# Physical and Chemical Inventions

Twenty-five instruments, devices, materials and processes whose principle is physical or chemical. Each rests on a named mechanism — a transport process, a defect, a reaction, a thermodynamic cycle or a mode — and each names the technical obstacle that has kept it unbuilt. None depends on a hypothetical material.

### I001 — Salt-Gradient Diffusiophoretic Stack

**What it does.** Concentrates dilute colloids, proteins or vesicles into a small volume using only a salt gradient, with no membrane, centrifuge or applied field.

**Principle.** Diffusiophoresis: in a solute concentration gradient, a charged particle drifts with the slip velocity of its diffuse double layer. The drift combines an electrophoretic term from the field set up by differing cation and anion mobilities, and a chemiphoretic term from the osmotic pressure gradient inside the double layer. A stack of co-flow stages, each with a fresh high-salt/low-salt junction, repeatedly sweeps particles toward a stagnation point, so concentration compounds stage by stage instead of saturating in one channel.

**Why it does not exist yet.** A single junction decays in seconds, so the drift stops before a useful concentration is reached. Sustaining it needs continuous co-flow with matched withdrawal, and the factor per stage is capped by the ratio of drift velocity to Brownian back-diffusion — only about an order of magnitude for 100 nm particles. Stacking multiplies the pressure drop and salt consumption, and the salt must be removed without re-diluting the product.

**First test.** A two-stage PMMA device, 200 nm carboxylate polystyrene, 0.5 M against 5 mM NaCl, 1 µL/min per stream; measure the stagnation-point concentration factor by fluorescence. Viability: above 20× in 10 min with under 10% wall loss.

**Nearest known capability.** Stirred ultrafiltration and tangential-flow filtration, both of which foul and require pressure.

### I002 — Marangoni-Renewed Liquid-Membrane Contactor

**What it does.** Extracts a solute from an aqueous feed into an immiscible solvent through a thin, continuously renewed liquid film, without dispersing the two phases.

**Principle.** Marangoni convection: solute transfer across a liquid–liquid interface changes the local interfacial tension. Where transfer is fast, the interface becomes unstable and rolls the film, refreshing the boundary layer. The device pins a thin solvent film in a micromachined grid between two aqueous streams and lets the interfacial-tension gradient do the stirring, so mass transfer is limited by partition and reaction rather than by diffusion through a stagnant film.

**Why it does not exist yet.** The same gradient that renews the film can emulsify it into the feed by spontaneous emulsification, and the film ruptures when the pressure difference across the grid exceeds the capillary pressure. Solvent loss into the raffinate and fouling of the grid are the practical walls, and the mass-transfer gain is largest exactly where the film is thinnest and most fragile.

**First test.** A 200 µm grid, water/1-octanol, acetic acid transfer; measure the overall mass-transfer coefficient with and without an imposed interfacial-tension gradient, set by changing feed pH, and the solvent carryover. Viability: the coefficient at least three times the stagnant-film value with under 50 ppm solvent in the raffinate over 24 h.

**Nearest known capability.** Packed extraction columns and membrane contactors.

### I003 — Thermo-Osmotic Gel Press

**What it does.** Dewaters a gel, sludge or biomass by driving water through a charged nanoporous membrane with a small temperature difference, instead of mechanical pressure.

**Principle.** Thermo-osmosis: water confined in a charged pore moves along a temperature gradient because its excess enthalpy differs from that of the bulk fluid. The effect is largest when the double layers overlap, so a membrane with 20–50 nm pores and a fixed surface charge acts as a thermal pump for water while rejecting dissolved salts and colloids. The direction is set by the sign of the surface charge and the enthalpy of adsorption, so the membrane can be chosen to pump toward the cold side.

**Why it does not exist yet.** The volumetric flux per unit area is tiny at a few kelvin, so the press needs a large membrane area or many stages. The effect reverses sign when the surface charge is neutralized by adsorbed organics, so the device can silently stop or run backwards. The cold side also concentrates solutes, and the resulting osmotic back-pressure cancels the thermal drive at high solids.

