Ch 20 · Non-Commutativity — The Bridge Domain

20.1 Three Feedstocks, One Fuel Specification

Sustainable aviation fuel (SAF) is not one chemical process — it is a certification standard, ASTM D7566, that currently recognizes eleven distinct production pathways to a single drop-in jet-fuel specification. Three dominate the real production volume, and each one is a different physical route to the same hydrocarbon target.

HEFA (Hydroprocessed Esters and Fatty Acids) starts from lipids — used cooking oil, animal fat, waste greases — and runs them through two catalytic steps: hydrodeoxygenation strips the oxygen from the fatty-acid chain over a sulfided NiMo or CoMo catalyst, then hydroisomerization over a platinum-zeolite bifunctional catalyst branches the resulting paraffins for cold-flow performance. FT-SPK (Fischer-Tropsch) gasifies biomass to syngas, builds it back up into long paraffin waxes over a cobalt or iron catalyst, then hydrocracks and isomerizes those waxes to jet-range molecules over a second bifunctional metal-zeolite catalyst. ATJ (Alcohol-to-Jet) starts somewhere else entirely — fermentation. Sugars are converted by yeast or engineered bacteria into an alcohol (ethanol or isobutanol), which is then dehydrated to an alkene and oligomerized into jet-range chains over an acid zeolite, and finally hydrogenated.

Pathway	Feedstock	Where K, F act	Substrate type
HEFA	Waste lipids (UCO, tallow)	NiMo/CoMo deoxygenation → Pt/zeolite isomerization	chemical (D2-style)
FT-SPK	Gasified biomass (syngas)	Co/Fe Fischer-Tropsch → Pt/zeolite hydrocracking	chemical (D2-style)
ATJ	Sugars → fermented alcohol	engineered enzyme (fermentation) → acid-zeolite dehydration/oligomerization	hybrid: biological + chemical
SIP (farnesane)	Sugars → engineered yeast isoprenoid	terpene synthase (fermentation) → hydrogenation	biological (D3-style)

Only ATJ requires both an enzymatic step and a zeolite-catalytic step to reach the same molecule class the other pathways reach chemically or biologically alone. That's the reason this chapter exists.

20.2 The Feedstock Economics — Mandate Meets the Rural Supply Chain

Before any catalysis, SAF is a policy-forced demand curve landing on a feedstock base that is disproportionately rural, and often in the developing world. The EU's ReFuelEU Aviation regulation mandates a SAF blend share rising from 2% in 2025 to 6% by 2030 and 70% by 2050 — a legally required demand floor, independent of price. HEFA cannot fill it alone: even routing 100% of the world's collected used cooking oil exclusively to aviation would meet only 3–8% of projected 2030 SAF demand. That ceiling is precisely why the mandate is pulling the market toward agricultural and forestry residues — rice straw, wheat straw, cotton stalks, palm and corn residues — feedstocks concentrated in exactly the regions least represented in the SAF industry so far.

The opportunity is real and has been studied directly, not just asserted. Adha et al. (2026) evaluate Indonesia's agricultural-residue potential (palm, rice, and corn residues) as domestic SAF feedstock; Mukherjee & Sanjeevaiah (2026) make the case for India building a resilient biomass value chain for SAF around crop residue and low-cost renewable power, framing it explicitly as a low-carbon rural growth opportunity, not just a decarbonization exercise. Collecting and aggregating residue that would otherwise be burned or wasted, and selling it into a mandate-guaranteed premium market, is a genuine new rural income stream where none existed.

The risk is just as real and deserves the same weight, not a footnote. Aidam et al. (2026) study the food–fuel nexus for SAF transition in Sub-Saharan Africa directly, and the structural risk factors are well documented across the feedstock-economics literature more broadly: insecure land tenure, smallholder-dominated agriculture without collective bargaining power, weak storage and transport infrastructure, climate-sensitive yields, and thin local policy and financing systems. None of these are arguments against using rural biomass for SAF. They are exactly the conditions under which a mandate-driven price premium gets captured by whichever actor already has the capital and infrastructure to aggregate feedstock at scale — which is not automatically the smallholder growing it.

Stated in this chapter's own vocabulary, without inflating it into a new theorem: sustainability certification (ASTM, CORSIA, ReFuelEU's own criteria) is itself a K-like gate, deciding which biomass streams the mandate's price premium is allowed to reach. Whether that gate is built to admit smallholder-aggregated residue on fair terms, or only large vertically-integrated supply chains that can absorb certification's cost and complexity, is a design choice made by policy and financing structures — not a law of chemistry. Getting that design right is a legitimate, separate engineering problem from anything in §20.4 onward, and this book does not pretend the catalysis chapters answer it.

