From 1 kg to Commercial Scale: Why Small-Molecule API CDMO Projects Fail During Scale-Up

Many small-molecule API scale-up problems arise not because the target molecule cannot be synthesized, but because the laboratory route cannot be converted into a safe, reproducible, impurity-controlled and economically viable GMP process.

That distinction matters. A medicinal chemistry route can deliver grams or kilograms of API with excellent chromatographic purity and still be a weak candidate for clinical or commercial supply. Manufacturing suitability depends on the operating window, impurity fate, physical properties, equipment fit, process safety, analytical control, raw-material strategy and documentation. Yield matters, but yield is only one part of manufacturability.

This article focuses on chemically synthesized small-molecule APIs. It does not address biologics, antibody-drug conjugates, peptides, oligonucleotides, cell therapies or finished-dose manufacturing except where the boundary is useful.

What “Scale-Up” Actually Means

Scale-up is often described as a move from 1 kg to 10 kg, 100 kg or commercial batches. In practice, the more important change is the objective of the work.

Early synthesis proves that the molecule can be made. Clinical GMP manufacture must supply material with appropriate controls for human studies. Late-stage development must show that the route, impurity profile, analytical methods and unit operations can support pivotal and commercial expectations. Commercial manufacture must run repeatedly under an approved control strategy, with defined responsibilities, change control and supply continuity.

ICH Q7 describes API manufacturing as covering receipt of materials, production, packaging, relabeling, quality control, release, storage and distribution, and explains that GMP expectations become more stringent as the process proceeds toward final API steps (FDA, Q7A Good Manufacturing Practice Guidance for Active Pharmaceutical Ingredients). For clinical-trial APIs, Q7 also recognizes that process and test procedures may remain flexible as process knowledge increases.

Commercial scale is not a fixed 1,000 kg target. A highly potent oncology API, a chronic high-dose medicine and an orphan-drug API may require very different annual quantities. “Commercial” means the process can support the product’s approved quality, regulatory and supply commitments at the demand level that actually applies.

FDA’s process validation guidance frames validation as a lifecycle activity that includes process design, process qualification and continued process verification (FDA, Process Validation: General Principles and Practices). A sound small-molecule API CDMO scale-up program should therefore retire defined risks at each stage, not simply increase batch size.

Why the Laboratory Route May Be the Wrong Manufacturing Route

A laboratory synthesis is often optimized for speed, available reagents and proof of structure. A manufacturing route must be optimized for controlled repetition.

Several laboratory conveniences become liabilities at CDMO scale. Column chromatography may be practical for grams but unsuitable when it drives large solvent volumes, long cycle times and difficult waste handling. Extreme dilution can suppress side reactions in the lab while destroying plant productivity. Very low-temperature chemistry may fit a laboratory reactor but exceed a plant’s cooling capacity or lengthen addition times enough to change impurity formation.

Hazardous reagents, unstable intermediates, expensive ligands, air-sensitive catalysts and repeated solvent swaps may all work in a development notebook. They still may be poor choices if they create safety risk, high cost, long cycle time, unreliable purification or difficult regulatory justification.

Process metrics help reveal this difference. The ACS Green Chemistry Institute Pharmaceutical Roundtable defines process mass intensity as total input mass divided by bulk API output mass, including water, and uses it to benchmark resource efficiency in API synthesis (ACS Green Chemistry Institute, Process Mass Intensity Calculation Tool). A high-yield step run in very high solvent volumes with chromatography may be less attractive than a lower-yielding step that crystallizes cleanly and purges impurities predictably.

Solvent choice is also a manufacturing decision. ACS GCI’s solvent-selection guidance includes reaction performance, isolation, crystallization, safety, operability, recycling, waste treatment, environmental compliance and total cost burden as considerations (ACS GCI Pharmaceutical Roundtable, Solvent Selection). A solvent is not only a reaction medium. It affects plant fit, fire risk, residual solvent control, crystallization, drying and waste treatment.

Route selection should therefore consider the complete process: starting-material strategy, telescoping potential, impurity rejection, solvent recovery, isolation, equipment needs, safety margins and expected regulatory change burden. Individual-step yield is useful information, but it is not a route-selection criterion by itself.

