfers to a 3 MWh skid at C/4 grid rate (Phase 1 directly tests this); (2) whether 30-year calendar aging holds up under the electrolyte-drift, seal-degradation, and CO₂-ingress envelope (accelerated stress tests + Year-10/20 service intervals address this); (3) whether hyperscaler procurement actually prices $/kWh-cycle over $/kWh upfront (Stage 0 monitors RFP language; the eight-surface revenue structure holds even if the carbon-credit surface goes to zero). The card “What evidence would change your mind and kill the program?” in Section F names the five specific kill-criteria.

The full engineering case

Three open theses (per the proposal’s Storage Landscape, “What We Have Not Yet Proven”): (1) Skid-scale geometry — current distribution, edge effects, electrode-stack mechanics at 3 MWh module size; Phase 1 validates at ≥1 Ah pouch; Phase 2 extrapolates to commercial skid via finite-element modeling cross-checked against half-cell measurements. (2) 30-year calendar aging — electrolyte drift (CO₂ ingress, water activity shift), seal degradation, PBA framework stability in chloride over decades; accelerated stress tests + sealed make-up + Year-10/20 refurbishment protocol. (3) Customer LCOE adoption — whether hyperscaler procurement actually prices $/kWh-cycle over upfront $/kWh; Stage 0 monitors public RFP language; the eight-surface revenue structure carries an explicit conservative case where the carbon-credit surface goes to zero and the program still floors at Phase 3 license-back royalty + retained Plantweb attach revenue.

Section A

The Chemistry

What is this battery, why won’t it catch fire, why hasn’t anyone done it before, and why the prior magnesium-ion attempts don’t apply.

A1 · The basics

You might ask

What is this battery, in one sentence?

In one sentence

A non-flammable, neutral-pH aqueous battery that can run for 120,000 charge-discharge cycles — roughly 30× the life of today’s lithium-iron-phosphate grid storage — built from commodity salt and a copper-iron Prussian-blue cathode.

The plain-English version

It stores grid electricity. The chemistry uses salt water (specifically magnesium chloride) instead of the flammable lithium electrolytes most batteries use, which is why it can’t catch fire. The two electrodes are made of materials that let magnesium ions move in and out repeatedly without breaking down, which is why it lasts so much longer than a lithium battery.

The full engineering case

Aqueous Mg²⁺/Ca²⁺-ion battery with a covalent organic polymer (Hex-TADD-COP) anode and a copper-iron Prussian Blue analogue (CuFe-PBA) cathode in neutral-pH MgCl₂ + CaCl₂ electrolyte. Published February 2026 by Chen et al. (City University of Hong Kong) in Nature Communications. Reported performance: 120,000 cycles at 20 A/g, 2.2 V full-cell voltage, 48.3 Wh/kg electrode-level. The chemistry combination — aqueous neutral-pH electrolyte + COP radical-anion (n-type) anode + CuFe-PBA open-framework cathode — is unprecedented and structurally bypasses the failure modes that ended prior Mg-ion programs.

You might ask

How does a battery actually work?

In one sentence

A battery stores chemical energy by separating positively charged ions on one side and negatively charged electrons on the other; closing a circuit lets the electrons flow back, which is electricity.

The plain-English version

Two electrodes (anode and cathode) sit on either side of an electrolyte that lets ions pass but not electrons. Charging pushes ions and electrons apart; discharging lets them recombine, and the electrons take the long way around through whatever you’ve plugged in. The chemistry of the electrodes determines how much energy you can store, how fast you can move it, and how many cycles you can repeat before things wear out.

The full engineering case

At the anode during discharge: oxidation releases electrons. At the cathode: reduction accepts them. Between them, the electrolyte conducts ions to balance charge. Aqueous Mg-ion specifically: Mg²⁺ intercalates into the CuFe-PBA cathode framework during discharge and out during charge; the COP anode hosts the corresponding electrons via reversible radical-anion formation. The voltage (2.2 V) is set by the redox potential difference between the two electrodes; the energy density is set by how much active material participates per unit mass; the cycle life is set by how reversibly the structure handles repeated ion shuttling.

I asked this

What exactly did the Hong Kong paper find?

In one sentence

Chen et al. demonstrated a working aqueous magnesium-ion cell that ran 120,000 charge-discharge cycles at high current with near-100% coulombic efficiency — the first time anyone has shown this combination of chemistry, cycle life, and safety simultaneously.

The plain-English version

A research team at City University of Hong Kong combined three components that nobody had combined before: a neutral-pH salt-water electrolyte, a polymer anode that stores charge through radical chemistry rather than physical insertion, and a copper-doped Prussian blue cathode. The result ran longer than any aqueous battery on record — far past the point where lithium cells would have failed — without catching fire, leaking, or losing meaningful capacity.

The full engineering case

Nature Communications, February 2026. Cell architecture: Hex-TADD-COP anode (covalent organic polymer, radical-anion / n-type storage mechanism) + CuFe-PBA cathode (Prussian Blue analogue with Cu/Fe redox centers) + 4.0M and saturated 5.8M MgCl₂/CaCl₂ aqueous electrolyte at pH 7. Reported metrics: 120,000+ cycles at 20 A/g, 2.2 V full-cell voltage, 48.3 Wh/kg at electrode level, near-100% Coulombic efficiency in steady state, pH stable in 4.91–7.02 window over cycling. Independently corroborated for the broader aqueous Mg-ion platform by Wu et al. (BUCT / PKU Shenzhen, JACS Feb 2026, doi:10.1021/jacs.5c21656) using a Ta-doped MoO₃ anode — same chemistry family, different electrode pair, zero author overlap.

You might ask

What does “aqueous” mean and why does it matter?

In one sentence

“Aqueous” means water-based; for a battery this means the electrolyte is salt dissolved in water rather than lithium salts dissolved in flammable organic solvents.

The plain-English version

Lithium-ion batteries use organic solvents that ignite at high temperatures — that’s the source of every laptop and EV battery fire. Water doesn’t burn. Aqueous batteries trade some energy density for a fundamentally non-flammable safety profile, which is exactly the trade you want for grid storage that doesn’t move and sits next to expensive infrastructure or people.

The full engineering case

The thermodynamic stability window of water at standard conditions is ~1.23 V; beyond that, water electrolyzes into H₂ + O₂. Operating aqueous cells above 2 V requires “water-in-salt” regimes where extreme salt concentrations coordinate free water to ions and suppress splitting. Chen et al.’s 4.0 M and saturated 5.8 M MgCl₂ electrolytes operate in this regime, extending the practical window to 2.2 V. The remaining gas evolution is non-flammable (H₂+O₂ at concentrations far below combustion limits, manageable with standard catalytic recombiners from the VRLA lead-acid industry).

A2 · Why it’s safe

You might ask

Why won’t this catch fire like phones and EVs do?

In one sentence

The electrolyte is water and salt; water doesn’t burn, so the thermal-runaway chain reaction that destroys lithium cells cannot start.

The plain-English version

Lithium battery fires happen when an internal short heats the cell, the organic solvent decomposes and ignites, that releases more heat, more solvent decomposes, and so on until 800°C+ flame. With water as the electrolyte, there’s no fuel for that chain. The worst case is an internal short that produces some hot water, some hydrogen and oxygen gas in concentrations too dilute to burn, and the cell stops working — no flame, no thermal runaway, no propagation to neighboring cells.

The full engineering case

Li-ion thermal runaway requires: (1) a heat-generating event (internal short, mechanical damage, overcharge), (2) decomposition of organic carbonate solvent above ~150°C, (3) oxygen release from the cathode at higher temperatures, (4) self-sustaining combustion with cell-to-cell propagation. Aqueous Mg-ion eliminates step 2 entirely (water replaces organic solvent) and step 3 (PBA cathode doesn’t release lattice oxygen). A puncture or short produces localized H₂/O₂ gas evolution at sub-flammable concentration ratios in the water-in-salt regime, plus some local heating that water absorbs as latent heat of vaporization. No flame, no runaway, no bunker walls required.

You might ask

Is the electrolyte toxic?

In one sentence

No — magnesium chloride is FDA food-grade and used in tofu coagulation, food processing, and road de-icing; calcium chloride is similarly benign.

The plain-English version

The salt in this battery is the same salt that’s in the ocean, in your bag of road de-icer, and in the tofu coagulator at any Asian grocery. It’s rated GRAS (Generally Recognized As Safe) by the FDA. Spilling the electrolyte on a parking lot is comparable to dumping pickle brine: a cleanup nuisance, not a hazmat event.

The full engineering case

MgCl₂: FDA GRAS, FAO/WHO permitted food additive (E511), no acute or chronic toxicity at exposure levels relevant to a battery service event. CaCl₂: same regulatory profile. Disposal under standard waste protocols per Chinese GB 18599-2020 (the standard the Chen et al. cell qualified under), well below RCRA Subtitle C thresholds in the US. The Cu leached from the PBA cathode is the only material that requires recovery for both ecological and economic reasons; that’s addressed by Emerson’s Aqueous Electrolyte Stewardship Service (Rosemount inline ICP detection + electrowinning recovery, ~95% efficient).

You might ask

What happens if a cell or skid is punctured in the field?

In one sentence

Electrolyte leaks into a containment tray, the local cell stops working, and the rest of the skid continues operating — no fire, no evacuation, no cascading failure.

The plain-English version

A skid is a sealed system designed with a containment bund underneath. A puncture causes the affected module to disconnect and the electrolyte to drain into the bund where it can be pumped back into service or disposed of. The neighboring modules keep running because they’re electrically isolated. This is the opposite of lithium fires that propagate cell-to-cell across an entire container.

The full engineering case

Standard BESS containment architecture applies: each module sits over a leak-detection bund (Rosemount level switches), modules are electrically and hydraulically isolated via Fisher valves, breach triggers immediate module-level disconnect and bund-fill alert through DeltaV. Recovered electrolyte returns to service after particulate filtration (no decomposition products to remove, since nothing combusted). No flammable vapor cloud, no toxic gas release, no need for fire suppression infrastructure. The total “hazmat event” is comparable to spilling 100 gallons of brine on a slab.

A3 · The skeptics’ wall

I asked this

Aqueous batteries always fail — what about Aquion?

In one sentence

Aquion’s business-model failure mode is genuinely parallel to ours (aqueous, stationary-only, non-battery-native entrant, $/kWh-cycle pitch) — the chemistry was a different ion family, but the trap is real and deserves honest mitigation, not denial.

The plain-English version

A reviewer who has watched battery startups die will read this proposal and think: aqueous, stationary-only, low energy density, non-battery-native company, $/kWh-cycle pitch. That is exactly Aquion’s structure. We don’t win this argument by saying “different chemistry” — the chemistry differed (Na-ion vs Mg-ion), but the failure pattern transfers. Honest comparison: Aquion raised $190M and went Chapter 11 in March 2017 because LFP’s cost-down curve outpaced their own scale-up. Their cells worked fine; their unit economics didn’t. Three differences mitigate (not eliminate) this proposal’s exposure: (1) Phase 1 validates chemistry for $1.4–1.8M before any major capital commits — vs Aquion deploying $50M+ before their first commercial cell; (2) Plantweb subscription is a structurally different revenue surface that Aquion never had access to and LFP integrators still don’t; (3) projected 8,000–25,000 cycles at C/4 (rate-derated, see Chemistry section) versus Aquion’s ~3,000-cycle ceiling. The chemistry-difference defense is misdirection; the stage-gating and subscription-moat differences are the actual answer.

The full engineering case

Aquion architecture: Na⁺ in Na₂SO₄ aqueous electrolyte, Mn-oxide cathode, carbon anode; ~25 Wh/kg, ~3,000 cycle ceiling. Whitacre’s post-mortem identified the binding constraints: (a) capital burn rate vs LFP’s cost-down curve (LFP fell from $1,200/kWh in 2010 to $130/kWh by 2017); (b) low energy density drove BoS cost share to ~60% of installed capex; (c) no recurring-revenue moat — once sold, the cells generated no further margin to Aquion. Honest parallel: this proposal targets the same aqueous-stationary-only niche from a non-battery-native company. Honest differentiation: (1) Stage-gated capital — $1.4–1.8M Stage 0/1 commit before any product-build expense vs Aquion’s $50M+ pre-commercial; (2) Subscription-moat differential — Plantweb attach generates 30-year recurring margin on every deployed skid, structurally absent from Aquion’s flat hardware-sale model; (3) Cycle-life delta — projected 8,000–25,000 cycles at C/4 (rate-derated from Chen’s 120,000 at 20 A/g) vs Aquion’s ~3,000; at $/kWh-cycle the 3–8× ratio holds even at the conservative bound; (4) License-back floor — Phase 3 carries an explicit option to license IP and BoM to CATL/BYD if vertical scale-up returns negative under Phase 2 actuals, preserving 8-figure floor revenue that Aquion never structured. Risk 5 in the proposal (“The Aquion Trap”) names the parallel explicitly and treats this card’s four mitigations as the full response.

I asked this

What about Pellion and Toyota’s magnesium-ion programs?

In one sentence

Pellion and Toyota both worked on non-aqueous Mg-ion and hit a passivation wall that doesn’t exist in water-based chemistry.

The plain-English version

Both teams used organic solvents (like the ones in lithium batteries) with magnesium salts dissolved in them. The problem: magnesium deposits in organic solvents form a passivation layer on the anode — a kind of chemical scab that blocks further ion movement. Pellion (an MIT spinout) gave up and pivoted to lithium-sulfur, then shut down. Toyota’s program remains in the lab after more than 15 years. Chen et al.’s breakthrough was moving Mg-ion into aqueous neutral pH, where this passivation chemistry physically cannot happen.

The full engineering case

Pellion Technologies (2010–2018, MIT spinout, Gerbrand Ceder co-founder) and the Toyota Research Institute Mg program (decade+) both worked with Mg(TFSI)₂-type salts in glyme/ether solvents. Mg²⁺ deposition in these solvents forms a passivating layer (typically MgO/MgF₂/decomposition products) that blocks subsequent Mg ion transport — the classical non-aqueous Mg problem since the 1990s. Solutions attempted (Grignard-based electrolytes, halogen-free salts, novel additives) made progress in coulombic efficiency but never scaled. Chen et al.’s aqueous neutral-pH MgCl₂ system entirely sidesteps this failure mode: in water, Mg²⁺ shuttles between hydrated states without forming the passivating film. The Pellion/Toyota wall is structurally impossible in this chemistry.

I asked this

If this is so good, why hasn’t anyone done it before?

In one sentence

The specific combination of aqueous neutral-pH electrolyte + COP radical-anion (n-type) anode + Cu-doped Prussian Blue cathode wasn’t published until February 2026; nobody assembled these three pieces in this configuration before.

The plain-English version

All three components had been studied separately in other contexts. PBA cathodes were used in sodium-ion batteries (Natron Energy is scaling those now). COP polymers were explored as organic electrodes for various chemistries. Aqueous Mg-ion was attempted at low voltage. But this specific three-part recipe — CuFe-PBA + Hex-TADD-COP + neutral-pH MgCl₂ — was new. That’s what makes the Feb 2026 result legitimately first-of-kind, not a refinement of something existing.

The full engineering case

Each component has independent prior art: Na-ion PBA cathodes (Natron, Faradion); COP-based organic electrodes (multiple academic groups since ~2017); aqueous Mg-ion at sub-2V (various, none with strong cycle life). The novelty is in the system-level co-design: (1) Cu doping of the PBA framework specifically enables Mg²⁺ redox shuttling at 2.2V (uncommon for Cu-free PBAs), (2) the COP’s radical-anion (n-type) storage mechanism — reversible reduction of the polymer’s pyrazine sites, with Mg²⁺ coordinating — gives the 120,000-cycle stability, (3) the neutral pH avoids the dendrite and corrosion failure modes of acidic or alkaline aqueous variants. Each individual choice has prior art; the combination, and its 120,000-cycle result, do not.

I asked this

Where did this chemistry come from — is it a fluke of 2026?

In one sentence

It’s the payoff of a thirty-year lineage, not a fluke: aqueous rechargeable batteries (1994) → Prussian-blue electrodes (Stanford, 2011, 40,000 cycles) → organic-anode + Prussian-blue in aqueous magnesium (2017) → the finished cell (2026). The pieces are old and proven; 2026 is when they were finally assembled.

The plain-English version

Every part of this cell stands on a decade or more of prior work. In 1994 John Dahn showed you could build a rechargeable battery in plain salt water instead of flammable solvent. In 2011 Yi Cui’s Stanford lab turned Prussian blue — the blueprint pigment — into an electrode that cycled 40,000 times, and Prussian-blue analogues became the workhorse of water-based batteries. In 2017 a team paired an organic anode with a Prussian-blue cathode in aqueous magnesium — the exact recipe CityU would perfect, nine years early. So when two independent teams finished adjacent halves in February 2026, they were completing an arc, not getting lucky. That pedigree is what turns ‘a promising 2026 paper’ into ‘an independently verified field’ — and it’s why Phase 1 is a de-risked replication, not a leap into the dark. (The full arc is in the Reading Textbook, chapter 18.)

The full engineering case

Timeline with primary sources: (1) 1994 — Li, Dahn & Wainwright, Science 264:1115 — first aqueous rechargeable lithium cell; proved water can replace flammable organic electrolyte. (2) 2011 — Wessells, Huggins & Cui (Stanford), Nat. Commun. 2:550 — copper hexacyanoferrate aqueous electrode, 83% capacity over 40,000 cycles, the Prussian-blue-aqueous cornerstone; matured to a full open-framework cell in 2014 (Pasta/Cui, ncomms4007). (3) 2012 onward — the aqueous-zinc explosion (Xu/Kang, Angew. 2012). (4) 2017 — the direct architectural blueprint: an organic anode + Prussian-blue cathode in aqueous Mg²⁺ (ACS Energy Lett. 2:1115), nine years before Chen. (5) Feb 2026 — CityU publishes the finished cell; a mainland-China consortium publishes the cobalt self-healing cathode three days later. Field maturity: aqueous batteries broadly are industrializing (Natron, Amazinc); the aqueous-divalent-organic branch is young, fast-accelerating, and pre-commercial — Emerson would enter at the frontier, not chase a crowded field.

You might ask

Has Chen’s specific cell been independently replicated?

In one sentence

Not yet for the specific COP+CuFe-PBA+MgCl₂ system; the broader aqueous Mg-ion platform was corroborated by Wu et al. (BUCT/PKU Shenzhen, JACS Feb 2026) using a different anode, validating the platform but not the exact cell.

The plain-English version

There is one other Feb 2026 paper from a Chinese university group showing 75,000 cycles in aqueous magnesium-ion using a different anode chemistry (Ta-doped MoO₃). That confirms aqueous Mg-ion can run ultra-long-life cycles in principle, but doesn’t replicate Chen’s specific combination. Phase 1 of this proposal is exactly the work to become the second lab in the world that has validated Chen’s cell at grid-relevant rates.

The full engineering case

Wu et al. (BUCT/PKU Shenzhen, JACS 2026, doi:10.1021/jacs.5c21656): 75,000 cycles in aqueous Mg-ion with Ta-doped MoO₃ nanotube anode — different electrode pair than Chen, zero author overlap with the CityU group. This corroborates the broader aqueous Mg-ion platform but not Chen’s specific COP+CuFe-PBA combination. A third, independent February-2026 result — the cobalt-doped self-healing copper-hexacyanoferrate cathode (JACS 2026, doi:10.1021/jacs.5c20391; a mainland-China consortium — Inner Mongolia UST, CAS-SINANO, Tongji — with zero overlap with either CityU or the Wu group, published within three days of Chen) — corroborates the cathode half from yet another direction, and is the enabling science behind the Phase-2 “capacity mode” fork. Three unrelated labs converging on adjacent pieces of the same chemistry within a single month is the signature of a field reaching maturity, not a lone claim. Phase 1 of this proposal is designed to be the second independent replication of Chen’s specific cell at C/4 (grid-relevant rate), not just the 20 A/g extreme. If Phase 1 holds, Emerson owns the only validated cell of this type outside CityU.

A4 · Performance specifics

You might ask

How long do these last in years?

In one sentence

15–30 years per skid — bounded by calendar aging, seal degradation, and balance-of-plant MTBF, not by cycle count; the cycle envelope from Chen et al. (rate-derated to a projected 8,000–25,000 grid-rate cycles) is comfortably ahead of every other failure mode.

The plain-English version

Cycle life is theoretical math. In practice, the cells themselves can run far longer than the surrounding equipment (frame, plumbing, electronics, electrolyte seals), so the design target is permanent infrastructure on a 15–30 year horizon with periodic service. That’s 2–3× the life of a typical lithium grid skid, which gets fully replaced every 8–10 years.

The full engineering case

Cycle envelope (rate-derated): projected 8,000–25,000 cycles at C/4 + 80% DoD (per the “Chen ran 120,000 cycles at 20 A/g” card), which at one cycle/day yields 22–68 years of pure cycle life. Cycle-count is not the binding limit; calendar fade is. Real-skid life is bounded by: (a) electrolyte calendar drift (CO₂ ingress, water activity shift; ~20–30 yr design with sealed make-up); (b) seal degradation on primary stainless containment (~20–30 yr); (c) balance-of-plant electronics MTBF (BMS, inverters refresh at Year 10 and Year 20); (d) economic depreciation. The proposal’s 30-year operational target aligns with Emerson’s general infrastructure-product life span (Rosemount transmitters routinely 30+ years). Service intervals: electrolyte top-up + Cu recovery at Year 10 and Year 20; cathode refurbishment via aqueous re-dissolution + re-precipitation at Year 20–25; full skid refurb or second-life redeployment at Year 25–30.

You might ask

How fast can they charge and discharge?

In one sentence

Chen et al. published at 20 A/g (very fast); grid stationary storage operates at C/4 to 1C (much slower), and Phase 1’s job is to confirm cycle life holds at those grid-relevant rates.

The plain-English version

The published paper used extreme charge rates for accelerated testing. Grid storage doesn’t need that — it needs to dispatch over 4 to 8 hours. The open question Phase 1 answers: does the 120,000-cycle life still hold when the cells are run at gentler 4-hour discharge rates the way they’d actually be deployed? Industry expectation is yes (slower is usually easier on a cell, not harder), but it must be measured.

The full engineering case

20 A/g is a very high C-rate, typically used for accelerated cycling because it compresses test time. Grid-stationary deployments target C/4 to C/2 (4-hour to 2-hour discharge). Phase 1 explicitly tests C/10 through 1C to map performance across the deployment envelope. The hypothesis: slower rates reduce ohmic stress and electrode strain, so cycle life at C/4 should equal or exceed the 120,000-cycle benchmark, with somewhat higher energy efficiency. If the cell fails this test (cycle life craters below ~5,000 at C/4), Phase 1 ends at the kill-gate and Emerson has spent $1.4–1.8M to learn that, against a $1.7B median-case revenue upside.

You might ask

What about energy density — how does that compare?

In one sentence

~48 Wh/kg electrode-level (~8–15 Wh/kg at the full skid) — about 1/3 to 1/5 of LFP at the skid level; the customer pays for that delta through footprint, slab, and shipping, and our $/kWh-cycle math accounts for it.

The plain-English version

Energy density doesn’t appear in the customer’s bill directly, but it shows up indirectly as footprint, slab tonnage, shipping mass, and electrolyte capex. A 3 MWh aqueous skid is roughly 2.5× the footprint and ~1.5× the slab cost of a Tesla Megapack at the same capacity. We don’t hide that delta; we beat it on $/kWh-cycle. Over 8,000–25,000 grid-rate cycles at 80% DoD, the LCOE math (see Section C) still shows a 3–8× advantage on lifetime delivered energy. The trade is real, the customer math is favorable, and the only application this chemistry is categorically wrong for is vehicles — which is why it’s stationary-only by design.

The full engineering case

Electrode-level: 48.3 Wh/kg per Chen et al. Cell-level (packaging, current collectors, electrolyte): ~20–30 Wh/kg. Skid-level (honest): 8–15 Wh/kg accounting for the ~15–30 t of aqueous electrolyte mass per 3 MWh module — the binding cost driver at scale. LFP cell-level ~150 Wh/kg; LFP skid-level ~50–80 Wh/kg. VRFB system-level: 15–25 Wh/kg — the incumbent long-duration grid chemistry sits in the same density band as aqueous Mg/Ca-ion and the market prices it at $300/kWh anyway, because stationary grid customers don’t buy Wh/kg. LFP is the right density benchmark only for vehicles; for the market this chemistry actually serves, VRFB is the parity comparison. For a 3 MWh stationary skid: LFP container ~10–15 tons; aqueous Mg-ion equivalent ~30–50 tons. Customer-side cost flow-through: footprint (~2.5× LFP), slab tonnage (~1.5×), one-way shipping mass (~3×), electrolyte capex (linear with mass). These offsets get explicit treatment in the $/kWh-cycle math (Section C card “What’s ‘$/kWh-cycle’…”), where the 8,000–25,000 projected cycles at C/4 + 80% DoD yield $0.018–$0.020/kWh-cycle LCOE compared to LFP at $0.078/kWh-cycle — a 3–4× advantage at the conservative cycle-life bound, 13× at the thesis bound. Energy density is the cost; cycle life is the offset; the trade is positive under any defensible Phase 1 outcome that meets the kill-gate floor.

You might ask

What’s the round-trip efficiency?

In one sentence

Projected 70–80% round-trip efficiency — comparable to LFP, significantly better than iron-air’s ~40–50%, and decisive for daily-cycle applications where every percent compounds.

The plain-English version

If you put 100 kWh in, you get 70–80 kWh back out — the rest is lost as heat. Lithium-iron-phosphate is similar. Iron-air loses about half because it has to convert chemicals through an oxygen-breathing reaction. For daily charge-discharge applications, those efficiency differences compound enormously over 30 years; for long-duration backup (where the battery sits charged for days waiting for a storm), efficiency matters less.

The full engineering case

Projected ~70–80% DC round-trip efficiency at C/4 based on the chemistry’s low overpotential (intercalation-style cathode, radical-anion anode — both relatively reversible processes). Voltage hysteresis target: <250 mV at mid-SOC. Measured value from Chen et al. paper is not directly comparable due to the 20 A/g test rate; Phase 1 will measure RTE at C/10 through 1C as part of the validation campaign. LFP comparison: 85–92% RTE typical. Iron-air: ~40–50% due to round-trip oxygen evolution / reduction thermodynamics. For 4–8 hour daily-cycle stationary, aqueous Mg-ion is structurally competitive with LFP and structurally dominant over iron-air.

I asked this

What’s coulombic efficiency, and why did you tier the threshold instead of using one number?

In one sentence

Coulombic efficiency (CE) is the fraction of electrons that come back out of the cell on discharge versus what went in on charge — a single fixed threshold ignores that formation cycles always run dirty and would euthanize cells the field would have accepted.

The plain-English version

Every electron that enters a battery during charge should come back out during discharge. The ratio is CE, and it should approach 100%. But the first ~50 cycles always look messy because the cell is settling: solid-electrolyte interphase forming, electrolyte wetting the porous electrodes, hardware degassing. A blanket “CE ≥99.5% from cycle 1” rule would kill cells that recover to 99.95% by cycle 200 and then run forever. So Phase 1 tiers the threshold: ≥97% during formation (cycles 1–50), ≥99.0% during break-in (50–200), ≥99.5% in steady-state. That mirrors how cycler engineers actually evaluate cells.

