Product designation CMP-PRC-EL-100 is an electronic/EL grade aqueous formulation supplied for post-chemical mechanical planarization cleaning of semiconductor substrates. The liquid is a clear, low-viscosity acidic solution with nominal pH of 2.6–3.2 at 25 °C, density of 1.02–1.06 g/mL, and dynamic viscosity below 1.2 cP. The product is filtered through a 0.05 µm-rated membrane and packaged in fluoropolymer-lined containers under Class 4 cleanroom conditions per ISO 14644-1:2015. The formulation combines a carboxylic-acid chelating system with a non-amine pH modifier and is intended to remove ceria, silica, alumina, and tungsten residues from dielectric and metal films after oxide, tungsten, or barrier CMP. The product is not diluted with technical-grade water and does not contain added surfactants, amine corrosion inhibitors, or alkylphenol ethoxylates.

The component release is governed by electronic-grade raw material specifications derived from ASTM D5127-13(2020) Type E-1.2 and SEMI C63 guidelines. Lot-specific certificates of analysis report critical transition metal concentrations, anion burden, particle counts, and total organic carbon. The absence of amine-based additives is deliberate: post-CMP cleaning solutions containing primary or secondary amines can generate copper-amine complexes, alter surface zeta potential, and leave nitrogen-containing residues that degrade interconnect reliability.

What Impurity Limits Are Applied to the Electronic/EL Grade Cleaner?

The electronic/EL grade specification imposes narrower trace impurity windows than technical-grade aqueous cleaners. Table 1 summarizes lot-release upper control limits and corresponding analytical techniques. Values are measured on the neat product after 0.05 µm filtration and are not to be interpreted as post-dilution bath values.

These limits align with the low-ionic-strength design required for single-wafer cleaning where carryover of chloride or sulfate above 50 ppb can increase copper corrosion and porous low-k damage. Because the product is acidic, stainless-steel wetted components are excluded; distribution systems use PTFE, PFA, or high-density polyethylene and avoid elastomeric seals containing zinc oxide or sulfur-cured vulcanizates. Extractable screening of packaging components is performed with the neat product for 14 days at 40 °C, and total non-volatile residue from packaging must remain below 0.5 ppm.

In production-scale 300 mm copper/low-k post-CMP cleaners, the product is typically diluted with ultrapure water of resistivity 18.2 MΩ·cm to a concentration of 5–15 vol%. Dilution is performed in a dedicated chemical blend/delivery system with point-of-use particle filtration at 0.05 µm or finer. The bath is delivered to double-sided brush scrubbers, megasonic immersion tanks, or single-wafer spray tools at a temperature of 20–60 °C and a contact time of 30–120 s. In brush scrubbers, PVA brush pressure is maintained between 0.5 psi and 1.5 psi to avoid slurry particle re-embedding. After chemical contact, the wafer is rinsed with ultrapure water until final rinse resistivity returns to 18.2 MΩ·cm, followed by N₂-assisted drying.

A common failure mode observed on manufacturing lines is re-deposition of ceria particles when the brush rinse is shortened below 30 s or when bath life exceeds 48 h in high-throughput oxide CMP. Monitoring of bath pH and total organic carbon is required because chelating capacity decreases with metal loading and airborne CO₂ absorption. In immersion tools, bath replacement every 24 h is recommended for cobalt-containing wafer lots because dissolved cobalt accumulates in the acidic solution and can produce metal-induced residue defects on subsequent oxide wafers.

Operating Envelope for Brush, Spray, and Immersion Cleaning

The optimal operating window is constrained by the need to remove oxide slurry residues without etching dielectric films or corroding exposed metal plugs. At 10 vol% dilution and 40 °C, typical thermal silicon dioxide etch rate is below 0.5 Å/min, while TEOS and dense SiOC blanket etch rates are below 1.0 Å/min depending on porosity. Copper etch rate at the same condition is below 0.2 Å/min in the absence of dissolved oxygen but increases if the bath is saturated with air. For this reason, process lines often operate the cleaning bath with a nitrogen blanket or degasified water to maintain dissolved oxygen below 100 ppb.

When the product is used in immersion tools, acoustic energy is usually supplied at 0.8–1.2 W/cm² and 1.0 MHz; higher power density can generate cavitation damage on patterned low-k structures. In single-wafer spray applications, nozzle pressure is maintained between 0.3 MPa and 0.7 MPa, and wafer rotation speed is set between 500 rpm and 1,500 rpm. Operating below 20 °C slows chelation kinetics and reduces particle lift-off efficiency, while operating above 60 °C can accelerate low-k surface modification and increase solvent evaporation from the process bowl.