What Physically Changes During Scale-Up

A 10-liter process cannot be reproduced in a 10,000-liter reactor by multiplying every charge by 1,000. Chemistry and transport phenomena do not scale in the same way.

A review of scalable reactor design for pharmaceuticals and fine chemicals explains the underlying problem: chemical rate constants are scale-independent, while physical parameters and phenomena are not (Graham Caygill, M. Zanfir and A. Gavriilidis, Scalable Reactor Design for Pharmaceuticals and Fine Chemicals Production. 1: Potential Scale-up Obstacles, OPRD, 2006). Heat transfer, mixing, gas-liquid mass transfer, suspension behavior and discharge are all equipment- and scale-dependent.

For a geometrically similar vessel, volume increases roughly with length cubed, while heat-transfer area increases roughly with length squared. A useful simplification is:

A / V is proportional to 1 / L

Where A is heat-transfer area, V is vessel volume and L is a characteristic vessel dimension. As the vessel becomes larger, available heat-transfer area per unit volume falls. A lab reaction that appears only mildly exothermic can become difficult to control if heat release is faster than the plant can remove it.

Mixing also changes. A reagent that disperses quickly in a small reactor may form local high-concentration zones in a large vessel before full mixing occurs. That can increase side reactions, over-reaction, local pH excursions, decomposition or uncontrolled precipitation. Scale-up criteria may include heat-transfer capability, mixing time, power per unit volume, impeller tip speed, suspension quality, Reynolds number, gas-liquid mass transfer or residence-time distribution. No single rule works for every process; the right criterion depends on what controls reaction outcome and product quality.

Gas-liquid reactions add another constraint. Hydrogenations, oxidations and carbonylations may be limited by mass transfer rather than intrinsic kinetics. Changing reactor geometry, agitation, pressure or gas sparging can change conversion, selectivity and impurity profile.

Quenching can be equally scale-sensitive. A small quench may be added quickly with visible control. At plant scale, the same quench may release heat, generate gas, form salts, foam, emulsify or create a thick slurry. Solids introduce additional constraints: slurry transfer, filterability, cake compressibility, washing efficiency and drying time can dominate batch cycle time.

Plant fit is often where a theoretically scalable route becomes commercially awkward. A process may require a glass-lined reactor, pressure capability, low-temperature utility, corrosion-resistant metallurgy, contained charging, a centrifuge instead of a nutsche filter, or dryer capacity in the correct GMP train. A chemically elegant route that does not fit the receiving site becomes a schedule, cost and change-control problem.

Process Safety as a Scale-Up Decision

Process safety is not a late checklist item. It can decide whether a route should move forward.

Reaction calorimetry measures heat generated during intended chemistry. Thermal screening and decomposition studies evaluate what can happen if material accumulates, cooling is lost, a dosing line blocks, an addition is delayed or a quench is mishandled. In pharmaceutical process safety practice, adiabatic temperature rise is a key measure:

Delta T_ad = Q_rxn / (m x C_p)

Where Delta T_ad is the adiabatic temperature rise, Q_rxn is the heat released by the reaction, m is the reacting mass and C_p is heat capacity. A pharmaceutical process-safety review by Allian, Shah, Ferretti, Brown, Kolis and Sperry describes the use of thermal and reaction hazard evaluation through the API lifecycle and notes that safety studies may lead to process changes, specialized equipment, scale reduction or process redevelopment (Ayman D. Allian et al., Process Safety in the Pharmaceutical Industry—Part I: Thermal and Reaction Hazard Evaluation Processes and Techniques, OPRD, 2020).

The practical questions are specific. Can the plant remove heat at the planned addition rate? What is the maximum temperature of the synthesis reaction? What happens if a reactive intermediate accumulates? Does the reaction generate gas? Can the vent system handle it? Is a delayed exotherm credible? Does the quench create pressure, foam or insoluble salts? Are emergency scenarios understood well enough for the intended plant?

If the answers are weak, the process may need slower addition, dilution, different solvent, alternative reagent form, lower operating temperature, semi-batch control, continuous flow, or a new route. ICH Q13 provides scientific and regulatory considerations for continuous manufacturing of drug substances and drug products (FDA, Q13 Continuous Manufacturing of Drug Substances and Drug Products). Continuous processing can improve heat and mass transfer for selected chemistries, but it is a process-design choice, not a default solution.

