Ideal solubility

In the special case where the energy of the solute-solvent bond is equal to the energy of the solvent-solvent bond, then solute-solvent bonds may form with no change in intermolecular energy (i.e. ΔH_mix = 0) and dissolution is said to be ideal. Ideal dissolution (although unlikely in reality) leads to ideal solubility and is an interesting theoretical position because it can be described in thermodynamic terms that allow calculation of the dependence of solubility with temperature.

From Equation 23.1, if ΔH_mix = 0 then ΔH_f is equal to ΔH_sol. Incidentally, since ΔH_f must be positive (i.e. endothermic) ΔH_sol must also be positive for ideal dissolution. For a process to occur spontaneously the Gibbs free energy (ΔG) must be negative. The familiar thermodynamic relationship for dissolution is:

(23.4)

where T is temperature. ΔG_sol is most likely to be negative when ΔH_sol is negative but, as noted above, ΔH_sol is frequently positive for dissolution. This means that for dissolution to occur spontaneously, the driving force must be an increase in entropy.

Equation 23.3 shows that solubility has the attributes of an equilibrium constant. This being so, it is possible to apply the van’t Hoff equation (Eqn 23.9), yielding:

(23.5)

Making the assumption that ΔH_f is independent of temperature, then integrating Equation 23.5 from T_m to T results in:

(23.6)

where T_m is the melting temperature of the pure drug and T is the experimental temperature.

Equation 23.6 is very useful in preformulation, since it allows prediction of ideal solubility at a particular temperature if the melting temperature and enthalpy of fusion of the pure drug are known. This is why determination of melting point and enthalpy of fusion are the next physicochemical parameters to be determined during preformulation.

Determination of melting point and enthalpy of fusion using differential scanning calorimetry

The energy changes discussed above can be measured by differential scanning calorimetry (DSC). In DSC, the power required to heat a sample in accordance with a user-defined temperature program is recorded, relative to an inert reference. The heating rate (β) can be linear or modulated by a mathematical function. When the sample melts, energy will be absorbed during the phase change and an endothermic peak will be seen, Figure 23.1. The enthalpy of fusion is equal to the area under the melting endotherm while the melting temperature may be determined either as an extrapolated onset (T_o) or the peak maximum (T_m).

Invariably, real solutions do not show ideal behaviour because the assumptions made earlier that ΔH_mix = 0 and that ΔH_f is independent of temperature are not always valid. A negative (exothermic) enthalpy of mixing increases solubility, while a positive (endothermic) enthalpy of mixing reduces solubility. Table 23.3 lists the experimentally measured solubilities for aspirin and paracetamol in a range of solvents. Note that solubility in water is by far the lowest of any of the solvents shown, while solubilities in tetrahydrofuran (THF) and methanol approach ideality in the case of aspirin, and exceed ideality in the case of paracetamol.

The reason that so many solvents, and water in particular, display such non-ideal behaviour is because of significant intermolecular bonding resulting from their chemical structure and properties. The three primary chemical properties are the dipole moment, dielectric constant and capacity for forming hydrogen bonds.

A molecule has a dipole when there is a localized net positive charge in one part of the molecule and a localized net negative charge in another. Such molecules are said to be polar. Water is an example of a polar molecule. Drugs that have dipoles or dipolar character are generally more soluble in polar solvents.

Dielectric properties are related to the capacity of a molecule to store a charge and are quantified by a dielectric constant. Polar solvents may induce a dipole in a dissolved solute, which will increase solubility. The dielectric constants of a number of commonly used pharmaceutical solvents are given in Table 23.4. It can be seen that water has a high dielectric constant (78.5) relative to that of methanol (31.5) even though both are considered to be polar solvents.

Hydrogen bonding occurs when electronegative atoms (such as oxygen) come into close proximity with hydrogen atoms; electrons are pulled towards the electronegative atom creating a reasonably strong force of interaction. A drug that has a functional group capable of hydrogen bonding with water (such as —OH, —NH or —SH) should have increased aqueous solubility.

