Elemental Impurities (ICH Q3D) for API Buyers

Elemental impurities under ICH Q3D create more confusion in procurement than almost any other quality requirement, and the confusion is usually the same one: buyers read Q3D as a testing mandate. It is not. ICH Q3D asks for a risk-based control strategy — a documented argument that every elemental impurity of concern in your finished drug product stays below its permitted daily exposure, supported by whatever mix of process knowledge, supplier data and analytical verification actually justifies the conclusion.

That distinction has real money attached. A control strategy built on route knowledge and periodic verification costs a fraction of a full panel run on every lot, and it survives an inspection better, because it demonstrates that you understand your process. A specification that simply says “heavy metals: conforms” demonstrates nothing at all, and the old compendial heavy metals colorimetric test it descends from was retired precisely because it did not measure what most people assumed it measured.

If you buy APIs or advanced intermediates, most of your Q3D exposure sits in one place: the transition metal catalysts your supplier deliberately used to make the molecule. That is the part buyers most often miss, and it is where this guide spends the most time.

What ICH Q3D Actually Requires

ICH Q3D requires a documented, risk-based evaluation of elemental impurities in the drug product, with controls sufficient to keep each element below its permitted daily exposure. It does not prescribe a test list. The guideline is explicit that the assessment is the deliverable and that testing is one of several possible controls.

The Assessment Is the Deliverable

The Q3D risk assessment is a written document, and it has three parts. First, identify: which elemental impurities could plausibly be present, from which sources. Second, analyze: what levels would be expected from each source, drawing on process knowledge, published data, supplier information and analytical results. Third, evaluate: compare the total expected level against the established permitted daily exposure and decide whether additional controls are needed.

Most first-time assessments fail on the second step. A team lists all twenty-four elements, tests a batch, finds everything below quantitation, and declares the job done. That is a data point, not an assessment. It says nothing about lot-to-lot variability, nothing about what happens if the supplier changes catalyst source, and nothing about why the levels are low. An assessor reading it learns that one batch was clean. An assessor reading a good assessment learns why every batch will be.

Scope Sits at the Drug Product, Responsibility Sits Everywhere

Q3D applies to the finished drug product. That is a genuine source of confusion for API buyers, because an API is not a drug product, and strictly the guideline does not impose a specification on it. In practice the drug product manufacturer cannot complete their assessment without knowing what is in the drug substance and in every excipient, so the requirement propagates upstream through supply agreements and quality agreements rather than through the guideline text.

This is why the request arrives at you as a purchasing question. Your customer needs a number and a justification; you need it from your supplier; your supplier needs it from whoever ran the catalytic step. Everyone in the chain is passing along an obligation that formally lands on the last party. Build it into the quality agreement rather than negotiating it lot by lot — the same discipline that belongs in any chemical supplier qualification checklist.

Where the Regulatory Text Lives

The primary reference is the ICH Q3D guideline on elemental impurities, which carries the classification system and the PDE tables. In the United States the compendial implementation is USP General Chapter <232> for the limits and <233> for the procedures, published by the United States Pharmacopeia. FDA’s own position is set out in its guidance for industry on elemental impurities in drug products. Read <233> in particular before you write a specification — it defines the validation expectations that your supplier’s data has to meet.

The Classification System, and What Each Class Means for You

Q3D sorts elements into four classes by toxicity and by the probability of their being present. The class determines whether you must assess an element always, conditionally, or only for certain routes of administration.

Class	Elements	When assessment is required
Class 1	As, Cd, Hg, Pb	Always, for all routes. Human toxicants with no legitimate use in manufacturing.
Class 2A	Co, Ni, V	Always, for all routes. Relatively high natural abundance makes presence probable.
Class 2B	Ag, Au, Ir, Os, Pd, Pt, Rh, Ru, Se, Tl	Only if intentionally added at any step. Low natural abundance, so absent unless you put them there.
Class 3	Ba, Cr, Cu, Li, Mo, Sb, Sn	Depends on route. Generally out of scope for oral unless intentionally added; assessed for parenteral and inhalation.