Impurity Control as a Development Strategy

“99% purity” is not enough to establish pharmaceutical suitability. Two API batches can have the same assay or area-percent purity and very different regulatory implications if the impurity identities, toxicological relevance, purge mechanisms or analytical detectability differ.

Impurities may originate from starting materials, reagents, catalysts, ligands, intermediates, side reactions, degradation, solvents, process aids, water, packaging-contact materials or carryover. ICH Q7 defines an impurity as any component in an intermediate or API that is not the desired entity, and defines an impurity profile as the description of identified and unidentified impurities present in an API (FDA, Q7A Good Manufacturing Practice Guidance for Active Pharmaceutical Ingredients). ICH Q11 addresses drug-substance development and manufacture, including process development and steps designed to reduce impurities in CTD Module 3 drug-substance sections (FDA, Q11 Development and Manufacture of Drug Substances).

Mutagenic impurities require particular care because trace-level materials may be toxicologically important. ICH M7(R2) addresses assessment and control of DNA-reactive mutagenic impurities that reside or are reasonably expected to reside in final drug substance or product (EMA, ICH M7(R2) Assessment and Control of DNA Reactive (Mutagenic) Impurities). Residual solvents are addressed by ICH Q3C(R9), which recommends acceptable amounts of residual solvents in pharmaceuticals based on patient safety (EMA, ICH Q3C(R9) Residual Solvents). Elemental impurities are addressed by ICH Q3D(R2), which describes a risk-based approach to assessment and control (EMA, ICH Q3D(R2) Elemental Impurities).

Nitrosamines show why impurity control is now lifecycle work. FDA’s revised 2024 guidance recommends steps for API and drug-product manufacturers to detect and prevent unacceptable nitrosamine levels, including small-molecule nitrosamines and nitrosamine drug substance-related impurities (FDA, Control of Nitrosamine Impurities in Human Drugs). EMA’s nitrosamine guidance for marketing authorization holders emphasizes risk evaluation, confirmatory testing, updates where needed and ongoing lifecycle monitoring (EMA, Nitrosamine Impurities: Guidance for Marketing Authorisation Holders).

A mature CDMO program studies impurity formation and purge across the route. Which impurities form upstream? Which are rejected by extraction, crystallization or distillation? Which track with the API? Which increase when temperature, hold time, water content, reagent age or mixing changes? Which starting-material impurities carry through? Specifications are necessary, but without process understanding they become a late detection system rather than a control strategy.

Why Crystallization Can Determine the Process

Crystallization is often treated as final cleanup. For many small-molecule APIs, it is a central manufacturing operation.

Solid form is a quality attribute. A review by Alfred Y. Lee, Deniz Erdemir and Allan S. Myerson describes solid-state form as a key quality attribute of crystalline products and emphasizes the need to understand and select the appropriate solid form during development (Lee, Erdemir and Myerson, Crystal Polymorphism in Chemical Process Development, 2011).

The crystallization process controls more than chemical purity. It can determine polymorph, salt form, solvate or hydrate formation, crystal habit, particle-size distribution, agglomeration, bulk density, filtration, washing, drying and residual solvent. Hsien-Hsin Tung’s industrial review of pharmaceutical crystallization discusses solubility, crystal form, morphology, crystallization kinetics, seeding, supersaturation, mixing time, mixing intensity and scale-up in nonuniform suspension environments (Hsien-Hsin Tung, Industrial Perspectives of Pharmaceutical Crystallization, OPRD, 2013). A later review on particle-size specification notes that nucleation, growth, breakage and agglomeration influence final particle size, and that cooling profile, seeding and agitation are important process variables (Fan Liu et al., Targeting Particle Size Specification in Pharmaceutical Crystallization, OPRD, 2022).

Common failure modes are familiar to process teams: oiling out instead of crystallizing, spontaneous nucleation before seed addition, polymorph conversion during drying, fines that blind a filter, sticky wet cake, solvent trapped in crystals, variable hydrate level and low bulk density that constrains dryer or container capacity. Particle properties may later affect drug-product processing, even though the immediate CDMO project is API manufacture.