Solubility as a function of temperature

Equation 23.6 indicates that the heat of fusion should be determinable by experimentally measuring the solubility of a drug at a number oftemperatures (since a plot of lnx₂ versus should be linear and have a slope of ). Since ΔH_fmust be positive, Equation 23.6 suggests that the solubility of a drug should increase with an increase in temperature. Generally, this agrees with everyday experience, but there are some drugs for which solubility decreases with increasing temperature. This is because an assumption was made in deriving Equation 23.6, namely that ΔH_f was equal to ΔH_sol. However, as noted above and as demonstrated by the data in Table 23.3, ΔH_mix is frequently not zero. In cases where ΔH_sol is negative (i.e. the heat of solution is exothermic), solubility will decrease with increasing temperature. These effects are shown in Figure 23.2.

Examples of data from three drug molecules plotted this way are given in Figure 23.3. While such plots are frequently found to be linear, they are usually plotted over a very narrow temperature range and the heat of fusion so calculated is rarely ideal, although it can be considered to be an approximate heat of solution.

Solubility and physical form

If molecules in the solid state are able to align in different patterns (the phenomenon of polymorphism, Chapter 8), then it is highly likely that the strength of the intermolecular bonds, and hence crystal lattice energy, will vary. Two polymorphs of the same drug will thus have different melting temperatures and heats of fusion. Usually the stable polymorph has the highest melting point and greatest heat of fusion and so, from Equation 23.6, the lowest solubility. Any metastable forms will, by definition, have lower melting points, lower enthalpies of fusion and so greater solubilities. The amorphous form, by virtue of not possessing a melting point, will have the greatest solubility. It is therefore clear that the solid-state structure of a new drug candidate should be determined during preformulation.

As discussed above, solubility is defined as the equilibrium between dissolved solute and the solid form. Thus, if a saturated solution is prepared by dissolving a metastable form and the excess solid is removed by filtration, the solution will be supersaturated with respect to the stable form. Ultimately, the stable form will precipitate as the system re-establishes a position of equilibrium (Fig. 23.4). Formulating any drug in a (solid) metastable form thus involves an element of risk; the risk being that the stable form will appear during storage or post-dissolution. In either case solubility will be reduced with, potentially, a consequential reduction in bioavailability.

Measurement of intrinsic solubility

Initially, during preformulation, solubility should be determined in 0.1 M HCl, 0.1 M NaOH and water. These ‘unsophisticated’ choices are determined by the scarcity of material at this stage. Saturated solutions can be prepared by adding an excess of solid to a small volume of solvent, agitating with time and then filtering.

Ultraviolet (UV) spectroscopy is the first choice assay, for reasons of familiarity, cost, the small volume of solution needed and the fact that the majority of drugs contain at least one functional group that absorbs in the UV region (190–390 nm). Table 23.5 lists the UV absorbance maxima for a series of common functional groups (called chromophores).

Excitation of the solute with the appropriate wavelength of light will reduce the amount of light passing through the solution. If the original light intensity is I₀ and the amount of light passing through the sample (the transmitted light) is I, then the amount of light absorbed will be a function of the concentration of the solute (C) and the depth of the solution through which the light is passing (the path length, l). This relationship is usually expressed as the Beer-Lambert equation:

(23.7)

where ε is a constant of proportionality called the ‘molar extinction coefficient’.

Higher values of ε mean greater UV absorbance by the solute. Values of ε for a range of functional groups are given in Table 23.5; it can be seen that groups containing large numbers of delocalized electrons, such as those containing benzene rings, have much greater ε values that groups containing simple carbon-carbon double bonds. The absorbance of a chromophore can be affected by the presence of an adjacent functional group if that group has unshared electrons (an auxochrome). A list of common auxochromes and their effects on the molar extinction coefficients of their parent benzene ring is given in Table 23.6.

Measurements of UV absorbance (thus solution concentration) should be recorded until the concentration remains constant and at a maximum. Care should be taken to ensure that the drug does not degrade during testing, if hydrolysis or photolysis are potential reaction pathways, and also that the temperature does not fluctuate. If the measured solubilities are the same in the three solvents then the drug does not have an ionizable group. If solubility is highest in acid then the molecule is a (weak) base and if solubility is highest in alkali the molecule is a (weak) acid.

