Special Wet Etching Formulation Electronic/EL Grade is a submicron-filtered buffered fluoride mixture for wet chemical patterning of dielectric, transparent conductive oxide, and metallization layers in semiconductor and electroluminescent device production. Available buffer ratios are designated by the ammonium fluoride to hydrofluoric acid mass ratio of 5:1, 7:1, 10:1, and 20:1. The product is filled into double-walled fluoropolymer containers under ISO 14644-1:2015 class 5 conditions and is released only after inductively coupled plasma mass spectrometry for metal cations, ion chromatography for chloride and sulfate, and optical particle counting at 0.2 µm. The free HF content is controlled to ±0.2 wt% of the specified ratio, and the ammonium fluoride content is controlled to ±0.5 wt%. Wetted surfaces must be PFA, PTFE, or PVDF. Borosilicate glass, quartz, and 316L stainless steel are incompatible because free HF attacks the oxide network and releases contaminants into the bath.

The Electronic/EL grade designation is a contamination specification rather than a single composition. The same cation and particle ceilings apply across the four buffer ratios, while the density and viscosity shift with ammonium fluoride mass fraction. This allows the end user to select the ratio for the target dielectric etch rate without re-qualifying the trace-metal contamination budget of the wet bench. For gate-oxide stripping and via-opening steps, the 7:1 ratio is the most common reference; for field-oxide removal where etch rate is less critical, the 20:1 ratio is sometimes specified to reduce free HF vapour pressure in the cleanroom.

What limits lot-to-lot ratio control in electronic-grade fluoride etchants?

Ratio control in buffered fluoride etchants is limited by evaporation of free HF through the container headspace and by water absorption during use. A vented container stored at 30 °C can lose 0.3 wt% free HF over 90 days, shifting the thermal oxide etch rate by more than 10%. Field data from a 200 L recirculation bath using the 7:1 ratio show that batch-to-batch free-HF variation remains within 0.2 wt% over 12 consecutive lots when the bath is covered with a tight-fitting PFA lid and filtered at 0.05 µm. A general-purpose 7:1 mixture stored in a single-wall HDPE drum under the same conditions showed 1.8 wt% free-HF loss over the same interval. The specification limit is therefore set as a production requirement for maintaining oxide etch rate within ±5% of target on a 40 L immersion tool.

At 25 °C, the 7:1 ratio etches thermally grown SiO₂ at 90–110 nm/min when measured by spectral reflectometry on 100 mm monitor wafers. Buffered fluoride etchants of similar stoichiometry show an apparent activation energy in the range 11–15 kcal/mol over 20–35 °C. A temperature error of ±1 °C therefore alters the oxide etch rate by approximately 8%, which tightens the usable processing window to 25.0±0.5 °C for thickness-critical gate-oxide removal. A closed-loop PFA heat exchanger with proportional-integral-derivative control is required; unjacketed immersion baths show vertical thermal gradients of 4–6 °C in a 40 L tank and generate an oxide etch rate spread of 32% or more.

Phosphorus-doped silicate glass is etched deliberately faster in buffered fluoride chemistry than thermally grown silicon dioxide. The 7:1 ratio removes 6 wt% phosphorus-doped silicate glass at approximately 200–350 nm/min at 25 °C, while the underlying thermal SiO₂ loss remains near 100 nm/min. This selectivity is used in via-opening and passivation overetch steps in electroluminescent display panels. Silicon nitride is nearly inert; typical removal is below 5 nm/min at 25 °C, making the nitride layer a practical etch stop when the formulation is used as a post-via clean.

Ion chromatographic and particle retention benchmarks

Lot-release testing is performed at the point of filling. The specification below is for the 7:1 ratio at 20 °C; other ratios are normalised to the same contamination ceilings.

How the Electronic/EL Grade differs from general-purpose buffered etchants

The table below compares typical lot-release data for the Electronic/EL Grade and a general technical-grade 7:1 buffered fluoride mixture with the same nominal free HF. The difference is not total fluoride activity but the contamination profile that controls device stability.

Soluble sodium, potassium, and calcium are not removed by point-of-use filtration; they remain dissolved and become incorporated into growing dielectric films, causing flat-band voltage instability. The Electronic/EL Grade is therefore released against lot-specific analytical values rather than a “low-particle” description. Users should retain the lot certificate for every drum and re-validate cation levels after transfer into the chemical distribution loop, because extraction from dry PFA fittings can add 10–50 µg/kg of iron if the loop has been exposed to ambient air for more than 72 h.