**First test.** A 100 nm anodic alumina disc with a 5 K gradient, 1 wt% bovine serum albumin gel; measure water flux and protein rejection. Viability: above 0.5 L/m²·h at 5 K with over 95% protein retention and no sign reversal after 72 h.

**Nearest known capability.** Filter presses and electro-osmotic dewatering.

### I004 — Photophoretic Aerosol Enricher

**What it does.** Concentrates light-absorbing aerosol particles from a large air volume into a small one for analysis, using a light beam rather than a filter.

**Principle.** Photophoresis: an absorbing particle under directional illumination develops a temperature difference between its illuminated and shadowed faces. Gas molecules leaving the warm face carry more momentum, so the particle is pushed. In a weakly converging coaxial beam the force is toward the axis for absorbing particles and much weaker for transparent ones, so the beam acts as a species-selective funnel.

**Why it does not exist yet.** Photophoretic forces are large only at low pressure, where the Knudsen number is near one, or for strongly absorbing, low-density particles. At atmospheric pressure, natural convection from the beam and thermophoresis from the heated walls overwhelm them. The beam also heats the collection zone, so enriched particles evaporate or thermodeposit on the window.

**First test.** A 200 mW, 405 nm beam through a 20 cm cell at 0.2 atm, with 300 nm carbon-black and 300 nm silica particles; measure axial enrichment with a downstream counter. Viability: above 10× enrichment for the absorbing fraction, below 2× for silica, at 1 L/min sampling.

**Nearest known capability.** Filter sampling and impactors, which are blind to absorptivity.

### I005 — Acoustophoretic Nanoparticle Classifier

**What it does.** Separates nanoparticles below 100 nm by size in continuous flow, using a standing acoustic wave.

**Principle.** The acoustic radiation force on a particle scales with its volume and with the contrast in density and compressibility, while the opposing Stokes drag scales with diameter. In a standing wave the deflection of a particle from a flow streamline is therefore size-dependent, and a flow profile tilted relative to the pressure nodes converts that deflection into a lateral exit channel. Driving the resonator at tens of megahertz pushes the useful size range down toward 100 nm.

**Why it does not exist yet.** At the frequencies needed, viscous boundary-layer losses heat the fluid, and acoustic streaming becomes comparable to the radiation-force drift for small particles, scrambling the separation. The force also falls as the third power of radius, so a cut below 100 nm needs fields that cavitate water.

**First test.** A 20 MHz lithium niobate transducer with a 100/250/500 nm polystyrene mixture; measure outlet size distributions by dynamic light scattering at 10 µL/min. Viability: over 70% of 500 nm particles in the target outlet with under 20% of 100 nm particles co-collected.

**Nearest known capability.** Field-flow fractionation and size-exclusion chromatography.

### I006 — Pressure-Recovery Barocaloric Heat Pump

**What it does.** Cools or heats with the temperature change of a solid under hydrostatic pressure, recovering most of the compression work, so it can approach vapor-compression efficiency without a refrigerant gas.

**Principle.** The barocaloric effect: the entropy of a solid changes with pressure, so an adiabatic pressure increase warms it and a release cools it. Plastic crystals with orientational disorder, such as neopentyl glycol and layered perovskites, show large entropy changes at modest pressure. A pair of beds exchanging pressurized oil through a hydraulic intensifier recovers the compression work, and a heat-transfer fluid sweeps each bed at the two pressure plateaus.

**Why it does not exist yet.** Compressing a solid plus oil takes large work, and unless it is recovered the cycle efficiency collapses. Liquid pressure exchangers exist but must run at 100–300 MPa with low leakage and without mixing the hot and cold streams. The material must also survive millions of cycles without cracking or losing its hysteresis.

