# Biological, Chemical and Informational Inventions

Twenty-five devices, materials and processes whose principle is molecular, biological, chemical or informational. Each rests on a named mechanism — a catalytic site, a photoisomerization, a membrane pore, a redox couple, a binding equilibrium or a strand-displacement reaction. Where a biological system appears, the claim is about one of its molecules, never about the behaviour of an organism. None is a software product, and none depends on a hypothetical material.

### I026 — Thermal-window ligation ratchet

**What it does.** Joins two oligonucleotide fragments on a template without an enzyme, and enriches correct joins over mismatched ones by cycling through a narrow temperature window.

**Principle.** Chemical ligation requires the reactive ends to be held in a duplex. A single mismatch lowers the duplex melting temperature by a few degrees. If the cycle holds between the matched and mismatched melting temperatures, matched duplexes re-form and ligate on every cycle while mismatched ones stay melted. The ratio compounds geometrically, so fidelity comes from the temperature window rather than from a protein.

**Why it does not exist yet.** The activated phosphate needed for enzyme-free ligation hydrolyses on the same timescale as the annealing hold, so most of it is destroyed before it reacts. The usable window is only a few degrees wide for short duplexes and narrows as the duplex lengthens, which is exactly when discrimination improves.

**First test.** A 20-nucleotide template with 10 + 10 donors carrying a 5′-phosphorimidazolide. Measure per-cycle yield for a matched junction and for a mismatch at position 5 over ten cycles stepping 0.5 °C through 10–30 °C. Viability: matched yield above 30% and a matched-to-mismatched ratio above 10.

**Nearest known capability.** T4 DNA ligase, which needs a protein, a free 5′-phosphate and a fixed temperature optimum.

### I027 — Dwell-time fingerprint of modified bases

**What it does.** Identifies which covalent modification a DNA base carries from the shape of the polymerase dwell-time distribution at that position, without chemically converting the base.

**Principle.** A polymerase selects a base by shape and stacking in its catalytic site. A modification such as methylation, hydroxymethylation or oxidation changes the base's size, hydrogen bonding and stacking, which changes both the catalytic rate and the rate at which the incoming nucleotide dissociates. The dwell-time distribution therefore has a modification-specific median and tail; the median and the fraction of long pauses form a fingerprint read from one sequencing run.

**Why it does not exist yet.** Sequence context shifts dwell times by more than the modification does, so a fingerprint learned in one context does not transfer. The rare long pauses that carry the most information are also the events most easily produced by enzyme impurities and template damage, and enzyme-to-enzyme variation adds a second variance component.

**First test.** One synthetic template carrying the same four-base motif in five flanking contexts, each with four modifications, and at least 200 single-molecule dwell events per site. Ask whether a two-feature classifier separates the four modifications within a context and transfers across contexts.

**Nearest known capability.** Bisulfite sequencing, which destroys the base, and nanopore current signatures, which need a pore and a trained model.

### I028 — Pseudoknot-locked toehold

**What it does.** Suppresses leak in DNA strand-displacement reactions by hiding the toehold inside a pseudoknot that only a fully matched input can unwind.

**Principle.** A strand-displacement gate is off until an input binds a single-stranded toehold and invades the adjacent duplex. Leak comes from spontaneous breathing of that duplex. If the toehold is sequestered in a pseudoknot, the input must first unwind the pseudoknot, which costs more than unwinding a plain duplex, and a mismatch leaves the pseudoknot partly intact. The lock converts a small thermodynamic difference into a large kinetic one.

**Why it does not exist yet.** Pseudoknot folding needs magnesium and is slow, so locked and unlocked conformations coexist for hours. The magnesium that folds the lock also stabilises the duplex to be invaded, reducing the gain. The lock slows the correct input's on-rate as well, so leak suppression is bought with response time and the balance is unmapped.