20.3 Why D4 — The Bridge Domain

D1 (the riboswitch family), D2 (zeolite confinement, Chapter 18), and D3 (enzymatic catalysis, Chapter 19) are three physically distinct substrates, each instantiating Theorem 5.3 (Vol. I, §5.3: the operators C, K, F, U do not commute) on its own terms. SAF does not introduce a fourth substrate. It introduces something that hasn't appeared yet: a single manufacturing chain, ATJ, in which a D3-style non-commuting pair (an engineered enzyme's substrate-gating and catalytic-turnover step, Chapter 19's own K and F) hands its product directly to a D2-style non-commuting pair (a zeolite's pore-gating and confined-reaction step, Chapter 18's own K and F).

Concretely: fermentation converts sugar to isobutanol using an engineered decarboxylase — a D3 instance, gated by the enzyme's active site. That isobutanol is then dehydrated and oligomerized over an acid zeolite — a D2 instance, gated by the zeolite's pore and acid-site density. Neither step is optional and neither can substitute for the other; biology cannot oligomerize an alkene into a C10–C12 jet-range chain, and zeolite acid sites cannot ferment a hexose sugar. The chain is genuinely G_D3 then G_D2, in that order, because the input to the second stage does not exist until the first stage's F has run.

This is why the domain tier is D4 rather than a new fundamental substrate: it is the book's first chapter about the seam between two already-established domains, not about a new one. The honest content here is narrower than D1–D3's — it is not "does Theorem 5.3 hold in a new material," it is "what happens at the handoff, and does getting the order wrong inside either half cost you fuel quality the same way getting the order wrong cost Chapter 18's zeolite cage a product."

20.4 Theorem 20.1 — The Same Catalyst Cannot Skip the Order

The empirical case comes first. Guo, Guo, Suzuki, Wu, Yoneyama, Yang & Tsubaki (2020) ran isobutyl alcohol — the same molecule ATJ ferments — over acidic zeolite catalysts (H-Y, SAPO-34, H-MOR, Al-MCM-41) and identified the reaction explicitly as two steps on one catalyst bed: (i) dehydration of the alcohol to isobutylene, a threshold acid-catalyzed reaction, then (ii) oligomerization of the resulting small alkenes into liquid-range chains. Their own framing states plainly that pore size and acid-site density are "the two main factors controlling this catalytic dehydration and oligomerization reaction" — precisely a K (pore/acid gating) and F (chain-building fold) pair, on a single zeolite. Their 2021 follow-up, run over dealuminated zeolite Beta with funding from Japan Airlines and the Green Earth Institute, reports that dehydration and oligomerization catalyzed together, simultaneously, at the same strong acid site produces an inferior fuel compared to the two reactions catalyzed separately.

THEOREM 20.1 — CATALYTIC NON-COMMUTATIVITY, ATJ UPGRADING

Let K denote the zeolite's pore/acid-gating projection and F the oligomerization (chain-folding) operator, both realized on the alcohol-derived alkene stream. If K and F are forced to act at the same site with no separation in space, time, or catalyst identity — i.e., the system cannot express K∘F and F∘K as distinct, controllable orderings — then [K,F] is not merely nonzero, it is uncontrolled: the product distribution is whatever the co-located site happens to produce, rather than the ordering the engineer chose.

This sharpens Chapter 18 and Chapter 19's shared claim. There, non-commutativity was framed as a choice between two well-defined orderings (K∘F vs. F∘K), each individually reachable. Here, Guo et al.'s finding shows a third, worse regime: a design that fails to separate K and F at all, forfeiting the ability to choose either ordering, and getting a measurably inferior fuel as a result. Deliberately staging dehydration and oligomerization — in separate reactors, or on catalysts with spatially distinct acid and shape-selective sites — is the engineering answer to a theorem about operators that do not commute.

DOMAIN APPLICATION SLOT — D-SAF (bridges D-ZEO and D-ENZ, instantiates Theorem 5.3) κ* \leftrightarrow zeolite pore aperture / acid-site density controlling dehydration-oligomerization selectivity (Ch18's η*, re-instantiated) c* \leftrightarrow enzyme active-site gating tolerance in the upstream fermentation step (Ch19's c*, re-instantiated) λ \leftrightarrow combined turnover: fermentation flux (mol isobutanol / g cells / h) feeding zeolite space velocity (WHSV) μ max = -2 \leftrightarrow [open] curvature bound, untested at the D3\toD2 handoff Hill n \approx 3.64 \leftrightarrow [open] does site-separation degree follow the same sigmoid as Ch18's confinement-selectivity or Ch19's mechanism-switch predictions? Status: Domain mapping proposed here for the first time. Parameters in brackets are not yet measured — see §20.11, the open problem this chapter leaves for the camarada.

