# Pharmaceuticals and fermentation: from molecule to medicine: Making the API: multistep batch chemistry

Source page: https://www.materialinsights.org/learn/chemicals/pharmaceuticals-and-fermentation/p2/

[Understanding Chemicals](https://www.materialinsights.org/learn/chemicals/) / [Module 10: Pharmaceuticals and fermentation: from molecule to medicine](https://www.materialinsights.org/learn/chemicals/pharmaceuticals-and-fermentation/) Part 2 of 5

## Making the API: multistep batch chemistry

An API is usually built in many steps, each a separate reaction followed by isolation of a solid. The plant is a set of stirred tanks, filters and dryers used for one product after another. Three things decide cost: how much material survives each step, how pure the final crystals are, and how much work goes into proving the process to a regulator.

### 2.1 Each step loses material, and losses multiply

A synthesis is a chain of reactions. The product of step one is the raw material of step two, and so on. Each step has a yield, the fraction of the theoretical amount actually recovered as pure product. Because the losses compound, overall yield is the product of the step yields, not their average.

The arithmetic is unforgiving. Ten steps at 90% each give 0.9 to the power 10, or 35% overall. Twenty steps at 90%, roughly the length Pollak gives for atorvastatin from simple chemicals, give 12%. At 95% per step, twenty steps give 36%. Moving from 90% to 95% per step triples output from the same inputs.

This is why process chemists work so hard to shorten routes. Pollak's tactics include fewer steps (by not isolating intermediates), enzymatic steps, higher yields, cheaper raw materials and recycled solvents and catalysts. Pollak gives a rule of thumb that cost of goods should fall 10 to 20% each time annual production doubles.

The early steps matter most for raw material cost, because material lost late carries the value of all earlier steps with it. Pharmaceutical companies therefore buy advanced intermediates from fine chemical suppliers, the "Pharma starting materials and intermediates" and "Chiral and specialty building blocks" tiles, and keep only the last few steps in house.

Kent adds a modern complication: many new drugs contain several chiral centres, points where the molecule can exist in mirror-image forms that behave differently in the body. Regulators now require each mirror form (enantiomer) to be tested, so the right one must be made selectively. Chiral catalysts that produce mainly one form have largely replaced older methods of making both and separating them.

Data Overall yield for different step counts and step yields (arithmetic)

| Steps | 80% per step | 90% per step | 95% per step | 98% per step |
|---|---|---|---|---|
| 4 | 41% | 66% | 81% | 92% |
| 10 | 11% | 35% | 60% | 82% |
| 15 | 3.5% | 21% | 46% | 74% |
| 20 | 1.2% | 12% | 36% | 67% |

Overall yield = (step yield) raised to the number of steps. Atorvastatin is about 20 steps from simple chemicals (Pollak).
 Source: Pollak (2011), Fine Chemicals, pdf p. 28 (step count); arithmetic by the course  Worked example

#### Aprepitant: four steps, 54% overall

Kent's Scheme 11.1 shows the route to aprepitant, the anti-nausea drug Emend. A chiral reduction using 0.5 mol% of a ruthenium catalyst makes a key alcohol in over 98% yield and 91% enantiomeric excess; recrystallisation lifts chiral purity above 99%. The main sequence then has four steps at 73%, 84%, 91% and 96%. Multiplying: 0.73 x 0.84 x 0.91 x 0.96 = 0.536, the 54% overall yield Kent reports. One weak step (73%) costs more than the other three together.

- Kent (2012), Ch 11 Chemistry in the Pharmaceutical Industry · pdf pp. 403, 426-427
- Pollak (2011), Fine Chemicals · pdf pp. 28, 91

### 2.2 A worked synthesis: salicylic acid to aspirin

Shreve's flowchart for salicylic acid and aspirin is one of the few in the sources with raw materials per tonne, and it shows what bulk API chemistry looks like.

The route starts with phenol, from the "Phenol and acetone" tile, and "Caustic soda". Phenol and sodium hydroxide give sodium phenolate in water. The phenolate must be bone dry and finely divided for the next step, so the solution goes into a revolving, heated ball mill and is dried under vacuum at 130°C. Carbon dioxide is then admitted at 700 kPa and 100°C. It adds to the ring (a carboxylation known as the Kolbe-Schmitt reaction), first forming sodium phenyl carbonate, which rearranges to sodium salicylate.