Class 1 and Class 2A: The Always List

Four Class 1 elements — arsenic, cadmium, mercury and lead — are assessed in every product regardless of dosage form or route of synthesis. They have no manufacturing role, so any presence is contamination, typically from mined excipients, water or equipment. Class 2A adds cobalt, nickel and vanadium, which are assessed always for a different reason: they are abundant enough in ores, metals and mineral excipients that presence is probable rather than exceptional. Nickel in particular shows up from stainless steel contact and from nickel-based hydrogenation catalysts.

Class 2B: The Catalyst Class

Class 2B is the one that matters most to anyone buying a synthetically complex API, because it contains the entire platinum group: palladium, platinum, rhodium, ruthenium, iridium and osmium, plus silver, gold, selenium and thallium. These elements have low natural abundance, so Q3D takes the sensible position that they need not be assessed unless someone intentionally added them.

Read that clause carefully. Intentional addition at any step in the manufacture of the drug product includes every step of the drug substance synthesis, including steps run by a supplier three tiers upstream. If your route uses a Suzuki coupling, palladium is in scope and you cannot argue it out. The exemption protects you from assessing iridium in a product nobody used iridium to make; it does not protect you from assessing palladium in a product built by cross-coupling.

Class 3: Route-Dependent

Class 3 elements have relatively low oral toxicity, and their permitted daily exposures for oral administration are high enough that oral products generally do not need them assessed unless they were intentionally added. For parenteral and inhalation products, where the PDEs drop by one or two orders of magnitude, they come into scope. Copper is the one to watch: it is Class 3, it is cheap, and it appears in Ullmann couplings, click chemistry and various oxidations, so the intentional-addition trigger fires more often than teams expect.

PDE Versus Concentration Limit, and the Three Conversion Options

A permitted daily exposure is a toxicologically derived limit expressed in micrograms per day. A concentration limit is expressed in micrograms per gram. Converting between them requires an assumption about daily dose, and Q3D gives you three ways to make that assumption.

Why the Distinction Trips People Up

The PDE is fixed by toxicology and by route of administration. It does not change with your product. The concentration limit is derived, and it changes with the daily dose of the product. A given oral PDE in micrograms per day means one concentration limit for a 250 milligram tablet taken once daily and a very different one for a product dosed at four grams a day. Two suppliers quoting five parts per million palladium are making claims of different stringency if the products differ in dose. Always convert to micrograms per day before comparing.

The Three Options

Option 1 — fixed concentration across all components. Assume a maximum daily intake of ten grams of drug product and divide the PDE by ten to get a concentration in micrograms per gram that applies uniformly to every component. It is the simplest option and the most conservative for low-dose products. If every component meets the Option 1 concentration, the finished product passes regardless of formulation arithmetic. Its weakness is exactly its simplicity: a component may fail Option 1 while the product is comfortably compliant.

Option 2a — sum of components at actual daily dose. Use the real maximum daily dose rather than the ten gram assumption, and require that the sum of contributions from all components stays within the PDE. This is more permissive than Option 1 for any product dosed below ten grams a day, which is most of them, and it is the option most oral solid dose products land on.

Option 2b — individual component limits. Assign each component its own limit, allocating the PDE across components according to what each actually contributes. Useful when one component legitimately carries more of an element than the others — a mined excipient carrying arsenic, or the drug substance carrying residual palladium. It requires more arithmetic and more justification, but it avoids penalizing every excipient for one component’s profile.

Option 3 — finished product analysis. Test the drug product directly against the PDE divided by the maximum daily dose. Conceptually the cleanest, since it measures what the patient actually receives. Practically it is the least useful for a buyer, because it gives you no early warning: you find the problem after the batch is made, when the remedy is rejection rather than material selection.

