Residual solvents analysis by GC headspace is the standard way to show that the organic volatile chemicals left behind from synthesis, purification or formulation stay below the limits set by ICH Q3C and USP <467>. The sample is sealed in a vial, heated until the volatiles partition into the gas phase, and an aliquot of that headspace is injected onto a gas chromatograph, usually with flame ionisation detection (FID). Because only vapour reaches the column, non-volatile API and excipients never touch the inlet, which is why static headspace GC-FID has become the default technique for this test. This guide covers the solvent classes and limits, how to calculate a product-specific limit, how headspace methods are built and validated, and the pitfalls that most often cause out-of-specification or unreliable results.
What ICH Q3C and USP <467> require
ICH Q3C sets permitted daily exposures (PDEs) for residual solvents in drug substances, excipients and drug products. The current version is ICH Q3C(R9), which reached Step 4 on 24 January 2024. R9 was a minor revision to section 3.4 (analytical procedures): it adds that the impact of solvent volatility on the test method should be considered during validation. The previous revision, R8, added PDEs for 2-methyltetrahydrofuran, cyclopentyl methyl ether (CPME) and tertiary-butyl alcohol.
USP General Chapter <467> Residual Solvents applies these limits to products covered by USP or NF monographs and gives harmonised GC procedures. USP revised the chapter to align it with ICH Q3C(R9), with an official date of 1 August 2025. USP's own FAQ confirms that the General Notices allow an appropriately validated alternative method, so you do not have to use the compendial procedure verbatim. Section 3.4 of Q3C likewise says that harmonised pharmacopoeial procedures should be used where feasible, and that otherwise a suitable validated method may be chosen.
The three solvent classes
- Class 1 — solvents to be avoided: known or strongly suspected human carcinogens and environmental hazards. Use them only where their use can be strongly justified.
- Class 2 — solvents to be limited: non-genotoxic animal carcinogens, agents of other irreversible toxicity such as neurotoxicity or teratogenicity, or solvents suspected of other significant but reversible toxicities.
- Class 3 — solvents with low toxic potential: PDEs of 50 mg/day or more. Amounts up to 50 mg/day (5000 ppm, or 0.5%, under Option 1) are acceptable without justification.
Test only for the solvents that are actually used or produced in the manufacture of the drug substance, the excipients and the drug product. Calculate each solvent's limit from its PDE and the real maximum daily dose. Then make sure your headspace method is validated for the volatility and matrix of your sample, not just for a standard mixture.
Limits and how to calculate them
The table below lists ICH Q3C(R9) limits for solvents that come up most often in API synthesis. The concentration limits assume Option 1, which is explained after the table.
| Solvent | Class | PDE (mg/day) | Option 1 limit (ppm) |
|---|---|---|---|
| Benzene | 1 | — | 2 |
| Carbon tetrachloride | 1 | — | 4 |
| 1,2-Dichloroethane | 1 | — | 5 |
| 1,1-Dichloroethene | 1 | — | 8 |
| 1,1,1-Trichloroethane | 1 | — | 1500 |
| Chloroform | 2 | 0.6 | 60 |
| Pyridine | 2 | 2.0 | 200 |
| Hexane | 2 | 2.9 | 290 |
| Acetonitrile | 2 | 4.1 | 410 |
| N-Methylpyrrolidone | 2 | 5.3 | 530 |
| Dichloromethane | 2 | 6.0 | 600 |
| Tetrahydrofuran | 2 | 7.2 | 720 |
| N,N-Dimethylformamide | 2 | 8.8 | 880 |
| Toluene | 2 | 8.9 | 890 |
| Methanol | 2 | 30.0 | 3000 |
| Ethanol, acetone, ethyl acetate, 2-propanol, heptane, DMSO | 3 | 50 or more | 5000 |
Option 1: the default concentration limit
Option 1 converts the PDE into a concentration by assuming 10 g of product is taken each day:
Concentration (ppm) = (1000 × PDE) / dose, where PDE is in mg/day and dose is in g/day.
With a 10 g dose, methanol's PDE of 30.0 mg/day gives 3000 ppm. If every component of a drug product meets its Option 1 limit, the product passes without further calculation, whatever its daily dose.
