For most of the twentieth century, "heavy metals" testing meant sulfide precipitation and a colour comparison against a lead standard — non-specific, insensitive, blind to which element was present. ICH Q3D ended that. Adopted in 2014 and revised as R1 (2019) and R2 (April 2022), it replaced that test with toxicologically derived limits for 24 elements, a risk-based framework for deciding which of them matter, and element-specific analysis where warranted.
Q3D is not a test method. It is a risk assessment that sometimes ends in one.
PDE limits
The PDE is the maximum acceptable daily intake, derived from a NOAEL or MRL with modifying factors and a 50 kg body mass assumption. Table A.2.1 of Q3D(R2):
| Element |
Class |
Oral (µg/day) |
Parenteral (µg/day) |
Inhalation (µg/day) |
| Cd |
1 |
5 |
2 |
3 |
| Pb |
1 |
5 |
5 |
5 |
| As |
1 |
15 |
15 |
2 |
| Hg |
1 |
30 |
3 |
1 |
| Co |
2A |
50 |
5 |
3 |
| In |
2A |
100 |
10 |
1 |
| In |
2A |
200 |
20 |
6 |
| Tl |
2B |
8 |
8 |
8 |
| At |
2B |
300 |
300 |
3 |
| Pd, Ir, Os, Rh, Ru, Pt |
2B |
100 |
10 |
1 |
| With |
2B |
150 |
80 |
130 |
| At |
2B |
150 |
15 |
7 |
| That |
3 |
550 |
250 |
25 |
| Sb |
3 |
1200 |
90 |
20 |
| Not |
3 |
1400 |
700 |
300 |
| For |
3 |
3000 |
1500 |
10 |
| With |
3 |
3000 |
300 |
30 |
| Sn |
3 |
6000 |
600 |
60 |
| Cr |
3 |
11000 |
1100 |
3 |
Check your source on gold, silver and nickel. R2 corrected gold to 300/300/3 µg/day, silver’s parenteral value to 15, and nickel’s inhalation value to 6. Documents written before 2022 often still carry the old figures.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
He KED, or O₂ mass shift to m/z 91
|
|
|
|
|
|
|
|
He KED plus Mo correction, or 114Cd
|
Mercury shows severe memory and needs gold in the rinse and standards. Arsenic and selenium respond to residual organic carbon, so match carbon content between standards and samples. Internal standards must be matched by mass and ionisation energy — note that rhodium and iridium are themselves Q3D analytes.
Converting PDEs to concentration limits
PDEs are mass per day; suppliers and specifications speak in µg/g. Four routes across the gap:
-
Option 2b: component-specific limits summed as PDE ≥ Σ (Cₖ × Mₖ) — how a mined excipient at 3 µg/g arsenic, which fails Option 1 outright, stays in the formulation, offset by cleaner components.
Subtract contributions from equipment and the container closure system before allocating the remainder.
The risk assessment and the 30% control threshold
Four sources drive the assessment: intentionally added catalysts and reagents; unintentional presence in drug substance, water or excipients; manufacturing equipment; and the container closure system. Solid dosage forms are effectively exempt from the last; liquids and semi-solids are not, and the variables involved are the same ones driving an extractables and leachables programme.
The control threshold is 30% of the PDE; consistently below it, no further controls are needed. "Consistently" means three production-scale or six pilot-scale lots, with analytical and source variability factored in. One clean talc result is not a demonstration.
ICP-MS in practice
USP <233> and Ph. Eur. 2.4.20 carry the operational detail. ICP-MS is the default: sub-ppb detection, multi-element acquisition, and enough dynamic range to cover chromium at 11,000 µg/day and cadmium at 5 µg/day together.
Closed-vessel microwave digestion in nitric acid, usually with HCl and hydrogen peroxide, handles most matrices. HCl is not optional for the platinum group metals, gold, silver, mercury, antimony and tin. Silicates such as talc need hydrofluoric acid, and incomplete digestion of titanium dioxide is a common, quiet failure — monitor titanium as a completeness marker.
Interferences are where methods actually fail:
Validation anchors to J, the target-limit concentration in the prepared solution: 70–150% spike recovery at 0.5, 1.0 and 1.5 J; ≤20% RSD across six preparations at 1.0 J; ≤25% RSD across twelve for intermediate precision; specificity via a secondary isotope; LOQ at or below J.
Reference materials
Your data is only as traceable as the solution you calibrated against. Use an ISO 17034 accredited producer, with a certificate giving the certified value, expanded uncertainty, matrix and expiry. NIST 3100-series solutions are the usual SI anchor.
Q3D standards ship as several mixes because the elements are incompatible in one solution: silver precipitates as AgCl in chloride, the platinum group metals need HCl, mercury needs stabilising, antimony and tin hydrolyse in dilute nitric, osmium and ruthenium form volatile oxides. Combining at the point of use is fine; storing them combined is not.
Verify with an independently sourced second standard, and carry a spiked placebo through the full digestion rather than spiking the digest.
Pulling it together
Four parts, in order: a documented risk assessment; a concentration-limit strategy chosen deliberately; a validated ICP-MS procedure meeting USP <233>; and reference materials with real traceability. The analysis is the visible part — the risk assessment determines whether it was worth running.
