What “membrane grade” states, and what it does not

A cell name describes a production route

EPA’s supply-chain profile for sodium hydroxide sets out the manufacturing picture plainly. Sodium chloride is the primary raw material; the majority of caustic soda is produced by the chlor-alkali process, passing a direct current through a brine so that chloride is converted to chlorine at the anode while sodium ions and hydrogen collect at the cathode. The caustic is then separated from the solution by one of three methods: the diaphragm method, the membrane method, or the mercury cell method. EPA records the diaphragm method as the most common separation process used in North America.

Read that carefully and “membrane grade” resolves into what it actually is: the name of the separation step the caustic came out of. It is a production route, not a published specification, and the profile that describes the industry states no attribute limits at all. This matters for a mill because caustic is not a marginal purchase here—EPA puts pulp and paper manufacturing at 22 percent of North American sodium hydroxide consumption in 2015, against 3 percent for all water treatment applications combined. A designation that large a spend rests on should be written down as attributes, not as a word.

When a rule names a specification, it names an edition

21 CFR 184.1763 affirms sodium hydroxide as generally recognized as safe for direct use in food, and in doing so it does something a purchase order rarely does: it says which document the ingredient must satisfy. “The ingredient meets the specifications of the Food Chemicals Codex, 3d Ed. (1981), which is incorporated by reference.” One sentence, one document, one edition, pinned—and the edition it pins is from 1981. The Food Chemicals Codex is a purchased standard; it was not readable while this guide was written, so nothing here paraphrases its requirements or reproduces a limit from it.

The same section is also worth reading for its first paragraph, which records that sodium hydroxide “is prepared commercially by the electrolysis of sodium chloride solution and also by reacting calcium hydroxide with sodium carbonate.” Two routes, and they do not carry the same residual chemistry. If a route matters to a process—and in a mill it can—that is a question to put in writing, because a GRAS affirmation says nothing about it. Being affirmed as GRAS is not an approval of any particular supplier’s material, and nothing here states that an Alliance product is suitable, approved or recommended for a food-contact or any other application.

Where the mercury question actually lives

40 CFR part 63, subpart IIIII is the national emission standard for hazardous air pollutants for mercury cell chlor-alkali plants. It applies to each affected source at a plant site “where chlorine and caustic are produced in mercury cells,” and it reaches two affected sources: the production facility, meaning all cell rooms and ancillary operations, and the mercury recovery facility. An affected source is existing if construction commenced before July 3, 2002. The subpart is live: it was amended at 87 FR 27018 on May 6, 2022, with work-practice and electronic-reporting obligations dated November 7, 2022 and an emission limitation at § 63.8190(a)(2)(ii) applying on and after May 6, 2025.

That is where the mercury history sits in current federal law—as an air-emission and work-practice obligation on a producer, enforced through a title V permit. It is not an impurity limit on a delivered drum, and the continued existence of the subpart is not evidence about the provenance of any particular supply. If a mercury limit matters to a mill’s product or its own discharge, it has to appear on the purchase document the way any other requirement does: an attribute, a numeric limit, a named method, units and a reporting limit. A cell-technology word in a product name is not that.

Fifty percent is a band, and it does not stay put

The aggregated public record for sodium hydroxide describes the commercial article in commodity terms rather than specification terms: “Anhydrous (Rayon Grade) 99.0% minimum; Rayon Grade: 50% liquid; Regular Grade: 50% liquid, 47.7-51% purity.” Two observations follow. Rayon grade is another commercial designation sitting beside membrane grade with exactly the same problem—it names a market, not a document. And a nominal fifty percent is described there as a band roughly three points wide. A make-up calculation and an analytical control chart both depend on where in that band a delivery actually falls, which is a reason to ask whether a quoted figure is typical, a guaranteed limit, or a measured result for the lot.

The same record also explains why the number moves after delivery. Caustic solutions are hygroscopic and “when exposed to the air they absorb carbon dioxide, forming sodium carbonate,” and the stability entry is blunt about the consequence: containers “must be tightly closed to prevent conversion to sodium carbonate by carbon dioxide of air.” Carbonate is therefore an in-service impurity as well as a manufacturing one, and the vent arrangement on a storage tank belongs in the same conversation as the assay. Note the limits of this record too: it reports the melting point of the solid as 318 °C in one entry and 323 °C in another, from different agencies. Aggregated values describe the pure substance under someone else’s conditions; they are not measurements of a supplied lot, and this guide states no crystallisation temperature for any concentrated solution—that is a datum to request from the supplier for the exact product, with its basis and its method.

• Which document defines the designation being requested, and which attributes does it limit?
• Is the stated assay typical, a guaranteed specification limit, or a measured lot result—and on which basis and at what temperature?
• Which impurities does this process actually control, each with a method, units and a reporting limit?
• What is the crystallisation temperature for the exact product as delivered, and what heating and venting does the receiving arrangement provide?

Ask for the grade, and for the certificate

Alliance publishes the grade a product carries in its structured product data, and that record—not the marketing copy, and not the handle—is what a purchase document should quote. Where a grade is not stated for a product, the correct action is to ask rather than to infer one. A nominal grade or concentration does not establish interchangeability between products or between suppliers, and being merchandised under a paper and pulp collection is merchandising: an industry listing is not an application approval.

Alliance sends a Certificate of Analysis when a customer asks, at no charge. Put the request in the purchase order and name what it must cover: the analytes, the methods, the units and the reporting limit for each. What grade designations do and do not certify in general is treated at length in this library’s laboratory and pharmaceutical research collection, under “What a grade designation certifies, and what it does not”; this guide does not repeat it. This is source-based editorial guidance and independent specialist review is pending.

Requirement → evidence → decision boundary

An editorial checklist for your review—not a table of product specifications.

Revision record

1.0 · September 12, 2026 — Source-based editorial edition. No original testing or product qualification is claimed.

For pulp, paper and paperboard mills and their purchasing teams specifying bulk caustic, bleaching chemistry, coagulants and reducing agents. This collection does not select a bleaching sequence, set a dose or an application rate, design a stock or water circuit, or make a discharge, permit or waste determination.