Solubility should be measured at a (small) number of temperatures:

Note that possession of solubility data as a function of temperature allows (approximate) determination of the heat of fusion from Equation 23.6.

If the aim of the preformulation screen is to understand solubility in vivo, then solubility in bio-relevant media should be determined. Assuming oral delivery, typical media would include simulated gastric fluid (SGF), fed simulated intestinal fluid (FeSSIF) and/or fasted simulated intestinal fluid (FaSSIF). The use of these fluids is discussed further in Chapters 21 and 35 and details of their composition is given in Tables 35.2 and 35.3. Bio-relevant media tend to have higher ionic strengths and hence the risk of salting out via the common ion effect (see below) is greater.

Measurement of pK_a

Modern automated instrumentation is available that can determine pK_a values with very small (typically 10–20 mg) amounts of drug. This is extremely useful in the context of preformulation where material is scarce. Usually this instrumentation is based upon a potentiometric pH titration. The drug is dissolved in water, forming either a weakly acidic or weakly basic solution. Acid or base (as appropriate) is titrated and the solution pH recorded. A plot of volume of titrant solution added versus pH allows graphical determination of the pK_a, since when pH = pK_a the compound is 50% ionized. This method has the significant advantage of not requiring an assay.

Alternative methods for determining pK_a include conductivity, potentiometry and spectroscopy. However, if the intrinsic solubility has been determined, measurement of solubility at a pH where the compound is partially ionized will allow calculation of pK_a from the Henderson-Hasselbalch equations.

Determination of log P

Log P values can be determined experimentally or can be calculated from the chemical structure of the drug candidate using group additivity functions. For the latter approach there are numerous computer models and simulation methods available; selection will reduce to personal choice and familiarity and so these models will not be considered here. Rather, this text will focus on experimental determination. It is clear, however, that there is much value to be gained by comparing calculated and experimentally determined log P values. The value of the calculation approach is greatest when selecting a lead candidate from a compound library when it would simply be neither possible, nor practicable to measure the partitioning behaviour of the many thousands of compounds available.

Intrinsic dissolution rate

One assumption in the use of the Noyes and Whitney equation (described in Chapter 2, Eqns 2.3 and 2.4) is that the parameters of diffusion coefficient (D), the surface area of dissolving solid (A) and the thickness of the stationary solvent layer surrounding the dissolving solid (h) remain constant. Assuming a constant stirring speed and that the solution does not increase in viscosity as the solid dissolves, this is appropriate for D and h but A must always change as the solid dissolves (see Fig. 2.4). Also, if a tablet disintegrates, for instance, then A would increase rapidly at the start of dissolution before decreasing to zero, and there will be a concomitant effect on the dissolution rate.

If the sample is constructed such that A remains constant throughout dissolution, and sink conditions are maintained so that (S_t − C) ≅ S_t (see above), then the measured rate is called the intrinsic dissolution rate (IDR) (see also Chapter 2 and Eqn 2.6):

(23.25)

Wells (1988) suggests a method for measuring the IDR of a compound. A compact of the drug (300 mg) is prepared by compression (to 10 tonnes load) in an infrared punch and die set (13 mm diameter, corresponding to a surface area on the flat face of 1.33 cm²). The metal surfaces of the punch and die should be pre-lubricated with a solution of stearic acid in chloroform (5% w/v). The compact is adhered to the holder of the rotating basket apparatus using low-melting paraffin wax. The compact is repeatedly dipped into the wax so that all sides are coated except the lower flat face (from which any residual wax should be removed with a scalpel blade). Dissolution is recorded while the disc is rotated (100 rpm) 20 mm from the bottom of a flat-bottomed dissolution vessel containing dissolution medium (1 L at 37 °C). The gradient of the dissolution line divided by the surface area of the compact gives the IDR.

IDR as a function of pH

Measurement of IDR as a function of either pH or ionic strength can give a good insight into the mechanism of drug release and the improvement in performance of salt forms, since, for weak acids, substitution of Equation 23.14 into Equation 23.25 yields:

(23.26)

And for weak bases substitution of Equation 23.15 into Equation 23.25 yields:

(23.27)

In either case, the measured IDR will clearly be affected either by the pH of the medium or the microenvironment surrounding the solid surface created by the dissolving salt. The effect of pH on IDR is easily established by selection of the dissolution media. Standard media (0.1M HCl, phosphate buffers, etc.) can be used or, in order to get a more realistic insight into dissolution in vivo, simulated gastrointestinal fluids (as discussed earlier) can also be employed.