Option 2: using the real daily dose
When the maximum daily dose is known and smaller than 10 g, you can use the same formula with the actual dose. For a drug substance taken at up to 2 g/day, the acetonitrile limit becomes 1000 × 4.1 / 2 = 2050 ppm. Option 2 can also be applied to a finished product by adding up the solvent contributed by each component. ICH Q3C works through this example:
- Drug substance: 0.3 g at 800 ppm acetonitrile = 0.24 mg
- Excipient 1: 0.9 g at 400 ppm = 0.36 mg
- Excipient 2: 3.8 g at 800 ppm = 3.04 mg
- Total for a 5.0 g daily dose: 3.64 mg, which is below the 4.1 mg/day PDE, so the product conforms
Both the drug substance and Excipient 2 are above the 410 ppm Option 1 limit, yet the product still complies. Option 2 also requires that the solvent has been reduced to the practical minimum.
Class 3 and loss on drying
If only Class 3 solvents are present, Q3C permits a non-specific test such as loss on drying (LOD), with a limit of 0.5%. USP's FAQ adds that LOD is not appropriate when the Class 3 solvent content exceeds 0.5%. Above that level, or whenever Class 1 or Class 2 solvents are present, you need a specific chromatographic method.
How static headspace GC works
In static headspace, a known mass of sample is dissolved or suspended in a diluent inside a sealed vial (typically 10 or 20 mL). The vial is held at a set temperature until each solvent reaches equilibrium between the liquid phase and the gas phase above it. The concentration in the gas phase depends on two terms:
CG = C0 / (K + β)
Here C0 is the original concentration in the sample solution, K is the partition coefficient (concentration in the liquid ÷ concentration in the gas), and β is the phase ratio (gas volume ÷ liquid volume). Solvents with a small K, such as hexane or dichloromethane in water, enter the headspace readily. Polar solvents with a large K, such as methanol, ethanol or DMSO in water, stay mostly in the liquid, so sensitivity depends more on temperature and on the choice of diluent.
The pharmacopoeial procedures
- Procedure A: headspace GC-FID on a G43 (6% cyanopropylphenyl–94% dimethylpolysiloxane, "624-type") column. This is a limit test used to identify and screen solvents.
- Procedure B: the same approach on a G16 (polyethylene glycol, "WAX-type") column. It gives an orthogonal selectivity and is used to confirm peaks found by Procedure A, or when Procedure A gives co-elution.
- Procedure C: a quantitative test against a spiked or reference solution for solvents found above the limit by A or B.
USP describes A and B as limit tests and C as the quantitative test.
Developing a headspace GC method
A compendial screen is a useful starting point. However, a product-specific method is usually faster, more sensitive and easier to defend. Our gas chromatography page explains the underlying separation. The main decisions are listed below.
- Define the solvent list. Take it from the route of synthesis, the final crystallisation, cleaning solvents and excipient supplier declarations. Add any solvents that can form in the process, such as methanol from methyl ester hydrolysis.
- Choose the diluent. Water suits water-soluble salts. DMSO, DMF, N,N-dimethylacetamide (DMA) or 1,3-dimethyl-2-imidazolidinone (DMI) suit lipophilic APIs. Use headspace-grade diluent and check that it is free of interfering peaks. Avoid a diluent that is also on your solvent list.
- Set the sample amount and phase ratio. Typically 100–500 mg of sample in 1–5 mL of diluent. More sample improves sensitivity but can change matrix behaviour or exceed solubility.
- Optimise equilibration. Common starting conditions are 80–100 °C for 20–45 minutes, with aqueous diluents kept at the lower end of that range, safely below their boiling point. Raise the temperature or add salt to help high-K solvents, and confirm by experiment that equilibrium is actually reached.
- Choose the column and oven programme. A 624-type phase separates most Class 1 and Class 2 solvents. A WAX phase helps with alcohols, and a thicker film improves the retention of very light solvents.
- Choose the detector. FID gives a wide linear range and robust quantitation. Use GC-MS to identify unknown peaks or confirm co-eluting solvents. Our post on LC-MS vs GC-MS covers when mass detection is worth the extra cost.
High-boiling solvents
DMSO, DMF, NMP, DMA, ethylene glycol and similar solvents have high boiling points and partition poorly into the headspace, especially from DMSO or DMA diluents. You may need higher equilibration temperatures, a different diluent, or direct liquid injection GC with a suitable liner. Plan for this early rather than discovering low recovery during validation.