Isolation is therefore part of the API process design. A route that relies on an unreliable final crystallization is not robust merely because the reaction chemistry is high-yielding.

Analytical Development and Process Control

Analytical work begins before formal validation. Early methods must be fit for purpose: strong enough to guide chemistry, detect meaningful impurities and prevent false confidence. Later, methods need appropriate validation, transferability and lifecycle control.

ICH Q14 describes science- and risk-based approaches for developing and maintaining analytical procedures suitable for assessing the quality of drug substances and drug products (FDA, Q14 Analytical Procedure Development). ICH Q2(R2) provides a general framework for analytical procedure validation, including validation principles for analytical use of spectroscopic data (FDA, Q2(R2) Validation of Analytical Procedures).

For scale-up, analytical development must answer operational questions. What defines reaction endpoint? Is conversion measured by HPLC, GC, NMR, in-line spectroscopy or another method? Are unstable intermediates sampled safely and representatively? Are impurity response factors understood? Are reference standards available? Can the receiving CDMO reproduce the method? Are sample quench procedures and sample hold times controlled? Can the method distinguish the API from a close regioisomer, stereoisomer or degradation product?

ICH Q7 expects written in-process controls for steps that cause variability in quality characteristics, with critical in-process controls and monitoring stated in writing and approved by the quality unit (FDA, Q7A Good Manufacturing Practice Guidance for Active Pharmaceutical Ingredients). This is the difference between testing quality into a batch and building control into manufacturing. Release testing is essential, but it is too late to discover that an impurity formed three steps earlier and can no longer be purged.

Why Technology Transfer Fails

Technology transfer fails when teams transfer a synthetic procedure but not the knowledge required to run it.

WHO’s current technology-transfer guideline, TRS 1044 Annex 4, describes technology transfer as a documented, planned and systematic process that transfers knowledge, experience and documentation between units, with attention to organization, production, quality control, premises, equipment, qualification, validation and documentation (WHO, TRS 1044 Annex 4: WHO Guidelines on Technology Transfer in Pharmaceutical Manufacturing). Compared with a narrow “send the batch record” view, the WHO framing is broader: the receiving unit needs enough knowledge to reproduce and control the process, not merely follow instructions.

A robust transfer package should include the process description, flow diagram, master batch information, raw-material specifications, supplier information, equipment requirements, critical process parameters, proven acceptable ranges or development knowledge, critical quality attributes, in-process controls, analytical methods, impurity knowledge, safety data, cleaning considerations, deviation history, failed experiments, atypical observations, unresolved risks and change-control responsibilities.

Tacit knowledge is often the missing element. The reagent was added below the liquid surface. The slurry needed a specific seed age. The wet cake cracked if washed too quickly. A long hold changed the impurity profile. A lab analyst used an undocumented sample quench. These details may not appear in the formal procedure, yet they may control the process.

ICH Q10 identifies knowledge management as an element supporting the pharmaceutical quality system (FDA, Q10 Pharmaceutical Quality System). A 2021 PDA Journal article by Martin J. Lipa, Anne Greene and Nuala Calnan argues that weak knowledge transfer can limit the practical value of ICH Q8, Q10, Q11 and Q12 lifecycle concepts (Lipa, Greene and Calnan, Knowledge Management as a Pharmaceutical Quality System Enabler, 2021). In CDMO work, this risk rises when development records omit failed experiments, when analytical methods are immature, when engineering assumptions are undocumented or when receiving-site equipment differs materially from the sponsor’s pilot equipment.

Sponsor and CDMO Responsibilities

Outsourcing API manufacture does not remove the sponsor’s quality responsibilities. It changes how those responsibilities are executed and documented.

FDA’s 2016 guidance on quality agreements describes how parties in contract drug manufacturing can define, establish and document manufacturing activities to support CGMP compliance (FDA, Contract Manufacturing Arrangements for Drugs: Quality Agreements). ICH Q7 states that contract manufacturers, including laboratories, should comply with GMP; companies should evaluate contractors; written agreements should define GMP responsibilities; and changes in process, equipment, test methods, specifications or other contractual requirements should not be made unless the contract giver is informed and approves the change (FDA, Q7A Good Manufacturing Practice Guidance for Active Pharmaceutical Ingredients).