**First test.** Two gates with identical duplexes, one with a plain toehold and one with a pseudoknot lock; fluorescence onset for matched input and for one, two and three mismatches in 1–20 mM Mg²⁺. Viability: discrimination rises tenfold without the matched onset time rising tenfold.

**Nearest known capability.** Clamp and hairpin locks, which extend the toehold but do not use a competing folded structure.

### I029 — Amyloid strain memory

**What it does.** Stores a state in the packing of an amyloid fibril, written by seeding and read by a conformation-specific reporter.

**Principle.** Amyloid fibrils can adopt more than one stable packing of the same sequence. A seed of one packing templates the same packing in fresh monomer, so the conformation is inherited. Two packings that differ in β-strand register expose different surfaces: one binds a dye or an antibody, the other does not. Writing is adding a seed, reading is the surface signature, and erasing is melting the fibril.

**Why it does not exist yet.** Strains are metastable and can switch under growth conditions, so a written state has a finite retention time that has not been measured for a designed pair. The oligomers that accompany growth are toxic to the cells that would produce the monomer, which rules out the cheapest production route, and readout is bulk, so the state is averaged over the population.

**First test.** Two strains of one short peptide, distinguished by a thioflavin-T fluorescence ratio and by protease fragmentation, each seeded into fresh monomer at 1% w/w; measure the daughters' dye ratio. Viability: daughters match their parent for five serial passages, with the states separated by more than the passage-to-passage variance.

**Nearest known capability.** DNA melting memory and photochromic films, which store more states but need light or heat to read.

### I030 — Interfacial activation switch

**What it does.** Confines a catalytic reaction to a lipid membrane surface by making the catalyst inactive in water and active only when a lid binds the membrane.

**Principle.** Some lipases carry a surface loop that occludes the active site in water and moves aside when the enzyme adsorbs to a lipid interface. The loop's position controls access to the catalytic triad, so the rate depends on the bound fraction rather than on bulk concentration. Coupling a catalyst to this lid makes the rate depend on available membrane area, and therefore on where the membrane is.

**Why it does not exist yet.** The lid must be fully closed in water, and most designed lids leak. The membrane must survive the substrate and the products, and adsorption must be strong enough that the bound fraction is near one at usable lipid concentrations. The bound fraction is hard to measure separately from the catalytic step, so the origin of the gain is unknown.

**First test.** A lipase variant with a redesigned lid, assayed with a fluorogenic substrate at seven liposome concentrations from 0 to 5 mM lipid; compare the rate with the fraction bound, measured by fluorescence correlation spectroscopy. Viability: rate proportional to bound fraction and water rate below 5% of the bound rate.

**Nearest known capability.** Immobilised enzymes, which confine a reaction to a solid but cannot be placed by self-assembly.

### I031 — Asymmetric lipid leaflet

**What it does.** Builds a vesicle whose two leaflets have different lipid compositions, and holds the asymmetry long enough to use it.

**Principle.** Lipid flip-flop is slow for lipids with large headgroups because the headgroup must cross the hydrophobic core, and that slowness is what lets asymmetry persist. Asymmetry can be imposed by forming the vesicle from a mixture in which one lipid is initially outside, then modifying its headgroup so the two leaflets differ in charge. The trapped state relaxes over hours, and the relaxation rate is the measurement.

**Why it does not exist yet.** The modification must be fast compared with flip-flop and must not permeate the bilayer; most reagents that react with a headgroup also cross the membrane and symmetrise the composition. The resulting asymmetry is invisible to bulk assays, so it must be measured leaflet by leaflet with quenchers or labelled lipids, and the labels perturb packing.

**First test.** Vesicles of 1-palmitoyl-2-oleoyl phosphatidylcholine with 1% amine-labelled lipid; react the outer leaflet with a membrane-impermeant succinimidyl ester, then measure the labelled fraction by a quenching assay and by mass spectrometry after extraction. Viability: outer-leaflet labelling above 80% with asymmetry persisting four hours at 25 °C.