**First test.** A single-bed servo-hydraulic bench with 5 g of neopentyl glycol, 0–100 MPa at 1 Hz; measure the adiabatic temperature change, the work recovered per cycle, and the crack density after 10⁴ cycles. Viability: above 15 K, above 70% work recovery, no visible cracks.

**Nearest known capability.** Vapor-compression and magnetocaloric cooling.

### I007 — Thermal-Wave Adsorption Chiller

**What it does.** Produces cooling from 60–90 °C heat with a solid adsorbent, heating the bed through its own metal support and regenerating it in a traveling thermal wave.

**Principle.** In an adsorption cycle a porous adsorbent takes up vapor at low pressure, and heating desorbs it at higher pressure; the vapor condenses and later evaporates to provide cooling. Coating the adsorbent directly on a resistive metal mesh allows a narrow moving hot zone instead of heating the whole bed, so the desorbed vapor drives a concentrated pressure wave and the heat input per unit of refrigerant falls.

**Why it does not exist yet.** Adsorbent beds are poor heat exchangers, so conventional cycles need long cycle times and lose most of the input heat to the bed's thermal mass. Direct Joule heating solves the heat-transfer problem but creates hot spots that degrade the adsorbent where the local temperature overshoots. The thermal wave also steepens into a shock and then stalls.

**First test.** A 100 g silica-gel-on-nickel-foam bed, 80 °C regeneration, 15 °C evaporator; measure specific cooling power and COP, and the axial temperature profile at 0.1 Hz. Viability: above 300 W/kg and COP above 0.6, with the wave front sharper than 2 cm.

**Nearest known capability.** Silica-gel/water adsorption chillers with hot-water regeneration.

### I008 — Hydrate Cold Store with Memory-Effect Seeding

**What it does.** Stores cold at −15 to −25 °C as a pumpable slurry of clathrate hydrate, for load-shifting in refrigeration.

**Principle.** Clathrate hydrates: water and a guest molecule form an ice-like crystal whose dissociation temperature can be tuned by the guest, and whose latent heat is a large fraction of ice's. The memory effect — a melt that has not been heated far above the dissociation temperature re-forms hydrate much faster on the next cooling cycle — is used to avoid the long induction time that otherwise makes the slurry impractical.

**Why it does not exist yet.** The memory effect decays with time and with any excursion above the memory-loss temperature. Without it, induction can take hours at the subcooling available, so the store cannot follow a diurnal cycle. The slurry is also non-Newtonian and the hydrate particles agglomerate, blocking the heat exchanger.

**First test.** A 1 L stirred vessel, 20 wt% tetrabutylammonium bromide, cycled between 12 °C and −5 °C; measure induction time over 20 cycles with and without a controlled memory-keeping warm-up, and the slurry viscosity at 20 vol% hydrate. Viability: induction under 10 min on every cycle and viscosity below 1 Pa·s at 500 s⁻¹.

**Nearest known capability.** Ice-slurry storage and chilled-water tanks.

### I009 — Continuous Deracemization Crystallizer

**What it does.** Converts a racemic suspension into enantiopure crystals in continuous flow, without chiral additives.

**Principle.** Viedma ripening: grinding a suspension of a conglomerate near saturation dissolves the smaller crystals and grows the larger ones, and because the two enantiomorphs interconvert in solution, the population drifts to a single handedness. In continuous operation a classifier returns large crystals to the mill and removes product, so deracemization is driven by size-dependent solubility, the Gibbs–Thomson effect, rather than by a seed.

**Why it does not exist yet.** The process works only for conglomerate-forming systems, a minority of racemates, and the grinding that drives it also generates fines that dissolve and re-nucleate the losing enantiomer, capping purity. Scale-up needs a mill that abrades without fracturing product crystals, and the residence-time distribution must be narrow or the steady state smears out.

**First test.** A 200 mL continuous stirred mill, 10 wt% racemic glutamic acid, 40 °C; measure the enantiomeric excess of the product stream and the crystal size distribution at a 5 h residence time. Viability: above 95% ee at steady state with over 80% yield and no fines accumulation.

