Elemental Impurities by ICP-MS

The old USP ⟨231⟩ Heavy Metals test was a visual colorimetric comparison: the sample was ashed or treated, sulfide was precipitated at controlled pH, and an analyst compared the resulting brown-black turbidity by eye against a lead standard. Its defects were structural. It produced one lumped number rather than element-by-element data and could not distinguish lead from bismuth. Recoveries were poor and variable — mercury and arsenic were frequently lost during ashing and volatile species escaped entirely. The endpoint was a subjective visual judgment. And it was largely blind to the elements that matter most in modern synthesis: palladium and the platinum-group catalyst residues, and cadmium.

Effective 1 January 2018, ⟨231⟩ was omitted from USP–NF and all monograph references to it were deleted. A supplier still citing "heavy metals by USP ⟨231⟩" is citing a chapter that has not existed for years. It was replaced by ⟨232⟩ for limits, ⟨233⟩ for procedures, and ⟨2232⟩ for dietary supplements, with ⟨232⟩ aligned to ICH Q3D.

For a synthetic research material the practical question is not "are there heavy metals" but "which of these twenty-four elements are present, at what µg/g, with what quantitation limit." That is the question an element-resolved ICP-MS panel answers and a lumped sulfide test never could.

Sample preparation. USP ⟨233⟩ permits neat, direct aqueous, direct organic, and indirect (digested) solutions. For a solid organic material — which is what a lyophilized peptide or a synthetic small molecule is — closed-vessel microwave acid digestion is the working method. Roughly 0.1–0.5 g is weighed into a PTFE or TFM vessel (⟨233⟩'s example procedure uses 0.5 g) and predigested with concentrated nitric acid, typically 5–10 mL of trace-metal-grade or sub-boiled acid, with the initial exotherm allowed to subside before sealing. The vessel is ramped to roughly 180–220 °C and held 15–30 minutes; pressures of 40–60 bar are normal.

Hydrochloric acid is added where mercury must be stabilized or where the platinum-group metals — Pd, Pt, Au, Ru, Rh, Ir — must be brought into solution. Nitric acid alone will not reliably dissolve them, and mercury is lost or plates out without chloride. Hydrofluoric acid is used only for silicate-bearing matrices, with boric acid complexation or an HF-resistant introduction system. The digest is diluted to a fixed final volume, giving a known dilution factor that is carried through to the reported result.

Contamination control dominates trace work. Acid-leached PFA or polypropylene labware only — borosilicate glass leaches B, Na, Al, and Zn. Sub-boiled or trace-metal-grade acids. Clean-bench handling. And a method blank digested through the entire procedure on every batch, reported alongside the sample.

Interference management is where competence in this technique is visible:

• ⁷⁵As is mono-isotopic and sits under ⁴⁰Ar³⁵Cl⁺. Managed by helium collision/kinetic energy discrimination, or by oxygen reaction-mode mass shift to ⁷⁵As¹⁶O⁺ at m/z 91.
• ⁵²Cr suffers ⁴⁰Ar¹²C⁺ and ³⁵Cl¹⁶O¹H⁺; managed by He KED or by measuring ⁵³Cr.
• ¹¹¹Cd and ¹¹⁴Cd suffer ⁹⁵Mo¹⁶O⁺, ⁹⁸Mo¹⁶O⁺, and ¹¹⁴Sn⁺; managed by isotope selection, oxide-ratio tuning (CeO/Ce below roughly 2%), mathematical correction, and KED.
• Pb is radiogenically variable, so ²⁰⁸, ²⁰⁷, and ²⁰⁶ are summed — standard practice.
• Hg has severe memory and carryover; managed with gold in the diluent and rinse (typically 1–2 mg/L) plus HCl, an extended rinse, and monitoring the blank following the highest standard.
• Os volatilizes as OsO₄ from oxidizing media; either the digestion is adjusted or the limitation is accepted and documented.

Internal standards matched by mass and ionization potential — commonly Sc, Ge, Rh, In, Tb, Ho, Lu, Bi — correct for physical matrix effects and instrument drift. ⟨233⟩ requires drift of not more than 20% for each target element, checked against bracketing standards.