The full engineering case

CE = Q_discharge / Q_charge per cycle. In a Mg²⁺ aqueous system the formation phase consumes coulombs for (a) wetting and saturating COP anode radical-anion sites, (b) forming an SEI-analogue at the PBA/electrolyte interface, (c) initial gas evolution and outgassing from separator and hardware. Chen et al.’s own published data show CE climbing from ~96% at cycle 1 to >99.95% by cycle 500 and holding through cycle 120,000. Phase 1 kill-gate (Protocol §3.4) uses the same tiered envelope; a cell failing the steady-state floor is pulled for forensic post-mortem (FT-IR ATR on COP, ICP on PBA leach, XRD on harvested electrodes). A blanket 99.0% gate from cycle 1 would have euthanized Chen’s own cells during their first 30 cycles — methodologically wrong.

I asked this

Why split the cycle-life claim into “demonstrated” and “projected”?

In one sentence

“Demonstrated” means real cycles run on a real cell; “projected” means the chemistry’s published cycle count extrapolated to our skid geometry — keeping them separate is the only honest way to talk about a battery before you’ve built one at scale.

The plain-English version

Chen et al. ran 120,000 cycles on a 0.5 cm² coin cell at high current. That’s demonstrated. Phase 1 will run a few hundred cycles on bigger pouches at realistic C-rates — also demonstrated. The 120,000 number does not automatically transfer to a multi-MWh skid; geometry, current distribution, edge effects, and 30-year calendar aging all change the answer. The proposal cites both numbers, labels which is which, and treats the projection as a thesis to be tested — not a sales claim. Claim 7 in the proposal was retitled “Inherent Safety Now, Lowest-Carbon Bet for the Second Decade — a Testable Thesis” precisely to remove the implication that scale is guaranteed.

The full engineering case

The proposal splits the cycle-life claim three ways. Demonstrated (literature): 120,000 cycles at 20 A/g, 0.5 cm² coin cell, CityU lab (Chen et al., Nature Communications Feb 2026). Phase 1 demonstrated (target): ≥500 cycles at C/10 through 1C on pouches ≥1 Ah capacity, with the tiered CE envelope above and capacity retention ≥90% at the kill-gate review. Projected commercial: retention curve extrapolated from formation + break-in + steady-state CE measurements crossed with calendar-aging models from accelerated thermal/voltage stress tests. The third number is the only one used for $/kWh-cycle (LCOE) economics; it is explicitly bracketed as a thesis dependent on Phase 1 results and re-baselined after Phase 1 actuals. This split is also why “What We Have Not Yet Proven” sits as its own subsection in the Storage Landscape — reviewers see exactly which claims are demonstrated and which are still in front of us.

You might ask

Chen ran 120,000 cycles at 20 A/g — how does that translate to C/4 grid rate?

In one sentence

The headline 120,000 cycles was at lab-rate 20 A/g (~200–400 C, orders of magnitude above grid C/4); Tafel-corrected extrapolation to grid-realistic C/4 projects 8,000–25,000 cycles at 80% retention — still 2–6× LFP and the actual basis for the $/kWh-cycle math, NOT the 120,000 headline.

The plain-English version

The sharpest reviewer question is: “you cite 120,000 cycles, but that was a coin cell at 20 amps per gram. How does it translate to a 3 MWh skid running at a 4-hour discharge rate?” Honest answer: it doesn’t translate directly. Polarization at higher current accelerates the chemistry’s natural fade mechanisms; at lower current (C/4 vs 20 A/g), polarization is much smaller, so calendar aging becomes the dominant fade mode instead of rate-driven mechanical stress. Tafel extrapolation of Chen’s published per-cycle fade slope, corrected for reduced polarization and increased calendar time, projects 8,000–25,000 cycles at C/4 and 80% retention — still 2–6× LFP’s 3,000–6,000 cycles at the same retention floor. The $/kWh-cycle economics in Section C use the LOWER bound (8,000), not the headline 120,000. The Phase 1 kill-gate measures C/4 data on pouches and answers which end of the projection is closer to truth.

The full engineering case

Chen et al. 2026 protocol: ~0.5 cm² coin cell, active-mass loading ~2–5 mg/cm² PBA (estimated from TEM + binder ratio), 20 A/g charge + discharge to 1.5 V cutoff, 120,000 cycles, 81.2% retention at endpoint (mean steady-state fade ~0.0156% per 100 cycles). Translation to C/4: Tafel-corrected per-cycle fade = (rate factor 0.3–0.5 for reduced polarization-driven PBA lattice + COP backbone stress) × (calendar factor 1.5–3.0 for increased time at full SOC per cycle) ≈ 0.004–0.012% per 100 cycles at C/4. At 80% retention floor: 8,000–25,000 cycles projected at C/4 / 80% DoD. Phase 1 kill-gate: 500 cycles measured on ≥1 Ah pouches at C/4 + 80% DoD; kill if measured fade slope >2σ above the 0.012% upper bound. Honest framing: 120,000 = laboratory benchmark; 8,000–25,000 = the projection that funds the $/kWh-cycle economics; Phase 1 tells us which bound is closer to truth. Comparison: LFP at 80% retention = 3,000–6,000 cycles (Argonne ANL benchmark, Crenna et al. 2021).

You might ask

Does this battery grow dendrites like lithium and zinc do?

In one sentence

No — aqueous Mg-ion thermodynamics actually prevents Mg-metal plating at the anode (hydrogen evolution wins), which eliminates the dendritic short-circuit failure mode that plagues lithium, sodium, and zinc batteries; it’s a structural safety feature, not a workaround.

The plain-English version

Lithium batteries fail catastrophically when the anode grows microscopic metal “dendrites” — tree-like spikes that pierce the separator and short the cell, creating the internal heat that starts thermal runaway. The same problem plagues sodium and zinc batteries at lower energy density. In aqueous Mg-ion, this failure mode is thermodynamically impossible: at the operating voltage, hydrogen evolution from water reduction wins the competition against Mg-metal plating, so no Mg° ever forms at the anode. The COP anode itself is an organic radical-anion (n-type) host, not a metal-deposition surface. The dendrite shorting pathway that kills the lithium, sodium, and zinc families is structurally absent here.

The full engineering case

Half-cell thermodynamics at the Hex-TADD-COP anode operating window (~0 V vs Ag/AgCl, pH 7): (a) Mg²⁺ + 2e⁻ → Mg°, E° = −2.37 V vs SHE; (b) 2H₂O + 2e⁻ → H₂ + 2OH⁻, E° = −0.83 V vs SHE at pH 7. The hydrogen evolution reaction wins by 1.54 V; in 4.0 M MgCl₂ (decreased water activity but still abundant), HER remains overwhelmingly favored. No Mg° forms; no dendritic morphology; no internal short pathway. COP anode mechanism is N-centered radical-anion storage — reversible n-type reduction of the pyrazine site (Hex-TADD-N + e⁻ → Hex-TADD-N^−•), charge-balanced by Mg²⁺/Ca²⁺ coordination, not metal deposition. Independent confirmation: Wu et al. (BUCT/PKU, JACS Feb 2026) with Ta-doped MoO₃ anode shows the same absence of Mg-plating in aqueous Mg²⁺ at neutral pH. Comparison to incumbent failure modes: Li-ion dendrites (cause of UL 9540A propagation testing failures); Na-ion dendrites (slower formation, same end state); Zn aqueous (well-documented dendrite-driven shorting, the main reason commercial Zn remains primary-cell not rechargeable). Aqueous Mg²⁺ sidesteps this entire failure mode class.

You might ask

Whose prior work is this building on?

In one sentence

Six prior bodies of work meet at this cell — water-in-salt electrolyte (Suo 2015), Mg-metal-anode chemistry (Aurbach group), aqueous Mg-ion PBA cathodes (Wang group), aqueous Na/K-ion PBA at scale (Ji group, Natron, Faradion), radical-polymer organic electrodes (Schon & Tilley), and stationary-storage LCA (NREL / PNNL) — Chen et al. is the integration, not a from-scratch discovery.

The plain-English version

A reviewer asking “is this really new?” deserves the honest map of prior work. Six bodies of literature converge at the cell Chen et al. published in February 2026: (1) Water-in-salt electrolyte — Suo, Borodin, Xu et al. Science 2015 established that extreme salt concentrations widen water’s electrochemical stability window; this is the foundational reference for any pH-7 high-concentration aqueous claim. (2) Multivalent Mg chemistry — Doron Aurbach’s group at Bar-Ilan University has been the canonical source on Mg-electrolyte interfaces since the 1990s; their work explains why non-aqueous Mg-metal anodes passivate and why aqueous Mg-ion sidesteps that failure mode. (3) Aqueous PBA cathodes — the Wang group at U Maryland and the Ji group at Oregon State established Prussian-Blue-analog cathodes in aqueous Zn / Na / K systems (~2017–2022); Natron Energy and Faradion scaled PBA Na-ion to commercial deployment. (4) Mg-ion in PBA frameworks — multiple academic groups (Whittingham at Binghamton, Cabana at Illinois Chicago) demonstrated Mg²⁺ intercalation in PBA at sub-2V before Chen’s 2.2V result. (5) Radical-polymer organic electrodes — the Schon, Tilley, and Nishide groups (Imperial College, Wisconsin, Waseda) built the prior art for TEMPO / nitroxyl / radical-cation polymer electrodes; the Hex-TADD-COP anode descends from that lineage. (6) Stationary-storage LCA — NREL ATB, PNNL Energy Storage Technology Center, EPRI BESS reports, and Argonne ANL benchmarks frame the cycle-life and lifetime-carbon math. Chen et al.’s contribution is the system-level integration of these six bodies into one cell that holds 120,000 cycles — not a discovery from zero.

The full engineering case

Foundational references: Suo, Borodin, Xu et al. “Water-in-salt” electrolyte enables high-voltage aqueous lithium-ion chemistries, Science 347, 938–943 (2015) — foundational WiSE concept. Aurbach group, Bar-Ilan U — the canonical body on Mg-electrolyte interfaces and Mg-metal anode chemistry; key reviews 2000–2020 establish the non-aqueous Mg passivation problem this aqueous chemistry sidesteps. Wang group, U Maryland — aqueous WiSE Zn / Na / Mg PBA cathodes (multiple Nat. Mater. and JACS 2017–2019). Ji group, Oregon State — aqueous K-ion / Na-ion PBA cathodes; Mn / Cu / Fe-hexacyanoferrate stability in aqueous halide. Whittingham (Binghamton) + Cabana (UIC) — prior Mg²⁺ in PBA below 2V. Schon, Tilley, Nishide groups — TEMPO and nitroxyl radical polymer electrodes; structural ancestors of Hex-TADD-COP. Natron Energy — PBA Na-ion commercial deployment; the empirical proof that PBA cathodes hold up in aqueous service at scale (5,000+ cycles in commercial BESS service). Faradion — PBA Na-ion patent family + scale-up trajectory. NREL ATB / PNNL ESTC / Argonne ANL — stationary-storage LCA benchmarks (Crenna et al. 2021; Dai et al. 2019). EPRI BESS reports / NFPA 855 / UL 9540A — fire safety, indoor-deployment envelope, and propagation-testing standards. Chen et al. 2026 + Wu et al. 2026 sit at the integration point of all six lineages; this proposal stands on the same shoulders and cites them explicitly.

Why does the battery run on BOTH magnesium and calcium?

In one sentence

Chen 2026’s cell stores both — Ca²⁺ actually inserts more readily than Mg²⁺ into the CuFe-PBA cathode, and both share the same storage sites; the dual-cation electrolyte is a feature, not a compromise.

The plain-English version

Chen et al. 2026 published the cell in an MgCl₂ + CaCl₂ electrolyte at pH 7 — the calcium is not an accident, it is part of the working chemistry. Both divalent cations target the same storage sites in the copper-iron Prussian-blue cathode, and both are hosted reversibly by the Hex-TADD-COP anode. Calcium is actually the better-behaved of the two: higher selectivity for the cathode framework, faster kinetics, lower desolvation penalty. The electrolyte is intentionally dual-cation and the two divalents share the work.

The full engineering case

The CuFe-PBA cathode framework preferentially inserts divalent cations at the “center sites” (Cu redox); monovalents populate different corner sites (Fe redox). Measured PBA ion-selectivity sequence (Tinker et al. 2024, Small Structures sstr.202400317; “Selectivity mechanisms,” Matter 2025 S2590-2385(25)00618-6): Rb⁺ > K⁺ > Na⁺ > Ba²⁺ > Sr²⁺ ≈ Ca²⁺ > Mg²⁺ > Li⁺. Within the divalent target couple, Ca²⁺ intercalates more readily than Mg²⁺ and carries faster kinetics (lower desolvation penalty). Chen 2026 used 4.0 M MgCl₂ and saturated 5.8 M mixed MgCl₂/CaCl₂ (Nat. Commun. s41467-026-69384-2); the mixed-divalent electrolyte is on the paper. Concentrated CaCl₂ is also a stronger water-activity depressant than MgCl₂ — the mixed electrolyte keeps (or improves) the water-in-salt window advantage that makes the whole 2.2 V cell work.

Isn’t calcium an impurity you have to remove?

In one sentence

No — this is the keystone flip. The measured ion-selectivity sequence shows the real poisons are the monovalents (Na, K), not calcium; aqueous Mg/Ca-ion keeps Mg + Ca and rejects Na/K, deleting the industry’s hardest single purification step.

The plain-English version

Every conventional Mg-ion battery treats calcium as a chemical twin that has to be separated from magnesium — the “hardest step” in battery-salt purification, adding ~$400–600/ton. But the measured selectivity sequence for CuFe-PBA cathodes is Rb⁺ > K⁺ > Na⁺ > Ba²⁺ > Sr²⁺ ≈ Ca²⁺ > Mg²⁺ > Li⁺. The monovalents (Na, K) outcompete the divalents for the cathode’s storage sites — they are the site-blocking poisons. Ca²⁺ sits above Mg²⁺ in the sequence, meaning it inserts productively alongside Mg. So the correct electrolyte spec isn’t “separate Ca from Mg.” It’s “keep the divalents, reject the monovalents.” That is a cheaper, standard separation — NaCl and KCl crystallize out first from mixed brine and stay in the mother liquor while the divalent chlorides are collected. Same physical chemistry every salt-works uses to split NaCl from Mg/Ca bitterns.

The full engineering case

Practical implications for Track B purification: (a) monovalent reject via fractional crystallization (NaCl/KCl crystallize first from mixed-salt bittern), (b) Ba/Sr reject via sulfate precipitation (BaSO₄/SrSO₄ highly insoluble; captured as barite/celestine solids, sold via drilling-mud market per API spec — see Beyond the Battery §02), (c) transition-metal reject via chelating ion exchange, (d) no Ca/Mg separation stage — the two divalents run together as the productive electrolyte. Net purification cost drops ~25–40% by deleting the process chemistry (oxalate precipitation + Ca-selective chelating resin + extra recrystallizations) with the largest scale-up risk. Contamination-tolerance thresholds per ion are set empirically by Phase-1 Gate 4 (the deliberate contamination rig: cells spiked with Na/K/Ba/Sr at controlled ppm, capacity retention measured, purification spec defined). Sources: PBA selectivity sequence — Tinker et al. 2024, Small Structures (sstr.202400317); “Selectivity mechanisms,” Matter 2025 (S2590-2385(25)00618-6).

What actually carries the charge — magnesium or protons?

In one sentence

Honest: it is a live literature debate. A 2021 paper argues protons carry the charge in aqueous PBA cathodes (not Mg²⁺); Chen 2026 claims divalent intercalation. Phase 1 confirms via ex-situ ICP + XRD + XPS at charged and discharged states — and either answer supports the mixed-divalent electrolyte.

The plain-English version

A sharp reviewer will ask: are you sure it is the magnesium moving, not just protons from the water? A 2021 paper (PubMed 33512766) argues that Mg²⁺ does not actually insert into aqueous Prussian-blue-analogue cathodes — protons carry the charge and Mg²⁺ shuttles in the electrolyte only. Chen 2026’s Nature Communications paper argues the opposite: that divalents genuinely intercalate and drive the 2.2 V, 120,000-cycle result. This is real scientific uncertainty, and Phase 1 is designed to settle it. Ex-situ ICP (elemental composition of the harvested electrodes at charged and discharged states) + XRD (lattice signature of divalent vs proton insertion) + XPS (surface oxidation states) — three independent techniques that produce a definitive answer at low cost. Either result matters. If Mg²⁺ is the carrier, the mixed-divalent bet is exactly right. If protons are the carrier, cation identity barely matters — the mixed-divalent bet becomes more justified because any cheap divalent brine will do.

The full engineering case

Charge-carrier ambiguity is a real methodological gap in the aqueous PBA-cathode literature. The 2021 counter-interpretation (Lipson, Cabana et al.) argues that “Mg²⁺” cycling in aqueous PBA at neutral pH reflects proton insertion (H⁺ from water dissociation) with Mg²⁺ acting as a spectator ion. Chen 2026 argues genuine divalent intercalation and cites capacity retention across a pH-4.91–7.02 window over cycling as inconsistent with pure proton-coupling — but the paper does not perform the ex-situ ICP measurement that would settle it directly. Phase-1 Gate 4 protocol: (1) coin cells charged and discharged, disassembled in argon glovebox, cathode harvested; (2) ICP-OES of the harvested cathode confirms Mg/Ca elemental ratio changes between charge states; (3) XRD confirms lattice-parameter signature (divalent vs proton insertion produce distinguishable framework distortions); (4) XPS of the surface confirms redox-couple activity. Combined, these settle the carrier question. Strategic implication: if the carrier is proton-coupled, the aqueous Mg/Ca electrolyte still works (charge shuttling is proton-coupled but the cell cycles) — and the mixed-divalent bet’s argument shifts from “we skip Ca/Mg separation because Ca is productive” to “cation ratio is basically irrelevant, use any divalent brine you can source.” That is if anything a stronger argument for the mixed approach. Owning this question pre-empts the sharpest reviewer.

What if the mixed-valent electrolyte doesn’t co-cycle?

In one sentence

It is a ladder, not a cliff: mixed → Ca-rich → pure-Ca → pure-Mg, in rising cost. No single point of failure. See If the Battery Fails §01 for the full map.

The plain-English version

Chen 2026’s published cell is the mixed cell (mixed MgCl₂+CaCl₂). So the anchor rung isn’t “the mixed-valent bet” — it’s “does Chen’s mixed reproduce at grid-relevant C-rate and our cell geometry?” If yes, the frontier (running that proven mixed cell on already-mixed brine, skipping Ca/Mg separation) is unlocked. If reproduction at grid rate shows unexpected behavior, the ladder steps down. Rung 2: go Ca-rich — Ca is the better-behaved divalent (higher selectivity, faster kinetics than Mg), and a Ca-dominant electrolyte still avoids the hard Mg/Ca separation. Rung 3: go pure calcium — standalone Ca cycle life is a plausible bet (selectivity + kinetics argue for it) but currently unproven; Phase 1 tests it as a first-class outcome in the parallel ratio series. Rung 4: go pure magnesium — pricier (adds Ca/Mg separation ~$400–600/ton) and standalone Mg cycle life is also unproven at Phase-1 rates; the expensive fallback, not the only option. Every rung is a candidate with known costs; only Rung 1 (mixed) has Chen’s coin-cell data behind it. The Phase-1 parallel test measures all three simultaneously.

The full engineering case

Rung 1 (mixed Mg/Ca): the target thesis — measured selectivity sequence supports co-cycling; Chen 2026 demonstrated the aqueous chemistry works with mixed 4.0/5.8 M MgCl₂+CaCl₂ electrolyte. Phase-1 Gate 3 measures ratio-dependent cycling (100:0 / 75:25 / 50:50 / 25:75 / 0:100 Mg:Ca) against the tachydrite phase-diagram constraint — CaCl₂·2MgCl₂·12H₂O precipitates above ~22°C in Mg-rich regimes; the working composition + temperature stays outside that field (Kurnakov 1938 through present). Rung 2 (Ca-rich): if Mg under-utilizes in mixed, Ca-dominant runs on Ca kinetics — retains cost advantage. Rung 3 (pure Ca): if mixed truly does not cycle at any ratio, pure CaCl₂ from Solvay-process waste (world Solvay soda-ash produces ~2 kg CaCl₂ byproduct per kg Na₂CO₃; currently landfilled or discharged) — the cheapest single-cation option; skips Mg entirely. Rung 4 (pure Mg): fallback to published Chen chemistry with Ca/Mg separation added to Track B (oxalate precipitation + Ca-selective chelating resin: Lewatit TP-260 or Dowex M4195); ~$400–600/ton added purification cost; still profitable, still non-flammable, still 120,000-cycle target. Every rung a known chemistry with a known cost — the fallback ladder pattern is the whole reason $1.8M can be sized as a call option rather than a bet. Full stage-by-stage resilience: If the Battery Fails.

Where does the calcium come from — and is it sustainable?

In one sentence

Two proven US commercial routes (TETRA Technologies limestone+HCl for battery-relevant purity + OxyChem’s Ludington MI brine at commodity scale) plus desal-brine circular sourcing at Phase 4. Calcium’s supply is more resilient than magnesium’s single-vendor Nedmag story.

The plain-English version

CaCl₂ has a two-track sourcing story that mirrors and improves on the magnesium side. Track A-Ca (prove on known-good, battery-relevant purity): TETRA Technologies makes CaCl₂ from food-grade HCl + high-purity limestone with, in their own words, “very low alkali metals, iron, and other impurities” — explicitly battery-relevant. TETRA is the Nedmag-of-calcium: a US specialty producer already making the grade this chemistry needs. TETRA also makes calcium bromide clear-brine fluids, which ties straight into the bromine byproduct stream. Alternatively, OxyChem (Occidental) — the world’s largest CaCl₂ producer — runs Ludington, MI brine at scale. Domestic. Track B-Ca (cheap / domestic / circular): industrial CaCl₂ (OxyChem brine, abundant) purified in-house on the same Track-B skid + the desal-brine circular source at Phase 4 (the same plant pulls Ca alongside Mg). The key takeaway: multiple large domestic US producers already make food/battery-grade CaCl₂ — a more resilient supply than Mg’s single-vendor (Nedmag) story on the Mg side.

The full engineering case

Track A-Ca (Nedmag-parallel, battery-relevant purity): TETRA Technologies (onetetra.com/industrial-chemicals/calcium-chloride) — CaCl₂ made from food-grade HCl + high-purity limestone, explicitly marketed for low-impurity applications; the closest US analog to Nedmag’s battery-grade Mg supply. OxyChem / Occidental (oxychemcalciumchloride.com) — largest CaCl₂ producer globally, Ludington MI brine, food-grade and industrial-grade lines. Both domestic, both scale, both immediately qualifiable as Phase-1 feedstock. Track B-Ca (cheap + circular): commodity OxyChem brine purified in-house on the Track-B skid (the same monovalent/Ba/Sr/transition-metal reject flow works for Ca or Mg identically), and desal-brine circular sourcing at Phase 4 (Ca²⁺ ~0.4 g/L in seawater → ~0.8 g/L in reject brine → ~150 metric tons Ca/day at Carlsbad scale, discharged today, alongside the 492 t/day of Mg). Honest correction: Solvay-process byproduct CaCl₂ is real chemistry (~2 kg CaCl₂ per kg Na₂CO₃; the Solvay soda-ash industry produces ~30 Mt/yr) but a source we can’t confirm as a commercial high-purity supply — keep it as “a circular source to explore”, not a headline. Supply resilience: lead with TETRA + OxyChem + desal. Ca has no supply-chain vulnerability, no export controls, no critical-mineral-list dependency — a stronger supply story than Mg (which depends on Nedmag’s single Veendam facility or NM solar evaporation with 1,500-mile trucking). This is why the fallback ladder’s Rung-3 pure-Ca path is not just viable — it is arguably the safer commercial beachhead.

Does the battery degrade, or is it permanent?

In one sentence

It degrades — exceptionally slowly (120,000 cycles at >80% retention, vs LFP’s 4–6k) — but it is not permanent, and we won’t claim it is. The stronger truth: the cells age, but the closed-loop material recovery means Cu/Mg/Ca don’t leave; the battery is finite, the material loop is effectively permanent.

The plain-English version

Yes, it degrades — just exceptionally slowly. The real fade paths: copper slowly leaches from the CuFe-PBA cathode (the same phenomenon we recover as revenue), the organic COP anode slowly dissolves into the aqueous electrolyte, and the electrolyte pH drifts (from 7.0 down to ~4.9–7.0 over cycling, per Chen 2026 itself — so it does move) with slow water consumption. 120,000 cycles at >80% capacity retention is extraordinary compared to LFP’s 4–6k, but it is finite. The stronger truth than “permanent” is this: the cells age, but because of the closed-loop recovery, the materials don’t leave. Copper, magnesium, calcium all get recaptured and recycled at end-of-life. The battery isn’t permanent; the material loop effectively is. That ties degradation straight to the zero-discharge ethos — the atoms stay in the system, perpetual even as cells wear.

The full engineering case

Documented and hypothesized fade mechanisms: (a) Cathode (CuFe-PBA): Cu leaching from framework over cycling — the same phenomenon our electrowinning + closed-loop recovery converts to revenue (~99.5% dendritic Cu powder, ~$13/kg). Also lattice strain from repeated divalent insertion (Ca²⁺ at 1.00 Å is larger than Mg²⁺ at 0.72 Å — mixed cycling may stress the framework more than single-cation; see next card). (b) Anode (COP): slow organic dissolution of the covalent organic polymer into the aqueous electrolyte; addressable by cross-linking chemistry, unproven at 30-year timescales. (c) Electrolyte: pH drift documented in Chen 2026 (7.0 → 4.91–7.02 post-cycling), water consumption from small 2.2 V overpotential-driven electrolysis. All three fade paths are measured by Phase 1’s capacity-retention + Coulombic-efficiency + ex-situ characterization protocol. The 120k-cycle number is at 20 A/g extreme current; C/4 grid-relevant retention is Phase 1’s headline measurement. Honest posture: extraordinarily long-lived, not permanent. The material loop is effectively permanent because Cu/Mg/Ca all recycle at end-of-life; the cells themselves are finite. Both statements are true; only one is safe to lead with.

When a cell wears out, is it scrapped or restored — and how do you know exactly what to top up?

In one sentence

Restored, not scrapped: the copper, cobalt, and brine are recovered (designed for >90%) and the cathode is re-precipitated from its own metals, so you top up only single-digit-percent losses — and ICP + inline Rosemount analytics read the exact per-metal inventory, so every top-up is measured, not guessed. The skid frame and electronics are 30-year infrastructure.

The plain-English version

A worn cell isn’t a brick — it’s a bag of recoverable ingredients. The dissolved copper (and, in the capacity variant, cobalt) gets electrowon back; the Mg/Ca chloride brine gets cleaned and reused; and the cathode itself is re-dissolved and re-precipitated — you rebuild the Prussian-blue crystal from its own recovered metals. Because recovery is high, the “new materials” bill at restoration is small: single-digit-percent top-ups of the metals, a little brine, and (the most likely wholesale refresh) the cheap commodity organic anode. What can’t be done in a garage: the cathode rebuild + cell reassembly needs a facility. But because these cells are non-flammable and the electrolyte is non-hazardous, spent modules ship safely and cheaply — no hazmat, no fire risk — so the practical model is swap-and-return: swap the module in the field (home, riverside, mountainside), the spent one rides back to a regional hub for re-precipitation, and it re-enters the fleet. The opposite of lithium’s shred-smelt-landfill. And you know precisely what to top up because you can read the water: ICP (inductively-coupled plasma) analysis measures every dissolved metal — copper, cobalt, iron, magnesium, calcium — to parts-per-billion, and Rosemount inline sensors trend the electrolyte continuously. The same Emerson instruments that run Phase 1 meter the 30-year material inventory.