A quality agreement should define responsibilities for deviations, investigations, change control, release testing, stability, analytical-method transfer, regulatory support, complaints, recalls, data integrity, subcontracting and record access. It has limits. A legal document cannot compensate for a sponsor that does not understand its process or a CDMO that treats development gaps as late-stage commercial surprises.

For EU supply chains, EMA’s GMP/GDP Q&A on outsourced activities emphasizes written contracts, clearly defined responsibilities, robust communication, access to relevant contracts and records, audit expectations and supply-chain diagrams where applicable (EMA, Guidance on Good Manufacturing Practice and Good Distribution Practice: Questions and Answers). The practical message is consistent across regulators: technical, quality and commercial responsibilities must be explicit and operational, not left to informal escalation.

Why the Lowest Quotation May Not Be the Lowest-Cost Option

A low CDMO quotation can be legitimate. It can also be incomplete.

Total project cost includes process development, analytical development, raw materials, manufacturing, equipment modifications, containment, waste treatment, solvent recovery, failed batches, investigation time, qualification or validation work, stability studies, regulatory documentation, change management, delay cost and supply interruption risk.

A quotation that excludes route optimization may look attractive until chromatography, low productivity or impurity carryover blocks Phase III supply. A proposal that assumes sponsor-supplied analytical methods may be cheap until method transfer fails. A price that relies on a single-source starting material may be acceptable for early clinical supply but fragile for commercial planning. A CDMO with open reactor capacity may still be the wrong fit if the critical step needs faster heat removal, better solids handling, different metallurgy or stronger containment.

The commercial evaluation should separate batch price from project risk. The better proposal is often the one that identifies the missing development work early, prices it transparently and explains which risks must be retired before GMP or commercial commitment.

Hypothetical Case: A Five-Step API That Worked at 1 kg

The following case is illustrative. It is not based on a named company, disclosed commercial project or public inspection record.

A sponsor has produced 1 kg of a five-step small-molecule API under non-GMP conditions. The route reports good overall yield and 99% HPLC purity. During CDMO technical evaluation, the team identifies five risks that could affect GMP scale-up.

The route may remain viable after these studies. If the impurity cannot be purged, the exotherm cannot be safely controlled, or crystallization remains unpredictable, the right decision may be to change the route before pivotal or commercial supply. The value of the exercise is that the decision is made from process evidence rather than from optimism about a successful 1 kg batch.

How a Sponsor Should Evaluate a Small-Molecule API CDMO

A technically serious CDMO evaluation should test the ability to convert chemistry into controlled manufacture.

These weights should be adjusted for the molecule and development stage. A hazardous, exothermic process deserves heavier safety and engineering weighting. A poorly crystallizing API deserves heavier solid-state weighting. A late-stage transfer deserves heavier GMP, regulatory and change-control weighting.

Questions Sponsors Should Ask Before Awarding an API CDMO Project

1. Which steps in the current route do you consider non-scalable, and why?
2. What purification steps rely on chromatography, extreme dilution, solvent swaps or low-temperature operation?
3. What calorimetry and thermal-screening data do you require before the first GMP batch?
4. What is the expected heat-removal limitation in your proposed plant equipment?
5. How will you test whether mixing or addition rate affects impurity formation?
6. Which impurities are expected to form, which are known to purge and which are unresolved?
7. What is your plan for potentially mutagenic impurities, residual solvents, elemental impurities and nitrosamine risk assessment where applicable?
8. What solid forms are known, and how will you control polymorph, solvate, hydrate or salt form?
9. What seeding strategy, supersaturation profile and filtration assumptions are you using?
10. Which analytical methods are ready for GMP use, which need development and which require transfer or validation?
11. How will endpoint, in-process controls and sampling procedures be defined?
12. Which exact reactors, filters, centrifuges and dryers would be used for each step?
13. What equipment modifications, containment measures or cleaning-validation issues do you foresee?
14. Which raw materials are single-sourced, long-lead, hazardous or quality-sensitive?
15. What failed or atypical sponsor experiments do you need before quoting a realistic program?
16. How are deviations, investigations, OOS results and change controls communicated to the sponsor?
17. What information will be included in the technology-transfer package and batch records?
18. What assumptions are excluded from the quotation and could later become change orders?
19. Who owns regulatory writing support for Module 3 drug-substance sections?
20. What would make you recommend changing the route rather than scaling it?

