Meta title: JP vs EP/USP Colloidal Silica: The Limits That Decide Your Audit (58 chars) Meta description: Japanese Pharmacopoeia allows ≥98.0% assay and ≤12% ignition loss. European and US monographs do not. Here is the item-by-item gap Korean buyers test against. (154 chars)

Primary keyword: colloidal silicon dioxide pharmacopoeia compliance Secondary: JP vs USP silica limits, silicon dioxide E551 specification, pharmaceutical excipient silica audit, Korean excipient import requirements, fumed silica COA

A supplier sends a certificate. The paper is clean: assay, pH, loss on drying, heavy metals, microbiology — all present, all "conforms". The buyer in Seoul reads it and asks one question: which pharmacopoeia produced these numbers?That single question separates two product worlds that look identical on the pallet. Same white powder, same 10 kg bag, same 200 m²/g surface area. Different acceptance logic.

Japanese Pharmacopoeia and the European/US monographs describe the same material, but they do not draw the same lines. Where JP accepts, EP and USP may reject — and a Korean pharmaceutical or oral-care buyer typically tests against all four (Ph. Eur., USP/NF, JP, IP), not the loosest one.

This article lays out the actual numerical gaps, then shows what a serious test panel and documentation set has to contain to survive that audit.

Below are the monograph limits as they appear on a European certificate delivered to a Korean consignee, cross-referenced against a Japanese-standard technical data sheet.

Read the first three rows again. This is where the argument is won or lost.

Assay. JP opens the door at 98.0%. EP and USP open it at 99.0% and close it at 100.5%. A lot that reports 99.2% passes JP comfortably and passes EP/USP — but only just. A lot at 98.4% is fully JP-compliant and fails EP and USP outright. Same certificate, opposite verdicts.

Loss on ignition. JP permits up to 12.0%. EP permits 5.0%. USP permits 2.0%. The spread is six-fold. A product at 6% ignition loss is a legitimate JP material and is non-compliant in Europe and the United States.

Loss on drying. JP allows 7.0%; USP allows 2.5%. For a powder sold as moisture-sensitive, a 7% allowance is a different product promise than a 2.5% one. There is no single "more strict" pharmacopoeia. JP is lenient on assay and ignition loss, but adds tests the others omit (volume test, iron, calcium, aluminium). EP/USP are tight on assay and ignition loss but silent on those identity-adjacent tests.

The strictest set is the union, not the intersection — which is exactly why Korean buyers ask for four monographs at once.

Korean excipient import review — whether at a pharmaceutical company or a toothpaste manufacturer supplying a regulated brand owner — typically asks six things. Only one of them is "does it meet JP".

1. Four-monograph compliance, tied to the specific grade. Not "our silica range complies", but "this grade, this specification number, this version". A monograph claim must be bound to a document, not to a product family.

2. Elemental impurities with numbers, not adjectives. A certificate that writes "Conforms" for arsenic, lead, mercury and cadmium is technically valid — but it is not what an audit wants. Auditors ask for: measured value, applicable limit, analytical method, and detection or quantification limit. The absence of cadmium data is the most common gap in silica documentation, because cadmium is not a headline parameter in most silica monographs but is required under ICH Q3D risk assessment.

3. Microbiology — per batch or periodic? "Tested on a regular basis" is a legitimate statement and a red flag at the same time. If the answer is periodic, the buyer needs to know the frequency, the sampling plan, and which batches the report covers. TAMC, TYMC and Gram-negative bacteria are three separate questions, and many certificates answer only the first.

4. GMP scope, not GMP existence. The question is not "do you have a GMP certificate" but "does the certificate cover this product, on this line, at this site". A GMP document covering cosmetic raw materials does not automatically speak for a pharmacopoeial excipient line.

5. A customer-specific COA, batch by batch. Format aligned to the buyer's template, test items mapped to the applicable monograph, actual results per item, and suitable for onward submission to their customer. This is the requirement that most often decides whether a supplier becomes a long-term vendor.

