September 2022

Crystallisation and Screening Theory - Solid Form Screening

Crystal structure

Crystals are distinguished from amorphous solids by possessing both short and long range regular structure, minimal repeating units within this matrix are termed unit cells. Crystals are of lower intrinsic energy than amorphous forms and are stabilised by a number of different intermolecular forces:

• Ionic interactions (in salts)
• Lipophilic interactions
• Pi-pi interactions (stacking aromatic rings)
• Dipole stabilisation (common with heterocycles)
• H-bonding (see cocrystal document for more discussion)

There are 7 crystal classes; triclinic, monoclinic, orthorhombic, tetragonal, trigonal, hexagonal and cubic.

Based on the symmetry elements present crystals are divided into 230 space groups.

Crystallisation methods


Supersaturation is the condition where the concentration of solute in solution is greater than would be present at equilibrium with undissolved solid. It is a necessary condition for crystallisation to occur. It is described by a number of useful definitions:

• Degree of Supersaturation C = C-Ceq
• Supersaturation Ratio        C = C/Ceq
• Relative Supersaturation    C = C/Ceq

Supersaturation can be achieved by a number of practical methods. These include:

• Thermal cycling
• Anti-solvent addition
• pH manipulation
• Evaporation of solvent
• Dissolution of amorphous forms

Any supersaturated solution is metastable and may crystallise spontaneously. Nucleation (initial formation of a complex of sufficient size to promote crystallisation) occurs initially followed by crystal growth. The rate of spontaneous nucleation is, however, kinetically driven and may be very slow under conditions of low supersaturation. Once nucleation occurs the solute will crystallise until the equilibrium solubility is reached. This is illustrated for a typical cooling crystallisation.

The area between the equilibrium solubility curve and the curve where spontaneous nucleation occurs is defined as the metastable zone or MSZ. Note that in this zone there is a degree of supersaturation so any crystals present will grow. The width of the MSZ will depend on a number of factors, most importantly the rate of cooling and the presence of promoters of nucleation such as dust particles, so measurements of MSZ width should be interpreted with care, especially with changes of scale.

Nucleation and seeding

Spontaneous nucleation is an unreliable method to use for practical crystallisations, control of the process is achieved by seeding of crystalline material into crystallisations at minimum supersaturation, then maintaining supersaturation by, for example, slow cooling, evaporation or addition of anti-solvent. Initial rapid cooling, as happens with natural cooling, will result in high initial supersaturation and there is a risk of spontaneous nucleation and a loss of control.

Linear cooling, common in programmable reactors, has a similar risk. Non-linear cooling with a slow reduction in temperature in the first phase will give consistent particle size and a controlled and reliable result.

Seeding has the effect of narrowing the MSZ to achieve controlled nucleation combined with an acceptable crystal growth rate.

It is occasionally necessary to use a seed of a different crystalline compound as a template for the compound of interest. This can give otherwise unobtainable polymorphs and is of more academic than industrial interest.

Van Weimarn noted that the particle size achieved from a crystallisation was inversely proportional to the relative supersaturation as defined above, maintaining low relative supersaturation will give larger, more easily filtered particles.

Crystallisation and purity

A very useful property of crystallisation is that it tends to exclude impurities with the consequence that isolation and purification are achieved in one step.

The efficiency of purification depends on a number of factors:

» Similarity of impurity to API, if the impurity is very similar in structure there is only a small energy penalty for inclusion in the crystal.
» Relative concentration of the impurity to the API. This is the explanation for sequential recrystallization leading to an increase in purity.
» Degree of supersaturation. Rapid recrystallization at high supersaturation has the effect of overwhelming the other factors and results in limited improvement in purity.

Impurities and crystallisation

The first question that should be asked before crystallisation studies is ‘How pure is the material?’ Impurities, even at low levels, can have a profound effect on crystalline form. It is often noted that impurities change the shape of the crystals produced, this effect arises from different faces of the growing crystal presenting different functionality to solution and impurities binding preferentially to one face, preventing further growth.

Impurities in the form of ‘hazes’, polymeric material or tars can strongly inhibit nucleation and can also be responsible for coating seeds and preventing the progression of crystallisation. A combination of treatment with activated carbon and filtration is usually effective in removing such contaminants.

Ostwald Theory of Stages

This theory suggests that polymorph transformation takes place through a series of stages where each stage has a small energy barrier rather than a spontaneous one-step drop into a much more stable form.

In the diagram we see stepwise transformation through a series of polymorphs A, B and C to a stable crystal. This supports the use of elevated temperatures in slurrying and thermal cycling, overcoming kinetic barriers to transformation and making it more likely to avoid the isolation of metastable forms.