If the drug is an acid or base, then the self-buffering effect as dissolution occurs should not be ignored. In particular, the saturated concentration of solute in the diffusion layer often means that the pH in the medium immediately surrounding the dissolving solid differs significantly from that of the bulk solvent and will lead to deviations from the ideal behaviour predicted by Equations 23.26 and 23.27. A schematic representation of the buffering effect of salicylic acid is shown diagrammatically in Figure 23.11.

IDR and the common ion effect

The common ion effect (discussed in Chapter 2) should not be ignored, especially for hydrochloride salts, as the chloride ion is often present in reasonably high concentrations in body fluids (0.1 M in gastric fluid and 0.13 M in intestinal fluid). For this reason, fed and fasted simulated intestinal fluids should contain 0.1 and 0.2 M Cl⁻ respectively.

Hence, when the concentration of Cl⁻ in solution is high, the solubility advantage of choosing a hydrochloride salt is diminished. Li et al (2005) demonstrated the effect of chloride concentration on the IDR of haloperidol salts and showed that dissolution of the hydrochloride salt was slower than that of a either the phosphate or mesylate salt.

Salt formation

A salt is formed when an acid reacts with a base, resulting in an ionic species held together by ionic bonds. In principle, any weak acid or base can form a salt, although in practice if the pK_a of the base is very low, the salt formed is unlikely to be stable at physiological pHs. Stephenson et al (2011) note that no marketed salt exists for a drug with a pK_a below 4.6. They suggested that 5 is a general value below which salt formation is unlikely to be effective. Because salts usually dissociate rapidly upon dissolution into water, they are considered electrolytes. Sometimes a drug sounds from its name that it is a salt, but it may in fact be a single entity bound via covalent bonds, in which case electrolytic behaviour does not apply (e.g. fluticasone propionate).

Acids and bases can be classified as strong through to extremely weak, based on their pK_a (Table 23.8). When strong acids react with strong bases the reaction tends to completion, as both species will be fully ionized, and this is known as neutralization. For example:

(23.28)

In this instance, the salt formed will precipitate once it is present at a concentration beyond its solubility. However, most drug candidates are either weak acids or bases, in which case their character is usually based on the Brønsted-Lowry definition: an acidic compound is a proton donor and a basic compound is a proton acceptor. The removal of a proton from an acid produces a conjugate base (A⁻) and addition of a proton to an acceptor produces a conjugate acid (BH⁺).

(23.29)

(23.30)

Note that the Brønsted-Lowry definition requires acidic species to have an ionizable proton but does not require basic compounds to possess a hydroxide group; simply that they can accept a proton (thus the theory does not consider KOH and the like to be a base but a salt containing the basic OH⁻ moiety). In the case of a weak base (B) reacting with a strong acid, the conjugate acid and conjugate base may then form a salt:

(23.31)

When a salt dissolves in water it will dissociate. Assuming dissolution of a basic salt then the species in solution is the conjugate acid. The conjugate acid can donate its proton to water, reforming the free base:

(23.32)

All of the reasons for the change in solubility of salts are encompassed in Equation 23.32. A basic salt contains the conjugate acid of the drug. Upon dissolution, the conjugate acid donates its proton to water and the free base is formed. The solute is thus the free base, but the pH of the solution in which it is dissolved has reduced because of the donated proton. Recall that the solubility of weak bases increases as the pH of solution reduces. Thus, dissolution of a basic salt increases solubility because there is a concomitant reduction in pH of the solution.

The pH of a solution of a dissolved acid is given by:

(23.33)

and, because the acid species is BH⁺, then:

(23.34)

A similar situation occurs for the reaction of a weak acid with a strong base:

(23.35)

Upon dissolution of an acidic salt the conjugate base is formed:

(23.36)

The conjugate base can then accept a proton from water, re-forming the free acid and increasing the pH of the solution:

(23.37)

The solute is thus the free acid, but the pH of the solution in which it is dissolved has increased because of the generated hydroxide ion. Recall again that the solubility of weak acids increases as the pH of solution increases.