Option	Basis	Best fit	Main limitation
Option 1	PDE divided by 10 g assumed daily intake	Low-dose products, simple portfolios	Over-conservative; can fail compliant components
Option 2a	Sum across components at actual daily dose	Most oral solid dose products	Needs full component data whenever formulation changes
Option 2b	Individual limit per component	One component dominates a given element	Higher justification burden per component
Option 3	Direct finished product testing	Verification and troubleshooting	No early warning; failure is found late

A practical note for buyers: choose the option before you write purchase specifications, not after. The option determines whether you need a tight number on the API or a tight number on the sum, and those drive very different supplier conversations.

Where Elemental Impurities Actually Come From

There are six realistic entry routes, and they are not equally likely. Ranking them correctly is what turns a generic assessment into a useful one.

The Six Sources

1. The drug substance. Usually the dominant source for synthetically complex molecules, almost entirely because of catalysts. Covered in detail below.
2. Excipients. The dominant source for simple, high-dose products. Mined and mineral-derived excipients — talc, titanium dioxide, calcium phosphate, magnesium stearate, certain iron oxides — carry Class 1 elements at variable levels because they come out of the ground. Suppliers increasingly provide Q3D data packages; ask for the underlying data range, not just a conformance statement.
3. Water. Generally minor for purified water and water for injection, but not zero, and it matters at high water-to-solid ratios in the process.
4. Manufacturing equipment. Stainless steel contact contributes chromium, nickel, iron and molybdenum, with the contribution rising under acidic conditions, at elevated temperature and with long contact time. New or newly passivated equipment leaches more than seasoned equipment. Glass-lined reactors shift the profile rather than eliminating it.
5. Container closure systems. More significant for liquids and semi-solids than for solid oral dosage forms. Elastomeric closures, glass and certain colorants are the usual contributors, and the relevant data normally comes from extractables and leachables work rather than from Q3D testing.
6. Catalysts and reagents deliberately used in synthesis. The one buyers most often miss, and the one most likely to produce a finding.

Why Catalysts Dominate the Risk

Every other source on that list is contamination, present at trace level by accident. A catalyst is present by design, at a level thousands of times above what Q3D permits in the final product, and the entire question is how much of it was removed.

Consider the arithmetic. A palladium-catalyzed cross-coupling might charge one to five mole percent palladium relative to the limiting substrate. On a molecule of moderate molecular weight that is roughly one to five thousand parts per million of palladium in the reaction mixture. The Q3D oral PDE for palladium is 100 micrograms per day, which for a typical dose translates to a permitted concentration in the low parts-per-million range on the drug substance. The process therefore has to achieve a purge of roughly a thousand-fold, sometimes more, and it has to achieve it reproducibly.

That is entirely achievable — routine, even — but it is achieved by design, and the design has to be documented. A supplier who cannot tell you their catalyst loading and their measured purge is telling you they have not characterized this.

The Chemistries That Carry the Highest Q3D Risk

Cross-coupling is the largest single contributor in modern pharmaceutical synthesis. Suzuki-Miyaura, Buchwald-Hartwig, Negishi, Sonogashira and Heck couplings all use palladium, and coupling steps often sit late in a route where fewer downstream purifications remain to purge the metal. Boronic acid couplings in particular are ubiquitous — reagents such as 4-Bromophenylboronic acid (CAS 5467-74-3) and 2-Pyridineboronic acid (CAS 197958-29-5) appear in an enormous fraction of current medicinal chemistry routes, and every one of those couplings puts palladium into the process. Our boronic acids category covers the range; the Q3D consequence is the same across it. The synthesis background sits in our post on boronic ester synthesis in drug discovery.

The halide partner matters too, for reasons that are more kinetic than regulatory. Aryl iodides such as 3-Iodopyridine (CAS 1120-90-7) undergo oxidative addition readily and often permit lower catalyst loadings than the corresponding chlorides, which directly reduces the purge burden. Bromides such as 4-Bromoindole (CAS 52488-36-5) sit in between and are the common commercial compromise. When you are choosing between building blocks from the halogenated compounds category, catalyst loading is a legitimate selection criterion and not only a cost one — it is one of the few places where a procurement decision directly improves a regulatory position.

