How Does Aqueous Potassium Hydroxide Etch Selectivity Dictate Bulk Micromachining Geometry?

In silicon micromachining, aqueous potassium hydroxide solutions are employed as anisotropic etchants because the etch rate on the {111} crystal plane is approximately two orders of magnitude lower than on the {100} and {110} planes. Electronic/EL grade KOH is blended with Type E-1 deionized water produced under ASTM D1193-06 to form a working bath with a KOH concentration of 20–30 wt%; isopropyl alcohol is optionally added at 10–20 vol% to reduce hydrogen bubble adhesion and suppress pyramidal hillock formation on {100} surfaces. The formulation addition ratio is therefore 200–300 g of KOH solids per 1 L of final solution for a 20–30 wt% bath, with IPA introduced after full dissolution and cooling to below 40 °C to limit vapor-phase alcohol losses. The solution is recirculated through 0.1 µm polypropylene or PTFE membrane cartridges before entering the etch tank. Compliance for this wet-chemical step is anchored to ISO 14644-1:2015 Class 5 for the cleanroom environment, ASTM D1193-06 Type E-1 for rinse water, and REACH (EC) No 1907/2006 for the registration, handling, and disposal of corrosive potassium hydroxide. Trace metal specifications for electronic/EL grade KOH are verified by inductively coupled plasma mass spectrometry according to ISO 17294-2:2016, with attention to sodium, iron, chromium, and nickel because these impurities remain on exposed dielectric films after drying and shift flatband voltage in MOS structures.

On a production-scale wet bench constructed from polypropylene or PVDF, the etch bath is held at 80 °C ± 0.5 °C using indirect quartz or PTFE-lined heating. At this set point, the {100} silicon etch rate is typically 1.0–1.4 µm/min, while the {111} etch rate remains below 0.02 µm/min, giving an anisotropy ratio sufficient for cavity sidewalls to follow the {111} planes. Silicon nitride or thermally grown silicon dioxide films serve as hard masks; the selectivity of the KOH etch to silicon over thermally grown SiO₂ is approximately 200:1 under these conditions, although the silicon dioxide mask is consumed at 3–5 nm/min, which requires an initial mask thickness of at least 500 nm for a 200 µm through-wafer feature. Exothermic dissolution of solid KOH pellets is performed in a chilled make-up vessel with continuous agitation because localized boiling at the pellet surface alters local concentration and can generate caustic aerosols. The working bath is monitored by titration of total alkalinity, density measurement, and intermittent ICP-MS for metal accumulation. The bath is replaced or reconditioned when dissolved silicon concentration exceeds 20 g/L or when surface roughness Ra on test wafers rises beyond the process control limit. A limitation observed on recirculating lines is carbonate absorption at the air-liquid interface, which forms potassium carbonate and causes microloading, non-uniform etch depth across a 200 mm wafer cassette, and an increase in Ra. Nitrogen blanketing of the tank and closed-circuit filtration reduce carbonate ingress, but titrimetric verification of hydroxide and carbonate fractions remains necessary every 8 h of continuous operation.

On manufacturing lines, the dominant failure mode is non-uniform etch depth due to temperature stratification in the tank. Because the etch rate increases with temperature at approximately 0.1–0.2 µm/min per °C in the 70–90 °C window, a vertical gradient of even 1 °C across a cassette produces measurable cavity depth differences. Recirculating wet benches therefore use ceramic or quartz immersion heaters with ±0.5 °C PID control and upward flow distribution through the wafer cassette. Pump selection is also constrained by the solution density of approximately 1.25–1.35 g/cm³ and by the erosive potential of dissolved silicate species; magnetic-drive polypropylene pumps with PTFE bushings are preferred over mechanical seals. In addition, the accumulation of dissolved silicon raises viscosity and can deposit potassium silicate on heater surfaces, reducing heat transfer efficiency and requiring periodic acid cleaning with hydrofluoric acid or citric acid solutions. Process control includes inductively coupled plasma optical emission spectroscopy or ICP-MS for silicon and trace metals, along with refractometric KOH concentration measurement. KOH etching is restricted to pre-metal modules because potassium ions are highly mobile in silicon dioxide and shift threshold voltage in MOS transistors; post-metal wafers therefore use alternative etch chemistries. End products from this step include bulk-micromachined pressure sensor diaphragms, accelerometer proof masses, inkjet printhead nozzle plates, microfluidic mixing chambers, and gyroscope resonators produced on 150 mm and 200 mm silicon wafers.