The pH of a solution of base is given by:

(23.38)

and since A⁻ is a conjugate base then:

(23.39)

where pK_w reflects the ionization potential of water (and is equal to 14 at 25 °C).

Several consequences arise from this discussion. One is that salt formation might not best be achieved in aqueous solution, since dissolution of a salt in water generally results in formation of the free acid or base. For this reason, salts are often formed in organic solvents. Secondly, the increase in solubility of a salt over the corresponding free acid or base is a result only of the change in pH upon dissolution. The intrinsic solubility of the free acid or base does not change. Thus, if salts are dissolved in buffered solvents there will be no difference in the solubility profile of the salt relative to the corresponding free acid or base, because the buffer will act to neutralize any change in pH.

Selection of a salt-forming acid or base

For salt formation to occur there must be a sufficient difference in pK_a between the acid and base (the reactivity potential). For the transfer of a proton from an acid to a weak base, the pK_a of the acid must be less than that of the weak base and vice versa. As a general rule, a difference in pK_a (ΔpK_a) of 3 is indicated (although salt formation can sometimes occur with smaller differences; for instance, Wells (1988) notes that doxylamine succinate forms even though ΔpK_a is 0.2). The reason for this pK_a difference is to ensure both species are ionized in solution, thus increasing the chance of interaction. If ΔpK_a lies between 3 and zero then knowledge of ΔpK_a per se is not predictive of whether salt formation will occur and if ΔpK_a is less than zero then co-crystal formation is the more likely outcome.

Thus, selection of a salt-forming entity starts with knowledge of the pK_a of the entity and the pK_a of the drug. The pK_a values of some of the most common salt forming acids and bases in water are given in Tables 23.9 and 23.10.

The top ten anions and cations by frequency for drugs in the 2006 USP are shown in Table 23.11. For basic drugs, the hydrochloride salt is the most common form. In part this is because the pK_a of hydrochloric acid is so low it is very likely that it will form a salt with a weak base. Hydrochloride salts are also widely understood and form physiologically common ions and so are acceptable from a regulatory perspective. However, they do have some disadvantages, including the fact that the drop in pH upon dissolution may be significant (which is not good for parenteral formulations). There are also risks of corrosion of manufacturing plant and equipment, instability during storage (especially if the salt is hygroscopic) and reduced dissolution and solubility in physiological fluids because of the common ion effect.

Stahl (2011) organizes salt-formers into three categories, which may be used as a guide to selection.

First class salt-formers are those that form physiologically ubiquitous ions or which occur as metabolites in biochemical pathways. These include hydrochloride and sodium salts and, as such, they are considered to be unrestricted in their use.

Second class salt formers are those that are not naturally occurring but which have found common application and have not shown significant toxicological or tolerability issues (such as the sulphonic acids, e.g. mesilates).

Third class salt formers are those that are used in special circumstances to solve a particular problem. They are not naturally occurring, nor are they in common use.

An additional factor to consider is that the salt formed should exist as a crystalline solid, to enable ease of isolation and purification. Amorphous salts are highly likely to cause problems in development and use and so should be avoided.

Salt screening

Once potential salt formers are selected they must be combined with the free drug in order to see which preferentially form salts. Since the potential number of permutations and combinations of salt formers and solvents is large, a convenient method for salt screening at the preformulation stage is to use a micro-plate well approach. A small amount of drug (~0.5 mg) in solvent is dispensed into each well of a 96-well plate. To each well is added a solution of potential counter-ion. It is possible to construct the experiment in the well plates so that the effect of solvent is examined in the x direction and the effect of counter-ion is examined in the y direction. Solvents should be selected carefully. Commonly used solvents are listed in Table 23.12.

After an appropriate length of time, the presence in each well of salt crystals is checked with an optical device (for instance a microscope or a nephelometer). If no crystals are seen, then the plate can be stored at a lower temperature. If the reduction in temperature does not cause precipitation, then as a last attempt the temperature can be increased to evaporate the solvent (although care must be taken in this case during subsequent analysis because the isolate may contain a simple mixture of drug and salt former, rather than the salt itself).