The salicylate is dissolved out of the mill, decolourised with "Activated carbon", and the free acid precipitated by adding "Sulphuric acid", leaving sodium sulphate in solution. The crude acid is purified by sublimation (vaporising and recondensing it).

To make aspirin, salicylic acid is refluxed with acetic anhydride in toluene at 88 to 92°C for 20 hours. The anhydride transfers an acetyl group to the acid's hydroxyl, releasing acetic acid. On cooling in aluminium tanks, aspirin crystallises as large crystals, which are filtered or centrifuged, washed and dried.

The numbers repay a look. Per tonne of salicylic acid Shreve lists 800 kg phenol. The theoretical need is 94/138 = 0.68 t, so the yield on phenol is about 85%. For a plant making 1,800 kg a day, Shreve gives 5,000 kg steam, 3,600 MJ of power and 53 work-hours a day. Per tonne that is 2.8 t steam, 2,000 MJ and 29 work-hours: labour-heavy by commodity standards. Bulk aspirin sold for $3.63/kg in 1981.

Process flowchart Salicylic acid and aspirin (Shreve 1984) Inputs **Phenol**800 kg per t salicylic acid **Caustic soda**850 kg per t salicylic acid **Carbon dioxide**500 kg per t salicylic acid **Zinc, zinc sulphate**10 kg and 20 kg per t **Activated carbon**20 kg per t **Sulphuric acid**to precipitate the acid (not quantified) **Acetic anhydride**0.57 t per t aspirin (stoichiometric, course arithmetic) **Utilities**about 2.8 t steam, 2,000 MJ power, 29 work-h per t salicylic acid Process steps 1 **Phenolate formation**

2 C6H5OH + 2 NaOH → 2 C6H5ONa + 2 H2O
 2 **Drying in heated ball mill**

Vacuum, 130°C, to a very dry fine powder
 3 **Carboxylation (Kolbe-Schmitt)**

CO2 at 700 kPa and 100°C; sodium phenyl carbonate rearranges to sodium salicylate
 4 **Dissolve and decolourise**

Salicylate dissolved out; activated carbon treatment; filter
 5 **Precipitation**

H2SO4 frees salicylic acid; Na2SO4 to waste; continuous centrifuge
 6 **Sublimation**

Purifies technical salicylic acid
 7 **Acetylation**

Reflux with acetic anhydride in toluene, 88-92°C, 20 h
 8 **Crystallisation and drying**

Cool in aluminium tanks; filter or centrifuge, wash, dry
 Outputs **Salicylic acid (intermediate)**0.77 t per t aspirin (stoichiometric, course arithmetic) **Aspirin (acetylsalicylic acid)**1 t **Acetic acid by-product**about 0.33 t per t aspirin (stoichiometric) **Sodium sulphate solution**waste **Recovered phenol**to recovery

Raw materials per tonne of salicylic acid are Shreve's; stoichiometric figures are course arithmetic (aspirin 180, salicylic acid 138, acetic anhydride 102 g/mol).
 Source: Shreve (1984), Ch 40 Pharmaceutical Industry, pdf pp. 821-823 (Fig. 40.2)  Worked example

#### Paracetamol in one step

"Paracetamol" (acetaminophen) is made, Shreve says, by treating p-aminophenol with a mixture of glacial acetic acid and acetic anhydride, the same acetylation as in aspirin but on an amine. Pollak's table of API reactions adds the upstream chemistry: partial hydrogenation of nitrobenzene, then N-acetylation. Because the final step is short and the molecule simple, cost is set by the intermediate p-aminophenol and by scale. Wittcoff notes that originator companies claim it costs them more to make than it costs a generic maker, because they load future research costs onto current products.

- Shreve (1984), Ch 40 Pharmaceutical Industry · pdf pp. 821-823
- Pollak (2011), Fine Chemicals · pdf p. 55 (Table 4.1)
- Wittcoff (2004), Industrial Organic Chemicals · pdf p. 649

### 2.3 Crystallisation and impurities decide whether a batch can be sold

In most chemical industries a product is sold to a specification of a few properties. An API is sold to a specification that covers the molecule, every impurity above a threshold, and the physical form of the solid. Crystallisation is the main tool for meeting it.