**Nearest known capability.** Symmetric vesicles, which cannot present different chemistries to the two sides.

### I032 — Membrane electron conduit

**What it does.** Carries electrons across a lipid bilayer with a lipophilic redox shuttle, coupling a reduction on one side to an oxidation on the other without a protein.

**Principle.** A quinone with a long isoprenoid tail partitions into the bilayer and diffuses laterally and transversely. It accepts electrons from a reductant at one interface, moves to the other, and donates them to an oxidant. The shuttle carries two electrons and two protons, so the flux also carries protons and builds a pH difference. The rate is set by the shuttle's membrane concentration, its diffusion coefficient and the driving force between the two couples.

**Why it does not exist yet.** The shuttle must partition strongly enough to give a useful membrane concentration but weakly enough to exchange with the aqueous reactants. Its reduced form reacts with oxygen, which short-circuits the conduit, and its potential must sit between the two couples, which restricts the chemistry that can be coupled. Radicals formed at the interface attack the membrane.

**First test.** A planar bilayer separating two buffered compartments, ferricyanide on one side, a lipophilic ubiquinone homologue in the membrane and a reductant on the other; measure the current across the bilayer. Viability: current scales linearly with quinone concentration and reverses when the couples are swapped.

**Nearest known capability.** Membrane-embedded electron-transfer proteins, which are hard to reconstitute and cannot be retargeted to arbitrary couples.

### I033 — Enzymatic hydrogen-tunneling D/H separation

**What it does.** Separates deuterated from protiated substrate at room temperature by running an enzyme whose hydrogen transfer proceeds by nuclear tunnelling.

**Principle.** In some enzymes the transferred hydrogen moves through the barrier rather than over it. The tunnelling probability falls exponentially with the square root of the particle's mass, so the rate for protium can exceed that for deuterium by more than the classical zero-point-energy factor. Passing a mixture through an immobilised enzyme reactor at partial conversion therefore depletes the light isotope in the product and enriches it in the unreacted stream.

**Why it does not exist yet.** The enzyme works in water, whose protons exchange with the substrate and erase the isotope difference; it must run in a low-water organic phase or in D₂O with a non-exchanging substrate. Throughput per enzyme molecule is low, and the enzyme's own proton inventory means the separation factor is not the intrinsic kinetic isotope effect.

**First test.** Soybean lipoxygenase immobilised on a support in a low-water organic phase with a fatty-acid substrate at natural deuterium abundance; measure the substrate's D/H ratio by isotope-ratio mass spectrometry at 20%, 50% and 80% conversion. Viability: the ratio rises monotonically and the single-pass enrichment exceeds measurement uncertainty threefold.

**Nearest known capability.** Cryogenic distillation and the Girdler sulfide process, which need large plants and low temperatures.

### I034 — Cross-linked enzyme crystal, size-selective

**What it does.** Uses the solvent channels of an enzyme crystal as a size filter, so the crystal catalyses only substrates that fit through the lattice.

**Principle.** Protein crystals are porous: the lattice leaves channels of defined width, and the active sites line those channels. Cross-linking with glutaraldehyde makes the crystals insoluble and mechanically robust without closing the channels. A substrate larger than the channel cannot reach the active site, so the crystal's rate on a mixture depends on molecular size rather than on the enzyme's intrinsic specificity.

**Why it does not exist yet.** Cross-linking costs activity because it reacts with lysines near the active site, and the loss varies between batches. The crystals fracture under stirring, which releases free enzyme and destroys the selectivity. Channel width is fixed by the crystal form, so the cut-off cannot be tuned without finding a new form, and product can block the channels.

**First test.** Cross-linked crystals of a model hydrolase and the same enzyme in free solution, at equal active-site loading; measure the rate on a small substrate and on one too large for the channel. Viability: crystal rate on the large substrate below 5% of free enzyme while the small-substrate rate stays above 50%.

**Nearest known capability.** Enzymes immobilised on mesoporous supports, which give recycling but not a sharp cut-off set by the protein lattice.