**Nearest known capability.** Batch Viedma ripening and diastereomeric salt resolution.

### I010 — Staged Freeze Desalinator

**What it does.** Concentrates brine or produces potable water by directional freezing, rejecting salt at the ice front and draining the brine stage by stage.

**Principle.** During directional freezing the ice front rejects dissolved salt, but only if it advances slower than salt can diffuse away — the constitutional supercooling criterion. A stack of thin freezing stages, each with its own cold plate and brine drain, keeps the local growth rate below that threshold, so each stage makes ice at a fixed salinity and the brine is concentrated in steps instead of being trapped.

**Why it does not exist yet.** The allowable growth rate falls as brine concentration rises, so the last stages must be very slow, and the ice still occludes brine in the last few percent. The cold plate also accumulates an insulating ice layer, so heat removal decays; scraping it adds mechanical complexity and wears the plate.

**First test.** A five-stage cold-finger rig, 3.5 wt% NaCl feed, growth held at 1 mm/min per stage; measure product ice salinity and brine concentration per stage. Viability: below 0.05 wt% salt in melted ice from stages one to three, brine above 15 wt% in stage five, and under 200 kJ per kg of product water.

**Nearest known capability.** Multi-effect distillation and reverse osmosis.

### I011 — Kirkendall Hollow-Fiber Fabricator

**What it does.** Makes hollow, nanoporous sulfide or oxide fibers in one step by exploiting the unequal diffusion rates of a metal cation and a chalcogen or oxygen anion.

**Principle.** The nanoscale Kirkendall effect: when a metal fiber is sulfided, the cation diffuses outward faster than sulfur diffuses inward. The outward cation flux leaves a supersaturation of vacancies, which condense into voids; the reaction front moves outward and leaves a hollow, porous shell. Because the voids form behind the front, the shell stays continuous while the core empties.

**Why it does not exist yet.** The void distribution is uncontrolled: pores coarsen and the shell spalls once the void fraction passes a percolation threshold, and a native oxide at the interface blocks diffusion. Drawing uniform sub-micron metal fibers with a clean surface, and holding the reaction temperature so the cation flux stays just above the void-coalescence threshold, is the manufacturing bottleneck.

**First test.** Electrospun 300 nm copper fibers sulfided in H₂S at 150 °C; measure the hollow fraction and shell porosity by electron microscopy and nitrogen adsorption, and bend a single fiber. Viability: over 80% hollow with connected porosity above 30% and survival at 2% bending strain.

**Nearest known capability.** Template-assisted nanotubes and dealloyed nanoporous metals.

### I012 — Complexion-Activated Sintering

**What it does.** Densifies ceramics at 300–500 °C below their usual sintering temperature by stabilizing a disordered grain-boundary film.

**Principle.** Grain-boundary complexions: distinct interfacial phases with their own thickness and composition. Some switch a boundary from a low-mobility, faceted structure to a high-mobility, disordered film. A dopant that forms such a complexion at the boundary, at the monolayer level, raises grain-boundary diffusion without forming a bulk liquid phase, so densification proceeds by diffusion while the bulk is still rigid.

**Why it does not exist yet.** The complexion is metastable. It exists only in a narrow window of temperature and dopant excess, and can dewet into discrete particles on cooling or during a long hold. The same film that speeds densification also promotes grain growth and can embrittle the boundary, so the process window is narrow and sensitive to oxygen partial pressure.

**First test.** Cold-pressed alumina with 0.3 mol% rare-earth dopant, sintered at 1200 °C for 2 h; measure density, grain size and boundary-film thickness by electron microscopy. Viability: above 97% density with grain size below 300 nm and a continuous 1–2 nm film that survives cooling.