Hydrogenation contributes palladium from palladium on carbon, platinum from platinum oxide, rhodium from Wilkinson-type and asymmetric catalysts, and nickel from Raney nickel. Heterogeneous catalysts are generally easier to remove by filtration, but leaching is real, particularly with chelating substrates — pyridines, thiols, phosphines and amines all bind metal and pull it into solution. Asymmetric hydrogenation is the harder case, because homogeneous rhodium and ruthenium complexes stay dissolved by definition.

Metathesis uses ruthenium in homogeneous form, and ruthenium removal is notoriously difficult. Routes with a ring-closing metathesis step frequently need a dedicated scavenging operation rather than relying on crystallization alone.

Ullmann and Chan-Lam chemistry uses copper, usually at higher loadings than palladium chemistry. Copper is Class 3 with a comparatively generous oral PDE, but it is intentionally added, so it needs assessing, and for parenteral products the limit tightens considerably.

Route selection is therefore a Q3D decision as much as a cost and yield decision, which is one reason process development and regulatory should share a room early. Our post on process chemistry optimization from lab to pilot covers the broader trade-offs, and fluorination chemistry routes and safety covers a parallel case where route choice drives a different compliance burden entirely.

Purge and Scavenging: What Actually Removes Metal

Metal removal is achieved by a combination of physical separation, chemical scavenging and crystallization, and realistic performance depends heavily on the metal, its oxidation state and the substrate.

Functionalized Scavengers

Silica or polymer resins bearing thiol, thiourea, triamine or phosphine functionality bind soft metals selectively and are removed by filtration. They are the workhorse for palladium and ruthenium, typically applied as a slurry treatment in solution at moderate temperature for one to several hours.

Realistic performance: a single well-matched scavenger treatment commonly takes palladium from hundreds of parts per million to single-digit or sub-part-per-million levels, a one to two log reduction and occasionally better. Two sequential treatments generally outperform one large charge. Selection is empirical; thiol resins suit palladium in a reduced state, thiourea and triamine types suit different oxidation states, and screening a small panel is faster than reasoning from first principles. Cost is not trivial at scale, and scavenger loading is a real line item in a multi-kilogram campaign.

Crystallization and Recrystallization

Often the most cost-effective single operation, because you are usually crystallizing anyway. A well-designed crystallization commonly delivers a five to twenty-fold reduction in metal content per pass, since metal complexes generally stay in the mother liquor. The catch is that it is substrate-specific and unreliable when the API itself chelates. A recrystallization that purges palladium beautifully from a neutral amide may barely move it for an aminopyridine.

Activated Carbon

Cheap, broadly applicable and modest in performance. Treatment typically delivers a two to ten-fold reduction for palladium and is less effective for ruthenium. Carbon also adsorbs product, so a yield loss of a few percent is normal and must be priced in. It is best used as a polishing step in combination with something else, not as the primary control.

Extraction and Chelating Washes

Aqueous cysteine, N-acetylcysteine, thiourea or EDTA washes can pull metal into an aqueous phase efficiently when the product tolerates them. Performance varies widely with pH and with metal oxidation state. Inexpensive at scale when it works.

Technique	Typical reduction per operation	Best for	Practical caveat
Functionalized scavenger resin	10x to 100x	Pd, Ru, Rh in solution	Cost at scale; requires screening
Crystallization or recrystallization	5x to 20x	Most metals, if the product is crystalline	Fails when the product chelates
Activated carbon	2x to 10x	Pd polishing	Product adsorption and yield loss
Chelating aqueous wash	5x to 50x	Pd, Cu, when the product is stable	Sensitive to pH and oxidation state

These ranges are indicative of typical process chemistry outcomes and vary substantially with substrate and conditions. Treat them as a starting point for design of experiments, not as a specification.