### I035 — Radical-pair magnetic isotope separation

**What it does.** Enriches a magnetic isotope by running a photochemical radical-pair reaction to partial conversion, using the nuclear-spin dependence of singlet–triplet interconversion.

**Principle.** A photogenerated radical pair starts in a singlet state and interconverts with the triplet at a rate set by hyperfine coupling to the nuclei. Nuclei with spin, such as carbon-13 or oxygen-17, couple more strongly than spin-zero nuclei, so pairs containing them reach the triplet sooner and a different fraction escapes the solvent cage to form product. Stopping at partial conversion leaves the unreacted starting material isotopically shifted.

**Why it does not exist yet.** The effect per pass is small and is erased by fast cage recombination, oxygen quenching and spin relaxation. The pair must live long enough for coherent interconversion, which requires rigid, deoxygenated media and restricts the chemistry to a few photolabile precursors. Optical penetration limits reactor thickness, so throughput is hard to raise.

**First test.** Photolyse dibenzyl ketone in a deoxygenated micellar solution to 20% conversion; measure the carbon-13 content of the recovered ketone and of the product by isotope-ratio mass spectrometry. Viability: the fractions differ beyond replicate reproducibility, and the sign reverses in a homogeneous solvent that allows faster recombination.

**Nearest known capability.** Cryogenic distillation and gas centrifugation for carbon-13.

### I036 — Boronate saccharide affinity separator

**What it does.** Separates sugars, glycopeptides or ribonucleosides on a packed bed by reversible covalent binding, eluting with a pH step instead of a salt gradient.

**Principle.** Boronic acids condense with cis-1,2-diols to form cyclic esters. Only the tetrahedral boronate anion binds, so the equilibrium is strongly pH-dependent: tight at pH 8.5, negligible at pH 5. Stability also depends on diol geometry, so fructose, glucose and galactose differ. A bed of immobilised boronate is therefore a pH-switchable affinity phase whose selectivity is chemical rather than based on molecular size.

**Why it does not exist yet.** Aqueous binding constants are millimolar, so useful capacity needs a very high ligand density, which swells and compresses the support. Silica dissolves at the binding pH, and the acidic elution can denature a protein product. Boronate also binds the glycans of the column's own coating, which shortens its working life.

**First test.** A 10 cm bed of 4-carboxyphenylboronate beads, 1 mg each of glucose and fructose, eluted with a step from pH 8.5 to pH 5, with refractive-index detection. Viability: baseline resolution, capacity above 20 mg/mL, and under 5% ligand loss over 50 cycles.

**Nearest known capability.** Hydrophilic-interaction and ion-exchange chromatography, which separate by polarity or charge and need a salt gradient.

### I037 — Self-immolative amplifier

**What it does.** Turns one molecular triggering event into the release of many reporter molecules by unzipping a polymer from one end.

**Principle.** A polymer whose backbone is a chain of self-immolative units — each a substituted benzyl carbamate or a phthalaldehyde repeat — carries a trigger at its head and a reporter on every unit. Cleaving the head generates a nucleophile that eliminates the next unit, exposing a new nucleophile, and the process propagates to the tail. The gain equals the degree of polymerisation, and release is fast because each step is intramolecular.

**Why it does not exist yet.** The polymer must be stable in water for the storage time, and the side-group link must be labile only after unzipping, which is a narrow stability window. Synthesis gives a distribution of chain lengths, so the gain is an average and the dose-response is smeared. The unzipped fragments can inhibit the readout.

**First test.** A 50-unit poly(benzyl carbamate) with a fluorogenic side group and a protease-cleavable head; measure fluorescence for one hour without trigger and after trigger addition. Viability: untriggered rate below 1% of the triggered burst per hour and at least 30 fluorophores released per trigger.

**Nearest known capability.** Enzymatic amplification, which needs a protein enzyme for every step and cannot be stored dry.

### I038 — Bistable reaction-network latch

**What it does.** Stores one state in the steady state of a reaction network with two stable states, and holds it without continuous input.