20.5 HEFA's Mandatory Order — Deoxygenate, Then Branch

Monteiro, dos Santos, Arcanjo, Cavalcante Jr., Fernandez-Lafuente & Vieira's 2022 review of HEFA jet-fuel hydroprocessing lays out why the two catalytic stages cannot be reordered in practice. Hydrodeoxygenation must run first: the fatty-acid feed carries carboxylic-acid oxygen that would foul or poison the acidic hydroisomerization catalyst if it arrived un-stripped, and the paraffinic backbone produced by deoxygenation is what hydroisomerization actually needs to branch. Running the steps in the opposite order — attempting to isomerize an oxygenated ester before deoxygenation — is not a slower alternative; it is not the reaction the second catalyst is built to do.

This is HEFA's version of Chapter 19's lid-gated enzyme (§19.4): a system where |B| = 1, only one ordering is chemically reachable, and the branch set collapses to a single path by construction — not because engineers chose it, but because the feed chemistry forces it. Cold-flow performance (how low a temperature the fuel tolerates before wax crystals form and clog a fuel line) is the practical stake: without hydroisomerization after deoxygenation, HEFA product is a straight-chain paraffin wax, useless as jet fuel regardless of how clean the deoxygenation step was.

20.6 The Enzymatic Half — Engineering the Upstream K

ATJ's first stage is squarely Chapter 19's domain. Xie, Begum, Gunn & Lindblad (2025) directed-evolved α-ketoisovalerate decarboxylase (KIVD) — the enzyme that performs the committed step toward isobutanol and 3-methyl-1-butanol — in cyanobacteria, using error-prone PCR and a high-throughput screen coupled to substrate consumption. This is the same directed-evolution logic Chapter 19 applied to Codexis's sitagliptin transaminase (§19.6): random mutagenesis, a screen against the target reaction, iterative improvement of an active site that gates which substrates the enzyme's F step can act on. The difference is the downstream use — instead of a pharmaceutical intermediate, the product is a jet-fuel precursor, fermented in a photosynthetic host that fixes its own carbon.

Whatever K and F the fermentation stage instantiates — and Chapter 19's Theorem 19.1 and Theorem 19.2 (branch multiplicity) apply to it exactly as written there — its output, not its internal mechanism, is what Section 20.3's zeolite stage receives. The bridge is at the molecule, not at the mathematics: D3's theorem does not need to be re-derived here, only re-used.

20.7 The Zeolite Family, Revisited — A Candidate Dataset for Chapter 18's Open Prediction

Tomasek, Lonyi, Valyon & Hancsok (2020) hydrocracked Fischer-Tropsch paraffin mixtures over a family of platinum-loaded zeolites and reported a clean activity ordering: Pt/H-beta > Pt/H-ZSM-5 > Pt/H-mordenite, tracking the relative number of accessible acid sites, with diffusion constraints limiting the tighter-pored ZSM-5 and mordenite frameworks.

Catalyst	Framework	Hydroisomerization activity	Limiting factor
Pt/H-beta	BEA (12-ring, large pore)	highest	none reported — most accessible acid sites
Pt/H-ZSM-5	MFI (10-ring, medium pore)	intermediate	diffusion constraints in narrower channels
Pt/H-mordenite	MOR (12-ring, but 1-D channels)	lowest	diffusion constraints, 1-D pore blocking

This is worth flagging explicitly rather than letting it pass. Chapter 18's own open prediction (§18.8, PREDICTION D-ZEO) asked whether CH₄ selectivity across CHA, MFI, and FAU frameworks follows the same Hill sigmoid, n ≈ 3.64, that the riboswitch domain measured — and noted plainly that "no dataset yet assembled." Tomasek et al.'s BEA/MFI/MOR hydrocracking-activity comparison is not that exact dataset (different frameworks, different reaction), but it is a real, published, multi-framework catalytic activity series of the right shape sitting in the SAF literature. Whether it can be refit to test Chapter 18's prediction, or whether it only shows that the idea generalizes to a different reaction family, is exactly the kind of check this book asks the camarada to make rather than assuming.

20.8 U — Unfolding into Flight: The Farnesane Bridge

The SIP pathway (Synthesized Iso-Paraffins) unfolds a fully biological route: engineered yeast ferments sugar through the isoprenoid pathway to farnesene, a C15 terpene, which is then simply hydrogenated to farnesane. Farnesane has a freezing point near −70°C — well within jet-fuel cold-flow requirements without any zeolite step at all — and its blend composition is codified in ASTM D7974, the standard test method for determining farnesane, saturated hydrocarbons, and hexahydrofarnesol content in SIP fuel blended with conventional jet fuel. SIP is D3 unfolding directly into U with no D2 handoff required — the cleanest single-domain case in this chapter, and the honest counterexample to the claim that SAF always needs a bridge.