What a Purge Argument Looks Like

The regulatory argument you want from a supplier reads roughly like this: catalyst charged at X mole percent, measured level after the reaction Y parts per million, after scavenger treatment Z parts per million, after crystallization W parts per million, across N batches with the range reported. That establishes a purge factor with margin, and it is what allows skip-lot testing or elimination of routine testing at the drug product stage. A single passing result on one batch establishes nothing except that one batch passed.

ICP-MS Versus ICP-OES: Choosing the Right Technique

Both techniques use an inductively coupled argon plasma to atomize and excite the sample; they differ in how they detect what the plasma produces. ICP-OES measures light emitted at element-characteristic wavelengths. ICP-MS separates ions by mass-to-charge ratio and counts them.

The Practical Difference Is Detection Limit

ICP-OES typically reaches detection limits in the low parts-per-billion range in solution, which after accounting for the 100 to 500-fold dilution that digestion normally involves corresponds to roughly sub-part-per-million to low-parts-per-million in the original solid. That is comfortably sufficient for catalyst residues, for Class 2A elements and for most oral products, where limits sit in the parts-per-million range.

ICP-MS reaches parts-per-trillion in solution, two to three orders of magnitude lower. You need that when the limit on the solid is very low: Class 1 elements in a high-dose product, parenteral or inhalation products where PDEs are much tighter, or any case where the required dilution pushes the analyte close to the ICP-OES detection floor. USP <233> permits either technique provided it meets the validation requirements for the specific limits being measured — the chapter is performance-based rather than technique-prescriptive.

	ICP-OES	ICP-MS
Typical solution detection limit	Low parts per billion	Parts per trillion
Best fit	Catalyst residues, Class 2A, oral products at typical doses	Class 1 at low dose, parenteral, inhalation
Interference profile	Spectral overlap between emission lines	Polyatomic and isobaric mass interferences
Relative cost per sample	Lower	Higher
Tolerance of dissolved solids	Higher	Lower; more dilution required

ChemContract operates ICP-OES within our analytical services group, which covers the great majority of catalyst residue and Class 2A work for oral products. Where a specification demands parts-per-trillion sensitivity, ICP-MS is the appropriate industry technique and the work should go to a laboratory equipped for it. Be direct with any laboratory about which limits you actually need to demonstrate — the answer determines the technique, and a laboratory that does not ask is not thinking about your problem. Our guide to outsourcing analytical testing in pharma covers how to structure that conversation.

Sample Preparation Is Where Results Are Won or Lost

Closed-vessel microwave digestion is the standard preparation. The sample is weighed into a PTFE or quartz vessel, treated with concentrated nitric acid — often with hydrochloric acid, hydrogen peroxide or hydrofluoric acid depending on the matrix — and heated under pressure until the organic matrix is destroyed and all metals are in solution.

Two failure modes account for most bad data. The first is incomplete digestion, which leaves metal locked in undissolved residue and produces a falsely low result; this is the classic failure with refractory palladium species and with silica-containing matrices. The second is volatilization loss during open digestion, which affects mercury most severely and also arsenic and selenium — the reason closed-vessel digestion is standard rather than optional. Aqua regia or added hydrochloric acid is generally necessary for platinum group metals, which nitric acid alone does not reliably dissolve.

The Validation a Buyer Should Require

USP <233> defines the performance criteria. For a limit test the requirement is demonstrated detectability at the target limit. For a quantitative procedure, accuracy, precision and specificity must be demonstrated, and the practical evidence is spike recovery.

Ask specifically for spike recovery data in the actual sample matrix, at levels bracketing the specification — typically at 50 percent, 100 percent and 150 percent of the limit. USP <233> sets an acceptance range of 70 to 150 percent recovery for each element at each level. Recovery outside that range in your matrix, even with a perfect calibration curve, means the method has not been shown to work on your material. A supplier who provides a certificate of analysis with a palladium number but no matrix spike recovery has given you a number of unknown reliability. Further detail on validation expectations sits in our guide to analytical testing methods for pharma buyers.