6. Traceability to the manufacturing site. Site name, address, and confirmation that the entity on the COA is the entity that made the material. Tolling or trading arrangements have to be declared, not discovered.

A "JP grade" answer satisfies item one, partially. It leaves five open.

A short certificate lists eight to twelve parameters. A full panel for a pharmacopoeial colloidal silica runs closer to twenty, and the extra rows are not decoration — each closes a specific failure mode:

Two items deserve emphasis because they are under-reported in the market.

Sieve residue. A typical acceptance figure is below 0.1% on 325 mesh, with well-controlled material reaching "NIL" (not detectable). For toothpaste, residual grit is the single most expensive defect — it is invisible in the COA and obvious in the consumer's mouth.

Carbon content. "Undetected" on a carbon test is a meaningful statement about process cleanliness, and it is frequently omitted from abbreviated certificates.

Two suppliers can report identical assay values and still be worlds apart at bid stage. The difference is the document set.

A certificate that is computer-printed and unsigned is still usable — but only because it references a controlled specification number and a batch code.

Without those two identifiers, the paper is marketing. With them, it is a record.

1. Which pharmacopoeial monographs does this grade, at this specification version, meet?

2. Are the limits on your TDS the monograph limits, or your internal tighter limits? State which.

3. For arsenic, lead, mercury and cadmium: value, limit, method, LOQ.

4. Is microbiology per batch or periodic? If periodic, what frequency and what sampling plan?

5. Does your GMP certificate name the site and the product line that makes this material?

6. Is elemental impurity testing performed under ICH Q3D, and is a report available?

7. Do you declare residual solvents and TSE/BSE status explicitly?

8. What is the 325-mesh sieve residue, typically and as a limit?

9. Can you issue a batch COA in our template, with per-item results, acceptable for onward submission?

10. What is the shelf life, and what is the re-test rule for material older than that?

A supplier who answers all ten in writing is qualified. A supplier who answers six is a sample source, not a vendor.

Is Japanese Pharmacopoeia compliance enough for export to Korea? No. Japanese compliance commonly satisfies one part of the file. Korean buyers generally require the four-monograph set, plus elemental impurity, microbiological and GMP evidence.

Why does the same batch give two different loss-on-drying values? Because the monographs specify different drying conditions and sample preparation. Divergent values across JP and USP columns are normal — provided each value is labelled with its method. Divergence without a method label is a data integrity question.

Can a JP-compliant lot fail USP? Yes. Assay at 98.4% passes JP (≥98.0%) and fails USP (≥99.0%). Ignition loss at 6% passes JP (≤12.0%) and fails USP (≤2.0%).

Does a "Conforms" entry on heavy metals meet audit requirements? Usually not for a regulated end use. Auditors want measured values, limits, methods and LOQs. "Conforms" is acceptable on a transport certificate; it is thin on a registration dossier.

Is high BET always better for toothpaste? No. BET drives thickening and thixotropy; it is not an abrasive measure. A grade optimised for thickening behaves differently from one optimised for cleaning. Match the parameter to the function, not to the headline number.

What sieve residue should a toothpaste-grade silica hold? Below 0.1% on 325 mesh is a common acceptance level; tightly controlled material reports "NIL". Ask for the limit, not just the result.

Does silicon dioxide need a cadmium declaration if the monograph is silent? For pharmaceutical and food-contact review under ICH Q3D, yes — the risk assessment covers cadmium regardless of whether the monograph lists it.

How long is typical shelf life, and what happens after it? 24 months in unopened original packaging under stated storage conditions is the market norm. Material past that date should be re-tested — at minimum for moisture — before use.

"JP grade" is a legitimate description and an incomplete answer. It describes one monograph out of four, on a material whose commercial value depends on the other three. The gap is not marketing language; it is three numbers — assay floor, ignition-loss ceiling, drying-loss ceiling — and a document set that either carries them or does not.

Buyers who understand this stop asking "does it meet the standard" and start asking "which standard, at what limit, measured how, on which batch". Suppliers who can answer that sentence are the ones who keep the account.