The full engineering case

What degrades (life-limiting): the cathode — imperfect copper re-deposition (the self-heal slows copper loss ~100× but not to zero), framework vacancy / coordinated-water loss, and residual lattice strain. The organic anode and brine are more durable; the anode is the likeliest wholesale-refresh item. (30,000 cycles is the published 90%-retention endpoint, not the death point; conventional 80% end-of-life sits beyond it.) Material recovery: Cu electrowon as ~99.5% dendritic powder (~$13/kg; the existing Zone-A recovery loop), Co co-recovered (valuable critical mineral), Fe/framework reclaimed via aqueous re-dissolution + re-precipitation, brine cleaned and recycled. Designed for >90% material recovery; top-up = the measured deficit (single-digit-% metals + brine) plus the organic anode. Metal monitoring — how the top-up is quantified: (1) ICP-OES / ICP-MS on periodic electrolyte samples (and the recovered-material stream) — per-element quantification of Cu / Co / Fe / Mg / Ca / Na / K to ppb; this exact inventory sets each top-up. Already the lab’s analytical backbone (the “Cu leach >15 ppm/100 cycles” kill-gate is an ICP measurement). (2) Rosemount inline sensors (toroidal conductivity, pH, Coriolis density) trend the electrolyte continuously in service — they flag when composition drifts and trigger a sample, though they don’t speciate per-metal alone. (3) The BMS infers cumulative copper loss indirectly from capacity-fade signatures. Together: continuous inline trending (Rosemount) → targeted ICP quantification → precise, measured top-up. Restoration siting: field-serviceable = electrolyte/brine top-up, a BMS mode-shift toward the endurance window to stretch life, and a compact/mobile electrowinning unit for large sites; facility-scale = cathode re-precipitation + cell reassembly (precipitation reactors + coating + assembly). A containerized mobile refurbishment skid is a plausible Phase-3 product (Emerson’s skid wheelhouse — only thinkable because the chemistry is non-hazardous). Honest posture: electrowinning is mature and re-precipitation is literally how the cathode is synthesized, so the loop is demonstrated in principle — but the end-to-end closed-loop refurbishment at scale, and the exact recovery percentages, are Phase-2/3 engineering targets, not yet proven. Recovery is high, not 100%. Ties to the degradation card and the zero-discharge ethos: the cells are finite; the material loop is effectively permanent.

Does the mixed magnesium/calcium version degrade faster than single-cation?

In one sentence

Chen 2026’s 120,000-cycle result is the mixed cell — so mixed is the anchor, not the unknown. What’s unproven is how ratio-dependent behavior looks across the Mg:Ca sweep, and how pure-Mg and pure-Ca standalone cycle life compare (both currently [H], unproven). The Phase-1 parallel ratio series measures all three simultaneously.

The plain-English version

Chen 2026 published the 120,000-cycle result on a mixed electrolyte (4.0 M MgCl₂ and saturated 5.8 M mixed MgCl₂+CaCl₂). So at the cell level, mixed is the proven anchor, not a novelty bet. What is unknown is (a) how Chen’s coin-cell result at 20 A/g translates to grid-relevant C-rates in our cell geometry, and (b) how the ratio-dependent behavior looks across the 100:0 / 75:25 / 50:50 / 25:75 / 0:100 Mg:Ca sweep. Pure-Ca and pure-Mg standalone cycle life are themselves unproven — both are plausible bets from selectivity + kinetics arguments, but the 120k number does not transfer to a single-cation cell. So the parallel ratio series isn’t “does mixed work?” — it’s “how well does Chen’s mixed transfer to our geometry, and are pure-Ca / pure-Mg viable single-cation alternatives?” Three questions, one experiment. See the fallback ladder for how the outcomes stack.

The full engineering case

The proven / bet / frontier hierarchy: Chen 2026’s mixed dual-cation cell is the proven anchor (120k cycles at 20 A/g, coin cell, mixed MgCl₂+CaCl₂). Standalone pure-Ca and pure-Mg cycle life at Phase-1 rates are bets tested for the first time in the parallel ratio series. The frontier is running the proven mixed cell on already-mixed brine feedstock (see next card and Phase 4). Failure-mode inventory across compositions: (1) lattice fatigue from ionic-radius mismatch (Mg²⁺/Ca²⁺ Shannon radii 0.72/1.00 Å) — measured by ex-situ post-cycling XRD; Chen’s coin-cell result implies this is manageable at his composition but Phase-1 tests the C/4 transfer + ratio sensitivity. (2) Tachydrite precipitation — CaCl₂·2MgCl₂·12H₂O forms above ~22°C in certain Mg:Ca regimes; a mixed-specific electrolyte failure mode. Mitigation: DeltaV recipe controls composition + temperature to stay outside the crystallization field (Kurnakov 1938 phase diagram, fully mapped). (3) Preferential Ca-cycling in mixed — per the selectivity sequence, Ca²⁺ may carry the electrochemistry disproportionately; if so, “mixed” performance approximates the Ca contribution — which is why pure-Ca cycle life is a first-class question, not an afterthought. (4) Kinetics tradeoff — Ca has lower desolvation penalty and faster migration; mixed with meaningful Ca fraction gets Ca kinetics benefit at Mg’s feedstock-cost benefit. Phase-1 Gate 3 measures all four simultaneously via the parallel ratio series (pure-Mg / pure-Ca / mixed compositions on the same 32-channel cycler) with per-config ex-situ ICP + XRD + XPS + DEMS characterization. Outcome resolves the ratio-degrade question with data, not hope. Also settles the charge-carrier question in the same experiment.

Why call it “saltwater” if the electrolyte is magnesium/calcium chloride, not sea salt?

In one sentence

Fair catch. Today’s electrolyte is a concentrated Mg/Ca chloride “water-in-salt” mineral brine — technically not sea salt (NaCl). The name points at the destination, not the beaker: the apex ambition is harvesting these minerals from the world’s most abundant waste stream, desalination reject brine, which is saltwater.

The plain-English version

Technically accurate: today the electrolyte is a concentrated Mg/Ca chloride water-in-salt brine, not NaCl seawater. A precise reviewer will note that and think “this isn’t saltwater.” The honest resolution is that the brand names the mission, not the beaker. Salt · Water · Power points at the destination: harvesting these minerals from the world’s most abundant waste stream, desalination reject brine, which is genuinely saltwater. So the brand keeps its aspiration (power from the minerals in saltwater/desal brine) while the technical copy calls the electrolyte exactly what it is — concentrated Mg/Ca chloride mineral brine. We’d rather raise the objection ourselves than have a reviewer raise it for us; the same honesty discipline as the energy-density and cost fixes.

The full engineering case

Technical precision throughout the corpus: the working electrolyte is 4.0 M MgCl₂ or saturated 5.8 M mixed MgCl₂/CaCl₂ at pH 7 — a “water-in-salt” regime (Suo et al. 2015 nomenclature) where extreme salt concentrations coordinate free water to ions and widen the electrochemical stability window to 2.2 V. This is mineral brine, not seawater. The Phase-4 Carlsbad mission is the point where the technology genuinely meets saltwater: desalination reject brine (~50 MGD at Carlsbad, ~70 g/L total dissolved solids, ~2.6 g/L Mg²⁺ and ~0.8 g/L Ca²⁺) is the feedstock for the divalent electrolyte and the byproduct-recovery platform. So Salt · Water · Power is descriptively accurate at the mission level (saltwater in, power out via the mineral loop) and directionally accurate at the beaker level (water-in-salt brine is technically salt + water, just Mg/CaCl₂ rather than NaCl). Technical copy uses the precise term (“concentrated magnesium/calcium chloride water-in-salt brine” or “mineral brine”); brand + mission copy uses the aspirational term. Same discipline as calling the cell “non-flammable aqueous” precisely (rather than looser “water battery”) while calling the whole endeavor Salt · Water · Power.

Section B

The Build

How Phase 1 actually validates the chemistry, what the skid looks like at scale, where Shakopee fits, and where it deliberately doesn’t.

B1 · The Phase 1 lab

You might ask

Where will Phase 1 happen?

In one sentence

A ~10,000 SF lab built inside the ground floor of the ~30,000 SF two-floor Shell Space at the northeast corner of Emerson’s Shakopee campus.

The plain-English version

Emerson’s Shakopee facility has an unbuilt two-floor shell space sitting empty in its NE corner. The author walked it on June 17 2026 with a tape measure: 135'-4" by 112'-4" per floor, ~15,200 SF per floor, ~30,000 SF for both floors together. Phase 1 builds a working battery lab in about ~10,000 SF of L1 (the ground floor) — four zones, every measurement instrumented — with ~4,500 SF of L1 buffer for Phase 2 expansion and all of L2 preserved as co-tenant space for the sanctioned testing programs already running there.

The full engineering case

Three-zone lab: Zone A (purification — electrolyte prep, ICP characterization, electrowinning bench), Zone B (assembly — glovebox, doctor blade, vacuum oven, electrode press), Zone C (testing — NI HPS-17000 cycler, NI PXI EIS, environmental chamber). Shell-space fit-out: walls, utilities, HVAC, fume hood, eyewash, fire suppression, 480V/200A power. Integrated zero-discharge infrastructure: solvent still, NaOCl in-line dosing, electrowinning rig, recovery-loop plumbing tied to DeltaV. Full floor plan in hub/04_floorplan; isometric in 05_isometric; 3D walkthrough in 07_walkthrough.

You might ask

What is the L2 reservation, and what does it actually cost?

In one sentence

A 12-month option from W26 commissioning on the upstairs ~15,200 SF, with two anchored candidates (10× Cu refinery + extended-cycle aging bay) at $2.4M NTE / $2.7M envelope — not part of the $4.2M Phase 1–3 ask and triggered only after three GO/NO-GO gates clear.

The plain-English version

Two real things would happen on L2 if the lab earns its next chapter. First, the kg/wk dendritic copper stream from Phase 2 scales 10× into a 75-kg/wk supply Rosemount sensor manufacturing actually consumes (tiered take-or-pay so procurement isn’t asked to swap LME Grade A 99.99% cathode for 99.5% dendritic without a structure). Second, a dedicated 10,000+ cycle calendar-life proof line — a second NI HPS-17500 plus walk-in env chambers — produces the multi-year aging dataset that BaaS pricing, voluntary-carbon-market 20-year credits, and any insurer-backed cell-leasing pilot all literally require. Competitors can’t fake time. Total $2.4M not-to-exceed (with $300K owner contingency built in, $2.7M envelope), broken into four visible lines so the number has bones showing: equipment $1.0–1.35M, L2 shell-to-lab buildout $450–700K, commissioning & qual $150–200K, contingency $300K. The structure is a 12-month option from Phase 3 commissioning with a kill-switch — if the Phase 2 50-cell pilot doesn’t hold ≥80% capacity at 2,000 cycles on ≥20/50 cells, L2 is released and capital stays uncommitted. BU sponsorship is pre-aligned with Rosemount (Nathan Pettus, primary) and NI (Luke Schreier, Test & Measurement, co-sponsor) during Phase 2.

The full engineering case

L2 envelope (field-measured June 17 2026, HGA + Dunham A220): 135'-4" × 112'-4" ≈ 15,200 SF gross, ~11,000 SF usable after the HVAC floor opening (10'-3" × 12'-4", ~300 SF fall-protection envelope), MEP core stack (~418 SF, walled floor-to-ceiling), Capsule Mark sanctioned co-tenant footprint (~600–900 SF east strip, Eric Guggisberg, safety review locked 4/14/2025 — stays put, Phase 4 layout designed around it), circulation around all three. Anchored picks: (1) 10× Cu refinery skid — 10kL jacketed vessel $220–300K + scaled agitator/scrubber $80–110K + automated cathode-swap line $80–120K + drum/QA station $40–60K + DeltaV controls tie-in $40–80K + L2 secondary containment floor coating $50–70K = $450–600K, plus tiered Rosemount take-or-pay LOI (99.5%+ premium / 99.0–99.5% discount for less-critical SKUs / sub-spec recycled back to electrolyte loop). (2) Extended-cycle aging bay — second NI HPS-17500 cycler $350–500K + two walk-in env chambers (calendar/cycle split) $120–180K + impedance spectroscopy station for periodic SOH $80–120K + isolated UPS + dedicated DAQ + Plantweb tie-in $50–80K + cell fixturing for 200–400 cells $40–80K + L2 partition + power $40–60K = $550–800K. M&V-grade instrumentation baked in at marginal cost (~$30–50K) replaces standalone VCM credit lab. Explicitly deferred: BESS grid-services characterization bench — honest scope is 5–10 kWh pack-level rig with 4-quadrant grid emulator ($900K–1.3M done right, not the cell-only cycler the early draft implied); W52 decision point gated on aging-bay output. Longer-horizon adjacencies under evaluation but explicitly not in the Phase 4 envelope: cell recycling/refurb bench (Phase 5 once Phase 1 cells return at end-of-life, ~W260+) and pilot manufacturing line seed (Phase 5 once Phase 3 partner-bench pipeline names customers). Long-duration prismatic format R&D is acknowledged as the genuine TAM for aqueous chemistry but is a Phase 5 strategic pivot contingent on pouch chemistry lock at W20 — format transition mid-program is too aggressive for Phase 4. Three W26 GO/NO-GO gates (all must clear): (1) Rosemount BU procurement-signed (not R&D liaison) tiered take-or-pay LOI on 10× Cu volume by W22; (2) Phase 2 emew line sustained ≥3 kg/wk at ≥99.5% for 8 consecutive weeks; (3) Phase 2 50-cell pilot produced ≥500 cycles on ≥40/50 cells with <15% capacity-fade variance AND NREL secondee delivered written aging-protocol recommendation by week 8 of residency. 3-of-3 GO at W27 with Cu refinery as immediate start + aging bay W30 start; 2-of-3 defer Phase 4 to W34–W40; 1-of-3 rescope to aging-bay-only; 0-of-3 release L2 reservation as Phase 3.5 consolidation. Plus a 4th non-technical gate: AHJ Stair N egress capacity review CLEARED at W6 of Phase 1 (not W26) — at L1+L2 combined occupant load (~110–135 calculated), Stair N alone likely fails IBC 1006.3.3 single-exit threshold; budget $8–15K consultant (Jensen Hughes or Aon FPE) at W6, fold a second stair into Phase 3 NW open zone scope if needed (~$250–450K, far cheaper folded into Phase 3 concrete than retrofit). Timing critical path: NI HPS-17500 cycler bay lead time is 22–28 weeks from PO (verified from Phase 2 bay 2 procurement), so the Phase 4 LOCK decision is actually W18–W22 (PO drop), with W26 the formal gate review, W42–W48 commission. Sanctioned-tenant MOU work (SYNCHROS — Kerry Schoenecke, NiM re-flash — Tim O’Connell) must start W14 (Phase 2 kickoff), not Phase 4 itself; budget $80K relocation capex in Phase 3. NREL secondee + Octopus Energy alumna are Phase 3 triggers (already lined up for Q2), producing Phase 4 W26 gate inputs (aging-protocol recommendation, Hex-COP cycle-stability paper) while sitting in Phase 3 — they do not wait for Phase 4. Octopus alumna’s deliverable is a written Phase 5+ commercial-pathway thesis, not near-term BESS revenue (chemistry is TRL 4–5; grid commercial relevance is TRL 7–8 — mismatching risks her credibility and the program’s).

You might ask

What is a coin cell and why do we start there?

In one sentence

A coin cell (CR2032 form factor) is a thumb-sized research battery that uses milligrams of material — the standard industry tool for validating new chemistry before any scale-up.

The plain-English version

A coin cell looks like a watch battery. It contains tiny amounts of the actual electrode materials, the actual electrolyte, the actual separator — everything that would go into a real cell, but at lab scale. You can build 480 of them in 9 months at low cost and test them across many chemistry variations and operating conditions. If the chemistry doesn’t work at coin-cell scale, it certainly won’t work at skid scale, and you’ve learned that for $1.4M instead of $200M.

The full engineering case

CR2032 form factor: 20 mm diameter × 3.2 mm height. Stack order (top to bottom): top cap, wave spring, SS spacer, SS 316L anode current collector, COP anode (~50–80 µm), Whatman GF/A glass-fiber separator (260 µm, 16 mm ø), CuFe-PBA cathode (~50–80 µm), Ti foil cathode current collector, bottom case. ~80 µL electrolyte. The Phase 1 campaign builds ~480 cells across a 4 × 12 sweep (4 chemistry variants × 12 conditions). Full cell-stack diagram and procedure documented in hub/03_protocol (CR2032 Cell Stack reference figure).

You might ask

What does “Phase 1 succeeded” actually look like?

In one sentence

One canonical table referenced everywhere — kill floor: ≥500 cycles at C/4 + 80% DoD with ≥80% retention; target: ≥1,000 cycles at C/4 with ≥90% retention + Cu leach <5 ppm/100 cycles + voltage hysteresis <250 mV; stretch: capacity-fade slope consistent with extrapolation to the projected 8,000–25,000 cycle envelope.

The plain-English version

Success isn’t a feeling. It’s three specific numbers measured by Emerson’s own instruments. The cell has to still hold 80% of its starting capacity after 5,000 charge-discharge cycles at the speed grids actually use. It has to leak less than a trace of copper per 100 cycles. And the voltage gap between charging and discharging has to stay tight enough to be efficient. Hit all three and Phase 2 is justified. Miss any one and Phase 1 ends at a clean kill-gate.

The full engineering case

Phase 1 success criteria — canonical, referenced from every other card that cites Phase 1 kill-gates (full table in hub/03_protocol Success Criteria): Cycle count & retention — KILL FLOOR ≥500 cycles at C/4 + 80% DoD with ≥80% capacity retention; TARGET ≥1,000 cycles at C/4 + 80% DoD with ≥90% retention; STRETCH capacity-fade slope <0.012% per 100 cycles consistent with the 8,000–25,000 cycle projection at 80% retention (per the “Chen ran 120,000 cycles at 20 A/g” card). Coulombic efficiency tiered — formation cycles 1–50 ≥97% (kill <90% or monotonic decline), break-in 50–200 ≥99.0% (kill <95% or slope >0.5 pp/100-cyc), steady-state 200+ ≥99.5% (kill <97% sustained over 50-cyc rolling mean). Voltage hysteresis at C/4 mid-SOC <250 mV (kill >400 mV). Cu leach <5 ppm per 100 cycles (kill >15 ppm/100). Electrolyte pH stability 4.91–7.02 (kill <4.0 or >8.0). Self-discharge <3%/month at 25°C (kill >10%/month). Cells/pouches ≥1 Ah on the NI HPS-17000 cycler. 90-Day Assessment Framework runs validation, integration scoping, and first-contact in parallel. If a card elsewhere in the FAQ cites a different number, this card overrides.

B2 · Scaling up

I asked this

How big and heavy is one production skid?

In one sentence

A 3 MWh production skid is roughly 30–50 tons — about 2–3× heavier than the equivalent Tesla Megapack — with most of that mass being electrolyte (salt water at ~1.4 g/cc).

The plain-English version

The skid is roughly the size of a shipping container with a 30–50 ton mass: 5–10 tons of steel frame, 15–30 tons of electrolyte (the bulk), 3–8 tons of cell hardware and active materials, and 2–5 tons of plumbing, valves, BMS, and instrumentation. Heavier than lithium for the same energy stored, but it sits on a concrete slab and doesn’t move once installed — mass doesn’t cost anything in stationary storage.

The full engineering case

Per-skid mass breakdown (3 MWh design): Frame/structural 5–10 t (carbon steel painted, stainless only for chloride-contact zones); Electrolyte 15–30 t (MgCl₂ solution ~1.4 g/cc, the heavy bulk; the salt is the host, not the fuel — doesn’t deplete); Cell hardware + active materials 3–8 t (Ti foil cathode collectors, SS 316L anode collectors, PBA cathode mass, COP anode mass, GF/A separator); Thermal/electrical balance 2–5 t (Fisher valves, Micro Motion flow, Zitara BMS, Cu interconnects, Rosemount instrumentation). Total 30–50 tons. Reference: Tesla Megapack 2 XL is 3.9 MWh / ~38.1 tonnes, ~$300–400/kWh installed (2026 estimate; last public price was Sept 2022 at $1.85M/unit at 100+).

I asked this

Why so much stainless steel? Where does all the mass come from?

In one sentence

Most of the mass is electrolyte (water + salt), not steel; stainless is only used in chloride-contact zones where chemistry demands it, with carbon steel for the rest of the frame.

The plain-English version

The 30–50 tons isn’t all metal. The electrolyte is 15–30 tons because it’s a thick salt solution and you need a lot of it to store megawatt-hours of energy. The frame itself is 5–10 tons of mostly painted carbon steel, with stainless used only where the metal actually touches the salty electrolyte (about 2–4 tons of stainless per skid). Stationary infrastructure designed for a 30-year service life uses honest materials that don’t corrode rather than thin tricks that wear out.

The full engineering case

Mass allocation per 3 MWh skid: electrolyte ~60% of total (water + dissolved salt at ~1.4 g/cc); frame steel ~15–20%; cell hardware + active materials ~10–20%; plumbing/electronics ~5–10%. Stainless 316L is used selectively for chloride-contact zones (anode current collectors, electrolyte tankage internals, fittings) at ~2–4 t per skid; the rest of the frame is painted carbon steel. Carbon vs stainless capex ratio at industrial scale is ~5:1, and the corrosion engineering is well-understood from oil & gas / chemical-plant practice. Coatings and cathodic protection extend non-stainless frame life to 30+ years routinely.

I asked this

Will skids be custom-built per customer or standardized?

In one sentence

Standardized 1–3 MWh modules, exactly like Tesla Megapack and CATL EnerC, with customer customization happening at deployment (how many stacked, grid integration), not at the skid level.

The plain-English version

The entire BESS industry has converged on standard containerized skids of 1–3 MWh that ship on a flatbed and drop onto a slab. Tesla Megapack is 3.9 MWh per unit. Powin Centipede, CATL EnerC, Wartsila Quantum — all the same modular pattern. Aqueous Mg-ion follows the same pattern: standardized factory-integrated skid, customer customization happens by quantity at the deployment site, not by building each skid one-off.

The full engineering case

BESS industry standardization: Tesla Megapack 2 XL 3.9 MWh / Megapack 3 4.0 MWh; CATL EnerC 6.0 MWh; Powin Centipede 5.0 MWh; Wartsila Quantum 2 MWh. Standard skid economics depend on factory throughput, so customization happens at site (DC-coupled stacking, AC inverter sizing, BMS configuration for grid services, Plantweb fleet tier). The proposal’s “land-and-expand” commercial pattern (customers commit to a small initial deployment, add additional skids onto the same DC bus and Plantweb fleet as appetite grows) explicitly exploits this modularity. Custom skid per deployment would defeat the factory economics that make Phase 4 commercial scale possible.

You might ask

What’s the timeline from lab to deployed product?

In one sentence

Phase 1 takes 13 months (coin cells), Phase 1.5 another 9 months (pouch cells), Phase 2 about a year (first skid prototype), Phase 3 takes 2–3 years (pilot manufacturing); first commercial deployments land in years 5–7 from now.

The plain-English version

First-of-kind battery chemistry doesn’t hit commercial scale in 18 months. Lithium took roughly 15 years from Goodenough’s lab to Sony’s first product, then another 25 years to grid scale. Form Energy is on roughly a 7-year arc from founding to Weirton trial production. This proposal is on a similar 5–7 year arc from Phase 1 commit to first commercial deployment, assuming chemistry validates at each gate.

The full engineering case

Phase trajectory (full table in “Building It at Scale” section of the proposal): Phase 0 today, decision to invest ($200–400K, distributed); Phase 1 months 0–13, R&D lab + coin cells ($1.0–1.8M, Shakopee); Phase 1.5 months 9–18, pouch cell prototype ($1.5–3M add’l, Shakopee extension); Phase 2 year 2, first skid prototype + manufacturing path research ($10–25M, hybrid build); Phase 3 years 3–5, pilot manufacturing line ($80–200M, decision point); Phase 4 years 5–10, commercial rollout 1–10 GWh/yr ($500M–$2B, multi-site or partner network). Total program ~$600M–$2.2B over a decade.

B3 · Supply chain

You might ask

Where does the magnesium chloride come from?

In one sentence

Two qualified domestic supply tracks: Nedmag pre-purified MgCl₂ (Zechstein deposits, Netherlands) and Intrepid Potash (Carlsbad, NM) requiring DeltaV purification on-skid, both at ~$300/ton vs lithium carbonate at $9,000–17,000/ton.

The plain-English version

MgCl₂ is a global commodity produced at roughly 1.7 million tonnes per year. The two supply paths Phase 1 evaluates are Nedmag (Netherlands, pre-purified, premium price) and Intrepid Potash (New Mexico, raw, lower price but needs a DeltaV-controlled purification step). Both are FDA food-grade, no export controls, no supply-chain vulnerability. Magnesium chloride is also recoverable from desalination waste brine at scale — a Phase 2 R&D track for the long term.

The full engineering case

Global MgCl₂ production ~1,700,000 tonnes/yr; market value ~$738M (2025); price $200–600/ton industrial grade. Qualified supply tracks for Phase 1: (1) Nedmag (Netherlands, Zechstein evaporite, pre-purified to battery grade); (2) Intrepid Potash (Carlsbad NM, raw MgCl₂ requiring DeltaV-controlled purification skid — an Emerson product opportunity in itself). Phase 2 desal-brine recovery: ~142M m³/day of MgCl₂-rich brine dumped globally (UNU-INWEH 2019); recovery pathway technically feasible but unproven at battery-grade purity, hence a Phase 2 question rather than a Phase 1 assumption. Seawater Mg content ~1.3 g/L is the long-tail feedstock if paired with selective extraction.

I asked this

Are we really sure we can pull battery-grade purity from desal waste brine?

In one sentence

No — the recovery pathway is technically feasible but unproven at battery-grade (~99.9%) purity, so the proposal explicitly treats it as a Phase 2 question and doesn’t depend on it for Phase 1 or 2 success.

The plain-English version

Desal brine is rich in magnesium chloride and the world dumps about 142 million cubic meters of it per day. In principle, recovering that magnesium is a great closed-loop story. In practice, getting it from raw brine to battery-grade purity (~99.9%) is a chemistry problem that nobody has solved at commercial scale yet. The proposal originally implied it was solved — it’s not. Honest framing now: Phase 2 R&D track, parallel to (not dependent on) the main chemistry validation.

The full engineering case

Desalination brine MgCl₂ content varies by source (~50–70 g/L in concentrated reject brine, ~3.7% of seawater). Recovery requires selective extraction (ion exchange or selective precipitation), purification to remove competing ions (Ca, K, SO₄, organics), and crystallization to battery-grade. Adjacent industry precedent: solar evaporation ponds (Great Salt Lake, Dead Sea) produce industrial-grade MgCl₂ but not battery-grade in current process. Phase 2 evaluates whether DeltaV-controlled selective extraction can reach the battery-grade threshold at competitive cost. Backup supply (Nedmag + Intrepid) covers Phase 1–3 entirely; desal recovery is optionality for Phase 4+ supply diversification.