ATJ, by contrast, needs both. Trace its full chain: C concentrates sugar feedstock into a fermentation broth; K (fermentation stage) gates which substrates the engineered decarboxylase accepts; F (fermentation stage) turns over sugar to isobutanol; the product crosses the bridge; K (zeolite stage) gates alcohol dehydration by pore and acid-site geometry; F (zeolite stage) oligomerizes the resulting alkenes into jet-range chains; U hydrogenates and unfolds the upgraded stream into a certified drop-in fuel. G_D3 then G_D2, end to end, is a molecule of sugar becoming a molecule that flies.

20.9 Interactive: Sequenced vs. Co-Located Catalysis

The bars compare, schematically, the qualitative finding Guo et al. (2021) reported: dehydration and oligomerization catalyzed together at one strong-acid site underperforms the same two reactions catalyzed with deliberate separation. No specific yield numbers are asserted here — the source reports a qualitative "inferior fuel" comparison, not the precise magnitude plotted below, which is illustrative only.

⊞ C₅+ Liquid Fuel Quality: Co-Located vs. Sequenced Catalysis (schematic)

co-located K,F (single acid site, order uncontrolled) sequenced K,F (separated sites, order controlled)

Illustrative only — see §20.4 for the qualitative source finding this schematic represents.

20.10 The g-Series of Way-Stations

Reading the chapter index as a roadmap once more: SAF's honest position on the g-series is the least advanced of this book's non-commutativity chapters, and that is worth stating in plain numbers rather than softening it.

Regime	SAF analogue	Status
g⁰ — Quiescent	Unprocessed feedstock: waste lipids, gasifiable biomass, fermentable sugar, all unconverted	observed
g² — Nascent oscillation	First certified pilot-scale pathway per feedstock (HEFA, FT-SPK, ATJ, SIP each separately demonstrated)	prototyped, all four ASTM-certified
g⁶ — Stable micro-cycle	Commercial-scale blending meeting a non-trivial share of global jet fuel demand	not yet reached — 2026 SAF supply is ≈0.8% of global jet fuel, per IATA
g³³ — Stability threshold	Routine, cost-competitive bridge pathways (ATJ-style D3→D2 chains) displacing HEFA's waste-lipid ceiling	early — feedstock and cost barriers dominant
g⁶⁴ — Circuit saturation	SAF as the default aviation fuel, fossil kerosene the exception	Axiom 9 — honest incompleteness

Unlike Chapter 19, where the sitagliptin transaminase had already reached g⁶ as a deployed industrial process, SAF as a whole has not. IATA's own 2026 figures put global SAF production at roughly 2.4 million tonnes against a total jet-fuel market where that is approximately 0.8% of demand — up from 0.6% in 2025, but explicitly described by industry reporting as "teeny, tiny and embryonic" relative to 2050 net-zero goals. Axiom 9 applies with unusual force here: this chapter does not claim SAF is close to saturating aviation fuel demand. It claims that the same non-commuting K and F that decide whether a zeolite makes good jet fuel or a fouled catalyst bed, or whether an enzyme accepts a bulky substrate or rejects it, also govern — recursively — whether each additional plant, each additional pathway, and each additional bridge between domains produces usable fuel or an inferior blend.

20.11 Open Problem for the Camarada

PREDICTION D-SAF — SITE-SEPARATION SELECTIVITY SIGMOID

Guo et al. (2020, 2021) established, qualitatively, that co-locating K (dehydration/pore-gating) and F (oligomerization) at one acid site produces inferior C₅+ selectivity compared to deliberately separated catalysis. Does the size of that quality gap scale with a measurable "site-separation degree" — for instance, the spatial distance between acid and shape-selective sites in a dual-catalyst bed, or the residence-time offset between two staged reactors — following the same dm³ sigmoid this book has proposed in the riboswitch domain (n ≈ 3.64, established), the zeolite domain (§18.8, open), and the enzyme domain (§19.9, open, competing against a mean-field n = 2)? A second, more immediately actionable task: Tomasek et al.'s (2020) Pt/H-beta, Pt/H-ZSM-5, Pt/H-mordenite hydrocracking-activity series (§20.7) is real, published, multi-framework data sitting adjacent to Chapter 18's still-unassembled CHA/MFI/FAU dataset — determine whether it can be adapted to test that prediction directly, rather than waiting for a purpose-built experiment.

Status: [ ] Pending — no purpose-built dataset yet assembled; one adjacent real dataset (Tomasek et al. 2020) flagged as a candidate · Genre: literature reanalysis + sigmoid fit · Level: D4+