Conclusion

Laboratory synthesis establishes chemical feasibility. It does not establish manufacturability.

Each development stage should retire specific risks: unsafe heat release, scale-sensitive impurity formation, weak analytical methods, unstable crystallization, poor filtration, equipment mismatch, raw-material fragility and unclear quality responsibilities. Leaving these questions to the first large GMP batch turns process development into batch failure investigation.

For sponsors selecting a small-molecule API CDMO, capacity, lead time and batch price are not enough. The stronger basis for selection is evidence of process understanding: how the CDMO evaluates the route, challenges assumptions, defines controls, transfers knowledge and explains the risks that must be resolved before the molecule can become a reliable GMP supply chain.

Frequently Asked Questions

What causes small-molecule API scale-up problems?

Scale-up problems often come from weak process robustness, heat removal, mixing, impurity control, crystallization, analytics, equipment fit or raw-material strategy.

Is high laboratory yield enough for API CDMO scale-up?

No. Yield is useful, but manufacturability also depends on safety, operating window, impurity purge, isolation, reproducibility and plant fit.

Why is crystallization important in API manufacturing?

Crystallization can determine solid form, purity, particle size, filtration, drying, residual solvent and downstream manufacturability.

What should an API technology-transfer package include?

It should transfer process knowledge: CPPs, CQAs, impurity data, safety data, methods, equipment needs, deviations, failed experiments and unresolved risks.

Does outsourcing remove sponsor quality responsibility?

No. Sponsors still need defined responsibilities, contractor oversight, quality agreements, controlled changes and technical governance.

References and Further Reading

• FDA, Q7A Good Manufacturing Practice Guidance for Active Pharmaceutical Ingredients
• FDA, Process Validation: General Principles and Practices
• FDA, Q11 Development and Manufacture of Drug Substances
• FDA, Q13 Continuous Manufacturing of Drug Substances and Drug Products
• FDA, Q14 Analytical Procedure Development
• FDA, Q2(R2) Validation of Analytical Procedures
• FDA, Q10 Pharmaceutical Quality System
• FDA, Contract Manufacturing Arrangements for Drugs: Quality Agreements
• FDA, Control of Nitrosamine Impurities in Human Drugs
• EMA, ICH Q3C(R9) Residual Solvents
• EMA, ICH Q3D(R2) Elemental Impurities
• EMA, ICH M7(R2) Mutagenic Impurities
• EMA, Nitrosamine Impurities: Guidance for Marketing Authorisation Holders
• EMA, Guidance on Good Manufacturing Practice and Good Distribution Practice: Questions and Answers
• WHO, TRS 1044 Annex 4: WHO Guidelines on Technology Transfer in Pharmaceutical Manufacturing
• Graham Caygill, M. Zanfir and A. Gavriilidis, Scalable Reactor Design for Pharmaceuticals and Fine Chemicals Production. 1: Potential Scale-up Obstacles, Organic Process Research & Development, 2006
• Ayman D. Allian et al., Process Safety in the Pharmaceutical Industry—Part I: Thermal and Reaction Hazard Evaluation Processes and Techniques, Organic Process Research & Development, 2020
• Alfred Y. Lee, Deniz Erdemir and Allan S. Myerson, Crystal Polymorphism in Chemical Process Development, Annual Review of Chemical and Biomolecular Engineering, 2011
• Hsien-Hsin Tung, Industrial Perspectives of Pharmaceutical Crystallization, Organic Process Research & Development, 2013
• Fan Liu et al., Targeting Particle Size Specification in Pharmaceutical Crystallization, Organic Process Research & Development, 2022
• ACS Green Chemistry Institute, Process Mass Intensity Calculation Tool
• ACS GCI Pharmaceutical Roundtable, Solvent Selection
• Martin J. Lipa, Anne Greene and Nuala Calnan, Knowledge Management as a Pharmaceutical Quality System Enabler, PDA Journal of Pharmaceutical Science and Technology, 2021