ICP-MS versus ICP-OES. ⟨233⟩ describes both, and permits alternative procedures that meet the chapter's validation criteria. For most target elements ICP-MS instrument detection limits in solution are in the low ng/L range. A 0.5 g sample digested to 50 mL is a 100-fold dilution, so a 0.01 µg/L solution detection limit corresponds to roughly 0.001 µg/g in the original solid — orders of magnitude below the parenteral Class 1 concentration limits, which is the margin you want. ICP-OES detection limits are typically 1–50 µg/L in solution, roughly 0.1–5 µg/g in a solid at that dilution. That is entirely adequate for the high-PDE Class 3 elements and for oral-route limits in clean matrices, but it does not reliably deliver adequate margin against parenteral limits for Cd, Hg, and As. For a general panel on an unknown organic solid, ICP-MS is the default and ICP-OES is the supplement, not the reverse.

PLACEHOLDER: Arcadia's actual elemental instrumentation — ICP-MS (quadrupole with collision cell? triple quadrupole?), ICP-OES, and microwave digestion system. Do not name any make or model the laboratory does not own; delete instrument-specific copy entirely if none is confirmed

⟨233⟩ defines J as the concentration (w/w) of the element of interest at the target limit, appropriately diluted into the instrument's working range. Every acceptance criterion keys off J.

Limit procedure, for a pass/fail determination against J:

• Detectability — the mean of three replicates of a sample spiked at J must fall within ±15% of the standard solution at J, and a sample spiked at 80% of J must read below the standard.
• Repeatability — RSD not more than 20% for six independent samples spiked at J.
• Specificity — each target element must be unequivocally resolved from the other targets and from the matrix.

Quantitative procedure:

• Accuracy — 70% to 150% recovery, mean of three replicates, at each of several levels spanning 50% to 150% of J.
• Repeatability — RSD not more than 20%, n = 6.
• Intermediate precision — RSD not more than 25%, n = 12, being the repeatability determination repeated on a different day, instrument, or analyst.
• Range — must span 50% to 150% of J.

These windows are wider than a chromatographic assay's, and deliberately so. A 70–150% accuracy tolerance is appropriate to trace elemental work carried through an acid digestion, and it is worth stating plainly to clients who arrive expecting HPLC-assay tightness. The correct response to that width is not to hide it but to report the spike recovery and duplicate RSD actually obtained, per element.

PDEs are µg/day, not concentrations. Converting a PDE to a concentration limit requires a dose assumption: concentration limit (µg/g) = PDE (µg/day) ÷ maximum daily intake (g/day). ICH Q3D provides four control options — Option 1 uses common concentration limits assuming a 10 g/day intake, Option 2a uses the actual daily intake, Option 2b applies component-specific limits summed across a formulation, and Option 3 analyzes the finished product against the PDE directly.

Under Option 1 the parenteral concentration limits work out to 0.2 µg/g for Cd, 0.5 µg/g for Pb, and 1.5 µg/g for As. Those numbers embed a 10 g/day human intake assumption that has no meaning for a research-use-only article. We therefore report measured µg/g with the per-element LOQ, and where any limit is quoted we state which route, which control option, and which daily-intake assumption generated it.

Further limits of the result:

• It is a total-element measurement, not a speciation measurement. It does not distinguish inorganic from organic arsenic, or Cr(III) from Cr(VI). Where speciation matters, a speciated method is required and must be requested separately.
• It sees only the elements on the panel. An element not determined is not excluded by this analysis; it is simply unexamined. The report names what was and was not in scope.
• Osmium may be under-recovered where oxidizing digestion conditions volatilize OsO₄. Where Os is in scope, the limitation is documented rather than silently absorbed into the result.
• It describes the aliquot digested. Elemental contamination can be heterogeneous, particularly where it arrives as discrete particulate from equipment wear or a catalyst filter breakthrough. Where the client supplied the sample, no statement about the batch is possible.
• It is not a safety determination. A µg/g result characterizes a research material. It does not establish suitability for any use, and no acceptance criterion published by Arcadia accompanies it.