What to Demand From an API or Intermediate Supplier

Ask for a risk assessment, not a certificate line. The single most useful thing a buyer can do is replace “provide heavy metals result” in the specification with a short list of documents that let you build your own assessment.

The Request List

1. A written elemental impurity risk assessment for the material. It should name the elements considered, state the rationale for those excluded, and identify entry points. One to three pages is normal. A supplier who has done this work can produce it in days.
2. Catalyst identity and loading for every catalytic step. Which metal, which ligand or support, what mole percent charge. Suppliers sometimes resist this as proprietary; a reasonable compromise is metal identity and a loading range without ligand detail, since the metal identity is what Q3D turns on.
3. Purge data across multiple batches. Measured levels at the relevant process points, with a batch range rather than a single figure. Three or more batches is a meaningful dataset; one is an anecdote.
4. The analytical method and its validation summary. Technique, digestion procedure, and spike recovery in the actual matrix at the specification level.
5. A change notification commitment. Written agreement to notify you before changing catalyst, catalyst supplier, scavenger, or any purification step in the last two stages. This is the clause that protects the assessment over time, and it belongs in the quality agreement. Our guide to cGMP contract manufacturing requirements covers where these commitments sit contractually.
6. Equipment train information for the final stages. Materials of construction for the last isolation and drying steps, sufficient to reason about any stainless steel contribution.

Red Flags in a Supplier Response

• A certificate of analysis reporting only “heavy metals: conforms” against the retired colorimetric test. That test is not Q3D-compatible and tells you nothing element-specific.
• A palladium result reported as not detected with no stated limit of quantitation. Not detected at what level, by what method.
• Refusal to name the catalyst metal on confidentiality grounds. Metal identity is not a route disclosure, and you cannot complete a Q3D assessment without it.
• Data from a single batch presented as a process capability claim.
• A method with no matrix spike recovery, or spike recovery run in a clean solvent rather than in the digested product.

Building the Requirement Into Sourcing

Handle this at qualification rather than at purchase order and the cost drops sharply. Add the six documents above to your supplier qualification package, treat the answers as a scored criterion, and you will discover which suppliers actually understand their process before you are dependent on them. Teams planning ahead will find this fits naturally into a 2027 chemical procurement budget framework, where analytical and qualification spend needs its own line item rather than being absorbed as an unplanned cost mid-program.

How Q3D Interacts With Q3A, Q3B and M7

Q3D is one of three parallel impurity frameworks, each covering a different impurity family under a shared risk-based logic. They are complementary rather than overlapping, and a complete control strategy addresses all three explicitly.

Framework	Impurity family	Basis for limits	Typical control
ICH Q3A and Q3B	Organic impurities and degradants in drug substance and product	Reporting, identification and qualification thresholds tied to daily dose	HPLC or GC specification, stability program
ICH M7	Mutagenic and potentially mutagenic impurities	Acceptable intake from carcinogenic potency, with a threshold of toxicological concern default	Purge argument, upstream control, or specification
ICH Q3D	Elemental impurities	Toxicologically derived permitted daily exposure by route	Risk assessment plus verification testing

The Shared Logic Is Purge

M7 and Q3D converge on the same practical argument: rather than testing the final API for something you know entered upstream, demonstrate that the process removes it, and control it at the stage where control is meaningful. M7 formalizes this through purge factor calculation for mutagenic impurities; Q3D reaches the same conclusion for metals through its risk assessment. A team that has built a good M7 purge argument already has the mental model for Q3D, and often the same process data.

Where They Genuinely Overlap

Rarely, but it happens. Some organometallic reagents and some metal salts raise both an elemental question under Q3D and a genotoxic question under M7 for the organic ligand or counterion. Hexavalent chromium is the classic case where speciation matters and total elemental analysis by ICP does not answer the toxicological question — a useful reminder that Q3D measures total element rather than species, and that where speciation drives toxicity you need a separate argument.