I asked this

Where does the salt actually come from — and isn’t New Mexico evaporation-pond extraction environmentally toxic?

In one sentence

Four candidate sources were evaluated (NM solar-pond, Dutch Zechstein, Great Salt Lake, US desalination-brine); Intrepid Potash (Carlsbad, NM) stays as Phase 1 primary because solar drying gives the lowest cradle-to-gate carbon, Nedmag (Veendam, NL) is qualified as the high-purity backup, Great Salt Lake (Compass / Cargill Ogden) was disqualified for press liability, and US desalination is named as a post-2028 horizon — not a Phase 1 or 2 sourcing path.

The plain-English version

Connor’s worry on a walk-through of v17: "isn’t New Mexico evaporation-pond environmentally toxic? Can we find better domestic sources? The dream is desalination brine but if the energy and carbon cost of purifying it swamps the gain, dig deeper." A panel of four voices (LCA scientist, mining chemist, supply-chain economist, hyperscaler-procurement skeptic) graded four feedstocks against (a) cradle-to-gate carbon, (b) water footprint, (c) landed $/ton at battery-grade ≥99.9%, (d) provenance / political risk. Verdict: Connor’s method instinct (solar-evaporation pond) was right; his toxicity instinct was not — MgCl₂ itself is FDA GRAS (food-grade; used as tofu coagulant and road de-icer). The real ESG flag at Carlsbad is Pecos River salinity loading, which USGS attributes overwhelmingly to natural geologic discharge at Malaga Bend (~172,000 tons of saturated brine per year, naturally), not to Intrepid’s extraction. That makes Pecos a chronic water-quality concern in a flowing river, not a Great Salt Lake–style ecological collapse story. Compass Minerals’ Ogden operation IS publicly tied to GSL collapse and was disqualified on press-liability grounds; carbon math irrelevant. Phase 1’s kg-scale procurement (~15–30 kg total) is invisible to the Pecos watershed regardless — Phase 2 supplier agreements pick up brine-discharge LCA reporting commitments. US desalination brine, including Carlsbad CA (50 MGD reject discharging ~492 metric tons of magnesium per day unrecovered to the Pacific), is technically real but commercially nonexistent — no US plant has a Mg-recovery pipeline as of June 2026, the working pilots are EU-based (SEArcularMINE, Sea4Value), and the Phase 1 economics break by four orders of magnitude (~$60–90 of Intrepid MgCl₂ versus a $200K–500K Carlsbad-CA pilot install). Stays in the proposal as a Phase 4 long-horizon co-location story (CA desal → CA Mg recovery → CA hyperscaler BESS), not a Phase 1–2 dependency.

The full engineering case

Ranking, with honest hedge. Per-kg cradle-to-gate carbon values below are third-party LCA modeling estimates, not supplier-published EPDs — Nedmag has filed an EPD (Hub document 701) and Intrepid’s 2024 + 2025 Sustainability Reports are public, but both are image-locked PDFs not yet text-extractable; Phase 1 procurement workflow includes supplier outreach to request verified per-kg GWP data. (1) Intrepid Potash, Carlsbad NM — KEEP as Phase 1 primary. Solar evaporation, ~0.1–0.25 kg CO₂-eq/kg MgCl₂ (modeled, panel estimate), ~$1,500–3,000/ton battery-grade equivalent landed Shakopee. Domestic. Mild ESG watch (Pecos), addressable via Phase 2 supplier-agreement disclosure language. Solution-mines subsurface brine; does not drain a surface lake. (2) Nedmag, Veendam NL — QUALIFY as high-purity backup. Zechstein solution-mining produces bischofite (MgCl₂·6H₂O) at intrinsic ~94–96% purity, reducing Track-B purification load on the DeltaV skid. Carbon ~0.7–1.3 kg CO₂-eq/kg (modeled; ~5–10× higher than NM due to gas-fired evaporation), ~$2,000–3,500/ton landed. Use selectively where FTO specifies the lowest impurity profile. (3) Great Salt Lake (Compass Minerals / Cargill Ogden) — DISQUALIFIED. Carbon math is fine (~0.1–0.2). Compass’s Ogden brine extraction is publicly tied to GSL ecological collapse (lake down ~50% from historic levels, sustained national press coverage). Any procurement skeptic finds the story in 60 seconds. Hard pass on provenance grounds. The Compass Minerals Wendover NV facility (different brine source than GSL Ogden) is named as a Phase 2 RFQ candidate. (4) US desalination-brine recovery — demoted to Phase 4 horizon. Carlsbad CA (Poseidon Water, 50 MGD product / ~50 MGD reject at ~2.6 g/L Mg²⁺) is the only US site with feedstock concentration and volume to matter; Tampa Bay is 10× smaller, FL/TX brackish plants are Mg-poor. No commercial Mg recovery pipeline exists at any US desal plant; EU pilots (SEArcularMINE on Mediterranean bittern, Sea4Value modular skids) demonstrate ~€1.62/kg Mg(OH)₂ floor cost at sub-100 t/yr scale and battery-grade purification adds HCl + crystallization + zone-refining stack on top. Phase 1 (~30 kg) Intrepid spot price: ~$60–90 delivered. Phase 1 Carlsbad-CA recovery skid install: ~$200K–500K capex + 12-month permitting + Poseidon commercial agreement. Economics break by four orders of magnitude. Phase 2 (~100 tons): Intrepid ~$300K; Carlsbad-CA pilot install multi-million. Phase 4 commercial scale-up co-located with West Coast desalination expansion and hyperscaler BESS load becomes the realistic horizon (2030+), not 2027. Pecos vs Great Salt Lake. USGS attributes Pecos baseline salinity primarily to natural saturated-brine upwelling from the Rustler-Salado formation boundary, especially Malaga Bend (~172,000 tons / yr at ~338,000 mg/L NaCl); USFWS High-Salinity project does not name Intrepid as a primary contributor. Intrepid’s only material public-record exposure is the August 2021 Fifth Judicial District ruling on water rights (5,700 ac-ft of unused rights forfeited after attempted sale to oil & gas) — a financial story (shares fell ~20% in two days), not an ecological-harm narrative. Sources cited in this answer: EPD Hub doc 701 (Nedmag bischofite); Intrepid Potash 2024 + 2025 Sustainability Reports; DitchCarbon Intrepid Moab scrape; USGS Pecos River Basin Salinity Assessment; USFWS High-Salinity Pecos Project; USGS Malaga Bend WRI-804; mining.com on Intrepid water-rights ruling; Wikipedia / SDCWA on Carlsbad CA Desal Plant; Springer 2022 + MDPI 2025 Mg-from-brine reviews; ACS Sustainable Resource Mgmt 2025; SEArcularMINE and Sea4Value EU project pages.

I asked this

Will Emerson ever do every step of the supply chain in-house — including recovering the salt itself from desalination plants?

In one sentence

Yes — the long game (post-2028, Phase 4) is a co-located MgCl₂ recovery + battery-grade purification plant on the brownfield adjacent to the Bud Lewis Carlsbad Desalination Plant in California, processing reject brine that currently discharges ~492 metric tons of magnesium per day unrecovered to the Pacific Ocean — with a credible engineering path to net-negative carbon per delivered kWh, and ~$100–200M of federal + California incentives stackable against the build.

The plain-English version

Battery-grade MgCl₂ from raw seawater brine takes three industrial steps: bulk separation of Mg from competing ions (brucite precipitation), conversion to MgCl₂ via HCl, and polish purification to ≥99.9%. Phase 1 already builds step 3 (that’s what Track B is) on existing Emerson DeltaV + Rosemount instrumentation. Phase 4 adds step 1 — the brucite-precipitation plant — and bolts it onto the polish step Phase 1 already mastered. The site is the former Encina Power Station brownfield, ~95 acres directly adjacent to Bud Lewis Carlsbad Desal, already industrially zoned, already routed past Poseidon Water’s brine outfall pipe. Connor’s instinct on this is right: the long-term ambition is full vertical integration, and Carlsbad CA is the only US site with the brine concentration and volume to matter. The honest caveats: the production carbon per kg of MgCl₂ from desal brine is genuinely worse than NM solar evaporation (the sun does the bulk-separation work for free in NM), so the "carbon-negative" claim only holds at the per-delivered-kWh denominator and only under five specific engineering conditions (on-site solar PPA, RO waste heat, byproduct HCl, LCFS credits, atmospheric CO₂ uptake via the RSC-2023 NaOH→Na₂CO₃ pathway). The strategic value isn’t cheaper feedstock; it’s diversified supply, eliminating ~1,500 miles of Intrepid-NM trucking for West Coast hyperscaler BESS, and the lowest-carbon stationary-storage chemistry on Earth under post-2028 hyperscaler procurement frameworks. No Phase 1 capital is committed to any of this. Phase 4 only commits after Phase 1, 1.5, 2, and 3 land their kill gates. The whole long-game write-up lives at The Carlsbad Play.

The full engineering case

The three-step framework. Step 1 (bulk separation): raise brine pH >11 with NaOH or CaO, Mg precipitates as Mg(OH)₂ at 95–99% recovery (brucite precipitation; EU pilots SEArcularMINE and Sea4Value demonstrate ~€1.62/kg Mg(OH)₂ floor cost at sub-100 t/yr). Step 2 (conversion): react Mg(OH)₂ with HCl → MgCl₂ solution → crystallize as bischofite (MgCl₂·6H₂O). Step 3 (polish): recrystallization or zone refining to remove trace Na, K, Ca, Sr, B down to ≥99.9% (this is Phase 1 Track B). What changes between Intrepid path and Carlsbad path. The Intrepid path skips steps 1 and 2 because the sun pre-sorts Mg via solar evaporation in NM ponds, and Intrepid sells industrial-grade MgCl₂ (95–98% pure) already converted — only the Phase 1 Track B polish step is needed. The Carlsbad path adds the brucite plant on the front end; everything downstream is the same DeltaV + Rosemount stack Phase 1 builds. Carbon math, honest. Production LCA at CA 2030 grid intensity (~95 g CO₂/kWh, LBL projection): ~1.97 kg CO₂-eq per kg MgCl₂ realistic-case, ~0.90 best-case with on-site solar + waste heat. Intrepid solar baseline: ~0.1–0.25 kg/kg. Per kg of feedstock, brine recovery is 4–20× worse. But spread across the ~6,300–9,500 kWh of delivered grid storage that one kg of MgCl₂ represents over the cell’s 30-year cycle life (~3.3 kg cell × cell-level 20–30 Wh/kg × 120,000 cycles × 80% DoD — Chen’s 48.3 Wh/kg electrode-level derated for packaging, current collectors, and electrolyte mass, matching this FAQ’s energy-density card), the per-kWh production debit is ~0.21–0.31 g CO₂-eq/kWh — against LFP grid stationary at 50–100 g/kWh (Dai/Argonne 2019, Crenna 2021). The cycle-life advantage turns a per-kg production penalty into a per-kWh win of roughly 160–480× — two orders of magnitude, on the numbers we cite as honest for cell-level performance. Five conditions for genuine net-negative per kWh delivered: on-site solar PPA, waste heat from desal RO, byproduct HCl from co-located chlor-alkali, CARB LCFS or voluntary-carbon-market credit monetization for displaced LFP, atmospheric CO₂ uptake via integrating the published NaOH→Na₂CO₃ soda-ash pathway. Three of five → low-carbon. Four of five → net-zero. All five → net-negative. The incentive stack. Realistic ~$100–200M capex offset across §45X Advanced Manufacturing Production Credit (survived OBBBA, $5–8M/yr × 5 yrs), DOE Battery Materials Processing Grant ($50–100M one-time, Round 3 expected), California Competes Tax Credit ($10–20M, very low political risk, Peak Energy and MP Materials are recent precedents), CA Manufacturing Sales-Tax Partial Exemption ($2–6M, automatic), CEC Long-Duration Energy Storage Program ($10–25M if reopens). Requires ~2 FTE policy/grants capacity Phase 1 does not staff — Phase 2 hire. The honest gaps: no Poseidon Water MOU for third-party brine-recovery exists (Carlsbad CA is virgin territory; Saudi SWCC Ras Al Khair is the only operator with industrial-scale extraction); no Encina parcel option (NRG demolished Encina 2018-2022, City of Carlsbad redevelopment plan still in flux); no CARB LCFS pathway exists today for "battery feedstock from desal brine" (petition route is 18–36 months with no guarantee). All five gaps are nameable, none Phase-1-blocking. The full write-up: The Carlsbad Play.

You might ask

What about titanium foil supply — is that constrained?

In one sentence

Battery-grade 25 µm Ti foil has no public spot price and only three qualified global suppliers (Tokyo Steel, Baoji, ATI) — a real Phase 2 BOM-lock risk that needs an RFQ before any commercial commitment.

The plain-English version

Titanium foil is used as the cathode current collector. At 25 µm (very thin), only three suppliers in the world qualify. There’s no commodity index for it — you have to request a quote. Estimated cost is $150–400 per kg at multi-tonne volumes, with a 30–50% rolling premium versus thicker mill stock. Phase 2 must do an RFQ and qualify a second source before any Phase 3 BOM commitment.

The full engineering case

Ti sponge (input material) ~$12/kg per USGS Mineral Commodity Summaries 2026. Last US sponge plant closed 2024, so US supply is 100% import-dependent. 25 µm mill-roll foil estimated $150–400/kg at multi-tonne volumes (no public spot index; subject to RFQ). Three qualified suppliers globally: Tokyo Steel, Baoji Titanium (China), ATI (US). Phase 2 risk register: BOM lock requires (a) RFQ from all three qualified suppliers, (b) second-source qualification (preferably non-Chinese for export-control resilience), (c) thickness optimization study (whether 30 µm or thicker can substitute for cost savings without performance loss). Alternative: Cu cathode collector substitution is technically feasible if Ti supply tightens, with chemistry implications still in research literature.

I asked this

Does the water need to be emptied before transit?

In one sentence

No — skids ship full and sealed at low SOC; the DOT regs for non-spillable aqueous batteries (UN3496 or a chemistry-specific Special Permit) are written for exactly this case.

The plain-English version

Draining a 4,000–8,000 lb electrolyte system for every move would defeat the customer-mobility advantage that makes Battery-as-a-Service work. Instead, skids ship the way they operate: hermetically sealed primary containment, secondary bund, level-monitoring on both. The Department of Transportation already has rules for aqueous batteries — lead-acid skids move this way every day — and we slot into UN3496 (“storage battery, wet, non-spillable, electric”) or apply for a chemistry-specific Special Permit at PHMSA. Transport SOC is mandated <50% per the Operations Manual to keep cell voltage well below the open-circuit operating window.

The full engineering case

DOT classification pathway: 49 CFR 173.185 (“lithium battery exception”) does NOT apply — this is not lithium. Operative pathway is UN3496 (non-spillable wet storage battery) under 49 CFR 173.159(d), or a chemistry-specific Special Permit obtained through PHMSA’s SP&A request process. Primary stainless containment per Operations Manual §3.0 is hydrostatic-pressure-tested to 1.5× max operating head; secondary bund leak-rate-tested per ASTM D5162. Transport SOC mandated <50% (cell voltage below 1.5V, well clear of 2.2V operating window). Hazmat endorsement under 49 CFR 172.704 required for drivers transporting >1,000 kg gross battery mass; carrier holds and audits training records every 3 yr. Every shipped skid includes a Bill of Lading template, emergency-response sheet (Operations Manual Emergency Procedure E-12), and 24-hour CHO contact line.

I asked this

Why does Phase 1 need a Chemical Hygiene Officer — isn’t a rotating safety lead enough for five chemists?

In one sentence

Federal law (29 CFR 1910.1450) requires any lab using hazardous chemicals to designate a named CHO with documented authority and time — “rotating safety officer” or part-time contractor fails that test under OSHA audit.

The plain-English version

The OSHA Lab Standard is unambiguous: there has to be a named Chemical Hygiene Officer with allocated time, signature authority on the Chemical Hygiene Plan, and the institutional power to halt work. The first draft of the Operations Manual rotated this role among the chemists. The compliance audit panel killed it immediately — under inspection, “everyone is the safety officer” reads as “no one is.” The current Phase 1 staffing names a 0.25 FTE CHO — same person every quarter, co-signature on every Chemical Hygiene Plan revision, with explicit power to stop experiments. That’s the difference between a clean OSHA file and a citation under General Duty Clause 5(a)(1).

The full engineering case

29 CFR 1910.1450(e)(3) mandates a designated CHO who “(i) works with administrators and other employees to develop and implement appropriate chemical hygiene policies and practices; (ii) monitors procurement, use, and disposal of chemicals; (iii) sees that appropriate audits are maintained; (iv) helps project directors develop precautions and adequate facilities.” Practical implementation per Operations Manual §1 + §6: 0.25 FTE allocation (6 named roles total at 4.75 FTE), CHO signature on every CHP revision (rev rows 2.0 and 3.0a–d show signed lineage), co-signature on every new SOP, and quarterly inspection cadence. Under inspection, OSHA pulls (a) the named appointment letter, (b) CHP revision history with CHO signatures, (c) inspection records, (d) training records per 1910.1450(f). All four are now provisioned and pre-printed; the Operations Manual at rev 3.0d (independently reviewed by a lab-ops engineer, a chemical hygiene officer, and a technical typographer) ratifies this construction.

I asked this

Why write a 30-year Operations Manual before Phase 1 even runs?

In one sentence

The manual is the safety case — if we can’t write a coherent maintenance, emergency, and refurbishment protocol for a 30-year asset today, we don’t yet understand what we’re selling.

The plain-English version

The exercise of writing a full Operations Manual — 6 roles, 12 emergency procedures, 30-year service schedule, refurbishment specs, OSHA cross-references — surfaces every soft spot in the proposal. If the answer to “what do you do if a skid leaks HCN” is hand-wavy, that’s a chemistry problem we’d rather find now than during an Emerson stage-gate review. Writing the manual proved we could answer every question. The customer-facing version is also a sales tool: a working facility operator can read it and decide whether to buy.

The full engineering case

The 30-year Operations Manual (hub/11_operations.html, rev 3.0d) ships with: (1) 6 named roles at 0.25–1.0 FTE with cell throughput targets; (2) three operating zones (lab / cell-build / cycler) with HVAC, PPE, and access matrix; (3) emergency procedures E-1 through E-12 including HCN evolution from ferrocyanide on thermal/acid exposure, NMP carcinogen handling, HNO₃ spill, arc-flash, confined-space, BBP exposure; (4) 30-year service schedule with 5-yr / 10-yr / 20-yr inspections including SNT-TC-1A Level II NDE, ASME Sec V Art 4/6 weld verification, FT-IR ATR forensic on COP anode; (5) standalone OSHA-compliant programs for Respiratory (1910.134), BBP (1910.1030), DOT shipping (49 CFR 173.185 / 172.704), EAP/FPP (1910.38/.39); (6) HCN reactor offgas scrubber spec (packed Pall rings, 10–15% NaOH, ≥3 sec residence, breakthrough alarm 0.5 ppm). Three independent reviewers — a lab-ops engineer, a chemical hygiene officer, and a technical typographer — read the manual at rev 3.0d and returned no redlines.

How is in-house purification cheaper than buying battery-grade salt?

In one sentence

Battery-grade MgCl₂/CaCl₂ commands $800–2,000+/ton because every incumbent must separate Ca from Mg (the hardest step); skipping Ca/Mg separation cuts purification cost ~25–40%. Raw brine feedstock at $200–350/ton vs lithium at $9,000–17,000/ton is a 20–50× structural feedstock advantage before any processing.

The plain-English version

Battery-grade Mg/Ca chloride from a specialty supplier (Nedmag) runs $800–2,000+ per ton, and most of that cost is one specific processing step: separating calcium from magnesium. The two are chemical twins (both Group 2, both divalent, similar ionic radii), and industrial-scale separation requires oxalate precipitation, Ca-selective chelating resins, and multiple recrystallization passes — expensive, slow, high-risk chemistry. Because the aqueous Mg/Ca-ion battery keeps both divalents in the electrolyte, that entire step is skipped. Track B handles the easy separations (Na/K reject via fractional crystallization, Ba/Sr reject via sulfate precipitation, transition metals via chelating IX) and delivers a mixed Mg/Ca electrolyte at ~$200–350/ton feedstock cost plus a moderate polish. Compare to lithium carbonate at $9,000–17,000/ton and the raw-feedstock cost advantage is 20–50× before any processing math.

The full engineering case

Nedmag battery-grade MgCl₂ cost breakdown (2024 pricing, industry sources): raw brine + evaporation ~$150–250/ton; primary purification (transition metals, sulfate) ~$50–100/ton; Ca/Mg separation ~$400–600/ton (oxalate + chelating resin + recrystallization); final polish + battery-grade QA ~$100–200/ton; margin ~15% → ~$800–1,400/ton FOB Nedmag Veendam; add ~$100–200/ton logistics to Shakopee → ~$900–1,600/ton delivered. Track B (in-house Shakopee, mixed-divalent target): raw brine or Nedmag intermediate-grade feedstock ~$150–350/ton; monovalent + Ba/Sr + transition-metal rejection ~$50–100/ton; no Ca/Mg separation stage; final polish + QA under DeltaV recipe control ~$50–100/ton. Total ~$250–550/ton — 40–65% below Nedmag delivered. The cost gap widens further at Phase-4 Carlsbad scale where feedstock is desal reject brine (near-zero raw cost, sourced from the pipe rather than mined). Raw-feedstock comparison to lithium: LiCl / Li₂CO₃ battery-grade $9,000–17,000/ton (2024 exchange pricing); MgCl₂+CaCl₂ mixed at $200–350/ton = 20–50× structural feedstock advantage on the input side before any of the processing math.

How do you know which byproducts are actually recoverable?

In one sentence

Feedstock assay — a Phase-1 Gate-0 task. Certificates of analysis from Intrepid + the desal operator plus one independent ICP-OES + halide panel maps the entire byproduct portfolio with data, before we build anything.

The plain-English version

We don’t guess. Before the lab flow is finalized, Phase 1 runs a “Gate-0” feedstock characterization: request certificates of analysis (CoAs) from Intrepid (potash brine supplier) and the desal operator (Carlsbad brine, if Phase-4 partnership progresses), plus commission one independent verification assay via a contract lab. The panel measures every element and anion of commercial interest — Mg, Ca, Na, K, Li, Sr, Ba, Fe, Mn, Cu and trace metals via ICP-OES/ICP-MS; Cl, Br, SO₄, I via ion chromatography. Total cost: hundreds to low thousands of dollars. Total output: the byproduct portfolio map (which minerals are present at recoverable grade) + the purification specification (which impurities, at what ppm) — two deliverables from one test. The highest-information, lowest-cost task on the whole board.

The full engineering case

Gate-0 protocol: (1) request CoAs from Intrepid Potash (specific-well assays are proprietary but released under NDA to qualified evaluators) and, if Phase-4 partnership is active, from Poseidon Water (Carlsbad plant); (2) if either CoA is unavailable or dated, commission independent contract-lab verification (turnaround 2–4 weeks, ~$1–3k per full ICP + halide panel from a certified environmental lab — Eurofins, ALS, Test America); (3) analyte panel: cations by ICP-OES (major: Mg, Ca, Na, K; minor: Sr, Ba, Li; trace: Fe, Mn, Cu, Cd, Pb, Zn, Ni, Cr, Al); anions by IC (Cl, Br, SO₄, I, F, NO₃, PO₄); boron by ICP-MS; (4) output deliverable 1 — the recovery portfolio map (each mineral present at ≥ recoverable-economics threshold gets a “yes” tag for the Zone-D pilot design); (5) output deliverable 2 — the purification spec (each impurity above target-grade threshold gets a “reject” unit-op assignment in Track B). Lithium and bromide are explicitly called out because their economics are the most concentration-sensitive — the assay directly settles whether they are recoverable, converting speculation to fact for near-zero cost. Runs in parallel with a commercial track (offtake LOIs, cert roadmap, LIMS-loaded specs, Emerson BU alignment, partner outreach) so the launch pipe is ready the moment the chemistry validates.

Can the pilot lab really refine all these minerals to grade?

In one sentence

No — and we don’t claim to. Phase 1 proves the recovery principle and does the core separations at pilot scale. Full mineral-specific refining to industry/food/pharma grade is a modular Phase-2/3 build; for specialized streams (lithium needs DLE, bromine is Class 8/6.1), the capital-efficient move is to sell intermediate concentrates to refiners.

The plain-English version

This is the guardrail on the poly-mineral story. Phase 1 lands two things: (1) reproduce Chen 2026 for the battery, and (2) prove the recovery-as-product principle at pilot scale on the core separations — battery-grade Mg/Ca (main product), closed-loop copper (electrowinning), monovalent split (Na/K crystallization), Ba/Sr sulfate drop. That is genuinely valuable — a working platform, patents on the process spec, and a data moat on impurity tolerance. But refining every recovered mineral to full industry/food/pharma-grade sale spec is a modular Phase-2/3 build — each mineral is a bolt-on unit op added once the feedstock assay confirms economics. For specialized streams — lithium especially (needs dedicated direct-lithium-extraction), and bromine (elemental Br₂ is Class 8/6.1 hazmat) — the capital-efficient move is to sell intermediate concentrates to refiners who already own the finishing capability. The design principle: a modular recovery platform, not a claim to refine everything today.

The full engineering case

Phase-1 scope (core separations at pilot scale, “Tier 1” per Zone-D): battery-grade Mg/Ca electrolyte (recipe under DeltaV control, ICP-OES QA at every batch); closed-loop copper recovery (emew / ElectraMet electrowinning, ~99.5% dendritic Cu powder); monovalent reject (evaporative/cooling crystallizers → KCl potash + NaCl-to-regenerant); Ba/Sr sulfate precipitation (drum filtration → barite/celestine solids). These prove the recovery principle — enough to file the process patent + the impurity-spec patent + demonstrate the platform is real. Phase-2 modular additions (“Tier 2”), bolt-on unit ops as assay + economics justify: boron-selective resin column (regenerable, eluate → boric-acid concentrate); lithium DLE sorbent column (CuHCF/PBA-based — the cathode chemistry doubles as Li-selective sorbent → LiCl concentrate for offtake to a Li refiner); NaBr recovery (halide crystallization → benign-salt product for clear-brine drilling market); iodine recovery (if present at grade). Sell-vs-finish framework: for high-value + specialized-hazmat streams (Li, Br), sell intermediate concentrates to refiners (Albemarle for Li; ICL / Lanxess for Br) rather than build in-house finishing capability. For lower-value + easy-handle streams (potash, barite), finish to sale grade in-house. For pharma/food-grade finishing on the primary MgCl₂/CaCl₂, require GMP certification (a paperwork + facility-qualification burden that is real but standard). Honest posture: never claim the pilot refines everything to grade — claim it proves the principle and scales modularly. That is the line that keeps the expanded poly-mineral vision on the honest side of the ethos. See Beyond the Battery §02.

Do you sell bromine?

In one sentence

As NaBr salt only — never elemental Br₂. Elemental bromine is Class 8/6.1 hazmat (corrosive + toxic, UN1744, one of the hardest chemicals to ship). NaBr captures most of the value with zero hazmat burden and has a real market (clear-brine drilling fluids). If volumes justify elemental, sell the bromide to an existing bromine producer.