**Nearest known capability.** Liquid-phase and two-step sintering.

### I013 — Bipolar-Electrode Janus Particle Reactor

**What it does.** Makes anisotropic Janus particles continuously, with a metal cap on one side, by passing them through a uniform electric field instead of masking or templating.

**Principle.** Bipolar electrochemistry: a conducting particle in an electrolyte under a uniform field polarizes, so one pole sits above the oxidation potential and the other below the reduction potential; deposition or etching then occurs at the poles with no wire connection. A fluidized bed in a vertical field, with a transverse flow, exposes each particle to a field that depends on its height, so cap size is set by residence time.

**Why it does not exist yet.** The induced potential across a particle scales with its size and the field, so small particles need large fields, which drive electrolysis and bubble formation at the feeder electrodes and disrupt the bed. The cap thickness also varies with particle position and orientation, giving a broad distribution.

**First test.** A 5 cm fluidized bed of 100 µm polystyrene beads seeded with 50 nm gold, in acetonitrile with 10 mM HAuCl₄, 1 kV/cm; measure the capped fraction and cap-area distribution by electron microscopy. Viability: over 60% capped with a coefficient of variation below 30%.

**Nearest known capability.** Seeded growth on Pickering emulsions and lithographic masking.

### I014 — Knudsen Pump Cascade

**What it does.** Pumps gas without moving parts using only a temperature gradient across a porous membrane, stacking stages to reach a useful pressure ratio at low pressure.

**Principle.** Thermal transpiration: in a pore comparable to the gas mean free path, molecules leaving the hot side carry more momentum than those leaving the cold side, so a pressure difference builds up that drives flow from cold to hot — the Knudsen pump. The pressure ratio per stage is roughly the square root of the temperature ratio, so a cascade multiplies the ratio.

**Why it does not exist yet.** The effect vanishes above the Knudsen regime, so a cascade must exhaust into a region that is already partly evacuated. Stages leak back through their supports and through the membrane's larger pores, and the flow is set by the very small conductance of the pores, so the pump is useful only for micro-flows.

**First test.** A 25-stage cascade of 50 nm anodic alumina membranes, 40 K across each, inlet at 1 kPa; measure outlet pressure and leak rate with the heaters off. Viability: outlet/inlet ratio above 2 at 1 kPa with leak below 10% of pumped flow.

**Nearest known capability.** Turbomolecular and diaphragm pumps for micro-gas analysis.

### I015 — Charge-Clamped Pyroelectric Electron Source

**What it does.** Produces a reproducible pulsed electron beam, and soft X-rays from a target, from a heated pyroelectric crystal with no high-voltage supply.

**Principle.** The pyroelectric effect: a polar crystal such as lithium tantalate develops a large surface charge when its temperature changes, because its spontaneous polarization changes with temperature. In vacuum the charge can raise the surface potential to tens of kilovolts, enough to field-emit electrons or ionize residual gas. Cycling the crystal therefore produces a self-limited electron pulse whose dose is set by the temperature swing, and a biased grid clamps the surface potential so the dose is set by the grid rather than by the crystal.

**Why it does not exist yet.** The crystal fractures under repeated thermal cycling, and the surface charge is screened by surface conduction or by the emitted charge itself, so the potential collapses after a short pulse. Emission comes from edges and defects rather than uniformly, so the beam is hard to focus and the dose is not reproducible.

**First test.** A 1 mm lithium tantalate crystal, 20 K swing at 0.1 Hz, 10⁻⁴ Pa; measure charge per cycle with a Faraday cup and the beam spot on a phosphor. Viability: above 10 nC per cycle, reproducible to ±10% over 10⁴ cycles, with no fracture.

**Nearest known capability.** Pyroelectric X-ray generators and field-emission cathodes.

### I016 — Parahydrogen Hyperpolarizer

**What it does.** Boosts NMR signals of a target molecule by orders of magnitude in continuous flow, using parahydrogen and a reusable catalyst.