**Principle.** A network with positive feedback and a saturable sink has two stable steady states separated by an unstable one. If the feedback species activates its own formation and the sink is a Michaelis–Menten removal, the steady-state curve is S-shaped. A transient input above the threshold flips the network to the high state, which persists after the input is removed.

**Why it does not exist yet.** The two states must be separated by more than the network's noise, and the noise scales with the inverse square root of molecule number, so the latch needs a minimum volume and concentration. The feedback species must not be consumed by side reactions, and the sink must not saturate, which sets a narrow parameter window. Readout perturbs the state if it consumes the feedback species.

**First test.** A cell-free network with an autocatalytic RNA ligase and an RNase sink in a microfluidic chamber array; apply input pulses of increasing length and measure the fraction of chambers that stay high after the pulse. Viability: the fraction rises from below 10% to above 90% over a twofold change in pulse length and persists for ten chamber volumes of washout.

**Nearest known capability.** Electronic flip-flops and DNA strand-displacement latches, which need power or continuous strand turnover.

### I039 — Chemical differentiator

**What it does.** Produces a signal proportional to the rate of change of an input concentration rather than to its level, using two reporter species with different response times.

**Principle.** Two binders for the same input, one fast and one slow, report the input at two different lags. The difference between the fast and slow signals is approximately the input's derivative times the lag difference. If the two reporters have matched dynamic range and are read ratiometrically, the difference cancels the level and leaves the rate.

**Why it does not exist yet.** The derivative is a difference of two noisy signals, so noise is amplified by the ratio of the lag difference to the lag itself. The reporters must have matched affinities and matched readouts, which is hard with different chemistries. The slow reporter must not saturate, and its lag must be stable over the device's lifetime.

**First test.** A pair of fluorescent binders for the same ion with dissociation rates differing fiftyfold, in a flow cell; apply a step, a ramp and a sinusoid and compare the difference signal with the known derivative. Viability: the step gives a transient decaying to baseline, the ramp a constant offset, and the sinusoid phase lag within 15° of a true derivative over a decade of frequency.

**Nearest known capability.** Electronic differentiators and enzyme cascades, which need power or a continuous substrate supply.

### I040 — Chemical oscillator clock

**What it does.** Provides a timing reference from the phase of a chemical oscillator, so a downstream reaction can be triggered at a fixed delay without electronics.

**Principle.** A Belousov–Zhabotinsky mixture in a gel oscillates between oxidised and reduced states with a period set by the concentrations of bromate, acid and organic substrate. The cycle is a relaxation oscillation: bromide accumulates and inhibits the autocatalytic step, then is consumed and the step restarts. The phase is a clock, and a threshold detector on the redox state fires once per cycle.

**Why it does not exist yet.** The period drifts as the reaction consumes reactants and accumulates bromine, and depends strongly on temperature, so the clock loses phase over hours. The gel must vent gas without drying, and the redox couple that reports the phase must not perturb it. The oscillation stops if bromide escapes or the catalyst photobleaches.

**First test.** A ferroin-catalysed bromate oscillator in a 2 mm agarose slab at 25 °C with a platinum electrode reading the redox potential; record the period for 24 hours, then repeat at 20, 25 and 30 °C. Viability: period reproducible to better than 5% between runs at one temperature, with a stable calibrated activation energy.

**Nearest known capability.** Quartz oscillators and enzymatic timers, which are more stable but need power or a purified protein.

### I041 — Dynamic combinatorial library with a covalent trap

**What it does.** Finds a tight binder for a target by letting a library of fragments re-equilibrate in the target's presence, then freezing the best combination with a chemical reaction.

**Principle.** Reversible bonds — disulfides, imines or boronate esters — let fragments exchange continuously, so the library's composition shifts toward combinations that bind the target. The shift is thermodynamic amplification: the bound combination is stabilised by the binding free energy, so its concentration rises exponentially with affinity. An irreversible reaction then locks the amplified combination so it can be isolated.