In a representative immersion etch process, the 7:1 product is heated to 25.0±0.5 °C in a 50 L PFA vessel and recirculated at 20 L/min through a 0.05 µm polytetrafluoroethylene filter. A 100 mm thermally oxidised monitor wafer is processed for 120 s; spectral reflectometry confirms an etch depth of 112–132 nm, consistent with the 90–110 nm/min range after native-oxide breakthrough. The formulation shows an induction period of 5–10 s on native oxide, after which the etch becomes linear. A pre-rinse with 0.5 wt% aqueous HF at 25 °C for 15 s removes the induction period and improves within-lot thickness uniformity to ±3%.

Compared with unbuffered hydrofluoric acid at the same free HF concentration, the ammonium fluoride buffer reduces the free HF vapour pressure and stabilises the etch rate against moisture uptake. This is an operational advantage in open recirculation tanks, but it also means the product cannot be used as a direct substitute for 0.5 wt% HF pre-rinse solutions; the buffered chemistry has different selectivity between thermal SiO₂ and phosphorus-doped silicate glass. End users should not adjust the product by adding solid ammonium bifluoride without quantitative mixing control, because local precipitation can form and block the 0.05 µm filter train.

The product is also used for pre-diffusion cleaning of silicon surfaces after chemical mechanical polishing. Immersion at 25 °C for 60 s removes native oxide and residual slurry particles without increasing surface roughness as measured by atomic force microscopy. The low metal cation ceiling prevents mobile ion contamination in the subsequently grown gate oxide. The rinse line downstream of the etch tank should be filtered at 0.1 µm and fed with electronic-grade water meeting ASTM D5127-13 type E-1 limits.

When the formulation is blended with hydrochloric acid for indium tin oxide patterning

In electroluminescent display manufacturing, indium tin oxide films of 150–200 nm thickness are commonly patterned with mixtures of hydrochloric acid and nitric acid. The product can be blended at point of use with 37% electronic-grade hydrochloric acid to control fluoride activity. A representative 1:3 product:hydrochloric acid blend by volume at 40 °C etches ITO at approximately 30–50 nm/min while keeping underlying silicon nitride loss below 2 nm/min. The selectivity window is narrow; an increase in free HF above 0.8 wt% induces measurable attack on the glass substrate at the ITO edge. The bath is circulated through a 0.05 µm PTFE filter and temperature-controlled with a PFA heat exchanger. Borosilicate glass flowmeters are prohibited because the blend etches borosilicate glass at rates above 10 nm/min and releases sodium into the bath.

For aluminium/molybdenum metallisation stacks used in thin-film transistor source-drain electrodes, the product is not a primary metal etch. The aluminium/molybdenum stack is patterned with phosphoric acid-based isotropic etchants; the Electronic/EL Grade is then used for post-etch residue removal and native-oxide stripping before subsequent sputter or passivation steps. Immersion at 25 °C for 60 s has been observed on a 200 mm TFT line to reduce contact resistance at via interfaces by 12–18% relative to a deionised-water rinse only, as measured by Kelvin structures. Published data for this specific residue layer is limited, so the process window must be re-qualified for each metallisation composition and preceding dry-etch chemistry.

Operational boundaries are set by fluoride volatility and wetted-material compatibility

The product must be handled in exhausted wet decks with PTFE-lined ducting. The maximum recommended bath temperature is 35 °C for the 7:1 ratio; above this, free HF vapour pressure increases and open-tank composition drift exceeds the reference tolerance within 8 h. The formulation must not be mixed with concentrated sulfuric acid, alkaline developers, or ammonia-containing strippers in the same drain without neutralisation; exothermic mixing liberates hydrogen fluoride vapour and can exceed the capacity of a standard point-of-use scrubber. Skin contact requires immediate application of 2.5% calcium gluconate gel. Splash goggles and a face shield are mandatory. The wetted materials are limited to PFA, PTFE, PVDF, and specially conditioned polyethylene; borosilicate glass, quartz, titanium, and stainless steel are incompatible.

The product is supplied in 20 L fluoropolymer drums and 200 L fluoropolymer-lined overpacks. Each container is pressure-tested at 50 kPa and fitted with a Viton-free closure to avoid amine and metal ion extraction. The shelf life is 24 months from fill date when stored at 15–25 °C in the original container. After opening, the container should be purged with filtered nitrogen for 30 s and resealed; repeated opening under humid cleanroom air increases water uptake and lowers the free HF titre. A desiccated vent filter with a 0.05 µm PTFE membrane is recommended when the container is connected to a chemical distribution loop.