For semiconductor oxide removal, select aqueous hydrofluoric acid (HF) by matching the delivered composition to the qualified bath, target film and contamination budget. Native-oxide removal, blanket-film etching and patterned oxide removal need different acceptance evidence. Concentrated HF purchased for dilution is a feedstock; dilute HF and buffered oxide etch (BOE) are distinct composition choices. An equal HF assay does not make an HF–water solution equivalent to an HF–NH₄F mixture, and lower impurity numbers can simply reflect dilution. Compare composition definitions, contaminant contribution at use and representative etch results before accepting a purchasing specification. This page covers aqueous silicon-oxide removal, excluding anhydrous HF, vapor etching and silicon-etch mixtures.

Match the HF Product to the Actual Etch Task

Start with the process requirement rather than a catalog concentration. Record the target oxide, exposed materials, qualified bath composition, temperature, exposure range and relevant bath-use condition.

These are evidence requirements, not universal assignments of one concentration or etchant to each application. Retain the already qualified chemistry unless a separately validated process change supports an alternative.

The procurement consequence is specific: a supplier’s thermal-oxide etch result cannot establish performance on a different deposited oxide or multilayer stack. Williams, Gupta and Wasilik’s original etch-rate study documents film-dependent behavior, including changes after annealing. Its measurements apply to its tested materials and conditions. Microelectromechanical Systems, Journal of

Use that evidence to choose a representative test film. Keep published rates as references rather than transferring them into a purchasing guarantee.

For the wider relationship between chemistry, contamination and commercial evidence, see the semiconductor wet-process chemical qualification framework. The decision here is narrower: which HF specification is suitable for the intended oxide-removal step?

Compare Concentrated HF, Dilute HF and BOE on Different Bases

Water is an intentional component of aqueous HF. An anhydrous-solvent moisture specification is inappropriate here; concentration and its tolerance describe the water–HF composition relevant to purchasing.

In BOE, NH₄F contributes fluoride alongside HF. Matching the HF assay alone leaves a major composition variable uncontrolled. Compare both component assays and validate the finished mixture’s etch response.

For a volume ratio, identify both components in order and their starting concentrations. Do not interpret an unlabeled “10:1” as a finished HF percentage. Weight percentage and volume ratio cannot be interchanged without the relevant composition and density information.

Why does BOE require its own comparison? HF–NH₄F composition changes the acid–fluoride environment and the distribution of fluoride species. The published abstract of Kim, Lee and Lim’s 2024 study, indexed in EPA HERO, describes contributions from HF, F⁻ and HF₂⁻ under different compositions and pH conditions. Its model covered the studied conditions up to pH 5. That supports composition-specific evaluation, rather than a universal etch-rate prediction from nominal HF concentration.

For a proposed BOE, request both component assays and declared wetting additives. A change in buffer content or additive identity is a formulation change even when the HF assay remains unchanged.

Define What the HF Assay Actually Measures

A nominal concentration needs a tolerance, reporting basis and method definition. Ask whether a result represents HF mass fraction, a defined free-acid measurement, total acidity or total fluoride. These quantities are not automatically interchangeable, especially in buffered mixtures.

The SEMI C28 public listing addresses semiconductor HF; SEMI C29 separately addresses 4.9% HF. As checked on October 13, 2026, both listings show their 0618 revisions as inactive. SEMI states that inactive standards remain valid for use.

Specify the agreed revision and applicable grade. The public abstracts establish scope and status; they do not provide enough detail to reproduce every grade limit or verify a supplier’s conformance claim.

For concentrated feedstock, assess the offered assay range against the buyer’s approved dilution control. For purchased dilute HF or BOE, assess the finished product’s composition. A concentration inside the supplier’s release range may still be unsuitable for a narrower customer process window.

As a hypothetical comparison, changing feedstock from 49.0 to 48.0 wt% HF under an unchanged mass-based dilution factor lowers the calculated bath HF mass fraction by about 2.04% relative. This is a composition calculation, not an etch-rate prediction. Acceptance depends on the qualified bath window and the actual preparation control.

Compare Impurities at the Same Intended Use Concentration

A lower impurity number in a more dilute product can reflect dilution rather than a cleaner HF source. Compare concentrations on a consistent mass or volume basis, then account for the amount of each offered product required at the same intended bath composition.

Consider this illustrative calculation for one dissolved contaminant:

The second result is ten times lower per gram of solution. At the same final HF mass fraction, the offered liquids contribute the same contaminant concentration, assuming simple mass balance and no contaminant removal or addition during subsequent preparation.