In monocrystalline photovoltaic cell processing, saw-damage removal and random pyramid texturing are carried out in a heated alkaline bath in which electronic/EL grade KOH is maintained at 1.5–3.0 wt% in deionized water. A proprietary texturing aid is dosed at 0.5–2.0 vol% and the bath temperature is controlled at 75–85 °C, depending on wafer resistivity and the target pyramid size of 2–5 µm. The formulation addition ratio is therefore approximately 15–30 g KOH solids per litre of final bath, with the additive volume adjusted to maintain a consistent H₂ bubble detachment pattern on the wafer surface. This wet-chemical process is governed by downstream module certification under IEC 61215-1:2021 and by cleanroom and chemical purity controls under ISO 14644-1:2015; incoming electronic/EL grade KOH is specified for trace iron, sodium, chromium, and nickel by ICP-MS according to ISO 17294-2:2016 because residual metals reduce bulk minority carrier lifetime and increase surface recombination velocity. In addition, the European Union Restriction of Hazardous Substances Directive 2011/65/EU applies to the finished photovoltaic modules and therefore to the selection of auxiliary process chemicals.

The production sequence on an in-line texturing tool consists of a pre-clean step using megasonic deionized water, transfer of the wafer cassette into the KOH bath, and a timed etch of 20–40 min to remove 10–20 µm of damaged silicon from the wire-sawn surface and to form random upright pyramids. Bath alkalinity is maintained by conductivity and density feedback rather than fixed-volume addition, because evaporation losses concentrate the solution unevenly across a cassette of 400–800 wafers per batch and change the etch rate by up to 10 % over the campaign. Uniformity is affected by dissolved silicon accumulation, carbonate absorption, and additive depletion; therefore, online measurement of refractive index or density is used to trigger replenishment of KOH and additive. The main process conflict is the balance between pyramid size and additive depletion. Because the texturing additive is consumed by electrochemical reduction on the silicon surface and by drag-out, the additive concentration falls during a cassette campaign. Without replenishment, the pyramid size increases from the target 2–5 µm to over 8 µm, which reduces light trapping and increases rear surface recombination. Modern in-line tools use automatic dosing pumps with flow-based or density-based replenishment to maintain bath composition within ±0.2 wt% KOH and ±0.1 vol% additive. Bath lifetime is constrained by accumulated dissolved silicon and organic degradation products; typical bath exchange intervals range from 24 h to 72 h depending on throughput and overflow rinse recovery.

End product types from this alkaline texturing step include monocrystalline p-type PERC cells, n-type TOPCon cells, and heterojunction cells that use the same random pyramid morphology before amorphous silicon deposition. The textured wafers are subsequently diffused, coated, metallized, and laminated into IEC 61215-1:2021-certified modules for utility-scale, commercial rooftop, and residential photovoltaic systems.

Alkaline Electrolyte Blending for Zinc–Manganese Dioxide and Nickel–Metal Hydride Cells

Primary alkaline Zn–MnO₂ cells and sealed nickel–metal hydride cells use potassium hydroxide electrolyte solutions with a concentration of 30–45 wt% in deionized water. The formulation addition ratio in primary alkaline cells is typically 300–450 g KOH per litre of final electrolyte, followed by dissolution of zinc oxide at 1.5–3.0 wt% to suppress hydrogen gassing at the zinc anode and to lower the equilibrium potential. Gelled anode formulations additionally incorporate 1.5–2.5 wt% sodium carboxymethyl cellulose or crosslinked polyacrylic acid to suspend zinc particles and maintain electrical continuity through discharge. Electronic/EL grade KOH is required because chloride, iron, and copper contamination at concentrations as low as 1 ppm accelerates self-discharge, increases gas evolution inside sealed cells, and can compromise the pressure relief vent. Compliance for this application is governed by IEC 60086-1:2021 for primary battery dimensions, electrical performance, and safety, ANSI C18.1M for portable primary cells, and Regulation (EU) 2023/1542 concerning batteries and waste batteries for restrictions on mercury, cadmium, and lead. The electrolyte preparation area is typically maintained under ISO 14644-1:2015 Class 8 or better, depending on cell sensitivity.