**Why it does not exist yet.** The lock must be fast compared with re-equilibration and must not trap the unamplified library, which sets a narrow window between exchange and locking rates. The target must tolerate the exchange chemistry, fragment solubility caps amplification, and the locked products are a mixture, so identifying the winner needs a separation that does not itself shift the equilibrium.

**First test.** A disulfide library of 20 thiols with a model target at 1 mM total thiol and 10 µM target; equilibrate for 24 hours, lock with a thiol-reactive quencher, and compare product distributions with and without the target by liquid chromatography–mass spectrometry. Viability: one product enriched fivefold by the target, and no enrichment with a non-binding analogue.

**Nearest known capability.** High-throughput screening, which tests compounds one at a time and cannot amplify a weak binder.

### I042 — Transition-state-imprinted catalyst

**What it does.** Catalyses a bond-breaking reaction with a polymer cavity shaped around a transition-state analogue, in water.

**Principle.** A functional monomer and a cross-linker are polymerised in the presence of a stable analogue of the reaction's transition state. Removing the analogue leaves a cavity with the analogue's shape and with functional groups positioned where the transition state needs them. The cavity stabilises the transition state more than the ground state, lowering the barrier, and it orients the substrate.

**Why it does not exist yet.** Imprinting in water is difficult because water competes for the hydrogen bonds between monomer and template, so most imprinted catalysts are made in organic solvents and lose their advantage when transferred. The cavity population is heterogeneous, so only a fraction of sites have the right geometry and the average rate is diluted. Residual template can catalyse the reaction non-selectively.

**First test.** An imprinted polymer for hydrolysis of a model ester, made with a phosphonate transition-state analogue; measure the rate in water at pH 7 with the imprinted polymer, a non-imprinted polymer and the free template. Viability: at least tenfold acceleration over the non-imprinted polymer, which disappears when template is added to the assay.

**Nearest known capability.** Enzyme catalysis, which is faster but needs a protein and a folded structure.

### I043 — DNA-templated sequence-defined polymer

**What it does.** Writes a sequence-defined non-natural polymer by using a DNA template to order the monomers, then releases the product.

**Principle.** Each monomer carries a short oligonucleotide tag complementary to a codon on the template and a reactive group. When two monomers' tags hybridise to adjacent codons, their reactive groups are held close enough to react, joining the monomers in template order. Repeating the cycle with fresh monomers builds a polymer whose sequence is read from the template rather than from a synthesizer.

**Why it does not exist yet.** Each coupling must be near-quantitative, because a 20-step synthesis at 95% per step gives only 36% full-length product, and unreacted monomers must be washed out without hydrolysing the linkages. Background coupling in solution sets a floor on fidelity, and releasing the product without damaging it needs a chemistry orthogonal to every linkage in the chain.

**First test.** A 10-codon template and 10 monomers, each with a distinct mass tag; read the product by mass spectrometry after release. Viability: full-length product above 50% of total polymer and the most abundant error a single missing monomer rather than a mis-ordered one.

**Nearest known capability.** Solid-phase synthesis, which needs a synthesizer and harsh deprotection, and biological translation, which is limited to α-amino acids.

### I044 — Melt-signature archive integrity

**What it does.** Gives a DNA archive an intrinsic integrity check, so a mutated block is detected without sequencing the whole archive.

**Principle.** Each stored block is a duplex whose length and GC content place its melting transition at a designed temperature, and the blocks are spaced so their transitions do not overlap. A mutation changes a block's transition temperature and the shape of its melt curve. Reading the pool's high-resolution melt curve therefore gives a per-block integrity signal, with blocks resolved by temperature rather than by sequencing.

**Why it does not exist yet.** A single mismatch shifts the melting temperature by a fraction of a degree, comparable to the width of the transition, so the blocks must be short and the transitions separated by less than the curve can resolve. The fluorescent dye redistributes between duplexes as they melt, which distorts the curve, and polymerase errors from amplification are indistinguishable from storage damage.

