USP 233 in practice: the J value and the four preparation routes

Two chapters, two different jobs

The limits chapter and the procedures chapter are quoted as a pair and read as one document. They are not one. The limits chapter answers how much of an element a patient may receive per day, banded by route of administration. The procedures chapter answers a narrower question: what must a laboratory demonstrate before it may claim a material sits below such a limit.

Two details in the limits chapter matter more than the values. The limits are written for drug products, and the table of default concentration limits for components covers products dosed at 10 g per day or less; the chapter calls those defaults an aid to discussion between manufacturer and supplier, so a component limit is a default for negotiation rather than a specification. The second detail: the arsenic and mercury limits are written for the inorganic forms, a question the method chapter does not answer.

J is the number you validate against

J is the concentration of the element at the target limit, diluted into the working range of the instrument. The chapter's own example: an oral solid at 10 g per day, lead at a target limit of 0.5 micrograms per gram, and an ICP-MS whose linear range ends at 0.1 micrograms per millilitre. A dilution of at least 1:10 brings the sample inside that range, so J for lead becomes 0.05 micrograms per millilitre. The limit did not change; the working concentration did.

From there the chapter stops speaking in micrograms per gram. The standardization solutions are 2J and 0.5J, the sample solution is diluted until the target elements sit at not more than 2J, accuracy is tested from 50 to 150 per cent of J, and detectability uses a spike at the target and a second at 80 per cent of it. A laboratory that has not written J down for each element cannot prepare any of them.

Preparation becomes irreversible here. J depends on sample weight and final volume, so turning a 0.5 g portion into 0.25 g doubles J and moves every standard, spike and acceptance band with it. Changing the dilution after validation does not adjust the method, it invalidates it. J also settles feasibility: if J sits below the quantitation limit, no spiking rescues the method.

Four preparation routes

The chapter lists four forms of preparation and leaves the choice to the analyst. They differ in how many chances they create for contamination and loss.

Closed vessel digestion, the chapter notes, minimises loss of volatile impurities and comes with a worked procedure: dehydrate and predigest 0.5 g in 5 mL of freshly prepared concentrated acid, leave loosely covered for 30 minutes, add 10 mL, digest until complete, and repeat with 5 mL if needed. Concentrated acid here means nitric, sulfuric, hydrochloric or hydrofluoric acid, or aqua regia at about 3:1 or 4:1.

The rule that survives a real laboratory is the least ambitious one: if the material dissolves in an aqueous solvent and J sits well above the quantitation limit, take the direct aqueous route. Every digestion step is another vessel to clean, another blank to run, and another chance for the blank to become the largest number in the calculation. Mercury is the exception, where the chapter directs a stabilizer; and liquid samples should be weighed, not pipetted.

Matched matrix includes the stabilizer

A matched matrix has the same solvent composition as the sample solution: the same acids at the same concentrations, and the same mercury stabilizer. The blank is a matched matrix, and so are both standardization solutions. A stock at 1000 micrograms per millilitre diluted into water is not matched to a digest that is 5 per cent in nitric acid with chloride present, and the chapter directs that matrix-induced interferences be corrected, naming argon chloride interference with arsenic.

The requirement also marks the boundary of a certified reference material. The chapter requires materials from a national metrology institute or traceable to one. A trace elements in water material verifies the instrument and the calibration: its certificate states values traceable to the SI, an expanded uncertainty at 95 per cent confidence, and a matrix of about 2 per cent nitric acid. That matrix is the limit. It verifies the measurement chain, not a preparation that ends in a mixed acid digest, and it cannot serve as the matched blank.

What a limit procedure has to prove

A limit procedure answers a yes or no question, and the detectability test is built around that, not around accuracy. Four solutions are prepared: a standard at the target concentration, a sample spiked at the target, a sample spiked at 80 per cent of it, and an unspiked sample. For instrumental procedures, the mean of three replicate readings of the first spike must fall within 15 per cent of the mean of the standard, and the second spike must read lower, with all values corrected using the unspiked sample.

Read as a resolution test, that makes sense: a limit procedure does not need to know how far below the limit a material sits, only that a material at the limit separates from one at 80 per cent of it.

System suitability is equally spare. The 2J solution is run before and after the batch and the two results must agree within 20 per cent, which is the whole calibration design: two standards bracketing the batch and a drift check, with no multi-point curve required. The chapter adds a note that catches more batches than the acceptance criteria do: if the samples are high in mineral content, rinse the system for 60 seconds between them.

What a quantitative procedure has to prove

The last row is the one most often over-engineered: a five-point curve with a correlation above 0.99 is not what the chapter asks for. It asks for accuracy across the range and lets that one experiment stand in for three parameters. The corresponding trap sits in the first row, where the spike goes in before solubilization or digestion; a spike added to a finished digest says nothing about the preparation, which is where losses occur. Where this chapter differs from the general validation chapter, the criteria here take precedence.

Speciation is outside the chapter

The procedures chapter states that it does not include speciation, the determination of oxidation state or organic complex. The limits chapter explains how a total-element measurement still works: arsenic limits are based on the inorganic and most toxic form, and arsenic may be measured by a total-arsenic procedure on the assumption that all arsenic present is inorganic, with a form-specific procedure available to show that the inorganic fraction complies where the total exceeds the limit. Mercury follows the same logic, written for the mercuric form.

The number on the report is therefore a total, and usually a conservative one, since the total is at least as large as the inorganic fraction. That conservatism holds while the material contains no organoarsenic species; when a total-arsenic result fails, the way out is a form-specific method, not a larger dilution.

Where the standards enter

Three questions decide whether a material can carry the role of standard solution: does the certificate give an assigned value with an uncertainty, does it name the matrix including acid concentration and any stabilizer, and is its validity unexpired. A purity percentage and a catalogue number substitute for none of the three. It is the same test that applies to a standard used as a calibrant rather than as an identity reference, and the fields themselves are set out in the structure of a certificate of analysis.

A multi-element mix whose matrix differs from the digest biases each element in a different direction, which is why matched standardization solutions matter more than the brand of the stock. An opened solution also has its own clock, and the established hold time for a working solution decides whether yesterday's 2J solution is still usable. The weighing before all of it carries a minimum mass, covered in the arithmetic of minimum weight.

A checklist before the first run

• Write the target limit per element with the route, the maximum daily dose, and the edition.
• Calculate J for every element, confirm it is at least ten times the quantitation limit, and confirm the sample solution ends at not more than 2J.
• Choose the preparation route, then build the blank and both standardization solutions in the same acid, with the same mercury stabilizer.
• Add the mercury stabilizer after cooling, and weigh liquid samples rather than pipetting them.
• Spike before solubilization or digestion, never after.
• Bracket the batch with the 2J solution, check drift against the 20 per cent limit, and state whether the result is total element or speciated.

What none of this decides

None of it answers whether the test should be run at all. The risk assessment framework sets a control threshold at 30 per cent of the permitted daily exposure, and where the total from all sources stays below it, additional controls are not required; testing is one possible control. Turning a permitted daily exposure into a concentration limit is a separate step, worked through in the conversion from permitted daily exposure to a sample limit, and for elements driven by DNA reactivity rather than by a threshold, the mutagenic impurity route replaces both.