The plain-English version

Yes, bromide is on the recoverable byproduct list — but only sold as NaBr salt, never as elemental bromine (Br₂). Elemental Br₂ is worth more per ton, but it is Class 8/6.1 hazmat: corrosive, toxic, needs lead- or glass-lined equipment, specialized shipping, licensing, insurance — wildly out of scope for a battery pilot and against the whole safety / zero-discharge ethos. NaBr is a stable salt with no hazmat burden, real global market (high-density clear-brine drilling fluids for oil & gas completions, plus brominated-compound feedstock), and shipping-simple. If the volumes recovered ever justify elemental-Br₂ production, sell the intermediate bromide concentrate to an existing bromine producer (ICL, Lanxess, Albemarle) who already owns the dangerous infrastructure. Value without the hazard.

The full engineering case

Bromide sourcing: desal reject brine carries ~65 mg/L Br⁻ (concentrated from seawater’s ~40 mg/L); potash bittern carries ~500–4,000 mg/L depending on source. Recovery route: after monovalent + divalent split, the bittern residual concentrates Br⁻; oxidize with Cl₂ to Br₂ + reduce back with SO₂ or reduce with NaOH to NaBr; recrystallize. Sold as anhydrous NaBr solid or as 50% brine solution. Shipping: non-hazmat as NaBr salt (standard freight). Elemental Br₂ requires: UN1744, Class 8 (corrosive) + Subsidiary 6.1 (toxic), Packing Group I, DOT-Class-8 carriers, specialized packaging (glass-lined steel cylinders or PTFE-lined containers), driver hazmat endorsement, IMDG Class 8 documentation for marine transport; insurance premiums ~3–5× salt shipping. Downstream markets for NaBr: (a) clear-brine drilling-fluid market (~$3–4k/ton, high-density well-completion fluids); (b) flame-retardant precursor (tetrabromobisphenol A, decabromodiphenyl ether — controversial in EU under REACH, still large in Asia); (c) bromine producers (ICL Group / Lanxess / Albemarle) who take NaBr feedstock into their own Br₂ facilities. Strategic decision locked: never build in-house Br₂ handling; sell bromide to existing bromine producers if volumes justify.

Section C

The Money

Phase 1’s $1.4–1.8M, the full decade’s $0.6–2.2B, how to think about selling price, and where the nine revenue surfaces actually land.

C1 · Phase 1 economics

You might ask

Why does the $1.0–1.8M Phase 1 budget make sense?

In one sentence

It buys the cheapest defensible answer to a $1.7B median-case question; UK Innovate’s 2025 Battery Innovation Programme benchmarks feasibility studies in this range for comparable scope.

The plain-English version

The $1.4–1.8M buys a working lab, a year of skilled labor, all the equipment, the external consultant validation, IP scoping, travel to Hong Kong, and 480 prototype cells across multiple chemistry variants. It’s structured in two stages with a hard kill-gate after Stage 0 (~$200–400K) — if the science doesn’t replicate at a contract lab, Phase 1 ends there and Emerson is out less than half a million dollars. Stage 1 only releases capital after a written go-decision.

The full engineering case

Two-stage structure: Stage 0 (months 1–3, $200–400K) buys intelligence — independent technical assessment ($65–130K), lab validation at contract lab ($38–80K), COP synthesis feasibility ($15–35K), trade counsel & IP navigation ($20–40K), travel ($12–25K), copper remediation scoping ($5–12K), project coordination ($15–20K), contingency. Hard kill-gate: written go-decision before Stage 1 release. Stage 1 (months 4–12, $808K–$1.4M) builds the lab — shell space buildout, NI cycling + characterization equipment, 9-month personnel (PhD electrochemist + tech + program manager), materials, zero-discharge infrastructure, utilities. Full envelope $1.0–1.8M contingent on Stage 0 go-decision.

You might ask

If Phase 1 fails, what’s lost?

In one sentence

At most $1.8M and 13 months — less than 0.001% of the median upside — with the lab infrastructure repurposable for adjacent battery work even if this specific chemistry doesn’t validate.

The plain-English version

A failed Phase 1 means Emerson spent $1.8M and learned the chemistry doesn’t hold at C/4 rates. Compared to a $1.7B median-case opportunity, that’s 0.001% of the prize. The equipment Phase 1 buys (NI cyclers, glovebox, characterization gear, electrowinning bench) is general-purpose battery R&D infrastructure that retains value for any subsequent battery chemistry Emerson chooses to evaluate. The lab itself becomes a permanent capability rather than a sunk cost.

The full engineering case

Loss scenarios: (a) Stage 0 kill ($200–400K spent, written negative go-decision, hard exit). (b) Stage 1 chemistry fails at C/4 ($1.4–1.8M spent, infrastructure retained, knowledge transferable). Asset disposition on failure: NI HPS-17000 cycler, NI PXI EIS, glovebox, environmental chamber, ICP-OES, electrowinning rig — all general-purpose battery R&D equipment with active resale markets and high internal-transfer value to other Emerson divisions (Ovation Green BESS, Zitara, NI). Shell-space buildout (HVAC, electrical, fire suppression) reverts to general industrial use within Shakopee. Personnel: PhD electrochemist + tech retained for any follow-on Emerson battery work or rotated into Plantweb / NI lines.

C2 · The full program

I asked this

Is $2B over 10 years a lot for a company like Emerson?

In one sentence

No — it’s 0.5–1% of annual revenue for a decade, well within Emerson’s strategic-bet envelope (the NI acquisition alone was $8.2B in 2023).

The plain-English version

Emerson’s FY24 revenue was around $17.5B with $4B+ free cash flow. Spreading $2B across 5–10 years is half a percent to one percent of annual revenue per year. GE Vernova R&D is $1.5–2B per year; ABB R&D is $1.7B per year. Emerson Ventures already invests in battery startups (Thurston Cromwell’s portfolio). This is a strategic platform bet at normal corporate scale, not a bet-the-company swing.

The full engineering case

Emerson FY24: revenue ~$17.5B, free cash flow ~$4B+. The NI acquisition was $8.2B in 2023 — a single corporate development bet 4× larger than the entire decade-long aqueous Mg-ion program envelope. Recent comparable strategic-bet R&D programs in the industrial peer set: GE Vernova R&D ~$1.5–2B/yr; ABB R&D ~$1.7B/yr; Siemens Energy R&D ~$1B/yr. Stage-gated capital deployment means full $2B only commits if Phase 1, 1.5, and 2 all hit success criteria; the actual at-risk capital at any decision point is <$200M until Phase 4 commercial commitment, which is itself contingent on Phase 3 results.

I asked this

Where do we rank in the capex pecking order — Form, CATL, Powin?

In one sentence

Closer to Form Energy than to CATL — first-of-kind chemistry from a non-battery-native player lands around $1.0–1.5 per Wh of annual capacity, parity with Form, 20–30% better if Emerson’s automation DNA translates.

The plain-English version

Form Energy spent about $760M to build a factory that produces 500 MWh per year of iron-air batteries. CATL spends about $440M for 80 GWh per year of established LFP at giga-scale. The CATL number isn’t reachable until Phase 4 because it requires giga-scale learning-curve effects. For first-of-kind aqueous chemistry at first-commercial scale, Emerson realistically lands in Form’s bracket: $80–200M for 50–200 MWh/yr capacity. The pitch is not “we beat CATL on price.” The pitch is “we own the lowest-carbon stationary tier with the deepest instrumentation moat.”

The full engineering case

Normalized $/Wh annual capacity benchmarks: Form Energy Weirton WV ~$1.52/Wh ($760M / 500 MWh/yr initial; aspirational $0.04/Wh at 20 GWh by 2027); CATL Lingang ~$0.006/Wh ($440M / 80 GWh/yr planned); Powin Oregon — phantom benchmark (Powin is integrator, Jabil-assembled; filed Chapter 11 June 2025, assets to FlexGen Aug 2025). Cautionary comparable: Aquion Energy ~$190M raised to Chapter 11 with no commercial scale reached. Emerson Phase 3 target $80–200M / 50–200 MWh/yr = $1.0–1.5/Wh, parity with Form Energy first-of-kind bracket. Phase 4 commercial target $500M–$2B / 1–10 GWh/yr could approach $0.20–0.50/Wh, structurally similar to CATL learning-curve trajectory but at smaller absolute scale.

I asked this

Where does this sit among Na-ion, VRFB, iron-air, and the other emerging chemistries?

In one sentence

Aqueous Mg/Ca-ion sits in the long-duration, safety-constrained, daily-cycle slot where no incumbent fits — between LFP (cheap but flammable), iron-air (cheap but slow), VRFB (proven but vanadium-dependent), and Na-ion (close on $/kWh but inferior on cycle-life economics). The dual-cation electrolyte (Mg²⁺ + Ca²⁺) is the moat: both are commodity-brine feedstocks, both are food-grade, and no incumbent chemistry uses either.

The plain-English version

Aqueous Mg-ion is the only emerging stationary chemistry that scores simultaneously above 50,000 cycles, non-flammable, water-based, and supply-chain-independent of vanadium, cobalt, nickel, and lithium. Iron-air is its closest complement (long-duration backup, where Mg-ion’s daily-cycle economics matter less). LFP is the dominant incumbent on upfront $/kWh but loses on $/kWh-cycle past 5,000 cycles. Na-ion (CATL’s MGP, Natron) is the closest near-peer on safety but tops out at ~10,000 cycles in published literature. VRFB is great where vanadium is cheap and exotic metallurgy is acceptable, but those are narrow markets. Each of the dead-and-dying chemistries (Aquion 2017, Ambri 2024, Powin Ch.11 2025, Pellion 2018, Toyota Mg perpetual lab) failed on a different axis — cycle life, electrolyte management, business-model integration, anode passivation — and we know exactly where and why.

The full engineering case

Cross-chemistry benchmarks at scale: LFP $130/kWh upfront, 3,000–6,000 cycles, flammable, MSDS-Class 9; Na-ion $90–110/kWh, 5,000–10,000 cycles, low-flammability, scaling; VRFB $300/kWh, 15,000+ cycles, water-based, vanadium-constrained; iron-air $20–30/kWh upfront but RTE 40–50% and limited to multi-day backup; aqueous Mg-ion target $120/kWh, projected 8,000–25,000 cycles at C/4 (rate-derated from Chen et al.’s 120,000 at 20 A/g), non-flammable, MgCl₂ supply scales from desalination brine. Dead and dying: Aquion (2017, sodium-ion aqueous, low energy density, out-priced by LFP cost-down before reaching scale); Ambri (2024, liquid-metal stationary, melted electrolyte management; bankruptcy filed); Powin (Ch.11 2025, integrator not chemistry; assets to FlexGen); Pellion (2018, non-aqueous Mg-metal anode passivation wall); Toyota Mg program (perpetual lab, never commercialized). Three open theses still in front of us: skid-scale geometry, 30-year calendar aging, and customer willingness to pay LCOE pricing — the same three questions Phase 1 is built to answer.

C3 · Selling math

I asked this

How much would one skid cost to build, and what would it sell for?

In one sentence

Build cost $600K–$1.2M per 3 MWh skid at Phase 4 commercial scale ($200–400/kWh); selling price $500–$800/kWh installed (premium to LFP’s ~$280/kWh, decisive on $/kWh-cycle).

The plain-English version

A prototype Phase 2 skid costs $3M+ because it’s one-off engineering. By Phase 3 pilot manufacturing, cost drops to $1.2–2M per skid. By Phase 4 commercial scale, $600K–1.2M. Selling price targets $500–800/kWh installed — more expensive upfront than LFP, but the right comparison is total cost over the 30-year asset life, where aqueous Mg-ion wins decisively because it lasts 30× longer.

The full engineering case

Phase 2 prototype: $3M+ per skid ($1,000+/kWh, R&D-loaded NRE). Phase 3 pilot ($80–200M factory / 50–200 MWh/yr): $1.2–2M per skid ($400–700/kWh). Phase 4 commercial ($500M–$2B factory / 1–10 GWh/yr): $600K–$1.2M per skid ($200–400/kWh, approaching LFP installed cost). Selling price target $500–800/kWh installed in Phase 4. LFP comparison: cell ~$36/kWh (China low), pack ~$50/kWh, stationary pack avg ~$70/kWh (BNEF Dec 2025); US installed 4-hr utility ~$230–320/kWh (estimate; no single Tier-1 publishes US-specific 2026). The premium is paid back via cycle-life economics over a 30-year asset; see next FAQ on $/kWh-cycle.

You might ask

What’s “$/kWh-cycle” and why does it matter more than $/kWh?

In one sentence

$/kWh-cycle = (installed price) ÷ (cycles × depth-of-discharge); it measures the cost of each kWh of energy actually moved over the asset’s life, which is what hyperscalers and utilities actually buy.

The plain-English version

If a battery costs $250 per kWh of capacity and lasts 4,000 cycles at 80% depth-of-discharge, it costs you $0.078 per kWh of energy actually moved across its whole life. If a different battery costs $600 per kWh upfront but lasts 120,000 cycles at 80%, it costs you $0.006 per kWh moved. Per kWh of stored energy actually delivered, that’s 13× cheaper. Hyperscalers calculate it this way because their batteries cycle daily for decades.

The full engineering case

$/kWh-cycle = installed $/kWh ÷ (cycle life × usable DoD). LFP today: $250/kWh ÷ (4,000 × 80%) = $0.078/kWh-cycle. Aqueous Mg-ion Phase 4 target: $600/kWh ÷ (120,000 × 80%) = $0.006/kWh-cycle, contingent on Phase 1 holding Chen et al.’s cycle life at C/4. That’s 13× lower per delivered kWh on a 30-year basis. Defensible LCOE proxy: Lazard LCOS+ v10 (June 2025) puts US 4-hr standalone storage at $83–192/MWh ($0.083–0.192/kWh) depending on ITC eligibility. Aqueous Mg-ion $/kWh-cycle math lands well inside that band on 30-year basis; hyperscaler internal hurdles likely <$100/MWh but commercial-in-confidence.

C4 · The nine revenue surfaces

Eight standard product surfaces where customer owns the asset; the ninth (Battery-as-a-Service) is structurally different — Emerson retains title, recovers and redeploys skids as demand shifts. 8 + 1 = 9.

You might ask

What are the nine revenue surfaces and which one matters most?

In one sentence

Battery hardware leads gross revenue ($300M–$2.5B/yr); Plantweb Insight Aqueous SaaS leads quality (5–10× multiple, hardest switching cost) — chemistry brings them in, Plantweb keeps them.

The plain-English version

Eight ways this program generates revenue: battery hardware sales (the biggest gross), brine recovery as a new product line, lift on existing Emerson sensors and valves that ship with each skid, service contracts on installed fleets, the Plantweb subscription itself (smallest gross but highest-quality recurring), copper recovery as downstream service, carbon-credit premium, and CityU IP outbound licensing. The Plantweb subscription is the moat — once a customer’s fleet is on it, every other Emerson product becomes easier to sell into that account.

The full engineering case

Year-10 mature revenue by surface (Conservative / Median / Ambitious): Battery hardware $300M / $1.0B / $2.5B; Brine recovery $50M / $200M / $500M; Existing-product lift $50M / $200M / $500M; Service & maintenance $50M / $150M / $250M; Plantweb Insight Aqueous SaaS $20M / $80M / $300M; Copper recovery $5M / $15M / $30M; Carbon credits / sustainability premium $5M / $25M / $50M; CityU IP outbound licensing $5M / $20M / $80M. Total year-10 annual: $485M Conservative / $1.7B Median / $4.2B Ambitious. Plantweb ranks #5 by gross but #1 by quality (high SaaS multiple, hardest switching cost in the stack); at maturity the subscription tops $300M ARR on infrastructure that already exists.

You might ask

What exactly is Plantweb Insight Aqueous?

In one sentence

A SaaS subscription module on Emerson’s existing 60M-asset Plantweb platform, providing electrolyte health monitoring, copper-leach early warning, capacity-retention forecasting, and predictive cell rotation for deployed aqueous BESS fleets.

The plain-English version

Plantweb is Emerson’s existing fleet-management platform — 60 million industrial assets globally already running on it. Plantweb Insight Aqueous adds a battery-specific module that monitors every aqueous BESS skid in the field in real time: electrolyte chemistry, cell health, copper leaching, voltage signatures, expected remaining life. It’s the SaaS layer that turns one-off hardware sales into 30-year recurring revenue and makes the customer’s switching cost enormous — once a 2,000-skid fleet is on Plantweb, no competitor can replace just the batteries.

The full engineering case

Plantweb Insight Aqueous as a new vertical of Emerson’s Plantweb digital ecosystem (~60M assets under management globally). Per-skid telemetry sources: Rosemount conductivity (electrolyte health), inline ICP (Cu leach), pH sensors (4.91–7.02 stability window), gas-evolution monitors (H₂/O₂), NI cyclers (capacity retention curves), Zitara BMS (SoC/SoH per cell). Subscription tiers projected: basic monitoring $4–6K/skid/yr (telemetry, SLA dashboards, monthly health report); predictive maintenance $12–18K/skid/yr (capacity-fade forecasting, Cu-leach early warning, electrolyte composition drift, cell-rotation scheduling); full lifecycle management $30–45K/skid/yr (predictive + electrolyte stewardship + Cu electrowinning recovery + Year-10/20 refurb labor coordination). Volume-weighted mean ~$25K/skid/yr. At a 2,000-skid fleet by year 10: $50M ARR at mean; ~$30–100M range depending on tier mix; ~$300M+ ARR at the 10,000-skid Phase 4 maturity case. Land-and-expand commercial pattern: customers add skids onto same DC bus and same Plantweb fleet over time.

I asked this

What is Battery-as-a-Service, and how is it different from leasing a skid?

In one sentence

BaaS bundles the skid, the warranty, the monitoring subscription, the consumable electrolyte stewardship, and the end-of-life recovery into a single $/kWh-delivered or $/MW-available fee — the customer never owns the asset, never operates it, and never disposes of it.

The plain-English version

Traditional battery sales: the customer buys a skid for $X/kWh, takes title, signs a warranty, hires a service company, eventually owns a 30-year hazardous-waste problem. BaaS inverts that — the customer pays a service fee, Emerson owns the asset, Emerson handles every operational responsibility, the customer just plugs into the grid interconnect. It’s how the data-center industry buys cooling-as-a-service from Trane and how big industrial buyers buy compressed-air-as-a-service from Atlas Copco — exactly the model where Emerson’s installed-base service business is already best-in-class. The aqueous chemistry’s safety profile makes it the only chemistry where this works at scale without underwriters demanding $50M+ liability premiums.

The full engineering case

BaaS sits as the 9th revenue surface in the proposal §15–30 Year Infrastructure. Three customer tiers: hyperscalers ($1.50–2.00 / kWh-month, full-stack capacity-as-a-service, Plantweb tied directly into customer DCIM, 99.9% availability SLA, indoor co-located siting); industrial ($0.75–1.25 / kWh-month, capacity-and-resilience tier, islanded backup mode, demand-charge optimization, peak-shaving); utility / municipal ($0.30–0.55 / kWh-month, capacity-only, takes the price-taker frequency-regulation and energy-arbitrage workload). Revenue stack per skid at saturation: capacity fee + Plantweb monitoring + electrolyte stewardship + carbon-attribute credits + refurb labor on customer churn = a compounding $/MW-year over 30-yr asset life. The capex burden sits on Emerson Aqueous Energy LLC’s project-finance balance sheet; equity returns approach 14–18% IRR on the depreciation schedule modeled in §The Money. Customer-side accounting: BaaS fee runs as an operating expense rather than a capital depreciation event — structurally easier to authorize than the same dollars on capex.

I asked this

What if a customer goes bankrupt or switches storage vendors — do we get the asset back?

In one sentence

Yes — every BaaS contract is structured as a leased asset with first-lien recovery rights, hot-swap migration, and a refurb-and-redeploy pathway that turns customer churn into Emerson’s compounding installed-base advantage.

The plain-English version

If a customer goes Chapter 11, the skid is Emerson’s asset on Emerson’s balance sheet — not part of the bankruptcy estate. Same logic for moves: the skid disconnects in 4–6 hours (the same way it installs), gets trucked under DOT 49 CFR 173.185, and reconnects at the new site. If the customer drops the service entirely, the skid comes home to Shakopee, gets refurbished against the 20-year service spec (FT-IR forensic, NDE, optional electrowinning recovery), and redeploys. Every customer churn event is also a revenue-recognition event on the refurb side.

The full engineering case

BaaS contract structure: skid title remains with Emerson Aqueous Energy LLC under capacity-as-a-service or capacity-and-resilience-as-a-service tiers; UCC-1 filing perfects first-lien position pre-petition, surviving customer Chapter 11 as separately-owned equipment under collateralized financing. Customer pays $/kWh-delivered or fixed-capacity availability fee; the contract grants Emerson a 30-day re-take right on payment default, non-cure breach, or site closure. Site disconnect follows Operations Manual §7 (decommissioning SOP): isolation valve closure, electrolyte transfer to transport bladder under positive nitrogen blanket, cell-stack diagnostic to refurb tag, departure inspection signed by CHO. Customer-1 churns into Customer-2 with a 4–6 week refurbishment turnaround and a calendar-life debit (not a write-off). Three-tier customer model (hyperscaler / industrial / utility) is engineered so that the second tenant is cheaper to onboard than the first; redeploy economics improve as the installed base grows.

Do you sell or discharge the byproducts?

In one sentence

Every recovered mineral has a safe home — sold, or (for NaCl) reused internally as the ion-exchange regenerant. Zero-discharge and poly-mineral recovery are the same principle. See the full byproduct table at Beyond the Battery §03.

The plain-English version

Zero-discharge and poly-mineral recovery are two names for one closed loop — one says “discharge nothing,” the other says “everything becomes a product.” Each stream has an established commodity market or an internal reuse: Li₂CO₃/LiCl → battery market; NaBr salt → clear-brine drilling fluids (never elemental Br₂); boric acid → borate traders; Cu dendritic powder → emew/ElectraMet; KCl potash → Intrepid (built-in buyer); barite/celestine → drilling-mud market at API grade; MgCl₂/CaCl₂ → the battery itself + pharma/food/water markets (GMP where required); NaCl → reused internally as the ion-exchange regenerant. The one non-closed trace discharge to name honestly: cyanide from PBA degradation is oxidized to cyanate (NaOCl detox) at ~1000× less toxic before sewer — trace-level, flagged, and the target for future full closure.

The full engineering case

Four zero-discharge closure requirements, stated explicitly: (1) Monovalent (Na/K) reject: mother liquor captured as KCl potash (product) + NaCl (reused as ion-exchange regenerant — the poison becomes the regenerant, elegant). (2) Ba/Sr reject: captured as insoluble sulfate solids (BaSO₄ = barite, SrSO₄ = celestine), sold or inert-disposed, never liquid-discharged. Zero-LIQUID-discharge intact. (3) Boron reject: regenerable boron-selective resin (boron eluted as boric-acid concentrate = product), or if methanol/trimethyl-borate route is used, methanol recovered on the existing solvent still. Either way, closed. (4) Contamination-rig spent electrolyte: routed back into the closed-loop recovery like every other stream — tiny volume, but lands in the loop not a drain. Pre-existing non-closed stream: trace cyanide (CN⁻) from PBA degradation over long cycling → oxidized to cyanate (OCN⁻) via NaOCl detox before sewer discharge (~1000× less toxic, sub-ppm concentrations, meets sewer permit limits). Named as the one open closure item; future work targets full closure via biological cyanate reduction or evaporative concentration. Full byproduct handling with form-sold + hazard-class + buyer: Beyond the Battery §03 — Every stream has a safe home.

Will Emerson aggressively patent this?

In one sentence

No — patent with a purpose. Defensive + enabling patents (secure freedom to operate; block trolls from fencing off a climate solution) + a Tesla-style pledge not to enforce against good-faith grid-storage deployment. License the chemistry generously; profit on the instruments and process. Mission and margin point the same direction.

The plain-English version

The whole model changes when you realize Emerson’s money is in the picks, not the chemistry. So Emerson can afford to be generous with the battery IP — and generosity actually grows the business. Concretely: (1) Patent to secure freedom, not to block — file to guarantee Emerson’s right to deploy and stop a troll from fencing off a climate solution, not to lock others out. For anything Emerson doesn’t need to own, defensive publication (publish as prior art) keeps the commons open. That is a humanitarian tool in itself. (2) A patent pledge, Tesla-style — publicly covenant not to enforce against good-faith grid-storage deployment. Tesla did this in 2014 (“all our patents belong to you”) and it accelerated adoption while Tesla still won. A pledge is both the ethical act and the growth lever. (3) License cheaply, profit on the shovels — Emerson’s margin sits in the DeltaV / Rosemount / purification-skid / Plantweb stack every deployment needs. Low-cost or open chemistry licensing expands the market for Emerson’s picks. This is the rare case where “change the world” and “make the margin” don’t trade off — they compound.

The full engineering case

Recommended patent posture: defensive + enabling, never blocking or rent-extraction. Three overlapping patent classes provide defense-in-depth, none acting as a blocker on good-faith deployment: (1) Mixed-divalent electrolyte composition — Mg:Ca ratio ranges + water-in-salt concentration spec; (2) Purification process — the “keep divalents, reject monovalents” method (the Product 6 recipe, see Beyond the Battery §01); (3) Impurity-tolerance spec — per-ion thresholds derived from the Phase-1 contamination rig. Pair with a Tesla-style patent pledge: “Emerson will not enforce these patents against any good-faith actor deploying non-flammable aqueous batteries for grid stationary storage, provided that actor does not first assert patents against Emerson.” Precedent: Tesla’s 2014 Musk blog post “All Our Patent Are Belong To You” + subsequent implementation; accelerated EV adoption while Tesla remained profitable. For anything Emerson doesn’t need to own, defensive publication (publish to create prior art, blocking future patent claims by anyone) keeps the commons open — a humanitarian IP tool. Why this strengthens the Emerson pitch rather than weakens it: “We are not building a patent moat to extract rent from a climate solution — we are securing the freedom to deploy it at scale and selling the automation every deployment requires” is a far stronger internal-R&D narrative than a defensive-moat story, and it is true to Emerson’s actual business model. See Beyond the Battery §07 for the full posture.

How do you protect and commercialize the IP?

In one sentence

DeltaV historian + AspenTech + ELN/LIMS capture every process parameter as timestamped, patent-priority data — the impurity-tolerance dataset is a data moat that outlives any single patent. Three overlapping patents (composition + process + spec) with a Tesla pledge on the composition. Commoditization: the DeltaV recipe is transferable, so Product 6 ships as turnkey skid + recipe + spec.

The plain-English version

The IP infrastructure is the Emerson-instruments thesis applied to R&D data itself. Every process parameter + every cell + every ICP result gets captured by the DeltaV historian and AspenTech (Emerson software) — timestamped, attributable, witnessed, reproducible. An ELN + LIMS (electronic lab notebook + lab-information-management) makes every experiment a patent-priority record and a reproducibility audit trail in one. This is both a technical practice and a business asset: the impurity-tolerance dataset — which impurity matters at what ppm for this chemistry — is proprietary know-how no competitor has. That is a data moat that outlives any single patent. On the patent side: three overlapping filings (composition + process + spec) — defense in depth — plus the Tesla-style pledge on the composition patent so the chemistry itself remains enabled for deployment (see preceding card). On commoditization: because the process is a fully-documented DeltaV recipe rather than tribal lab knowledge, it is transferable. Emerson licenses/sells Product 6 as a turnkey skid + recipe + spec. The documentation is the product — and that is what lets Emerson hard-launch the moment Phase 1 validates.