**First test.** Ten 40-base-pair blocks with melting temperatures spaced one degree apart, pooled at equal concentration, with one block mutated in 10% of its copies; compare melt curves on a high-resolution instrument. Viability: signal-to-noise above three against the unmutated pool with correct block assignment.

**Nearest known capability.** Sequencing the archive, which is definitive but expensive, and PCR assays, which need a primer pair per block.

### I045 — Nitrate-to-ammonia electrocatalytic converter

**What it does.** Converts nitrate dissolved in water into ammonia at ambient temperature and pressure on a copper electrode, with the ammonia captured as a salt.

**Principle.** Nitrate adsorbs on copper and is reduced stepwise through nitrite, nitric oxide and hydroxylamine to ammonia. The first electron transfer is rate-determining, and the branch between ammonia and nitrogen gas depends on the coverage of adsorbed nitric oxide and on the local pH. A gas-diffusion electrode holds a thin electrolyte film, so nitrate reaches the catalyst quickly and the local pH can be kept high enough to suppress hydrogen evolution.

**Why it does not exist yet.** Hydrogen evolution competes for the same electrons, ammonia crosses to the anode and is oxidised there, and copper reconstructs under reaction so the active facets disappear. At the low nitrate concentrations of real water the current efficiency collapses because transport, not catalysis, becomes limiting.

**First test.** A copper gas-diffusion electrode in 0.1 M potassium nitrate at −0.4 V against a reversible hydrogen electrode; measure faradaic efficiency and ammonia formation rate. Viability: above 80% efficiency at 100 mA/cm² with under 5% ammonia crossover in 24 hours.

**Nearest known capability.** Haber–Bosch synthesis and biological nitrogen fixation.

### I046 — Sugar-glass vitrification

**What it does.** Stores a reagent at room temperature in a dry sugar glass that preserves its activity for months.

**Principle.** Trehalose and similar sugars form a glass when water is removed, and the glass's high viscosity arrests the motions that denature a protein. The sugar also replaces water's hydrogen bonds at the protein surface, so the folded state is stabilised without water. The glass transition temperature sets the storage limit: below it the matrix is a solid and the reagent is frozen in place.

**Why it does not exist yet.** The glass must be dry enough that water does not plasticise it, but drying itself can denature the protein, so the drying path matters as much as the final composition. Trehalose crystallises at high humidity, which collapses the glass and releases the reagent, and rehydration must be fast enough that the reagent dissolves before it aggregates.

**First test.** Lysozyme in a trehalose glass dried at 30 °C and stored at 40 °C and 75% relative humidity for four weeks; measure activity against a freshly dissolved standard. Viability: above 80% activity retention at four weeks and no crystallisation by X-ray diffraction.

**Nearest known capability.** Freeze-drying with lyoprotectants, which needs a freeze-dryer and a cold chain.

### I047 — Two-threshold mechanophore peak-stress record

**What it does.** Records the highest mechanical stress a solid has experienced, as a two-level readout, without electronics.

**Principle.** A mechanophore is a molecule whose bond breaks faster under force. The rupture rate is exponential in force, so the fraction broken is a steep function of the peak force and is cumulative. Embedding two mechanophores with different force thresholds in the same polymer gives two fractions, which together bracket the peak stress. The record is read after unloading, so the device needs neither power nor interrogation during the event.

**Why it does not exist yet.** The force on a chain in a bulk polymer is not the applied stress; it is amplified by entanglements and by the network's strain, so calibration depends on architecture and loading rate. The mechanophore is also a weak point that nucleates fracture, which caps its concentration, and the released fragment must be detectable at the concentrations reached.

**First test.** Poly(methyl methacrylate) films with 0.05 mol% of each of two spiropyran mechanophores, loaded in uniaxial tension to five peak stresses and unloaded; measure the absorbance of the merocyanine band. Viability: each signal monotonic in peak stress, thresholds separated by a factor of two, and repeated loading to a peak reproducible within 10%.