The production process starts with chilled dilution in a jacketed stainless-steel or polypropylene mixing vessel because the dissolution of solid KOH is strongly exothermic and can increase local temperature above 90 °C if uncontrolled. The solution is cooled to 20–25 °C, vacuum-degassed below –0.08 MPa gauge to remove dissolved oxygen, and filtered through 0.5 µm polypropylene cartridges before injection into the anode cavity or slurry mixing vessel. In cylindrical cell manufacturing, the gelled anode slurry is metered into a copper alloy current collector, a separator soaked with electrolyte is inserted, and the cell is crimped with a nickel-plated steel can. In NiMH cell production, the potassium hydroxide electrolyte is typically 30–35 wt% with added LiOH at 1–2 wt% to improve high-temperature charging efficiency, and the filling is performed under controlled humidity to prevent carbonate pick-up. Process equipment must be compatible with strong alkali; EPDM, PTFE, and polypropylene are acceptable, while aluminum and galvanized steel components are excluded because they corrode rapidly. An operational boundary is the increase in electrolyte viscosity below 10 °C, which impairs dosing accuracy; storage and dosing lines are therefore maintained at 20 °C ± 2 °C with nitrogen blanketing to prevent potassium carbonate precipitation. Failure modes on manufacturing lines include crystallisation in dosing lines when the concentration exceeds 45 wt% and local cooling occurs, as well as zinc slurry gassing when the ZnO content falls below the solubility threshold for the working temperature.

End product types include cylindrical primary alkaline cells in LR6 and LR03 formats, prismatic alkaline cells for industrial instruments, and sealed NiMH cells for cordless power tools, emergency lighting, and hybrid vehicle battery packs. In all sealed systems, the KOH electrolyte remains immobilized within the separator and anode gel, and the fill volume is calibrated to the available void volume to prevent leakage at end of discharge.

Molten potassium hydroxide baths maintained at 350–450 °C are used to reveal threading dislocations and polarity domains on (0001) GaN epitaxial layers because the etch rate is strongly polarity-selective: the nitrogen-polar surface etches rapidly while the gallium-polar surface remains largely intact during short immersion times of 5–30 min. For this destructive metrology step, the bath consists of electronic/EL grade KOH pellets loaded at 100 wt% into a nickel, quartz, or platinum-lined crucible; for aqueous defect etching used on some production monitor lines, KOH is dissolved to 1–10 M in Type E-1 deionized water and heated to 80–120 °C under ultraviolet illumination. The formulation addition ratio is therefore either an anhydrous molten bath or an aqueous solution prepared by dissolving 56–560 g KOH per litre of final solution. Compliance for this metrology step is anchored to ISO 17294-2:2016 for trace metal verification, ISO 14644-1:2015 Class 5 for wafer handling in inspection areas, and REACH (EC) No 1907/2006 for high-temperature corrosive bath operation and waste neutralization. Electronic/EL grade purity is critical because residual transition metals and alkaline earth elements in technical-grade potash form interfacial reaction products that obscure etch pit boundaries and reduce the repeatability of defect counting.

The production process is destructive and is performed on monitor wafers pulled from the epitaxial growth lot, not on device wafers after front-end processing. Wafers are immersed in the molten KOH bath with a temperature uniformity of at least ±2 °C across the crucible, then quenched in deionized water and rinsed in an ultrasonic bath to remove potassium residues. Inspection is performed with Nomarski differential interference contrast microscopy or scanning electron microscopy; etch pit density is counted over a 100 µm × 100 µm grid and classified against acceptance limits that range from 10⁶ cm⁻² to 10⁹ cm⁻² depending on device architecture. The procedure also distinguishes gallium-polar from nitrogen-polar material on a patterned wafer surface, which is relevant for GaN-on-silicon power transistor processes where inverted polarity islands lead to increased leakage current. A known limitation is that the molten bath ages through evaporative losses of potassium and absorption of atmospheric water, so the etch temperature and bath replacement schedule must be calibrated against a reference wafer with a known dislocation density. Published data for this specific molten KOH configuration is limited to wafer-level metrology studies rather than high-volume production standards, so acceptance thresholds are generally derived by the epitaxial facility rather than imposed by an external standard.