Once a potential salt has been identified preparation can be undertaken with slightly larger sample masses (10–50 mg). XRPD may be used to get a preliminary idea of polymorphic form, while melting points may be determined with a melting point apparatus, hot-stage microscopy (HSM) or DSC. Examination with HSM, if operated under cross-polarized filters, allows visual confirmation of melting and any other changes in physical form during heating, while analysis with DSC provides the heat of fusion in addition to the melting temperature (and so allows calculation of ideal solubility). Additional analyses with TGA and DVS will provide information on water content and hygroscopicity (see below). All of these experiments can be performed if around 50 mg of salt is available.

Solubility of salts

It is not a simple matter to predict the solubility of a salt. In particular the common ion effect cannot be ignored, especially when dissolution and solubility in biological media are considered. There are many empirical approaches in the literature for estimating solubility of salts, but most require knowledge of the melting point of the salt, a value most reliably determined by preparing the salt and melting it (in which case, the salt is available for solubility determination by experiment). This section will thus consider the underlying principle of solubility pH-dependence, based on ionic equilibria and assumes that solubility would be determined experimentally using the actual salt.

Dissolution of salts

Salts have the potential to increase dissolution rate because the saturated concentration in the boundary layer is much higher than that of the free acid or base

For acidic and basic drugs, solubility is pH-dependent. Accordingly, the Noyes-Whitney model predicts that dissolution rate must therefore also be pH-dependent, with the solubility of the solute at the pH and ionic strength of the dissolution medium being the rate-controlling parameter. From the same argument, when the pH of the dissolution medium is around that of pH_max the dissolution rates of the free acid or base and its salt should be the same (because their solubilities are roughly equal at this point). There are, however, numerous examples where this is not the case, for instance doxycycline hydrochloride and doxycycline; sodium salicylate and salicylic acid; and haloperidol mesylate and haloperidol.

These differences suggest that the pH of the solution into which the solid is dissolving (i.e. the boundary layer) is materially different from that of the bulk solvent (and so the solubility of the dissolving species is different from that expected in the bulk solvent). The difference in pH between the boundary layer and bulk solvent arises because the boundary layer is a saturated solution and because dissolution of acids, bases or salts will result in a change in pH; when saturated, the pH change is maximised. Nelson (1957) first noted this correlation during a study of the dissolution of various theophylline salts; salts with higher diffusion layer pH had greater in vitro dissolution rates and, importantly, faster in vivo absorption.

The pH of the boundary layer at the surface is termed the pH microenvironment (pH_menv) and is equal to the pH of a saturated solution of the dissolving solid in water. The Noyes-Whitney equation still governs the dissolution rate, but the solubility value is not that of the solute in the dissolution medium but that in a medium of pH_menv. As the distance from the surface of the dissolving solid increases, the pH approaches that of the bulk medium (shown earlier in Fig. 23.11).

Effect of salts on partitioning

Ionized species do not partition into organic solvents or non-polar environments. Thus, while solubility may be enhanced by formation of a salt, there is a considerable risk that partitioning will decrease (example data for partitioning of the sodium salt of ibuprofen are given in Table 23.13). There is thus a compromise to be reached between increasing solubility while maintaining bioavailability and it may well be the case that on this basis, the most soluble salt is not taken forward for development.

Polymorphism

When a compound can crystallize to more than one unit cell (i.e. the molecules in the unit cells are arranged in different patterns) it is said to be polymorphic (Chapter 8). The form with the highest melting temperature (and by definition the lowest volume) is called the stable polymorphic form and all other forms are metastable. Different polymorphs have different physicochemical properties, so it is important to select the best form for development. A defining characteristic of the stable form is that it is the only form that can be considered to be at a thermodynamic position of equilibrium (which means that over time all metastable forms will eventually convert to the stable form). It is tempting therefore to consider formulating only the stable polymorph of a drug, since this ensures there can be no change in polymorph upon storage. The stable form might, however, have the worst processability (for instance, the stable form I of paracetamol has poor compressibility, while the metastable form II has good compressibility), or lowest bioavailability (for instance, the presence of the B or C forms of chloramphenicol palmitate dramatically reduces bioavailability). Selection of polymorphic form is not necessarily straightforward although if the stable polymorph shows acceptable bioavailability then it is of course the best option for development.