**Nearest known capability.** Strain gauges and piezoresistive sensors, which read instantaneous strain and need wiring and power.

### I048 — Photoacid pH pulse

**What it does.** Creates a fast, spatially confined pH jump in solution by releasing protons from a photoacid with a light pulse.

**Principle.** A photoacid such as a hydroxypyrene derivative becomes much more acidic in its excited state, and on relaxation it transfers a proton to the surrounding water. The result is a pH jump of one to two units within nanoseconds, confined to the illuminated volume. A reaction whose rate is pH-dependent can therefore be switched by light rather than by adding acid and base.

**Why it does not exist yet.** The jump is small and transient, because protons recombine with the conjugate base as the dye returns to its ground state. The photoacid must be regenerated without accumulating photoproducts, and its excited state must not react with the substrate. Buffer capacity, not photoacid concentration, sets the size of the jump.

**First test.** A hydroxypyrene photoacid with a fast pH indicator in a stopped-flow cell; fire a nanosecond laser pulse and record the indicator's absorption at microsecond resolution. Viability: a shift above 0.5 units with a relaxation time at least tenfold longer than the mixing time.

**Nearest known capability.** Stopped-flow mixing of acid and base, which is slower and not spatially confined.

### I049 — Antifreeze glycopeptide ice-growth inhibitor

**What it does.** Stops ice crystals from growing during freezing and thawing, so a frozen reagent keeps its structure.

**Principle.** Antifreeze glycopeptides adsorb to the prism faces of ice through hydrogen bonds from their hydroxyl and amide groups. Adsorption blocks the step edges where new ice would grow, so growth can proceed only between adsorbed molecules, which forces the surface to curve. Curvature raises the local melting temperature by the Gibbs–Thomson effect, and growth stops below the temperature at which the adsorbed molecules cover the surface.

**Why it does not exist yet.** The ice-binding face must match the ice lattice, so activity is lost with small changes in repeat length or glycosylation. The peptides are expensive to make, their effect saturates above a concentration, and adsorption is effectively irreversible on the timescale of a freeze–thaw cycle, so repeated cycling consumes them.

**First test.** A splat assay: freeze a microlitre drop on a cold stage, hold at −6 °C for 30 minutes, and measure mean crystal size by microscopy with and without peptide at 0.1, 1 and 10 mg/mL. Viability: crystal size at 1 mg/mL fivefold smaller than the control, and the effect does not reverse on warming.

**Nearest known capability.** Glycerol and dimethyl sulfoxide, which lower the freezing point but do not stop recrystallisation and are toxic at high concentration.

### I050 — Mussel-style oxidative wet adhesive

**What it does.** Cures an adhesive underwater by oxidising a catechol to a quinone that cross-links the polymer and couples it to the surface.

**Principle.** 3,4-Dihydroxyphenylalanine residues are oxidised to quinones, which react with amines and thiols on the polymer to form covalent cross-links and with nucleophiles on a mineral surface to form charge-transfer bonds. The reaction consumes oxygen, so it competes with water, and the cure is fastest where the oxidant is supplied. The result is a cross-linked network that adheres through covalent and coordination bonds rather than through drying.

**Why it does not exist yet.** The quinone reacts with water, so the activated species has a short lifetime and the adhesive must be mixed at the point of use. The cure rate is pH-dependent and too fast above pH 8, which limits the formulation window. The network is brittle, and further oxidation gives products that do not couple to the surface.

**First test.** Lap-shear joints of glass and polytetrafluoroethylene bonded underwater with a catechol-functionalised polymer and an oxidant, cured for 24 hours. Viability: shear strength above 1 MPa on glass, surviving 24 hours of immersion, and failing cohesively rather than at the interface.

**Nearest known capability.** Cyanoacrylate and epoxy adhesives, which need a dry surface and fail at the interface underwater.