**Principle.** Parahydrogen-induced polarization: hydrogen enriched in the para spin isomer is added across a double bond by a hydrogenation catalyst. The two protons start in a singlet state, and if the product's protons become magnetically inequivalent the spin order converts into net nuclear polarization. Because the polarization is created chemically rather than by a magnet, it can be generated at low field and transferred to a heteronucleus such as ¹³C or ¹⁵N for a long-lived readout.

**Why it does not exist yet.** The catalyst must be removed completely before detection, since metal residues broaden the NMR line, and the hydrogenation must be fast compared with spin relaxation. Continuous operation needs a catalyst that survives thousands of turnovers and a membrane that separates it without losing the hyperpolarized product.

**First test.** A flow reactor with a water-soluble rhodium catalyst and 50% parahydrogen, hydrogenating propargyl alcohol at 1 mL/min; measure ¹³C polarization against a thermal reference. Viability: above 5% polarization with under 1 ppm rhodium in the product and catalyst lifetime above 10⁴ turnovers.

**Nearest known capability.** Dissolution dynamic nuclear polarization and spin-exchange optical pumping.

### I017 — Electrochemically Driven Isoelectric Precipitator

**What it does.** Precipitates a protein at its isoelectric point by generating the pH shift electrochemically, with no added acid or base and no salt byproduct.

**Principle.** Water splitting at a bipolar membrane produces H⁺ on one side and OH⁻ on the other; a buffer flowing past the membrane is titrated to a pH set by the current. Because pH is set by charge rather than by mixing, the precipitation front can be moved along a channel so that nucleation and growth are separated. Protein precipitates where the local pH equals its isoelectric point, and the precipitate is harvested before it re-dissolves.

**Why it does not exist yet.** The membrane fouls with precipitated protein, and the local pH at the membrane surface can overshoot far beyond the bulk value, denaturing the protein before it can be collected. The protein solution is a poor conductor, so most of the current is carried by the buffer and the attainable pH change is set by buffer capacity.

**First test.** A bipolar-membrane channel, 5 g/L bovine serum albumin, 10 mA/cm²; measure the pH profile with a fluorescent indicator and protein recovery in the precipitate. Viability: above 85% recovery within ±0.2 pH units of the isoelectric point with under 5% loss of native secondary structure by circular dichroism.

**Nearest known capability.** Acid/base precipitation and isoelectric focusing.

### I018 — Mechanochemical Nitride Mill

**What it does.** Synthesizes refractory nitrides and other air-sensitive compounds at room temperature in a sealed ball mill, using the fresh metal surface created by fracture rather than bulk heating.

**Principle.** Mechanochemistry: repeated impact fractures the metal particles and exposes clean surface that reacts with the gas or with a solid nitrogen source. The reaction enthalpy is carried away by the mill body, so the product is nanocrystalline and often metastable. Because the reaction is fed by fresh surface, the rate is set by collision energy and gas pressure, not by the bulk diffusion that limits conventional solid-state synthesis.

**Why it does not exist yet.** The product contaminates with milling media, and the reaction is exothermic enough that the vial can overheat and sinter the product; as gas is consumed the pressure drops and the reaction stalls. Scaling from grams to kilograms changes collision energy per unit mass and heat removal, so a laboratory result does not transfer.

**First test.** A planetary mill with 10 g aluminium in 5 bar nitrogen and tungsten carbide media; measure conversion to AlN by X-ray diffraction and contamination as a function of milling time. Viability: above 95% conversion in 4 h with under 1 wt% contamination and crystallite size below 50 nm.

**Nearest known capability.** Carbothermal nitridation and self-propagating high-temperature synthesis.

### I019 — Acoustoelastic Stress Imager

**What it does.** Maps the residual stress field inside an opaque metal part by measuring the stress-induced change in shear-wave speed with a scanned array, without strain gauges.

**Principle.** Acoustoelasticity: the speed of an elastic wave depends on stress through the third-order elastic constants. For shear waves, the birefringence between two polarizations is proportional to the difference of principal stresses. A phased array measures time-of-flight differences with nanosecond precision, and a model of the part's geometry inverts the travel times into a stress map.