End product types from this inspection node include GaN-on-sapphire light-emitting diode epiwafers, GaN-on-SiC RF high-electron-mobility transistor wafers for 5 GHz and higher communication bands, and GaN-on-silicon power switching wafers for fast-charging adapters and data-centre power supplies. The KOH etch does not become part of the final device; it is used only as a quality-gate metrology tool before further front-end processing.

Tungsten Contact Planarization Requires Tight Alkali Dosing Windows

Chemical mechanical planarization slurries for tungsten damascene interconnects use electronic/EL grade KOH as a pH adjuster to hold the slurry in the 10.5–11.5 range, where tungsten oxidation and colloidal silica abrasion rates are balanced. The formulation addition ratio of KOH in the final slurry is typically 0.05–0.3 wt%, depending on the colloidal silica loading of 2–10 wt%, the oxidizer system, and the incoming slurry pH. The addition is made as a dilute KOH solution to avoid local pH spikes that can destabilize the colloidal silica dispersion. Compliance for this application is anchored to ISO 14644-1:2015 Class 4 for slurry preparation areas, ASTM E70 for glass electrode pH measurement of aqueous solutions, and ISO 17294-2:2016 for trace metal verification. The finished slurry is also required to meet the wafer fabrication facility's particle and metallic contamination specifications, which are typically aligned with ISO 14644-1:2015 cleanliness classes and with mass balance limits for alkali metals in front-end-of-line processing.

The production process begins with high-shear dispersion of colloidal silica in deionized water, followed by sequential addition of the tungsten oxidizer and dilute KOH under constant recirculation. The batch is then filtered through a three-stage cartridge train at 1.0 µm, 0.5 µm, and 0.2 µm to remove oversize particles and gel agglomerates before dispensing to the CMP tool. Polishing is performed on a multi-head rotary CMP tool with in situ pad conditioning; tungsten removal rates on patterned 300 mm wafers are typically monitored in the range 150–350 nm/min, while oxide erosion and tungsten dishing are measured by high-resolution profilometry after a 60 s post-polish buff step. The pH and conductivity of the slurry are logged continuously during polishing because carbon dioxide absorption from the cleanroom environment can lower the pH over time and reduce tungsten removal rate. Slurry pH drift during pad conditioning and wafer processing is a known production instability because the CMP tool contacts the slurry with carbon dioxide from the cleanroom atmosphere and with pad debris. In tungsten slurries formulated with KOH, a drop of 0.2 pH units can reduce tungsten removal rate by 10–20 %, while an increase above 11.5 leads to visible oxide erosion rings near wafer edges.

Some facilities therefore use pH-controlled day tanks with continuous nitrogen headspace and recirculation, and the KOH addition is trimmed by a dosing loop that reads a glass electrode in a side stream. The KOH concentration is not measured directly in the slurry; instead, total alkali is determined by conductometric titration and correlated to pH. Filterability of the slurry is sensitive to high local KOH concentration, which can dissolve silica particles and form soluble silicate; this is prevented by diluting the KOH to 1–5 wt% before adding to the slurry batch. An operational boundary is that KOH over-addition beyond pH 11.5 accelerates oxide erosion and causes tungsten plug recess, while under-dosing below pH 10.5 reduces tungsten removal rate and increases residue counts after cleaning. The compatibility of slurry distribution lines is also limited: EPDM and PTFE are suitable, but stainless-steel fittings in contact with high-pH slurry require passivation to avoid iron leaching.

End product types from this CMP step include DRAM wafers with tungsten contact plugs, 3D NAND wafers with tungsten word lines and vias, and logic wafers with tungsten local interconnects. The tungsten layer is planarized immediately after chemical vapour deposition, and the remaining tungsten plugs or lines are then capped with dielectric barrier films before subsequent metallization. In all cases, the KOH is present only as a slurry component and does not remain as a bulk residue on the finished wafer.