Crystallisation works because a growing crystal tends to admit only molecules that fit its lattice. Dissolve the crude product hot, cool it slowly, and the API comes out as crystals while most impurities stay in the liquid, the mother liquor. Repeating the process, recrystallisation, raises purity further; in Kent's aprepitant example it lifted chiral purity from 91% enantiomeric excess to over 99%.

The solid form matters as much as purity. Kent lists particle size, density, flowability, polymorphism, hygroscopicity (tendency to absorb water) and stability as critical properties for formulation. Polymorphism means the same molecule can crystallise in different arrangements, which dissolve at different rates and may even be covered by different patents. Crystallisation, filtration, drying and milling must therefore all be fixed well before the process is finalised.

The hardest impurity rules concern genotoxic impurities: traces of reagents or by-products that could damage DNA. Kent explains that these must be held below a threshold of toxicological concern of 1.5 micrograms a day of patient exposure. For a high-dose drug that can mean detecting below 1 part per million. A route that works well chemically may be abandoned if such an impurity cannot be removed or controlled.

Fermentation APIs face the same discipline in a different form. Shreve notes that penicillin leaves the fermenter only 5 to 10% pure, so extraction and crystallisation do most of the work of turning broth into a saleable product.
 Common misreading

#### "Pharma grade just means very pure"

Purity is necessary but not sufficient. A batch of API can be 99.9% pure and still be rejected if its crystal form, particle size or water content differ from the registered process, because these change how the tablet dissolves in the patient. Kent stresses that physical properties are as critical as chemical purity. This is why a generic maker cannot simply buy the cheapest API on the market: the solid must behave like the one in the approved dossier.

- Kent (2012), Ch 11 Chemistry in the Pharmaceutical Industry · pdf pp. 425-427
- Shreve (1984), Ch 40 Pharmaceutical Industry · pdf p. 840

### 2.4 Process development, scale-up and cGMP plants

Scale changes what matters. Pollak's comparison is neat: in the laboratory the measure of success is yield; in the plant it is throughput in kilograms per cubic metre of reactor per hour. In the laboratory reaction time is the bottleneck; in the plant it is heating and cooling, because a large vessel has little wall area per unit volume.

Scale-up proceeds in stages. Pollak gives bench reactors of about 10 to 60 L and pilot plants of about 100 to 2,500 L, laid out like the industrial plant. Before transfer, the process undergoes a hazard and operability (HAZOP) study and demonstration batches. Kent adds that statistical design of experiments and parallel robotic reactors now speed up development.

Industrial API plants are usually multipurpose. Pollak describes reactors of typically 4 to 6 m3, sometimes 1 to 10 m3, in stainless steel or glass-lined steel, rated for -20 to 200°C and 10 mbar to 3 bar. As a rule of thumb, one cubic metre of reactor makes 15 to 30 t of product a year for one step, so a bay of 4 to 6 m3 vessels handles about 100 t of a step a year. That matches demand: one third of the top 500 drugs are needed at 10 to 100 t a year; only 7% exceed 1,000 t.

All of this runs under current Good Manufacturing Practice (cGMP), the regulators' rules for quality. Kent explains that the batch concept suits cGMP because each batch can be accepted or rejected on its own. Under quality by design, the company maps a design space of process conditions known to give good product, and must stay inside it. Pollak adds that cGMP processes require formal validation: design, qualification and continued verification.