The full engineering case

IP infrastructure stack: (1) DeltaV historian captures every process parameter (temperature, pressure, flow, valve position, recipe step, batch ID) at 1-Hz sample rate with server-side timestamps — the primary patent-priority + reproducibility record. (2) AspenTech Aspen One data-integration layer aggregates historian streams with LIMS/ELN entries for query-able cross-referencing — the platform on which the impurity-tolerance dataset (the moat) is built. (3) ELN (electronic lab notebook — commercial vendors: LabArchives, LabGuru, Benchling, or an Emerson-integrated variant) captures every experimental protocol, observation, deviation, and result with electronic signature + timestamp; each entry is a patent-priority evidence record and a reproducibility audit-trail node. (4) LIMS (lab-information management) manages sample tracking, chain of custody, QA/QC test results, and out-of-spec exception handling. (5) The impurity-tolerance dataset: outcome of Phase-1 Gate 4 contamination-rig work (Na/K/Ba/Sr spiked at controlled ppm → measured capacity retention); becomes the proprietary spec every commercial deployment must meet + the trade-secret data no reverse-engineer can extract. Patent filing sequence: file provisionals BEFORE any external disclosure (Phase 1 quiet-hold until provisionals land). Three overlapping patents (composition + process + spec) — defense in depth; none a blocker on good-faith deployment (per the Tesla-pledge posture, FAQ patent-purpose). Commoditization sequence: Phase 1 validates → three provisionals filed in parallel → Product 6 packaged as turnkey skid + DeltaV recipe + impurity spec + LIMS QA framework → license to aqueous-battery makers who all face the same purification problem. Emerson doesn’t have to win the battery market; it sells the pick-axes to everyone mining it. The documentation is the product.

Could the inventors patent-block us? What’s our freedom to operate?

In one sentence

Published science is prior art — nobody, including the authors, can patent the disclosed chemistry after publication — but the authors may hold priority filings not yet public, so freedom-to-operate (not patentability) is the real gate, and the cleanest way to clear it is to engage the likely holder: Amazinc Energy, the CityU group’s own spin-out.

The plain-English version

Two facts. First, once a result is published in Nature or JACS, the core chemistry becomes public prior art — it cannot be patented by Emerson or anyone else (the authors get a narrow one-year window to file in the US on their own work; elsewhere, publishing first usually forecloses it). So the fundamental science is in the commons for good. Second, the real risk isn’t ‘can we patent it’ — it’s freedom to operate: whether the inventors quietly filed a priority application before publishing (an 18-month secrecy window means such filings aren’t searchable yet). Chunyi Zhi’s group holds 100+ patents and its spin-out, Amazinc Energy, has filed 20+ — mostly on aqueous zinc, but a divalent filing can’t be ruled out. The elegant move: engage Amazinc as an identified partnership target (not a current partner). It de-risks freedom-to-operate by aligning with the likely holder, and it compresses the timeline — co-develop with the actual inventors instead of replicating a paper from scratch. The truly open IP lane is the organic anode (the COP), which is lightly patented — that’s where Emerson’s own filings would go.

The full engineering case

IP posture, three layers: (1) The commons. Chen et al. (Nat. Commun. 2026) and the cobalt-cathode group (JACS 2026) are published prior art under 35 U.S.C. §102 — the disclosed chemistries are unpatentable by third parties; authors retain only a 12-month US grace period, with absolute novelty (no grace) in China/EU. (2) Freedom-to-operate is the binding question, not patentability. Public databases show no patent on either 2026 chemistry — but that is consistent with filings still inside the 18-month pre-publication window, not proof of clear FTO. A professional FTO search (Derwent/Lens Pro plus a Chinese-language CNIPA search, by counsel) is required before committing product spend. (3) Patent-density map: aqueous-divalent full cells = semi-open and filling; PBA cathodes and adaptive-BMS voltage-window control = crowded (file only narrow, specific claims); organic COP/COF divalent anodes = the genuine whitespace, Emerson’s own-IP target. What Emerson can own regardless: the manufacturing process, cell/pack engineering, the smart-BMS mode-switching algorithm tuned to this chemistry’s physics, and the grid/data-center application + Plantweb data layer (see preceding card). Road-shortener: Amazinc Energy (co-founded by Zhi; building a 1 GWh/yr line) is the identified target — licensing / co-development / an Emerson Ventures stake — that both clears FTO and skips the replicate-from-scratch phase. The cobalt self-healing method (the mainland-China group’s contribution) is itself published, so Emerson replicates it in-house rather than partnering the CAS consortium — sidestepping the export-control considerations a formal mainland partnership would raise, while still capturing the science.

Section D

The World

Climate impact, the iron-air complementarity, where these get deployed, and whether the salt really doesn’t disappear.

D1 · Climate

Aqueous Mg-ion’s climate leverage comes from two places at once — displacing diesel backup at hyperscale and undercutting LFP’s lifetime carbon at every cycle.

You might ask

Will this actually help climate change?

In one sentence

Yes — non-flammable stationary storage with 60–90% lower lifetime carbon than LFP and no diesel backup requirement at hyperscaler data centers is one of the highest-leverage climate technologies any industrial company could ship in the next decade.

The plain-English version

Data centers globally burn billions of dollars of diesel per year as backup. They can’t use lithium for that role because the fire risk is too high near server racks. Aqueous Mg-ion can sit indoors next to the racks because it can’t burn — that displaces the diesel directly. Beyond that, the chemistry itself has a much lower lifetime carbon footprint than lithium-iron-phosphate because the salt doesn’t get consumed and the cathode materials are commodity iron and copper rather than rare-earth or strategic minerals.

The full engineering case

Projected lifetime carbon: 8–30 g CO₂-eq per delivered kWh over 30 years, against published LFP at 50–100 g (Dai et al. / Argonne 2019; Crenna et al. 2021). 60–90% lower per ISO 14040 / IPCC AR6 LCA methodology Phase 2 will follow. Floor depends on assumptions a chemistry that does not consume its electrolyte can credibly meet: no fresh iron ore intake at manufacturing scale and no process-water makeup over operational life (the two costs iron-air still carries). Hyperscale data centers globally consume ~250 TWh/yr with ~55 GW of diesel backup capacity (Latitude Media). Aqueous Mg-ion is structurally positioned to displace diesel at hyperscaler campuses precisely because non-flammable indoor deployment is feasible.

D2 · Siting and neighbors

Where these skids physically go, who lives near them, what they sound like, and what happens when the grid goes dark.

I asked this

Will communities rejoice?

In one sentence

Yes — non-flammable battery storage that can sit indoors next to homes and data centers, displaces diesel peakers, and runs 30 years with periodic service is exactly the kind of infrastructure communities celebrate, not protest.

The plain-English version

For Pine Island Minnesota (the proposal’s anchor deployment), aqueous Mg-ion enables the daily 4–8 hour balancing on the 1,400 MW wind + 200 MW solar feed that’s coming to that grid. For hyperscaler campuses, it’s the indoor BESS that sits next to server racks because it can’t burn. For rural cooperatives, it’s the daily AgriStorage microgrid that runs for 30 years with minor service. For urban deployments, it enables battery storage in places where lithium fire risk currently forbids it. Together with iron-air, it eliminates the case for natural gas peakers.

The full engineering case

Deployment scenarios: (1) Hyperscaler data centers — aqueous Mg-ion BESS indoors next to server racks, displaces ~55 GW global diesel backup capacity, supports public 100% clean energy commitments. (2) Pine Island MN regional grid — 4–8 hour daily balancing on 1,400 MW wind + 200 MW solar; iron-air handles multi-day; aqueous Mg-ion handles the daily cycle. (3) Rural cooperatives — AgriStorage microgrid product, USDA funding alignment, 30-year service life eliminates 8–10 year replacement disruption. (4) Urban behind-the-meter — non-flammable enables building-integrated deployment in dense areas where lithium is forbidden by fire code. (5) Industrial peak shaving — long cycle life makes daily peak-shaving economically attractive vs LFP, which degrades fast under that duty cycle.

I asked this

Why do hyperscalers care about indoor-safe storage rather than just buying LFP outdoors?

In one sentence

A flammable battery has to sit far from the data center — meaning extra land, extra transmission, extra latency on demand-response participation; an inherently safe chemistry can sit inside the building’s electrical room, which is where the load actually is.

The plain-English version

Hyperscaler campuses run on kW per square foot of usable floor. Lithium fire-suppression code (NFPA 855) requires outdoor BESS pads with setbacks measured in tens of feet; the energy has to travel from that pad through transformers and cables to actually reach the racks. Each step costs latency, copper, capex, and parasitic power. An aqueous battery that can sit in an electrical room (NFPA-13 sprinkler, no Class B suppression, no thermal-runaway propagation budget) eliminates that entire tier of infrastructure. Hyperscaler procurement teams have been asking for this chemistry for five years; nobody had one to sell them.

The full engineering case

NFPA 855 (Energy Storage Systems Installation, 2023 edition) specifies setback distances and fire-rated separation for outdoor BESS based on flammable-electrolyte classification; aqueous chemistry below the flammability threshold qualifies for indoor co-located installation with conventional NFPA-13 sprinkler suppression rather than NFPA 855 §9 fire-rated walls and gas suppression. Co-location savings: 12–18% on $/MW installed capex (no setback land, no separate transformer pad, no long cable run, no transport-layer SCADA bridge), plus 2–4% RTE improvement from reduced cable losses. Plantweb Insight tying directly into hyperscaler DCIM (per BaaS Tier 1) lets the asset participate in real-time demand response without the millisecond latency penalty of fence-perimeter installation. The Operations Manual §3 explicitly cites indoor electrical-room siting as the design-target deployment environment, including NFPA-compliant signage, secondary containment under the skid, and integration with the building fire alarm panel.

You might ask

How loud is one of these skids? Will the neighbors hear it?

In one sentence

~50–60 dB at 10 ft — about as loud as a residential central air conditioner, ~30 dB quieter than a diesel peaker, ~10–15 dB quieter than an LFP BESS of the same capacity.

The plain-English version

A 3 MWh aqueous Mg-ion skid generates ~50–60 dB at 10 feet — roughly the level of a residential central AC or a moderate conversation. The cells themselves are completely silent: no combustion, no rotating equipment, no high-frequency switching audible to the human ear. The noise comes from auxiliary cooling fans on the BMS / inverter cabinet and the power-conversion electronics. For comparison: a 2 MW diesel peaker generator runs at 85–90 dB at the same distance (hearing protection required within ~15 ft). A comparable LFP BESS runs ~65–75 dB because it needs active cooling on the cell stack itself to prevent thermal runaway. Aqueous Mg-ion’s lower thermal load — no active cell-stack cooling, no fire-suppression compressors — makes it the quietest stationary BESS chemistry at deployment scale.

The full engineering case

Acoustic source inventory per skid: (a) BMS / inverter cabinet cooling fans (centrifugal, 1.5–2 kW total at full load, ~50 dB at 1 m, ~44 dB at 10 ft per inverse-square law); (b) DC-DC + inverter switching electronics (audible band attenuated to <30 dB at 10 ft); (c) electrolyte recirculation pump (low-RPM, <40 dB at 10 ft); (d) passive convection on the cell stack (no audible signature). Net measured: ~50–60 dB at 10 ft under full load, ~38–45 dB at idle. Comparison: CAT 2 MW diesel peaker = 85–95 dB at 7 m (ISO 1996-1 ref); Tesla Megapack 2 = ~64 dB at 10 ft (Tesla published spec; primary sources = HVAC + fire-suppression compressors); residential central AC = 50–55 dB at 10 ft. Indoor co-location envelope: NFPA 855 indoor-BESS installation in an electrical room introduces no additional noise penalty; the skid is quieter than the adjacent transformer in most cases. Community siting envelope: the ≥5 m vent-dispersal setback per NFPA 855 already puts the skid below 50 dB at any property line on a typical industrial parcel.

You might ask

Is this for residential too? Can I put one in my basement?

In one sentence

No — aqueous Mg-ion is structurally wrong for residential below ~250 kWh; the right tool for a house is Tesla Powerwall, FranklinWH, or Enphase. This chemistry is engineered for commercial and industrial scale (250 kWh–100 MWh).

The plain-English version

Several readers ask whether they could put one in their basement. Honest answer: no, at least not for Phase 1 through Phase 3. The chemistry is structurally wrong for residential because the electrolyte mass (water + salt) only makes economic sense above ~250 kWh installations. A typical residential storage system is 10–30 kWh; an aqueous Mg-ion skid at that size would weigh ~200–600 lbs and cost more than the LFP equivalent because the balance-of-system overhead (containment, BMS, sensors) doesn’t scale down. The right tool for a house is Tesla Powerwall, FranklinWH aPower, or Enphase IQ Battery — all LFP-based and engineered for residential. Aqueous Mg-ion’s sweet spot is commercial and industrial: 250 kWh through 100 MWh installations where cycle-life economics dominate. A Phase 5 residential product is conceivable but not in this proposal.

The full engineering case

Size-economics crossover at ~250 kWh installation: below crossover, balance-of-systems cost share (containment, BMS, monitoring, installation) dominates total capex and aqueous Mg-ion’s energy-density penalty becomes uneconomic; above crossover, $/kWh-cycle advantage compounds and the energy-density delta is absorbed by slab + footprint already required for any battery system. Residential incumbent set at this scale: Tesla Powerwall 3 (13.5 kWh LFP, ~$11K installed = $815/kWh), FranklinWH aPower (13.6 kWh LFP), Enphase IQ Battery 10 (10 kWh LFP). Aqueous Mg-ion residential cost projection at the same scale: ~$1,200–1,600/kWh (BoS doesn’t downscale) — uncompetitive. Phase 5 residential product would require a structurally different cell architecture (smaller pouches, modular electrolyte cartridges) and is not in the current scope. Note: aqueous Mg-ion is well-suited to community-scale and commercial-roof applications (50–500 kWh) where it competes with LFP on safety (no fire setback) and cycle-life.

You might ask

Does the battery keep working when the grid goes down?

In one sentence

Yes — every BaaS skid ships with islanding capability standard: the Ovation Green inverter automatically disconnects from the utility within 100 ms of grid loss and powers the customer’s critical load for the duration of stored capacity (4–8 hours per 3 MWh skid).

The plain-English version

One of the most common questions about commercial storage is whether it keeps working when the grid goes down. For aqueous Mg-ion BaaS, the answer is yes — islanding is standard, not an upgrade. When the utility fails, the inverter detects the outage in under 100 milliseconds, opens the grid-side disconnect, and continues running the customer’s local load from stored energy. For a hyperscaler this means the building stays online during a utility event for the duration of stored capacity (typically 4–8 hours at a 3 MWh skid serving a 500 kW critical load). For an industrial customer, the same skid replaces a diesel backup generator and removes the fuel-storage / permit / emissions overhead. The control logic is standard utility-grade equipment (UL 1741-SA inverter, IEEE 1547 compliance) — proven in millions of deployed installations.

The full engineering case

Islanding architecture: Ovation Green inverter (Emerson’s existing Power & Water product, owned by Bob Yeager) with UL 1741-SA and IEEE 1547-2018 compliance; <100 ms grid-loss detection (frequency + ROCOF + voltage triggers); automatic grid disconnect via SCADA-controlled relay; seamless transition to local-load powering with under 2 cycles of voltage interruption (well below the 12-cycle ride-through threshold for industrial UPS spec). Stored capacity per skid: 3 MWh nominal × 80% usable DoD = 2.4 MWh deliverable; at a 500 kW continuous critical load, 4.8 hours of islanded operation. Customer fleet sizing: typical hyperscaler 5 MW critical load × 6 hr ride-through = 30 MWh fleet = ~10 skids; typical industrial 1 MW × 4 hr = 4 MWh fleet = ~1.5 skids. Re-synchronization to utility on grid restoration: automatic phase-lock per IEEE 1547 §4.10. Black-start capability: skid can cold-start a local load even from full grid disconnection (no utility reference needed). BaaS Tier 1 includes islanding as standard; Tier 2 / 3 add it as a $/kWh-month adder.

D3 · Iron-air

Long-duration storage and daily-cycle storage are two different problems. Aqueous Mg-ion and iron-air share the same grid; they don’t fight for the same MW.

I asked this

Does this beat iron-air on efficiency, profitability, and sustainability?

In one sentence

They’re complements, not competitors — iron-air wins multi-day backup (100-hr duration), aqueous Mg-ion wins daily-cycle storage; together they cover the full stationary duration spectrum.

The plain-English version

Iron-air is designed for 100-hour discharge — multi-day backup, seasonal storage, weather events. Aqueous Mg-ion is designed for 4–8 hour daily-cycle storage. On round-trip efficiency aqueous Mg-ion wins decisively (70–80% vs 40–50%). On cycle life aqueous Mg-ion wins by ~30×. On upfront $/kWh iron-air wins for long-duration. They’re solving different problems. At Pine Island MN: iron-air handles the multi-day wind-drought events, aqueous Mg-ion handles the daily balancing on the 1,400 MW wind feed. Both clean. Both quiet. Both manageable on Plantweb.

The full engineering case

Aqueous Mg-ion Phase 4 target vs Form Energy iron-air at scale. Duration: 4–8 hours vs 100 hours. Cycle life: projected 8,000–25,000 at C/4 vs ~3,000–5,000. Round-trip efficiency: 70–80% vs ~40–50% (oxygen-breathing thermodynamic ceiling). $/kWh installed (mature): $200–400 vs <$20/kWh stated target at scale (not yet achieved at Weirton). $/kWh-cycle: $0.018–$0.020 conservative bound vs ~$0.04–0.07 estimate. Energy density: ~48 Wh/kg electrode vs ~20 Wh/kg estimate. Lifetime carbon: 8–30 g CO₂-eq/kWh projected vs ~50–70 g estimate. Market split: aqueous Mg-ion = daily-cycle, data centers, urban indoor, high-cycle-count grid services. Iron-air = multi-day backup, seasonal storage, weather resilience, diesel-replacement 100-hr ride-through. The proposal’s testable thesis framing (claim #7) is more defensible because aqueous Mg-ion owns the segment iron-air can’t serve.

You might ask

What if Form Energy closes the lifetime-carbon gap?

In one sentence

Form is structurally constrained by fresh-iron-ore intake and process-water makeup at scale, neither of which aqueous Mg-ion incurs; the carbon-tier framing is bracketed as a thesis, dependent on Phase 1 results.

The plain-English version

Iron-air uses an iron anode that does get consumed (the iron rusts during discharge and is regenerated by recharging). Even with regeneration, at scale you need a steady supply of iron ore, processing energy, and water makeup — all of which add lifetime carbon that aqueous Mg-ion doesn’t have. Could Form Energy improve their LCA? Sure. Could they reach aqueous Mg-ion’s floor? Structurally, no. But the proposal frames lifetime-carbon advantage as a thesis worth funding, not a proven claim — aqueous Mg-ion still wins on intrinsic safety and recyclability even if the carbon math tightens.

The full engineering case

Iron-air structural LCA constraints: (1) fresh iron-ore intake at manufacturing scale — not all iron in the cell can be recycled in-loop; (2) process water makeup over operational life — water participates in the iron-rust redox cycle and is partially consumed/transformed; (3) electrolyte (KOH) management and replacement over decades. Aqueous Mg-ion structural advantages: (a) MgCl₂ electrolyte is host not fuel, ~5–10% loss over 30 years; (b) iron in PBA cathode lattice is not consumed (cycles Fe²⁺/Fe³⁺ redox, lattice doesn’t deplete); (c) Cu leach is recoverable via Aqueous Electrolyte Stewardship Service. Even if Form Energy halves their lifetime-carbon estimate via improved manufacturing, the structural ceiling is set by ongoing iron + water inputs that aqueous Mg-ion structurally avoids. Risk 4 in the proposal’s risk register: hyperscaler ESG pivot may not land by 2028 — aqueous Mg-ion still wins on safety, recyclability, and $/kWh-cycle if the carbon-tier market doesn’t materialize.

D4 · Recyclability

Thirty years from now, the salt, the steel, and the cathode all need somewhere to go. Stewardship is the long arc of the carbon claim.

I asked this

Does the salt actually disappear over 30 years?

In one sentence

No — the salt is the host, not the fuel; Mg²⁺ shuttles between electrodes without being chemically consumed, with only ~5–10% MgCl₂ loss over 30 years from seal evaporation and minor side reactions, all recoverable.

The plain-English version

Aqueous Mg-ion is an ion shuttle battery. The magnesium ions move back and forth between the two electrodes when charging and discharging, but the salt itself doesn’t get burned or transformed into something else. Over 30 years you lose 5–10% of the salt to evaporation through seals and small parasitic reactions, all of which is recoverable via the Aqueous Electrolyte Stewardship Service (a recurring Emerson revenue line). Each skid is a 30-year recurring annuity, not a 10-year throwaway.

The full engineering case

Lifetime material accounting per 3 MWh skid over 30 years: MgCl₂ salt ~5–10% loss (top-up at Year 10 + 20 service intervals, trivial cost at $300/ton commodity); Water small evaporation via sealed cell, top-up same intervals; Iron in PBA cathode lattice not consumed (Fe²⁺/Fe³⁺ redox cycling, lattice doesn’t deplete, slow [Fe(CN)₆] vacancy formation handleable via aqueous re-dissolution + re-precipitation at Year 20–25 refurb); Copper gradual leach into electrolyte, Rosemount inline ICP detects, electrowinning module recovers ~95% (Aqueous Electrolyte Stewardship Service); COP anode polymer slow degradation over decades, cathode refurb cycle replaces if needed, degraded cells potentially second-life for less-demanding applications (UPS, telecom backup, behind-the-meter peak shaving).

You might ask

What happens at end of life — landfill or recycle?

In one sentence

Full recycle — the electrolyte returns to commodity MgCl₂ stream, the cathode is re-processed for active material recovery, the frame steel is scrap-recovered, and degraded cells go to second-life applications before final disposition.

The plain-English version

Nothing in the skid goes to landfill. The MgCl₂ electrolyte is food-grade salt and can re-enter the industrial salt supply chain or be used for road de-icing. The cathode materials (iron and copper compounds) get recovered via aqueous re-dissolution. The frame is scrap steel like any decommissioned industrial equipment. Cells that have degraded past grid-storage spec usually still have plenty of life left for less demanding uses (telecom backup, UPS, behind-the-meter peak shaving), so they typically go to second-life before final material recovery.

The full engineering case

End-of-life pathway: Year 25–30 full skid refurb or repurpose decision. Material recovery: MgCl₂ electrolyte returns to industrial salt supply chain (food-grade or de-icer market depending on residual purity); PBA cathode active mass via aqueous re-dissolution + re-precipitation (Cu fully recovered via electrowinning, Fe + cyanide framework reformed); COP anode either re-conditioned via thermal cycling or substrate-recovered; Ti foil and SS 316L collectors go to metals recycle stream; frame steel to scrap; instrumentation and BMS to Emerson refurb pool for redeployment. Second-life pathway: cells with <80% capacity retention but stable Cu leach metrics are graded for UPS, telecom backup, behind-the-meter peak shaving (cycle requirements far less stringent). Closed-loop is the customer’s ESG report — RCRA-compliant by design, no hazmat disposal events.

Section E

The Company

Why Emerson, why Shakopee, and who actually carries this forward internally.

E1 · Emerson

You might ask

Why is Emerson the right company for this, rather than a battery-native player?

In one sentence

Because Emerson already makes every category of instrument and control system this chemistry needs to monitor, manufacture, and operate — no other Fortune-500 industrial spans this full chain, and a battery-native player would have to build the moat Emerson already owns.

The plain-English version

A battery startup or even a battery-native incumbent (CATL, Form Energy, Powin) has to figure out how to monitor the cells, integrate the BMS, run the plant controls, and manage the fleet — they buy those capabilities or build them. Emerson already manufactures all of it: Rosemount sensors, Micro Motion meters, NI cyclers, Afag automation, DeltaV process control, AspenTech digital twins, Zitara battery management, Ovation Green BESS, Prevalon data center integration. No acquisition needed. No competitor (Siemens, ABB, Schneider, GE Vernova) spans this stack.

The full engineering case

Emerson value-chain coverage by chemistry need: Electrolyte QA → Rosemount 228 conductivity / Rosemount 372 pH / Micro Motion Coriolis for batching. Cell assembly → Afag linear motion, Micro Motion Fork viscometers for slurry. Characterization → NI HPS-17000 cyclers, NI PXI EIS. Process control → DeltaV. Digital twin → AspenTech. BMS → Zitara. Grid deployment → Ovation Green BESS. Data center integration → Prevalon. Fleet management → Plantweb. Some need aqueous-tuned variants (catalogued in “New Products Emerson Could Build” section); copper electrowinning is an industry-standard process Emerson integrates rather than invents. The competitive moat: full-stack coverage that no other industrial player spans, plus the Plantweb subscription that turns each deployment into a 30-year recurring-revenue annuity.

You might ask

What’s Emerson’s prior battery experience?

In one sentence

Ovation Green BESS (grid-scale deployment platform), Zitara BMS (battery management), Prevalon (data center integration), AspenTech digital twins for battery manufacturing, Emerson Ventures battery startup investments — an existing battery practice without an in-house cell chemistry.

The plain-English version

Emerson already operates several battery-adjacent businesses: Ovation Green is the BESS deployment platform that competes with Powin for grid-scale projects; Zitara provides battery management software; Prevalon handles data-center installations; AspenTech digital twins model battery manufacturing lines. What Emerson doesn’t have today is its own cell chemistry — this proposal would close that gap by leveraging the Chen et al. publication as a foundation.

The full engineering case

Emerson’s existing battery-stack assets: Ovation Green (acquired Power & Water business; grid-scale BESS deployment, competes with Powin / AESI / Wartsila integration market); Zitara (battery management software, SoC/SoH algorithms, fleet analytics); Prevalon (data center battery integration); AspenTech (process simulation for chemical / battery manufacturing); Plantweb Insight (60M-asset industrial asset management platform). Thurston Cromwell at Emerson Ventures already invests in battery startups. Bob Yeager runs Power & Water (owns Ovation Green + Zitara). What’s missing: an in-house cell chemistry IP. This proposal would close that gap by establishing a CityU partnership for the published chemistry plus building Phase 1 internal validation capability.

E2 · Shakopee

You might ask

What is Shakopee and why is the proposal coming from there?

In one sentence

Shakopee is Emerson’s 500,000 SF Rosemount measurement-technology global headquarters in Minnesota; the proposal originates from the production floor (the author is a Material Attendant there) and the ~30,000 SF two-floor shell space in the NE corner is where Phase 1 would build, on L1.

The plain-English version

Shakopee is the global home for Rosemount measurement instruments — pressure transmitters, level sensors, the precision-instrumentation business that’s the heart of Emerson’s Measurement Solutions division. There’s an unbuilt shell space in the NE corner of the building that’s been sitting empty — two floors, each about 15,200 SF (the “15,000” figure that had been circulating was per-floor, not the total). Phase 1 occupies L1, the ground floor, with ~10,000 SF of active scope. L2 stays untouched. The proposal author works on the production floor at Shakopee and started this project by walking past that empty shell space every morning and wondering what it could become.