**Why it does not exist yet.** The acoustoelastic constants are small: a 100 MPa stress changes wave speed by a fraction of a percent, comparable to the effects of texture, temperature and transducer coupling. Separating stress from texture needs a calibration on the same material or measurements along several propagation directions, which in a complex part is often impossible.

**First test.** A welded steel plate with a residual-stress field computed by finite element, a 5 MHz shear-wave array; compare the birefringence map with surface X-ray diffraction. Viability: agreement within 30 MPa over 100 × 100 mm with 10 MPa repeatability.

**Nearest known capability.** X-ray diffraction, which sees only the surface, and hole-drilling, which is destructive.

### I020 — Oscillating-Reaction Peristaltic Pump

**What it does.** Pumps liquid through a tube with no power supply and no valves, using a chemically oscillating gel that contracts and relaxes in a traveling wave.

**Principle.** The Belousov–Zhabotinsky reaction in a gel: the catalyst is attached to the polymer, so the reaction oscillates the polymer's oxidation state and, with it, the gel's volume. A row of gel segments with a phase lag set by their distance from a trigger produces a traveling contraction wave that squeezes the tube, as a peristaltic pump does.

**Why it does not exist yet.** The reaction consumes its bromate and malonic acid fuel and produces bromide, which eventually quenches the oscillation; the gel fatigues, and wave speed depends strongly on temperature. The reaction needs an acidic aqueous environment, so the tube must resist bromine and acid, and the carbon dioxide it produces forms bubbles that break the wave.

**First test.** A 1 mm silicone tube lined with five ferroin-gel segments fed with BZ solution; measure flow rate and inter-segment phase lag by video. Viability: stable flow above 10 µL/min for 8 h with a phase lag of 0.2–0.4 period.

**Nearest known capability.** Motor-driven peristaltic and osmotic pumps.

### I021 — Metal-Hydride Hydrogen Compressor

**What it does.** Compresses hydrogen from a low-pressure source to a higher pressure using heat instead of a mechanical piston, with no moving parts in the gas path.

**Principle.** Metal hydrides: alloys such as LaNi₅ or TiFe absorb hydrogen exothermically at low pressure and release it at high pressure when heated, because the plateau pressure depends exponentially on temperature through the van 't Hoff relation. A pair of beds alternating between cold absorption at low pressure and hot desorption at high pressure therefore acts as a thermally driven compressor.

**Why it does not exist yet.** The compression ratio per bed is limited by plateau slope and hysteresis, so one stage reaches only about an order of magnitude; multistage operation needs interstage cooling and a heat-management network. The alloy also pulverizes after repeated cycling, and the powder packs into a low-conductivity bed, so absorption and desorption become slow.

**First test.** A 200 g LaNi₅ bed, absorption at 3 bar and 20 °C, desorption at 90 °C; measure delivered pressure, cycle time and capacity over 100 cycles. Viability: delivery above 20 bar with a cycle under 10 min and under 10% capacity loss after 100 cycles.

**Nearest known capability.** Mechanical diaphragm and electrochemical compressors.

### I022 — Triboelectric Powder Separator

**What it does.** Separates a mixed plastic or mineral powder into its components by charging the particles against each other and deflecting them in an electric field, without liquids.

**Principle.** Triboelectricification: when two dissimilar insulators are rubbed, charge transfers in the direction of their positions in the triboelectric series. In a fluidized bed, particles of different polymers charge with opposite polarity, and a transverse field deflects them into separate bins. Charge is proportional to contact area and work-function difference, so residence time and low humidity maximize separation.

**Why it does not exist yet.** Charge decays with humidity and surface contamination, and fine particles are dominated by van der Waals adhesion, so they stick to the walls instead of following the field. The same charging makes the bed clump, and the field needed for small particles approaches the breakdown limit of air.