These are invented comparison inputs, not supplier data or acceptance limits. The calculation excludes additional water, buffer feedstocks, handling and package contributions. It also predicts neither wafer deposition nor electrical effects.

For purchasing, retain both the delivered-product result and its calculated contribution at use. Allocate the acceptable incoming-HF contribution with the process owner; dilution alone does not establish process acceptability.

Liquid impurity concentration and wafer contamination are different measurements. If the specification must protect a wafer endpoint, review process-relevant transfer evidence rather than assuming a fixed conversion. The chemical-to-silicon trace-metal transfer reference explains the effects of surface state, solution chemistry and rinsing without assigning a universal transfer factor.

Do not convert ng/L into ng/kg by assuming water density. Use the solution density at the stated condition, and confirm that reported results have been corrected back to the original delivered sample after laboratory dilution.

Require HF-Matrix Evidence for Critical Impurities

Elemental Contaminants

Define individual controlled elements and their limits. Na, K, Fe, Cu or other elements may belong in the panel when the process specification requires them; a generic “total metals” value cannot replace individual limits.

Require blanks, sample-preparation details, reporting limits, matrix suitability and an uncertainty or decision rule appropriate to borderline results. Thermo Fisher’s Application Brief 43197 demonstrates direct analysis in 48% HF and describes control of a fluoride-related interference affecting calcium measurement. It illustrates why an ICP-MS label alone cannot establish method suitability. The supplier still needs evidence for its actual matrix, element panel and reporting limits. documents.thermofisher.com

An illustrative purchasing limit of 0.10 ng/g cannot be confirmed by “ND” at a reporting limit of 0.50 ng/g. Keep that candidate on hold until decision-capable evidence is available. For a result near the limit, use the agreed measurement decision rule rather than treating the last reported digit as certain.

Particles, Ions and Organic Contribution

For particles, align size thresholds, cumulative or differential channels, counts per volume, sampling point and preparation. Require an HF-compatible measurement path and suitability evidence for the tested concentration. Chemical particle counts alone cannot identify wafer-defect causation. The particle-report review for electronic-grade batch release explains the necessary measurement comparisons.

For specified ionic contaminants, request methods demonstrated in the relevant fluoride matrix. For organic contribution, distinguish intentionally added wetting agents from unintended residues and select evidence against the process concern. A bulk organic result does not identify a particular compound.

These requirements belong in the purchasing specification only when they can change the material-selection decision. No universal impurity panel or numerical limit is established here.

Keep Analytical Conformity Separate From Etch Acceptance

A specification defines release requirements. An actual-lot COA reports the listed release results for its identified sample. A TDS may describe typical use or performance; an SDS addresses hazards and handling. None independently proves the customer’s oxide-removal endpoint.

For a new candidate, evaluate a representative sample against the qualified reference under matched bath composition, temperature, exposure, bath condition and rinse sequence. Select measurements that address the intended result: oxide removal, remaining-film thickness, within-wafer variation, adjacent-film loss, or surface contamination where relevant.

For thin native oxide, require a method capable of resolving the relevant removal endpoint. For patterned structures, include the actual feature and adjacent-material requirements. A method capable of measuring a thick blanket film may leave a different endpoint unresolved.

If etch behavior differs, check bath preparation, film deposition or annealing history, actual temperature and measurement repeatability before assigning the difference to the HF supplier. If removal passes while a contamination endpoint fails, the material remains unsuitable for that use.

A thermal-oxide monitor can support a defined screening comparison. It does not approve every deposited oxide, patterned feature or exposed material in the production stack.

Decide How Far the Evidence Allows the Candidate to Advance

Before commercial acceptance, require material representing the proposed manufacturing, dilution where applicable, filling and package configuration. A specially prepared evaluation bottle may support initial film-response testing while leaving routine delivered contamination unproven. Select commercial-lot coverage through the buyer’s protocol; no fixed lot count is prescribed.

The PFA and HDPE packaging validation guide covers the separate contact-system decision. A clean small bottle does not establish the performance of a drum, closure and dispensing assembly.

Changes in HF assay range, buffer composition, additives, dilution route, analytical method or commercial contact system require review of the evidence they affect. For full source approval, use the semiconductor wet-chemical supplier evidence assessment.

Next Step: Review the Application-Specific HF Specification

Select a specification only after its composition basis, contaminant contribution and required etch evidence align with the intended process. The immediate next step is a specification review.

Provide the proposed aqueous HF concentration or BOE composition, target oxide and exposed stack, critical acceptance criteria, and current qualification stage. Request the applicable specification, assay definitions and method/reporting basis. Include the proposed package identity when delivered-condition evidence affects this comparison, and confirm availability for the required grade separately.