Small crystals may bind together to form agglomerations, this can affect handling properties such as filtration rate. Agglomerations are best detected by particle size distribution analysis or visualisation using SEM. Stirring rate and the use of stirrers with high shear rates can help to minimise agglomerate formation.

Solvates and occluded solvent

There are 2 main mechanisms that result in the inclusion of solvent into a crystalline form. The first is solvate formation where the solvent is included as an intrinsic component of the crystal structure with defined stoichiometry. These are common and should be distinguished from genuine polymorphs (most solvents can be detected and quantified by 1H NMR) because they are not usually sufficiently stable to be considered for further development even if the solvent is non-toxic. Vacuum drying can lead to loss of the solvent and collapse of the crystalline form, although metastable ‘desolvated solvates’ are sometimes useful.

The second is occlusion of solvent during the crystallisation process where solvent is trapped in cavities within the solid created as the crystals grow. This process is more likely to occur at high supersaturation and can be reduced by operating at lower concentrations and slowing down the crystallisation process.

Occluded solvent can be surprisingly difficult to remove by vacuum drying, it is often suggested that grinding the solid before drying can help but this risks the creation of amorphous material.

Oiling out

Precipitation of an oil prior to crystallisation is a serious issue. As the oil forms a liquid/liquid extraction system is generated and impurities may extract into the oil and be incorporated into any crystalline material subsequently formed.

Again, operating at low supersaturation and using seeds will ensure the minimisation of oiling-out.

Optical resolution by crystallisation

1. Enantiomers and racemates
Consideration of the crystallisation of racemates gives us insight into how resolution of enantiomers might be achieved. There are essentially 2 options available as a racemate crystallises, firstly the racemate may crystallise where every crystal contains racemic compound. In about 10% of cases, however, the enantiomers crystallise separately, each individual crystal containing only 1 enantiomer (a conglomerate). Note that the solubility of the separate enantiomers will be identical. The diagram shows the melting point behaviour of conglomerates and racemic solids. Conglomerates have one eutectic point at 1:1 ratio whereas racemic solids have highest melting point at 1:1 ratio, in the racemic solid example addition of a small quantity of either enantiomer acts as an impurity, reducing the melting point.

In 1848 Pasteur discovered that sodium ammonium tartrate crystallised in separate left-handed and right-handed crystals (conglomerate), physically picked them apart and achieved the first resolution of enantiomers.

Despite the solubility of the enantiomers being identical, it is possible to seed a racemic solution with one enantiomer and crystallise one out preferentially in a process known as entrainment. Filtration of the first crop of one enantiomer leaves a solution with an excess of the other which may be isolated by seeding with the second enantiomer. In principle the solution can be reloaded with racemate and the process repeated. Entrainment has the advantage of recovering both enantiomers equally efficiently with essentially no loss and is used commercially on a large scale.

2. Diastereoisomeric salts
When a racemate of an API forms a salt with one enantiomer of an optically active acid the result is a pair of diastereoisomeric salts. These will have different melting point and solubility, the less soluble diastereomer may be crystallised out preferentially and the API recovered to give a method for resolution. This method is best suited to isolation of a desired enantiomer from a racemate.

Screening for useful resolutions involves using a number of diverse chiral acids derived from the chiral pool. Kits including commonly available acids cheap enough to be suitable for scale-up are available commercially.

It is important to have an analytical method available to check the crystallisation for enantiomeric excess, the most rapid and sensitive method is chiral HPLC and it is best to develop such a method before embarking on a screen.

Screening - scale considerations

The approach to scale in crystallisation screening is dependent on the amount and nature of the information required to be generated by the screen. A simple ‘yes/no’ can be obtained with a small scale screen, typically 5mg per well or less where the key information is whether the PXRD shows any well-defined peaks indicative of crystallinity. Where it is also necessary to gather DSC, NMR, DVS and TGA, however, it is obvious a larger scale experiment is required. A compromise is to perform initial screening at an intermediate scale, say 20mg, obtain enough data to select interesting forms for further characterisation, then scale up to 100mg+ for complete characterisation.


The following methods are useful in confirming crystallinity:

» DSC for a sharp melting point

Download PDF Graphic

Download the Crystallisation and Screening Theory PDF

Crystallisation and Screening Theory PDF






Ready to find out more?
Expertise dedicated to solve your challenges.

Our scientists partner with you to provide a consultative, flexible and dependable solution for the development, validation, and application of testing methods.

Let us help you to discover, develop, and manufacture medicines and ​products to improve the quality of life.

Let's discuss your project