For organic impurities, the companion work is impurity profiling and forced degradation. Our guide to impurity profiling and forced degradation covers the Q3A and Q3B side of the same control strategy, and the broader supplier-risk context sits in pharmaceutical supply chain de-risking.

A Practical Sequence for Buyers

If you are starting from nothing, work in this order. It produces a defensible assessment fastest and spends analytical budget last.

1. Establish the dosage form and the maximum daily dose. Everything downstream depends on these two numbers.
2. Choose your conversion option. Option 1 for simple low-dose portfolios, Option 2a for most oral products, Option 2b where one component dominates a given element. Write down which and why.
3. List every element in scope. Class 1 and Class 2A always. Class 2B for every metal intentionally used anywhere in the drug substance route. Class 3 by route of administration and intentional addition.
4. Ask every supplier the catalyst question in writing. Which metals, what loading, what purge data. The responses will sort your supply base quickly.
5. Gather excipient data packages. Most established excipient suppliers now provide them; request the underlying range rather than a conformance statement.
6. Test to fill the gaps, not to cover the list. Run confirmatory analysis where the risk assessment leaves genuine uncertainty. This is a much shorter list than the full panel, and it is defensible precisely because you can explain the selection.
7. Write the assessment and put it under change control. Revisit it whenever a supplier changes, a route changes, or a dose changes.

Applied consistently, this converts Q3D from a recurring testing bill into a one-time characterization effort with periodic verification, which is what the guideline intended in the first place.

ChemContract Research provides analytical services including ICP-OES for elemental analysis, alongside HPLC, GC, NMR, LC-MS and Karl Fischer, plus US-based custom synthesis from milligram to multi-ton and contract R&D for route development and metal purge optimization. If you are assessing a catalytic route for Q3D exposure, or need scavenging screened before a scale-up campaign, send us the route and the specification and we will return a scope and a quote within 24 hours.

Frequently Asked Questions

What does ICH Q3D actually require?

Q3D requires a risk-based control strategy demonstrating that elemental impurities in a drug product stay below their permitted daily exposure. It requires a documented assessment of every potential source, including drug substance, excipients, water, equipment and container closure. It does not require routine testing of every element in every lot where risk assessment justifies otherwise.

Which elements must always be assessed?

Class 1 elements arsenic, cadmium, mercury and lead must always be assessed because they are human toxicants with no legitimate manufacturing use. Class 2A elements cobalt, nickel and vanadium must also always be assessed because their natural abundance makes their presence probable. Class 2B and Class 3 elements are assessed conditionally, based on route and dosage form.

What is a PDE and how does it become a specification?

The permitted daily exposure is a toxicologically derived limit in micrograms per day for a given element and route of administration. Q3D gives three ways to convert it into a component specification: Option 1 applies a fixed concentration assuming ten grams of daily intake, Option 2 scales to the actual daily dose either summed across components or set per component, and Option 3 tests the finished product directly.

Why are catalysts the biggest Q3D risk in an API?

Because they are added deliberately and in quantity. A cross-coupling might charge one to five mole percent palladium, which is orders of magnitude above the residual level Q3D permits. Every microgram of palladium in the final API traces back to a deliberate charge, so the control strategy has to demonstrate purge rather than absence of contamination.

Is ICP-OES sufficient or do I need ICP-MS?

ICP-OES is sufficient for most catalyst residue and Class 2A work where limits sit in the parts-per-million range and sample dilution is modest. ICP-MS becomes necessary when detection limits at parts-per-billion are required, typically for Class 1 elements at low daily doses, parenteral products, or where matrix dilution pushes concentrations near the ICP-OES detection floor.

How does Q3D relate to ICH Q3A and M7?

They cover different impurity families under a common risk logic. Q3A and Q3B govern organic impurities and degradants with identification and qualification thresholds. M7 governs mutagenic impurities with acceptable intakes derived from carcinogenic potency. Q3D governs elemental impurities with toxicologically derived permitted daily exposures. A complete impurity control strategy addresses all three.