**First test.** A fluidized bed of a 50/50 mixture of 1 mm polypropylene and polyethylene terephthalate, 20 kV/m transverse field, dry air below 10% relative humidity; measure outlet purities. Viability: above 90% purity at 1 kg/h with under 5% wall loss.

**Nearest known capability.** Froth flotation and density separation.

### I023 — Sonochemical Leach Reactor

**What it does.** Extracts metals from ores or concentrates at lower temperature and acid strength by using cavitation to refresh the mineral surface.

**Principle.** Acoustic cavitation: bubbles collapsing asymmetrically near a solid surface drive a microjet that pits the surface, removes the passivating layer and exposes fresh mineral; the collapse also produces local hot spots and radicals that can oxidize sulfide minerals. Because the passivating layer is removed continuously, the leaching rate stays at the surface-reaction limit instead of decaying behind a growing product layer.

**Why it does not exist yet.** At the hydrostatic pressure needed to keep the acid from boiling, cavitation is suppressed, and at low pressure the reactor cannot be hot. The transducer erodes under acid at temperature, and sonochemical yield per unit volume falls on scale-up because the cavitation zone is confined to a thin layer near the horn.

**First test.** A 1 L flow-through reactor with a 20 kHz horn, chalcopyrite concentrate in 1 M sulfuric acid with 0.1 M ferric ion at 60 °C; compare copper extraction rate with a stirred control. Viability: three times the rate with horn erosion below 0.1 mm per 100 h.

**Nearest known capability.** Pressure oxidation and bioleaching.

### I024 — Electrostatic Coalescer for Water-in-Oil

**What it does.** Separates micron-sized water droplets from oil in a compact electric field, using dipole coalescence rather than gravity settling.

**Principle.** Electrocoalescence: a conducting water droplet in a non-conducting oil acquires an induced dipole in an alternating field. Neighbouring droplets attract, chain along the field lines, and coalesce when the field is switched off or reversed. Because the attraction scales with the square of the field and with droplet radius, a moderate field brings micron droplets together far faster than Stokes settling, and the coalesced droplets settle in a small gravity section.

**Why it does not exist yet.** At high water cuts the droplets form continuous chains that bridge the electrodes and short the cell; at low water cuts they are too far apart and coalescence is limited by collision frequency. The field also causes partial discharges and emulsifier fouling on the electrodes, and the AC frequency must avoid the relaxation frequency of the oil.

**First test.** A 1 L cell with 1 vol% water in a model oil, 2 kV/cm at 1 kHz; measure median droplet size and separation time. Viability: median above 100 µm after 60 s and residual water below 0.5 vol% after 10 min, with power below 1 kWh/m³.

**Nearest known capability.** Gravity separators and hydrocyclones.

### I025 — Fluidic-Diode Heat Pipe

**What it does.** Conducts heat in one direction only, using a passive fluidic diode in the liquid-return path of a two-phase capillary loop.

**Principle.** A heat pipe with a Tesla-type fluidic diode in its liquid return line. In the forward orientation, condensate returns to the evaporator through the low-resistance direction of the diode, and the device transports heat by latent exchange. In the reverse orientation, the diode presents a high hydraulic resistance, the evaporator is starved of liquid, and the device degrades to a plain conduction bar.

**Why it does not exist yet.** The diode's reverse resistance is finite, so some liquid still leaks back, and the wick's capillary pressure must exceed the diode's forward pressure drop. At high heat flux the liquid return is limited by wick permeability, and the diode adds volume and a nucleation site for vapor lock. Parasitic conduction through the wall also sets a floor on the reverse conductance.

**First test.** A 20 cm copper–water heat pipe with a Tesla diode in the return line, 50 W; measure effective conductance in both orientations and at several tilt angles. Viability: forward/reverse ratio above 20 with forward conductance above 2 W/K.

**Nearest known capability.** Heat pipes and mechanical thermal switches.