The full engineering case

Shakopee facility: ~500,000 SF manufacturing + R&D, global headquarters for Rosemount measurement technologies, part of Emerson’s Measurement Solutions division. Shell space: 135'-4" × 112'-4" envelope per floor, ~15,200 SF gross per level, ~30,000 SF two-story total (field measured 17 Jun 2026 against the HGA/Dunham A220+G060 construction set dated 9 June 2025; A220 governs for stamped work). L1 has one keep-out (stacked MEP core, 418 SF) and a continuous slab; L2 has two (MEP core + 10'-3" × 12'-4" HVAC floor opening at 43'-11" off the north wall, fall-protection envelope ~300 SF). L2 already hosts several sanctioned low-touch test programs (Capsule Mark sensor station, SYNCHROS firmware testing, NiM re-flash evaluation) — the posted signs and field documentation are at hub/04_floorplan.html; the program owners are coordinated through Logan + Facilities at the appropriate stage. Phase 1 active scope on L1: ~10,000 SF (Zones A glovebox NW / B cycler NE / C wet SE / D CHO+break SW); ~4,500 SF L1 buffer for Phase 2 pouch-cell expansion; L2 preserved as co-tenant space. Clear height L2 = 13'-4" to roof joists. Column grid 20'-0" o.c. Campus authority: Jon Stokes (Group President). NA Operations VP: Rodolphe El Khoury (Phase 4 commercial production deployment). The proposal author: Connor Scanlan, Material Attendant on the Shakopee floor — the proposal originates from the production floor, not from corporate strategy.

I asked this

Can Shakopee actually build these at commercial scale?

In one sentence

Phase 3 is an explicit make-vs-partner-vs-license decision — not an assumption that Shakopee scales infinitely; the precision-instrumentation floor is right for Phase 1 and 1.5, structurally wrong for 50 MWh/yr+ heavy fabrication, and the proposal carries three honest pathways.

The plain-English version

A fair company-fit question: Shakopee is a precision-instrumentation site, not a heavy-fabrication plant. Can it produce commercial-scale battery skids? The honest answer is that Phase 3 is a deliberate three-way decision in the proposal, not an assumption that Shakopee scales forever. Option A: Build — stand up a new Emerson heavy-fab site adjacent to Shakopee or in a Midwest partner location ($80–200M). Option B: Partner — integrate via Powin, AESI, or Wartsila who already do this work for lithium BESS ($20–50M). Option C: License — transfer the IP and BoM to CATL or BYD ($5–20M floor, but loses the Plantweb fleet moat). The decision happens at Phase 3 with real Phase 1 and Phase 2 data in hand, not as a Phase 1 commitment. Phase 1 (coin cells, ~480 builds) and Phase 1.5 (pouch cell prototype) DO fit Shakopee’s precision-instrumentation capability and run in the existing shell space.

The full engineering case

Phase 3 decision matrix (proposal §The Money + §Building It at Scale): (A) Build — new Emerson heavy-fab site, $80–200M for 50–200 MWh/yr pilot line, preserves full vertical integration including Plantweb attach, longest time-to-market. (B) Partner — integration partnership with Powin / AESI / Wartsila (Midwest-anchored battery integrators with existing lithium-BESS lines), $20–50M, retains chemistry IP + Plantweb attach, shorter time-to-market, splits margin on hardware. (C) License — transfer IP and BoM to CATL or BYD, $5–20M floor royalty preserves 8-figure revenue without manufacturing capital, sacrifices Plantweb fleet-moat advantage. Capability cascade by phase: Phase 1 (coin cells, ~520-cell campaign) fits Shakopee precision floor; Phase 1.5 (1–3 kWh pouch prototype) fits Shakopee + extension; Phase 2 (first 100 kWh module + skid prototype) is partial-Shakopee with cell + Plantweb integration on-site and heavy frame fab outsourced to a local Midwest steel partner; Phase 3 (50–200 MWh/yr pilot) triggers the make-vs-partner-vs-license decision; Phase 4 (1–10 GWh/yr commercial) follows the Phase 3 path. The proposal explicitly treats Phase 3 as the strategic-flexibility point, not as Shakopee scaling infinitely.

E3 · The champions

I asked this

Who actually carries this forward inside Emerson?

In one sentence

Strategic hero: Michael Muck (VP Global Operations & Supply Chain, Measurement Solutions); routing via Logan Woolery (Sr Manager Industrial Wireless Instrumentation, the Plantweb data path); campus blessing from Jon Stokes; Phase-4 scale-up partnership with Rodolphe El Khoury.

The plain-English version

The proposal targets four distinct internal stakeholders, each with a specific role at a specific phase. Michael Muck is the strategic hero who green-lit the prior Shakopee geothermal R&D pilot and runs the quarterly sustainability meetings — he has both the precedent and the budget power. Logan Woolery is the technical routing path because the Plantweb instrumentation moat is his territory. Jon Stokes is the campus authority who has to bless shell-space use. Rodolphe El Khoury becomes the scale-up partner downstream when Phase 4 puts a battery skid on an actual production line. All four named in the right roles, none claimed for a role they don’t actually have.

The full engineering case

Champion routing in claim #6 of the Executive Summary: Logan Woolery (Sr Manager, Industrial Wireless Instrumentation — the Plantweb data path) → Michael Muck (VP Global Operations & Supply Chain for Measurement Solutions; green-lit prior Shakopee geothermal R&D pilot; runs quarterly campus sustainability meetings). Jon Stokes (Group President) is the campus authority who blesses shell-space use. Rodolphe El Khoury (VP NA Operations) becomes the scale-up partner downstream when Phase 4 puts a battery skid on an actual production line. Corporate cover: Michael Train (SVP & CSO, former president of Rosemount); Bob Yeager (president, Power & Water, owns Ovation Green + Zitara); Thurston Cromwell at Emerson Ventures (already invested in battery startups). International IP linkage: Jennie Li (VP & GM Emerson China); COO Ram Krishnan (prior president of Climate Technologies in Asia, Hong Kong); CEO Lal Karsanbhai (Fed St. Louis chair, U.S.-China Business Council).

I asked this

Why Shakopee for Phase 1 rather than Rosemount or Austin?

In one sentence

Shakopee is the only Emerson site that combines a working battery-grade dry room, a chemistry team, halide-compatible safety infrastructure, and operational distance from Rosemount’s instrument lab — all in one campus that can host Phase 1 without disrupting an existing P&L.

The plain-English version

Phase 1 needs three things in the same building: a dry room with humidity control good enough for COP anode handling, a glovebox with HF-compatible scrubbing for the MgCl₂ electrolyte work, and a cycler bay with the electrical service for an NI HPS-17000. Shakopee already has the dry room and the gloveboxes from prior battery work; the cycler bay is a $250K buildout. Austin, Rosemount, and the other Emerson sites would need all three from scratch. Shakopee also has Stokes as group president — the right level of executive air cover for an R&D bet that competes with no existing Emerson P&L. Rodolphe enters at Phase 4 when battery skids start moving down production lines.

The full engineering case

Site-selection matrix from the proposal §Why Shakopee: Shakopee scores 5/5 on dry-room availability, 4/5 on glovebox HF-compatibility (one swap-out required for MgCl₂ tolerance), 3/5 on cycler-bay electrical service (250 kVA upgrade modeled in the Phase 1 budget), 5/5 on adjacent EH&S and waste-handling infrastructure; Austin scores 2/5, Rosemount 3/5, Boulder 1/5. Phase 1 sits outside the existing Shakopee P&L lines (Plantweb, Rosemount measurement) to prevent it from being mistaken for an incremental product investment; it gets its own stage-gate review against the demonstrated / projected / thesis claim split. Routing chain: Stokes (group president) sponsors the R&D commit, Logan Woolery handles day-to-day program management, Rodolphe (manufacturing) enters at Phase 4 when production-line integration begins.

I asked this

Five chemists and one battery scientist — is that really enough for Phase 1?

In one sentence

Phase 1 is a chemistry validation, not a product launch — five chemists running 40 cells per week through the kill-gate matrix, plus a CityU-trained scientific lead, is the team size that delivers a statistically meaningful answer in 18 months without diluting decisions.

The plain-English version

The Operations Manual lays out the cell budget: 8 cells/day × 5 working days × ~13 weeks per quarter × 4 quarters ≈ 520 cells over the campaign. That’s not a “make a lot of batteries” exercise; it’s a “test enough variants of the recipe to learn what works” exercise. Each cell goes through the same kill-gate (tiered coulombic efficiency, capacity retention, gas evolution, post-mortem forensic), and the scientific lead reads the data weekly. A bigger team would dilute chemistry decisions across more hands without speeding up the calendar — cycler bandwidth, not headcount, is the rate-limiting resource.

The full engineering case

Phase 1 team (Operations Manual §1): 5× Cell Scientists (1.0 FTE each, lab + cycler shifts), 1× Battery Scientist / Scientific Lead (1.0 FTE, CityU-trained per proposal §The Company), 1× Chemical Hygiene Officer (0.25 FTE), 1× Data Engineer (0.5 FTE, Plantweb integration + DeltaV calibration), 1× Operations Lead (0.5 FTE, supply + safety + admin). Total 4.75 FTE. Throughput: 8 cells/day × Mon–Fri = 40 cells/week × ~13 weeks/quarter ≈ 520 cells over the 4-quarter Phase 1 campaign. Cycler capacity: NI HPS-17000, 32 channels, C/10 through 1C testing in parallel. Statistical-power envelope: 8 recipe variants × 5 replicates × 13 calendar-aging conditions × 2 charge-rate variants = 1,040-cell theoretical matrix; 520 budgeted covers half. Phase 1 endpoint: statistical pass/fail on cycle-life retention thesis at ≥90% confidence.

You might ask

Who is Connor Scanlan, and why is a Material Attendant the principal author of a $1.8M R&D proposal?

In one sentence

Connor is a Shakopee Material Attendant whose 18-month independent study produced the proposal, Storage Landscape, Operations Manual, and this hub — Phase 1 hires a credentialed Battery Scientist as Principal Investigator; Connor serves as program manager and continuity owner, not the chemist.

The plain-English version

The proposal originates from the Shakopee production floor, not from corporate strategy. Connor read Chen et al. in February 2026, spent eighteen months on independent study — electrochemistry, BESS economics, OSHA compliance, Plantweb architecture — and drafted the proposal, the Storage Landscape, the 30-year Operations Manual, and the cards you’re reading. He is not the chemist. Phase 1 hires a credentialed Battery Scientist (PhD electrochemistry, CityU-trained or equivalent) as Principal Investigator. Connor’s Phase 1 role is program manager: the person who carries the proposal between the lab, the champions (Muck, Logan, Stokes), customers, and the public-facing story. If Connor leaves Emerson, the Battery Scientist owns the chemistry continuity; everything is documented at saltwaterpower.org so the program survives any single departure.

The full engineering case

Author of record: Connor Scanlan, Material Attendant, Emerson Shakopee, 2024–present. Domain preparation since Feb 2026: ~600 hours of independent study (electrochemistry through Newman & Thomas-Alyea + Bard & Faulkner; BESS economics through Lazard LCOE reports + NREL ATB; OSHA compliance through 29 CFR 1910 references; Plantweb architecture through internal Emerson documentation). Phase 1 organizational structure: Battery Scientist (PhD, 1.0 FTE) = Principal Investigator with chemistry decisions and Stage 1 kill-gate signature; CHO (0.25 FTE) = safety authority; Connor Scanlan (1.0 FTE if Phase 1 is funded) = program manager + Plantweb integration + champion routing + public-facing communication. Author succession plan: every artifact, dataset, and key relationship is documented at saltwaterpower.org with corresponding git history; intellectual continuity does not depend on Connor remaining at Emerson. Phase 4 transition: per the proposal, commercial scale-up reports to Rodolphe El Khoury (NA Operations) regardless of author status. Why this matters: hostile reviewer’s correct flag — the proposal’s intellectual provenance and the long-term continuity of its program manager are separate questions that deserve separate answers.

You might ask

How are you actually going to hire a CityU-trained battery scientist in Shakopee, Minnesota?

In one sentence

Three-tier pipeline: primary direct hire from CityU postdocs + US national-lab Mg/PBA researchers (Stage 0, 90 days); Plan B 12-month CityU visiting-scientist secondment at $250–300K; Plan C contract Phase 1 chemistry validation to Argonne or Sandia for $400–600K — Stage 0 closes with a signed term sheet for at least one pathway.

The plain-English version

Fair reviewer question: the aqueous Mg-ion talent pool in the US is essentially zero, and visa hires from Hong Kong typically choose Boston or the Bay Area, not the Twin Cities suburbs. The risk is real and the mitigation is layered. Primary hire: direct recruit from Chen’s CityU postdoc cohort (Chen has been supportive of the project and is willing to facilitate introductions), backstopped by US national-lab researchers at PNNL and Argonne with publication history in Mg²⁺ aqueous chemistry. Recruiter engaged at Stage 0; comp band $185–220K direct + relocation + sign-on + H-1B sponsorship. Plan B: 12-month visiting-scientist secondment from CityU at ~$300K/year direct + travel + Shakopee housing; covers Phase 1’s most critical period and simultaneously strengthens the IP relationship. Plan C: contract Phase 1 chemistry validation to Argonne, Sandia, or Voltaiq for $400–600K over 9–12 months — slower but preserves the program if neither primary hire nor secondment lands. Stage 0 deliverable: signed engagement letter for at least one of the three pathways.

The full engineering case

Candidate pipeline (Stage 0 weeks 1–12): direct outreach to CityU postdocs in Chen group (5–8 active candidates; arXiv + JACS author lists); US national-lab researchers (PNNL aqueous battery program, Argonne MERF, Brookhaven materials science; ~12–15 active candidates publishing on Mg²⁺ or PBA aqueous chemistry); Aurbach group alumni (Bar-Ilan U, Israel; the canonical multivalent chemistry source; 4–6 alumni in academic and industrial roles globally); industry refugees (Pellion alumni, Toyota MTRC Mg program alumni). Comp band $185–220K direct + $30–50K relocation + H-1B sponsorship. Plan B: formal MOU with CityU Chen group at Stage 0 (estimated 60-day negotiation); $250–300K direct + travel + housing in Shakopee for 12 months; covers Phase 1 weeks 1–52. Plan C: Argonne ANL Battery Manufacturing Facility (precedent for industrial chemistry validation at C/10–C/4 scale), Sandia (HSE focus), Voltaiq (industrial BESS chemistry consultant); $400–600K over 9–12 months; chemistry decisions reviewed by Connor + CHO + named external PI. Stage 0 closure requires a signed engagement letter for at least one of the three pathways before Phase 1 capital deploys.

Section F

The Risks

What kills this at each phase, the explicit Aquion-trap mitigation, and how the proposal caps the downside at each gate.

F1 · Technical risks

You might ask

What if the chemistry doesn’t validate at C/4 rate in Phase 1?

In one sentence

Phase 1 ends at the kill-gate, Emerson is out $1.4–1.8M total, the lab equipment is repurposable, and the program closes without further commitment.

The plain-English version

Phase 1 has hard kill-gates at Stage 0 ($200–400K spent) and Stage 1 ($1.4–1.8M spent). If the chemistry fails Chen et al.’s reported cycle life at grid-relevant C/4 rates, the program closes. Emerson’s downside is capped at the Phase 1 envelope, the lab equipment retains general battery R&D value, and the lessons learned go to Plantweb / NI / Ovation Green divisions. The whole point of staging is that you discover a failure at the cheapest possible cost.

The full engineering case

Phase 1 staged kill-gates: Stage 0 ($200–400K, months 1–3) returns written go/no-go; no Stage 1 capital releases without explicit go-decision. Stage 1 ($808K–$1.4M, months 4–12) chemistry validation against success criteria (capacity retention >95% at C/4, Cu leach <15 ppm/100 cycles, voltage hysteresis <250 mV mid-SOC, pH stability 4.91–7.02). Kill conditions trigger immediate program termination, equipment asset disposition (resale or internal transfer at full retained value), personnel rotation. Total at-risk capital at any point in Phase 1: bounded by $1.8M envelope. The whole staging philosophy is failure-cheap.

You might ask

What about gas evolution at skid scale?

In one sentence

Standard catalytic recombiners from the VRLA lead-acid industry handle it; the water-in-salt regime intrinsically suppresses evolution; Rosemount sensors monitor throughout to detect any anomalies.

The plain-English version

A small amount of H₂ and O₂ gas can form during charging because the 2.2V operating voltage is above water’s thermodynamic stability window. Two things keep this manageable: the “water-in-salt” chemistry suppresses gas formation by tying up free water, and catalytic recombiners (a standard component in lead-acid batteries since the 1960s) convert any H₂+O₂ back to water. Plantweb monitors gas levels in real time. Phase 1 measures the gas rate at coin-cell scale; Phase 1.5 measures at pouch scale; Phase 2 sizes the recombiner system for skid scale.

The full engineering case

Risk register Risk 2 (Phase 2): electrolyte loading at skid scale (80 µL → 100s of L) exposes gas evolution dynamics not seen in coin cells. Mitigations: (a) H₂/O₂ monitored throughout Phase 1.5 via Rosemount sensors so gas-evolution kinetics are characterized before Phase 2 scale-up; (b) catalytic recombiners standard from VRLA lead-acid industry (Pt/Pd-based, decades of field-proven reliability); (c) water-in-salt regime intrinsically suppresses evolution (Chen et al. operate at concentrations where free water coordination to salt ions raises the effective electrolysis voltage above 2.5V); (d) gas-evolution interlocks tied to DeltaV process control with automatic charge taper if rate exceeds threshold.

F2 · Market risks

You might ask

What if hyperscalers don’t pivot to lifetime carbon by 2028?

In one sentence

Aqueous Mg-ion still wins on intrinsic safety (non-flammable indoor BESS), still wins on 13× better $/kWh-cycle, still wins on recyclability — the lifetime-carbon thesis is upside, not the base case.

The plain-English version

Executive Summary claim #7 is explicitly labeled testable thesis rather than a verified claim — the lifetime-carbon advantage depends on hyperscaler ESG priorities pivoting toward 30-year lifecycle metrics. If they don’t, the chemistry still wins on the things it’s already proven: it can’t catch fire (indoor deployment, urban deployment, no fire-suppression overhead), it lasts much longer per cycle than LFP (decisive LCOE math), and it’s structurally recyclable. The carbon-tier market is a bonus opportunity, not a load-bearing assumption.

The full engineering case

Risk 4 in the proposal’s risk register: hyperscaler ESG pivot doesn’t land by 2028. Mitigation: claim #7 is already labeled “Inherent Safety Now, Lowest-Carbon Bet for the Second Decade — a Testable Thesis”; the “Safety Now” half is verifiable from chemistry (aqueous MgCl₂ physically cannot burn). Aqueous Mg-ion still beats LFP on intrinsic safety (no thermal runaway, indoor-deployable, no bunker walls or fire-suppression overhead), still recoverable for second-life applications, still 13× better $/kWh-cycle. The eight-surface revenue structure (battery hardware + brine recovery + existing-product lift + service + Plantweb subscription + Cu recovery + carbon credits + IP licensing) holds even if the carbon-credit surface goes to zero.

You might ask

What if LFP keeps getting cheaper and crushes us?

In one sentence

LFP wins on $/kWh upfront and will keep doing so; aqueous Mg-ion wins on $/kWh-cycle (LCOE basis) and on indoor / non-flammable / cycle-count requirements LFP cannot meet at any price.

The plain-English version

LFP installed cost dropped from $400+/kWh in 2020 to ~$250 today and may hit $150 by 2030. That continued cost-down is exactly what killed Aquion. The proposal’s answer is that aqueous Mg-ion doesn’t compete on upfront $/kWh — it competes on lifetime $/kWh-cycle, on intrinsic safety (LFP cannot go indoors next to server racks at any price), and on cycle-count duty (LFP degrades fast under daily deep-cycle stress that aqueous Mg-ion handles). Different markets, different math.

The full engineering case

Risk 5 in the proposal (The Aquion Trap, cross-phase): capital burn vs LFP learning curve. Mitigations: (a) Phase 1 validates chemistry for $1.4–1.8M, not $190M — the Aquion mistake was confusing “works in lab” with “works at LFP-comparable cost”; (b) Emerson’s pitch is $/kWh-cycle dominance over 30 years (13× advantage), not $/kWh parity; (c) the Plantweb SaaS moat is structurally different from anything LFP integrators can match, so eight-surface economics hold even at $/kWh disadvantage; (d) Phase 3 license-back option preserves IP value even if Emerson exits manufacturing entirely. Markets aqueous Mg-ion wins regardless of LFP price: indoor BESS (LFP forbidden by fire code in many urban contexts), high-cycle-count grid services (LFP degrades fast under daily deep-cycle stress), 30-year infrastructure (LFP requires full replacement every 8–10 years).

F3 · Strategic risks

You might ask

What if China restricts the CityU collaboration?

In one sentence

Chen et al.’s paper is published — the chemistry itself is public knowledge available to every researcher on Earth; CityU partnership matters for unpublished process know-how but isn’t strictly required for Phase 1 validation or eventual commercial production.

The plain-English version

Publication in Nature Communications is disclosure. The scientific findings (electrolyte, cathode, anode, performance) are now public knowledge. CityU may own patents on specific implementations (electrode ratios, synthesis procedures, cell architecture) — that’s what a tech-transfer partnership would license. If China restricts collaboration, Emerson can still proceed with Phase 1 validation against the published chemistry; the IP risk is in commercial production of the exact paper-described cell, which is a Phase 2–3 question that Trade Counsel addresses in Stage 0 of Phase 1.

The full engineering case

Per the proposal’s IP section: the material classes (Prussian blue analog cathodes, conducting organic polymer anodes, aqueous chloride electrolytes) are all established in the electrochemical literature with prior art stretching back decades. No entity can patent broad classes of materials. What CityU likely owns: specific electrode ratios, exact electrolyte concentration, particular cell architecture, unpublished process know-how. What nobody owns: the principle of Mg²⁺ intercalation into PBA lattices, the general class of Prussian blue analog cathodes, MgCl₂ as a battery electrolyte, the published performance data. Stage 0 budget includes $20–40K trade counsel + IP navigation for international IP attorney (China/HK specialty) + MOFCOM/CFIUS analysis + freedom-to-operate memo + CityU tech transfer office engagement + licensing term sheet draft.

You might ask

What if the Chen et al. paper is retracted?

In one sentence

Retraction would happen only if data fabrication is proven, which is a rare event for Nature Communications; the BUCT/PKU Shenzhen Wu et al. corroboration in JACS validates the broader aqueous Mg-ion platform independently; Phase 1 would still produce useful chemistry intelligence even if the specific cell turns out flawed.

The plain-English version

Peer-reviewed papers in journals like Nature Communications have rigorous review processes and retractions are rare. The Stage 0 external technical assessment specifically scrutinizes the paper’s methodology and data integrity before Stage 1 capital releases. Even if the specific cell turns out flawed, the broader aqueous Mg-ion platform is independently corroborated by Wu et al. (JACS Feb 2026), so Phase 1’s research investment still produces useful intelligence rather than a complete loss.

The full engineering case

Mitigation strategy: Stage 0 ($65–130K) external technical assessment includes independent battery electrochemistry consultant review of Chen et al.’s methodology, data quality, cell-fabrication reproducibility, and any anomalies that would suggest measurement artifacts. Stage 0 contract lab validation ($38–80K) replicates key paper claims at a separate facility before Stage 1 lab buildout. If Stage 0 surfaces material concerns, kill-gate triggers without Stage 1 commitment. Independent corroboration: Wu et al. (BUCT/PKU Shenzhen, JACS Feb 2026, doi:10.1021/jacs.5c21656) validates aqueous Mg-ion broader platform with a different anode, giving the chemistry family a second peer-reviewed data point even if Chen’s specific cell were retracted.

F4 · The Aquion trap

I asked this

Could this become another Aquion?

In one sentence

The failure mode is named explicitly in the risk register and structurally mitigated four ways: stage-gated capital, $/kWh-cycle pitch rather than $/kWh pitch, the Plantweb SaaS moat that no battery-native player can match, and the Phase 3 license-back option that preserves IP value even if Emerson exits manufacturing.

The plain-English version

Aquion raised $190M and went bankrupt because LFP got cheap faster than they could scale. This proposal explicitly mitigates that failure mode in four ways: (1) Phase 1 spends $1.4–1.8M to validate chemistry before any major capital commitment, vs Aquion’s $190M to first product; (2) the pitch is $/kWh-cycle dominance over 30 years, not $/kWh upfront where LFP wins; (3) Plantweb is a structurally different revenue surface that LFP integrators cannot match; (4) even if Emerson decides Phase 4 manufacturing isn’t the right path, the Phase 3 license-back option preserves IP value and Plantweb fleet revenue. The Aquion trap is real and named.

The full engineering case

Risk 5 in the proposal (cross-phase, The Aquion Trap): capital burn vs LFP learning curve. Aquion: ~$190M raised 2008–2017, Chapter 11 March 2017, assets sold $9.16M, failure mode per Whitacre = capital burn vs Li-ion cost-down curve + low energy density + balance-of-systems cost. Four-part mitigation: (a) Phase 1 validates chemistry for $1.4–1.8M, not $190M; the Aquion mistake was confusing “works in lab” with “works at LFP-comparable cost”; (b) Emerson’s pitch is not $/kWh parity, it’s $/kWh-cycle dominance over 30 years (13× advantage if chemistry validates); (c) Plantweb SaaS moat is structurally different revenue surface than LFP integrators can match, so eight-surface economics hold even at $/kWh disadvantage; (d) Phase 3 license-back option ($5–20M) preserves IP value and Plantweb fleet revenue even if Emerson exits manufacturing entirely. The failure mode is real and explicitly named.

I asked this

What if NI cancels the HPS-17000 cycler or doubles the price?

In one sentence

Named second-source (Maccor Series 4000) drop-in-compatible with our protocols, plus academic cycler-share LoIs for burst capacity — and since NI is now part of Emerson, the worst-case vendor scenario is internal escalation rather than market-rate negotiation.

The plain-English version

The HPS-17000 is the most capable cycler in its class — 32 channels, the right voltage/current envelope, integrated with LabView and Plantweb. Single-source-of-supply is a real risk. We mitigate it three ways. First, the test-protocol spec is written so the Maccor Series 4000 (Texas-based competitor) is drop-in compatible; switching costs about $180K and 6 weeks of calibration. Second, we have informal coverage agreements with two academic battery labs (CityU and a US partner) that can absorb our matrix at burst rate. Third, NI is now part of Emerson — so the worst-case vendor scenario is an internal escalation, not a market-rate negotiation.

The full engineering case

Primary cycler: NI HPS-17000, 32 channels, ±20A / ±100V envelope, C/10 through 1C bandwidth, integrated to DeltaV via OPC-UA; capital cost ~$680K (Operations Manual §8 supply schedule, 5-year refresh cycle). Secondary source: Maccor Series 4000, 64 channels at lower per-channel current, switch under SOP §8.3, total switching capex $180K + ~6 weeks calibration. Tertiary capacity: cycler-share LoIs with CityU (Chen group, existing relationship) and a domestic university partner under negotiation, billable $/channel-day, available for burst capacity. The single-vendor risk is named in proposal Risk Register §Risk 8 (“Cycler supply discontinuity”) with full mitigation; severity downgraded from High to Medium-Low by the 2023 NI/Emerson acquisition, which closes the worst-case adversarial vendor scenario. Calibration parity: every cell goes through a quarterly cross-cycler calibration cell tracked in Plantweb to guarantee the dataset stays comparable across instrument switches.