The product is compatible with polyvinyl acetal brush bodies and high-density polyethylene fluid lines at the recommended dilution range. Repeated exposure of polycarbonate sight glasses or polyurethane seals is not recommended because the acidic formulation can cause surface crazing and particle shedding after approximately 100 h of contact. In high-volume oxide CMP fabs, brush scrubber bowls are inspected for particulate accumulation after each 1,000-wafer lot, and point-of-use filter differential pressure is logged to detect gel-like organic residue from aged cleaning baths.

When Tungsten or Cobalt Plugs Constrain the Post-CMP Cleaning Window

The cleaning window narrows when tungsten or cobalt plugs are exposed because the acidic carboxylic-acid system can initiate galvanic coupling between copper interconnects and barrier metals. At 5 vol% dilution and 25 °C, the open-circuit potential difference between copper and cobalt in the diluted cleaner is below 50 mV, which limits corrosion current density to less than 0.1 µA/cm² in short-loop coupon tests. Increasing concentration to 15 vol% does not significantly increase the copper etch rate but does raise the cobalt dissolution rate, especially when the bath contains dissolved oxygen above 100 ppb. Process windows for cobalt-containing stacks therefore require the lower end of the dilution range and a contact time not exceeding 60 s.

For tungsten plug applications, the cleaner is compatible with tungsten plugs at 10 vol% and 40 °C, showing tungsten dissolution below 0.5 Å/min. However, the presence of residual peroxide from prior tungsten CMP must be minimized because mixed peroxide/acidic cleaner increases tungsten loss and generates soluble tungstate ions that redeposit on oxide surfaces. When wafer-level tungsten loss is a critical process control parameter, the cleaning sequence should include a pre-rinse of 15–20 s with ultrapure water to reduce peroxide carryover before the acidic cleaning step.

Published data for the effect of long bath aging on cobalt and tungsten corrosion in this specific formulation is limited. Monitoring of dissolved metal concentration by inductively coupled plasma mass spectrometry is required after 24 h of bath use for cobalt-containing lots. The product is not recommended for aluminium bond pad cleaning because the carboxylic acid system can induce aluminium pitting at pH below 3.0.

Compared with technical-grade aqueous cleaners, the electronic/EL grade product reduces trace-metal and particle burden at source. Technical-grade citric acid or acetic acid cleaners routinely contain sodium above 1 ppm, chloride above 1 ppm, and particle counts above 1,000 counts/mL at 0.1 µm; such levels can cause mobile ion contamination and gate oxide integrity failure in front-end-of-line cleaning. Reagent-grade solvents may offer lower particle counts but often contain plasticizer residues and are not filtered to 0.05 µm. The product differs from alkaline amine-based post-CMP cleaners by avoiding strong silicon dioxide etching and by reducing copper-amine complex formation. However, its acidic character makes it less effective for removing hydrophobic organic residues unless paired with a subsequent dilute hydrogen peroxide or ozonated ultrapure-water rinse.

The comparative values in Table 2 are typical ranges reported in supplier technical bulletins for electronic and non-electronic cleaning formulations; actual performance in a specific fab line depends on tool geometry, rinse configuration, and incoming CMP slurry composition. Product disposal and transport classification follow REACH EC 1907/2006 and CLP EC 1272/2008. The product is not classified as a flammable liquid under closed-cup flash-point testing above 93 °C.

Measuring Post-Clean Defectivity on Patterned Low-k and Tungsten Test Vehicles

On blanket TEOS wafers intentionally contaminated with ceria slurry, a 60 s brush clean at 10 vol% and 40 °C reduces added particle counts from approximately 50,000 counts at 0.1 µm to below 50 counts as measured by scanning surface inspection. In patterned copper/low-k test structures, the product reduces slurry residue defects by over 99% relative to a ultrapure-water-only control, while the post-clean contact angle change on SiOC remains below 2°. On tungsten CMP short-loop wafers, the formulation removes tungsten oxide residues without increasing the number of surface pits when compared to a dilute ammonium hydroxide reference.

Defect classification is performed using dark-field wafer scanners followed by SEM review; energy-dispersive X-ray spectroscopy is used to confirm that residual particles are not copper-containing corrosion products. For ceria removal, the residual particle composition is predominantly carbon-based foreign material rather than cerium oxide at the completion of the cleaning sequence. Process engineers should re-qualify the cleaner on product wafers after changing brush lot, PVA conditioning chemistry, or post-CMP queue time. The measured removal efficiency values are specific to the cited tool configuration and are not guaranteed across non-standard scrubber geometries or heavily aged slurries.