Validating the method
Q3C asks for validation in line with ICH Q2, and R9 now explicitly asks for solvent volatility to be considered. The level of validation depends on whether the method is a limit test or a quantitative test. Our ICH Q2(R2) validation guide covers the general framework. The table below shows how it is usually applied to headspace GC. The acceptance criteria are common in-house practice, not regulatory requirements.
| Parameter | What to show | Typical in-house criterion |
|---|---|---|
| Specificity | Each solvent is resolved from the others, from diluent peaks and from matrix volatiles | Resolution of at least 1.5 for critical pairs; blank free of interference |
| Detection or quantitation limit | Reliable response well below the specification limit | Quantitation limit at or below 10–30% of the limit, with S/N of at least 10 |
| Linearity and range | Proportional response from the quantitation limit to above the limit | r of at least 0.99 from the quantitation limit to 120–150% of the limit |
| Accuracy | Recovery of spikes in the actual sample matrix | 80–120% at limit-level spikes |
| Precision | Repeatability of six preparations at the limit | RSD of 10–15% or less at the limit |
| Robustness | Effect of equilibration time and temperature, sample mass and flow | Results within the precision range |
| Solution stability | No loss from vials while they wait in the autosampler | Recovery within accuracy limits over the run time |
For a compendial procedure used as written, perform a verification under USP <1467> rather than a full validation. At minimum, demonstrate specificity, sensitivity at the limit and system suitability in your matrix.
Common pitfalls and how to avoid them
- Losing volatile solvents. Weighing slowly, using open transfers or sealing vials poorly can lose dichloromethane, hexane or methanol before analysis. Crimp vials immediately, use fresh septa and check seals with a pressure test.
- Matrix effects. The API or its salt counter-ion can change partitioning, so a standard prepared in pure diluent does not match the sample. Use standard addition, or matrix-matched standards, when spike recovery is outside range.
- Diluent artefacts. Heated DMSO can produce dimethyl sulfide, and old DMF can contain dimethylamine. Run diluent blanks with every sequence.
- Co-elution. On a 624-type column some solvent pairs elute close together. Confirm suspicious peaks on a WAX column or by GC-MS before you report a failure.
- Thermal degradation. Some APIs release volatiles when heated, for example formaldehyde, acetaldehyde or alcohols from labile esters. Compare different equilibration temperatures to separate genuine residue from artefact.
- A solvent list that does not match the process. A route change, a new supplier or a reprocessing step can introduce a solvent that the method does not cover. Link the residual solvent specification to change control.
How SDS Labs can help
SDS Labs is an independent analytical testing laboratory in MIDC Mahape, Navi Mumbai, holding ISO/IEC 17025:2017 and ISO 9001:2015. Results are generated in line with 21 CFR Part 11 and cGMP requirements. Our analytical services include method development, validation and routine testing for raw materials, intermediates, APIs and finished formulations. These are supported by gas chromatography with FID, NPD and TCD detectors, and by GC-MS with EI and CI ionisation.
To discuss a residual solvent method or a set of samples, send your solvent list, route summary and specification through our contact page, call +91 90295 20144 or email info@sdslabs.in.
Frequently asked questions
What is the difference between Class 1, Class 2 and Class 3 residual solvents?
Under ICH Q3C, Class 1 solvents such as benzene and carbon tetrachloride are known or suspected human carcinogens or environmental hazards and should be avoided. Class 2 solvents such as acetonitrile, methanol and toluene have irreversible or significant toxicities and are limited by a PDE. Class 3 solvents such as ethanol and acetone have low toxic potential, with PDEs of 50 mg/day or more.
How do you calculate the residual solvent limit for a drug product?
Option 1 uses concentration (ppm) = 1000 × PDE (mg/day) ÷ dose (g/day) with an assumed dose of 10 g/day, giving, for example, 410 ppm for acetonitrile. Option 2 uses the actual maximum daily dose, or sums the solvent contributed by each component, and requires the total daily intake to stay below the PDE.
Why is headspace GC used for residual solvents instead of direct injection?
Static headspace injects only the vapour above a heated sample solution, so non-volatile API, salts and excipients never reach the inlet or column. This keeps the system clean, reduces matrix interference and gives good sensitivity for volatile solvents. Direct injection is still useful for high-boiling solvents such as DMSO, NMP or ethylene glycol that partition poorly into the headspace.
Can loss on drying replace GC for residual solvents?
Only in a narrow case. ICH Q3C allows a non-specific test such as loss on drying when only Class 3 solvents are present, with a limit of 0.5%. USP notes that loss on drying is not appropriate if the Class 3 solvent content exceeds 0.5%. Any Class 1 or Class 2 solvent requires a specific, validated chromatographic method.
What changed in ICH Q3C(R9)?
ICH Q3C(R9), adopted at Step 4 on 24 January 2024, was a minor revision to section 3.4 on analytical procedures. It asks that the impact of solvent volatility on the test method be considered during validation. The earlier R8 revision added PDEs for 2-methyltetrahydrofuran, cyclopentyl methyl ether and tertiary-butyl alcohol.