You might ask

Cyanide. You’re putting iron-cyanide cathode in salt water at 2.2V — under what conditions does this release HCN?

In one sentence

Free HCN evolution requires pH ≤4 OR temperature ≥~250°C; operating chemistry sits at pH 7, ≤2.2V, ≤50°C — fully outside both danger envelopes; the skid enforces three hardware bounds (pH cutoff at 4.5, thermal interlock at 85°C, packed-bed NaOH scrubber) that act long before HCN can evolve.

The plain-English version

This is the single most pointed safety question we get, and it deserves a direct answer. The CuFe-Prussian Blue cathode contains an iron-cyanide framework. Under acid (pH ≤4) or at high temperature (≥~250°C decomposition onset for CuFe-PBA in chloride), that framework can release free cyanide. Aqueous Mg-ion operating chemistry — pH 7, ≤2.2V, ≤50°C — sits comfortably outside both envelopes; Natron Energy’s Na-ion PBA cathode BESS deployments (>5,000 cycles in commercial service) show no published HCN release at operating conditions. The two failure modes that could trigger HCN evolution are (a) external building fire heating the skid above 100°C, and (b) electrolyte acidification from a contamination event. Three hardware bounds defend against both: (1) HCN reactor offgas scrubber on the cell stack (packed Pall rings, 10–15% NaOH recirculated, ≥3 sec residence, Drager downstream breakthrough alarm at 0.5 ppm); (2) pH monitor in the electrolyte loop with hard cutoff at 4.5; (3) thermal interlock at 85°C that disconnects the stack and floods the bund with neutralizer — 15°C below the closest HCN-relevant threshold.

The full engineering case

HCN evolution from K₄[Fe(CN)₆] / Mg₂[Fe(CN)₆] frameworks: thermodynamically, free CN⁻ release requires either (a) acid-catalyzed dissociation [Fe(CN)₆]⁴⁻ + 6 H⁺ → Fe²⁺ + 6 HCN (pK_a HCN/H⁺ = 9.21; kinetics negligible above pH 5), or (b) thermal decomposition above ~250°C (decomposition onset varies with framework substitution; CuFe-PBA stable to ~340°C in chloride per Wang et al. Nat. Mater. 2019). Operating envelope per Operations Manual §3: pH 7.0 ± 0.5 controlled by closed-loop dosing, voltage ≤2.2V controlled by BMS, temperature ≤50°C controlled by passive convection + active fault response. Worst credible scenario: external building fire raising skid surface temperature above 100°C; thermal interlock at 85°C activates >160°C before HCN decomposition threshold. Scrubber spec (Operations Manual §6): packed-bed Pall rings, 10–15% NaOH recirculated, ≥3 sec residence, Drager downstream colorimetric monitor with 0.5 ppm breakthrough alarm. Detection sensitivity: lab-grade ICP-MS on operating electrolyte aliquots (quarterly per CHP §6.4); field-grade pH + ORP + temperature continuous monitoring. Independent corroboration: Natron Energy’s Na-ion PBA cathode BESS deployments (commercial service since 2023, no published HCN incidents). This is the single biggest unanswered safety question if buried, which is why it gets its own card.

You might ask

What’s the freedom-to-operate position? Has anyone done a real patent landscape?

In one sentence

A signed third-party FTO opinion is a Stage 0 deliverable, NOT a Phase 1 line item — named patent counsel produces a landscape covering Natron, Faradion, HiNa, Altris (PBA cathode), prior COP electrode work, and the CityU patent family before any Phase 1 capital commits.

The plain-English version

A reviewer worth their fee will ask: “you cite Chen et al. — but does this specific CuFe-PBA + Hex-TADD-COP + MgCl₂ stack land cleanly between Natron’s PBA patents and CityU’s patent family?” The honest answer is: nobody knows yet, and the $20–40K trade-counsel line item in the proposal is the BUDGET for an FTO opinion, not an opinion itself. Stage 0 (the $1.4–1.8M pre-Phase-1 decision) requires a real opinion from a registered patent law firm (scope $30–50K, 6–8 weeks). Three outcomes possible: (a) clean space — the specific stack is novel enough to not infringe — proceed to Phase 1; (b) licensable encumbrance — Natron / CityU patent family covers part of the stack — negotiate license within Phase 1 budget; (c) blocking patent — design-around required — Phase 1 scope changes to alternative electrode chemistry (Mn-PBA cathode + alternate organic anode). Outcome (a) is the working hypothesis; (b) and (c) are stage-gated.

The full engineering case

FTO scope (Stage 0 deliverable, named external counsel — candidate firms: Womble Bond Dickinson / Cooley Energy Group / Foley & Lardner battery practice): (1) Natron Energy patent family (PBA cathode for Na-ion, ~50+ active patents; assess whether claims read on Mg²⁺ insertion); (2) Faradion patent family (Na-ion PBA + electrolyte, ~30 patents; UK + EP coverage); (3) HiNa Battery patent family (Mn-PBA, Chinese coverage primarily); (4) Altris patent family (Prussian White, Mn-PBA); (5) CityU patent family (Chen group, prior aqueous Mg work + the Feb 2026 Nat. Commun. method; specifically the CN patent application for CuFe-PBA + MgCl₂ aqueous); (6) prior COP electrode art (Tilley et al. Hex-TADD-COP for K⁺ aqueous; Schon, Tilley et al. TEMPO radical anodes); (7) general aqueous multivalent electrochemistry (Aurbach group, Bar-Ilan; Wang group, U Maryland WiSE). Deliverable: signed opinion, “clean to proceed at C/4 grid-rate aqueous Mg-ion BESS service” or named encumbrance + licensing strategy. Stage 0 kill-gate: outcomes (a) or (b) green-light Phase 1; outcome (c) triggers Phase 1 redesign before capital deploys. The line item is named, scoped, and signed off as a precondition — not deferred into Phase 1.

You might ask

A 3 MWh skid running at 99.5% Coulombic efficiency — how much hydrogen does that actually generate per day?

In one sentence

~1.5 cubic meters H₂ per day at full daily cycling — handled by a catalytic recombiner (VRLA lead-acid industry standard, sized for 2 m³/day with 33% margin), passive NFPA-compliant venting backstop, and a Plantweb-integrated 1 ppm H₂ sensor that alarms at 12.5% of design limit.

The plain-English version

A reviewer will multiply: “0.5% Coulombic efficiency loss × 3 MWh skid = how many liters of hydrogen per day?” The answer is real and large — about 1.5 cubic meters per day at full daily cycling. The hydrogen has to go somewhere. Standard design: a catalytic recombiner (technology that’s been in lead-acid VRLA batteries for fifty years) converts H₂ + ½O₂ → H₂O at the cell-stack head; sized for 2 m³/day, comfortably above the 1.5 m³/day expected at full cycling. Backstop: passive venting through NFPA-compliant relief valves to an outdoor dispersal point; the skid’s H₂ headspace concentration stays below 1% v/v by design (one quarter of the 4% lower explosive limit). If the recombiner fails, the vent prevents accumulation; if both fail simultaneously, H₂ reaches the LEL in 8–12 hours — long enough for the Plantweb gas-detection sensor (1 ppm resolution) to alert operations long before any safety threshold.

The full engineering case

H₂ evolution from parasitic water reduction at the COP anode, kinetically controlled. Steady-state η_CE = 99.5% per the tiered CE spec; at C/4 on a 3 MWh skid (12 MWh round-trip energy at one full cycle/day), 0.5% loss yields Q = 1.5 × 10⁶ C → ~27 mol H₂/cycle → ~1.5 m³ H₂/day at STP. Recombiner sizing (Operations Manual §3.6): Pt-coated alumina-pellet catalytic recombiner (Hoppecke / HBL Power, derived from VRLA industry); sized for 2 m³/day continuous throughput, ~33% margin; reaction H₂ + ½O₂ → H₂O at ambient; O₂ side comes from cathode oxygen evolution under the same parasitic envelope. Passive venting: NFPA 30 + NFPA 855 compliant pressure-relief valve at skid head, 1.5 psi cracking pressure, 3″ stainless tube to outdoor dispersal point ≥5 m from skid. H₂ headspace design limit: 1% v/v (= 25% LEL); LEL = 4% v/v. Detection: Plantweb-integrated catalytic-bead H₂ sensor (Drager Polytron 8200, 0–5% LEL range, 1 ppm resolution); alarm at 0.5% LEL. Failure-mode analysis: recombiner failure alone → vent handles full evolution; vent failure alone → recombiner handles 75%, accumulation reaches LEL in ~36 h, sensor alarms within 1 h; both fail simultaneously → LEL in 8–12 h, sensor alarms in <1 h. The control system maintains safe operation under any single-fault scenario.

You might ask

What evidence would change your mind and kill the program?

In one sentence

Five specific events between today and Stage 0 closure would change the go/no-go decision: Chen retraction; FTO returns a blocking patent; contract lab can’t reproduce Chen’s cycle-life signature within 2σ; hyperscaler LFP $/kWh-cycle drops below $0.10; or CityU collaboration is permanently restricted by export control.

The plain-English version

An honest stage-gate proposal includes a public statement of what would make us walk away early. Here are the five specific events between today and Stage 0 closure that would change the decision: (1) Chen et al. paper retraction or correction that invalidates the 120,000-cycle headline — estimated ~5% probability; we monitor the journal and the authors monthly. (2) Third-party FTO opinion identifying a blocking patent with no design-around — ~20%; mitigation is alternative electrode chemistry. (3) Stage 0 contract-lab validation that fails to reproduce Chen’s cycle-life signature within 2σ at 1 A/g over 1,000 cycles — ~25%, the highest single-point risk. (4) Hyperscaler procurement signal collapsing — LFP $/kWh-cycle dropping below $0.10 in published RFPs — ~15%; would compress our $/kWh-cycle advantage from 13× to ~3–4×. (5) Chinese export controls on aqueous-electrolyte battery research access — ~10%; CityU collaboration would shift to publication-only. Any one of these triggers a Stage 0 kill review; two simultaneous events triggers automatic Phase 1 cancellation. Showing the seams builds more trust than hiding them.

The full engineering case

Stage 0 kill-gate review (every 30 days for 6 months pre-Phase-1): (1) Paper status monitor: monthly check of Nature Communications correction log, Google Scholar citation forensics, RetractionWatch.com index; quarterly direct check with Chen author group on CityU collaboration platform. Trigger: any retraction notice or material correction (defined as ≥30% change in headline cycle life, voltage, or efficiency). (2) FTO trigger: third-party counsel opinion returns outcome (c) under the FTO card. (3) Contract-lab validation: Stage 0 contracts Argonne / Sandia / Voltaiq to run 1,000-cycle replication of Chen cell architecture at 1 A/g (lower than Chen’s 20 A/g, closer to grid-realistic); kill criterion = capacity retention >2σ below Chen’s published curve, OR capacity-fade slope at 1 A/g extrapolating to <5,000 cycles at C/4. (4) Procurement signal: monthly monitor of public hyperscaler RFP language (Microsoft, Meta, AWS, Google, Equinix); kill criterion = published LFP $/kWh-cycle below $0.10 across multiple hyperscaler procurements over 90 days. (5) Export control: Bureau of Industry and Security (BIS) Entity List + Commerce Control List monitoring; CityU + Chen group inclusion would trigger immediate collaboration restructuring. Composite kill: any single trigger pauses Phase 1 commit; any two triggers cancels Phase 1 within the same review cycle. Trust comes from publishing the kill criteria before we have to use them.

You might ask

What if Michael Muck leaves Emerson or de-prioritizes this program?

In one sentence

Champion routing is structurally distributed across four named stakeholders at different phases — Muck strategic / Logan routing / Stokes campus / Rodolphe scale-up — plus C-suite cover (Train, Krishnan, Karsanbhai) and a public intellectual-provenance record, so the program survives any single departure.

The plain-English version

A fair reviewer concern: the “Who actually carries this forward inside Emerson?” card names Michael Muck as the strategic hero. What happens if Muck retires, moves laterally, or de-prioritizes during Phase 1? Mitigation in three layers. First, the routing is distributed: Logan Woolery handles day-to-day program management under Plantweb instrumentation (not under Muck specifically); Jon Stokes blesses campus shell-space use (independent of Muck); Rodolphe El Khoury owns Phase 4 commercial scale-up (a separate succession entirely). If Muck leaves at any phase, three of four named champions remain in role. Second, the proposal’s intellectual provenance is documented at saltwaterpower.org with git-tracked revision history — any successor sponsor (Train, Yeager, Cromwell) can pick up where Muck left off without re-discovering the chain of reasoning. Third, the program has been positioned with corporate-level cover (CSO Michael Train, COO Ram Krishnan, CEO Lal Karsanbhai) so a champion departure at the VP level does not orphan the program at the C-suite level.

The full engineering case

Champion stack (proposal §The Company): VP-tier sponsors — Michael Muck (VP Global Ops / Supply Chain Measurement Solutions), Bob Yeager (President Power & Water), Jon Stokes (Group President). Middle-management routing — Logan Woolery (Sr Manager Industrial Wireless). Functional scale-up partner — Rodolphe El Khoury (VP NA Operations). C-suite cover — Michael Train (SVP & CSO, formerly president of Rosemount), Ram Krishnan (COO, previously president of Climate Technologies in Asia), Lal Karsanbhai (CEO, Fed St. Louis chair). Corp-dev optionality — Thurston Cromwell (Emerson Ventures), Jennie Li (VP & GM Emerson China, IP linkage). Departure scenarios: (a) Muck leaves Phase 1 → Logan + Stokes route, Train at C-suite ratifies, Yeager picks up Plantweb attach commercially; (b) Logan leaves → Muck routes directly to Stokes; (c) Stokes leaves → Train + Krishnan elevate program to corporate; (d) Rodolphe leaves Phase 4 → standard Emerson succession to incoming NA Operations VP. Single-point-of-failure across the four named champions: zero. Intellectual-provenance continuity: saltwaterpower.org + github.com/cscanlan/emerson-proposal provide a non-personnel-dependent layer that survives any departure on either side of the org chart.

F5 · Approvals & site-compliance

Phase 1 doesn’t need exemption from the Shakopee Shell Space rules — it needs conversion from staging to permanent R&D inside those rules. Named approvals, named owners, named timelines.

You might ask

Does this comply with Emerson’s existing storage and safety policy on the Shakopee Shell Space?

In one sentence

Yes — aqueous Mg-ion is non-lithium, non-flammable, and operates inside the posted policy envelope (no exposed power modules, no lithium outside a cover); the only formal ask is conversion of L1 from short-term staging to permanent R&D, which is exactly the outcome the 6-month rule was written to enable.

The plain-English version

The posted storage and safety rules for the Shell Space are: short-term storage only (6-month limit, red-tagged if it overstays), no exposed power modules, lithium batteries kept inside a cover, marked aisles clear, PPE and Lone Worker policy in force. Aqueous Mg-ion fits inside that envelope, not against it. The chemistry is non-lithium, non-flammable (water-based MgCl₂ electrolyte), and the Operations Manual already provisions a sealed-envelope HVAC pressure cascade with no exposed power. The one formal ask is conversion: take ~10,000 SF of L1 from the rotating-staging pattern (which is exactly what the 6-month rule polices against) and graduate it to permanent R&D. That ask is filed openly in the “Approvals Required” section of the proposal, with the storage exception first, the named approval path next to it, and the timeline third — peer-language, not supplicant-language.

The full engineering case

Posted policy (per the Shakopee Shell Space storage request form, field-confirmed 17 Jun 2026): (1) short-term storage only, 6-month limit, red-tag enforcement; (2) approved activities = storage and infrequent low-touch testing, both requiring Facilities approval; (3) safety = protective eyewear and footwear, Rosemount Lone Worker policy compliance, Blackline monitor or partnered work; (4) power = no exposed power modules; (5) lithium = batteries kept inside a cover; (6) circulation = marked aisles clear at all times. Chemistry compliance: aqueous Mg-ion is structurally non-lithium and structurally non-flammable (water cannot burn at the operating envelope; see “Why won’t this catch fire like phones and EVs do?” in §A Chemistry), so rules 4 and 5 are inside the chemistry posture, not strained by it. Operational compliance: the Operations Manual rev 3.0d (hub/11_operations.html) provisions a CHO at 0.25 FTE, marked aisles per §3 Lab Zones, a sealed-envelope HVAC pressure cascade per §4 Ventilation Envelope (negative on the lab, positive on the break room), and PPE protocols against 29 CFR 1910.132 + 1910.1450 throughout. The formal ask: conversion of L1 NE-quadrant from short-term staging to permanent R&D, named in §Approvals Required of the proposal. Existing sanctioned low-touch testing on L2 (Capsule Mark sensor station, SYNCHROS firmware testing, NiM re-flash evaluation, plus WPG calibration and Branson ultrasonic welding) establishes precedent that the facility already hosts sanctioned R&D in this space; we extend the pattern, we don’t introduce it. Existing L2 programs continue as co-tenants under a written MOU; program-owner outreach and MOU drafting routed through Logan + Facilities before Phase 1 commencement.

I asked this

What does Phase 1 actually need from Facilities and site management?

In one sentence

Three named approvals, three named approval paths, three named timelines — conversion of L1 from staging to permanent R&D, a co-tenant MOU with the L2 sanctioned programs, and an occupancy + code review for the change of use; all three are visible in the proposal’s §Approvals Required section, not buried.

The plain-English version

Phase 1 needs three things, each from a different gatekeeper, each on a different timeline:

(1) Conversion of L1 NE-quadrant from short-term staging to permanent R&D. Granted by Shakopee Facilities + site management. Target: end of Q1, with a dated clearance manifest attached (every red-tagged item gets a disposition — return to vendor, surplus auction, redeployed to another Emerson site — before Phase 1 starts). The 6-month posted storage rule was written to prevent dead long-term storage; conversion to a stamped R&D program is the exact outcome the rule was written to enable.

(2) Co-tenant MOU with the L2 sanctioned testing programs. The proposal does not displace existing sanctioned R&D; the floor plan and Operations Manual respect them as continuing tenants. Coordinate-point identification, program-owner outreach, and the MOU drafting route through Logan + Facilities before Phase 1 commencement — not through this document.

(3) Occupancy and code review for the change of use from staging+light-testing to continuously-occupied lab. Granted by AHJ (Shakopee building dept) + Emerson EH&S. Stage 0 deliverable, before Phase 1 fit-out. Triggers H-occupancy classification check at full skid-scale electrolyte inventory (IBC 2021 §415 / IFC §5004–5005); ventilation envelope per Z9.5 + ASHRAE 170 already spec’d in Operations Manual §4.

The full engineering case

Approvals matrix (full table in §Approvals Required of hub/01_proposal.html): Ask 1 (Conversion): Facilities + site management; end of Q1 with dated clearance manifest; reversion clause covers descope within 24 months back to compliant staging with the same posted-policy clearance SLA. Ask 2 (Co-tenant MOU): Facilities + Logan Woolery routing; written MOU before Stage 1 capital commits; preserves the existing L2 sanctioned testing precedent (Capsule Mark, SYNCHROS, NiM re-flash, WPG calibration, Branson weld) as a co-tenant relationship, not a displacement. Coordinate-point identification and program-owner outreach happen through Logan + Facilities at the appropriate stage, not through this document. Ask 3 (Occupancy + code): AHJ + Emerson EH&S; Stage 0 deliverable before fit-out; IBC 2021 §415 + IFC §5004–5005 govern H-occupancy classification, IMC §502 governs hazardous-materials exhaust, ASHRAE 170-2021 + Z9.5-2022 envelope already spec’d in Operations Manual §4 Ventilation Envelope. None of the three asks involves novel approval pathways; all three are paperwork actions Facilities and EH&S already process routinely. Naming them plainly in the proposal body, not the appendix, signals operating on a peer footing.

What happens to Emerson’s investment if the chemistry fails?

In one sentence

Bounded downside. The pilot builds durable capabilities (purification skid, brine recovery, ZLD, battery-materials QA platform, brine-validated instrumentation) — five sellable capabilities, two with humanitarian value. Total battery failure never returns zero. See If the Battery Fails for the gate-by-gate resilience map.

The plain-English version

A pitch that only describes success is a pitch you shouldn’t trust. If Phase 1 kills the chemistry, Emerson doesn’t walk away with nothing — the pilot builds durable capabilities that pay off regardless: (1) an automated, DeltaV-controlled high-purity salt purification skid, sellable into pharma / food / drinking-water remineralization markets (Nedmag’s existing business); (2) a poly-mineral brine-recovery platform, sellable independently as domestic critical-minerals + circular-economy infrastructure; (3) a desal-brine remediation business that turns hypersaline discharge into products; (4) a battery-materials QA + process-control + data platform that applies to every battery chemistry (LFP, sodium-ion, whatever wins), sellable to the whole battery industry regardless of which cell succeeds; (5) instrumentation validated for aggressive brine chemistry — a sales proof-point into mining, potash, and desalination. Two of those (safe drinking water + coastal ecosystem protection via desal remediation) are humanitarian goods on their own. The downside is bounded — this isn’t a single-chemistry gamble, it’s a portfolio option where even total battery failure returns durable, sellable capabilities. That is what a CFO wants to hear, and it is the substance behind the ~$200M worst-case revenue floor in the Financial Outlook.

The full engineering case

Bounded-downside framework: Phase 1’s ~$1.8M is priced as an option, not a gamble, because the downside branch returns durable capabilities regardless of chemistry outcome. Gate-by-gate resilience: Gate 1 (Chen 2026 doesn’t reproduce at grid-relevant C-rate) → chemistry dies; four capabilities remain (purification skid, poly-mineral recovery, battery-materials QA, brine-validated instrumentation) — see If the Battery Fails §02. Gate 2 (reproduces, mixed-valent under-performs at some ratios) → drop down fallback ladder to Ca-rich or pure-Ca; cell still works. Gate 3 (Track B in-house purification can’t hit battery grade) → fall back to Nedmag or TETRA battery-grade supply; purification skid becomes a pharma/water product. Gate 4 (bench works, scale-up fails) → Phase-2 discovery on Phase-2 capital; no Phase-1 loss; process IP + data platform + standalone markets remain. Total battery failure at any gate still leaves Emerson holding the five capabilities. Two are humanitarian on their own merits: (a) drinking-water remineralization via the automated purification skid (billions of people drink demineralized or desalinated water needing high-purity Mg/Ca to be healthy); (b) desal-brine remediation via the poly-mineral recovery platform (hypersaline discharge is killing coastal ecosystems; turning it into mineral products is real environmental infrastructure). Financial link: the ~$200M worst-case revenue floor on the Financial Outlook already prices these capabilities conservatively; the option premium of $1.8M is under 1% of even that worst case. And a separate axis: feedstock-supply resilience (a chemistry that runs on Mg-only, Ca-only, or mixed) means no single feedstock supply chain can halt production either — see If the Battery Fails §03 “Feedstock-supply resilience · a distinct upstream branch”. Full portfolio: Beyond the Battery; gate-by-gate map: If the Battery Fails.

Have a question we didn’t answer?

Every question made the proposal sharper. This page exists because Connor kept asking, and the answers kept holding up to scrutiny. Send the next one.

scan822@me.com

Asked. And answered.

How to read this FAQ

Every card has three layers — use the one that matches your question.

What this FAQ does not cover.

Where the open questions still are.

The Chemistry

What is this battery, in one sentence?

How does a battery actually work?

What exactly did the Hong Kong paper find?

What does “aqueous” mean and why does it matter?

Why won’t this catch fire like phones and EVs do?

Is the electrolyte toxic?

What happens if a cell or skid is punctured in the field?

Aqueous batteries always fail — what about Aquion?

What about Pellion and Toyota’s magnesium-ion programs?

If this is so good, why hasn’t anyone done it before?

Where did this chemistry come from — is it a fluke of 2026?

Has Chen’s specific cell been independently replicated?

How long do these last in years?

How fast can they charge and discharge?

What about energy density — how does that compare?

What’s the round-trip efficiency?

What’s coulombic efficiency, and why did you tier the threshold instead of using one number?

Why split the cycle-life claim into “demonstrated” and “projected”?

Chen ran 120,000 cycles at 20 A/g — how does that translate to C/4 grid rate?

Does this battery grow dendrites like lithium and zinc do?

Whose prior work is this building on?

Why does the battery run on BOTH magnesium and calcium?

Isn’t calcium an impurity you have to remove?

What actually carries the charge — magnesium or protons?

What if the mixed-valent electrolyte doesn’t co-cycle?

Where does the calcium come from — and is it sustainable?

Does the battery degrade, or is it permanent?

When a cell wears out, is it scrapped or restored — and how do you know exactly what to top up?

Does the mixed magnesium/calcium version degrade faster than single-cation?

Why call it “saltwater” if the electrolyte is magnesium/calcium chloride, not sea salt?

The Build

Where will Phase 1 happen?

What is the L2 reservation, and what does it actually cost?

What is a coin cell and why do we start there?

What does “Phase 1 succeeded” actually look like?

How big and heavy is one production skid?

Why so much stainless steel? Where does all the mass come from?

Will skids be custom-built per customer or standardized?

What’s the timeline from lab to deployed product?

Where does the magnesium chloride come from?

Are we really sure we can pull battery-grade purity from desal waste brine?

Where does the salt actually come from — and isn’t New Mexico evaporation-pond extraction environmentally toxic?

Will Emerson ever do every step of the supply chain in-house — including recovering the salt itself from desalination plants?

What about titanium foil supply — is that constrained?

Does the water need to be emptied before transit?

Why does Phase 1 need a Chemical Hygiene Officer — isn’t a rotating safety lead enough for five chemists?

Why write a 30-year Operations Manual before Phase 1 even runs?

How is in-house purification cheaper than buying battery-grade salt?

How do you know which byproducts are actually recoverable?

Can the pilot lab really refine all these minerals to grade?

Do you sell bromine?

The Money

Why does the $1.0–1.8M Phase 1 budget make sense?

If Phase 1 fails, what’s lost?

Is $2B over 10 years a lot for a company like Emerson?

Where do we rank in the capex pecking order — Form, CATL, Powin?

Where does this sit among Na-ion, VRFB, iron-air, and the other emerging chemistries?

How much would one skid cost to build, and what would it sell for?

What’s “$/kWh-cycle” and why does it matter more than $/kWh?

What are the nine revenue surfaces and which one matters most?

What exactly is Plantweb Insight Aqueous?

What is Battery-as-a-Service, and how is it different from leasing a skid?

What if a customer goes bankrupt or switches storage vendors — do we get the asset back?

Do you sell or discharge the byproducts?

Will Emerson aggressively patent this?

How do you protect and commercialize the IP?

Could the inventors patent-block us? What’s our freedom to operate?

The World

Will this actually help climate change?

Will communities rejoice?

Why do hyperscalers care about indoor-safe storage rather than just buying LFP outdoors?

How loud is one of these skids? Will the neighbors hear it?

Is this for residential too? Can I put one in my basement?

Does the battery keep working when the grid goes down?