Data Laboratory synthesis versus industrial-scale process (Pollak)

| Task | Laboratory | Industrial plant |
|---|---|---|
| Operator | Laboratory chemist | Chemical engineer |
| Measure of economy | Yield | Throughput (kg/m3/hour) |
| Units | g, mL, mol, min, h | kg, t, h, shift |
| Equipment | Glass flask | Stainless steel, glass-lined |
| Process control | Manual | Automatic |
| Critical path | Reaction time | Heating and cooling |
| Liquid handling | Pouring | Pumping |
| Liquid-solid separation | Filtration | Centrifugation |

 Source: Pollak (2011), Fine Chemicals, pdf p. 91 (Table 6.1)  In the data

#### Little economy of scale in API plants

Pollak observes that for products needed in at most a few tens of tonnes a year, made in multipurpose plants, there is little or no economy of size. The equipment is the same whatever the product; the plant simply runs a campaign of one product, cleans, and switches to another. Cost is driven instead by yield, cycle time and how fully the equipment is used. Pollak's own value-chain table makes the contrast: methanol and acetic acid come from dedicated continuous plants making millions of tonnes a year.

- Pollak (2011), Fine Chemicals · pdf pp. 35, 75, 90-91
- Kent (2012), Ch 11 Chemistry in the Pharmaceutical Industry · pdf p. 425

### See it in the table
[H₂O Water](https://www.materialinsights.org/commodities/water/)[Al Aluminium](https://www.materialinsights.org/commodities/aluminum/)[St Steel](https://www.materialinsights.org/commodities/steel/)[H2SO4 Sulphuric acid](https://www.materialinsights.org/chemicals/sulphuric_acid/)[NaOH Caustic soda](https://www.materialinsights.org/chemicals/caustic_soda/)[AC Activated carbon](https://www.materialinsights.org/chemicals/activated_carbon/)[CO2 Carbon dioxide](https://www.materialinsights.org/chemicals/carbon_dioxide/)[Tol Toluene](https://www.materialinsights.org/chemicals/toluene/)[Phe Phenol and acetone](https://www.materialinsights.org/chemicals/phenol/)[MeOH Methanol](https://www.materialinsights.org/chemicals/methanol/)[AcOH Acetic acid](https://www.materialinsights.org/chemicals/acetic_acid/)[KSM Pharma starting materials and intermediates](https://www.materialinsights.org/chemicals/ksm_intermediates/)[APAP Paracetamol](https://www.materialinsights.org/chemicals/paracetamol/)[Chi Chiral and specialty building blocks](https://www.materialinsights.org/chemicals/chiral_building_blocks/)

### Check your understanding

A 12-step route runs at 90% per step. A chemist raises one step from 60% to 90%. What happens to overall yield?
 Overall yield is a product of step yields, so replacing a 0.6 factor with 0.9 multiplies the whole by 0.9/0.6 = 1.5, whatever the other steps are.

In Shreve's salicylic acid process, why is the sodium phenolate dried in a ball mill before CO2 is added?
 Shreve states the phenolate must be finely divided and exposed to CO2 under pressure and heat; drying under vacuum at 130°C gives a very dry powder before CO2 at 700 kPa and 100°C.

Pollak's rule of thumb is 15-30 t of product per m3 of reactor per year for one step. Roughly what reactor volume does a 4-step API at 100 t/yr need, if each step occupies equipment in turn?
 Each step needs 100/30 to 100/15, or 3.3-6.7 m3-years; four steps need about 13-27 m3 of reactor time, a few multipurpose bays.

[← Part 1: The pharmaceutical value chain](https://www.materialinsights.org/learn/chemicals/pharmaceuticals-and-fermentation/p1/)[Mark read · Part 3: Formulation: turning powder into a dose →](https://www.materialinsights.org/learn/chemicals/pharmaceuticals-and-fermentation/p3/)

## Other sections

- [Module overview](https://www.materialinsights.org/learn/chemicals/pharmaceuticals-and-fermentation/index.md)
- [The pharmaceutical value chain](https://www.materialinsights.org/learn/chemicals/pharmaceuticals-and-fermentation/p1/index.md)
- [Formulation: turning powder into a dose](https://www.materialinsights.org/learn/chemicals/pharmaceuticals-and-fermentation/p3/index.md)
- [Fermentation: microbes as chemical plants](https://www.materialinsights.org/learn/chemicals/pharmaceuticals-and-fermentation/p4/index.md)
- [Enzymes, biologics and the economics](https://www.materialinsights.org/learn/chemicals/pharmaceuticals-and-fermentation/p5/index.md)
- [Key terms, quiz and sources](https://www.materialinsights.org/learn/chemicals/pharmaceuticals-and-fermentation/review/index.md)