Polymorphism screening

Polymorph screening at the preformulation stage is performed in much the same manner as described earlier for salt screening. Basic screening is achieved by crystallizing the drug candidate from a number of solvent or solvent mixtures of varying polarity. A small amount of drug (around 0.5 mg) is added into each well of a 96-well plate. To each well is added a small volume of each solvent or solvent mixture. After an appropriate length of time, the presence in each well of crystals is checked with an optical device (for instance a microscope or a nephelometer), using the strategies described previously for salt screening to facilitate crystallization.

X-ray powder diffraction (XRPD) provides structural data to identify and differentiate polymorphs. Figure 23.13 shows the powder diffractograms for two polymorphs of sulfapyridine; it is immediately apparent that each has a unique set of intensity peaks and so the forms are qualitatively different. The 2θ angles for each peak provide a ‘fingerprint’ for each form, while the intensities of each peak can be used as the basis for a quantitative assay for each form.

Differential scanning calorimetry (DSC) data differentiate polymorphs on the basis of their melting points and heats of fusion, thus providing thermodynamic information. This means DSC can identify which polymorph is stable and which are metastable. In addition, the heat of fusion can be used to calculate ideal solubility. Assuming there is only one polymorph present in a sample, and that it is the stable form, heating the sample in the DSC should result in a thermal curve showing only an endothermic melt, like that shown earlier in Figure 23.1. If the sample put into the DSC initially is a metastable form, then an alternate thermal curve is likely, Figure 23.14 (top curve). Here three events are seen; an endotherm followed by an exotherm followed by an endotherm. To what phase transitions can these events be assigned? The low temperature endotherm is easily assigned to melting of the metastable form. At a temperature immediately after the endotherm the sample is thus molten; but because the form that melted was metastable, and so at least one higher melting point form is available, the liquid is supercooled. With time, the liquid will crystallize to the next thermodynamically available solid form (in this case the stable polymorph). Crystallization is (usually) exothermic and so accounts for the exotherm on the DSC thermal curve. Finally, the stable form melts; the higher temperature endotherm. This pattern of transitions (endotherm-exotherm-endotherm) is a characteristic indicator of the presence of a metastable polymorph (indeed, if more than one metastable form is available, then an additional endotherm-exotherm sequence will be seen for each one). If the sample is cooled to room temperature and then reheated, often only the melting of the stable form in seen (Fig. 23.14, bottom curve). The combination of XRPD and DSC is very powerful and allows rapid assignment of polymorphic forms.

Amorphous materials

Several factors can make it difficult for molecules to orient themselves, in large numbers, into repeating arrays. One is if the molecular weight of the compound is very high (for example, if the active is a derivatized polymer or a biological material). Another factor is if the solid phase is formed very rapidly (say by quench-cooling or precipitation), wherein the molecules don’t have sufficient time to align. It is also possible to disrupt a pre-existing crystal structure with application of a localized force (for example, by milling). In any of these cases, the solid phase so produced cannot be characterized by a repeating unit cell arrangement and the matrix is termed amorphous (see also Chapter 8).

Because amorphous materials have no lattice energy and are essentially unstable (over time they will convert to a crystalline form) they usually have appreciably higher solubilities and faster dissolution rates than their crystalline equivalents, and so offer an alternative to salt selection as a strategy to improve the bioavailability of poorly soluble compounds.

Confirmation that a material is amorphous can be achieved with XRPD. In this case, no specific peaks as a function of diffraction angle should be seen; rather, a broad diffraction pattern, known as a ‘halo’, is the defining characteristic, as shown in Figure 23.15.

Particle size and shape

Particle shape is most easily determined by visual inspection with a microscope (some typical particle shapes are shown in Fig. 23.16). Usually a light microscope will suffice, unless the material is a spray-dried or micronized powder, in which case scanning electron microscopy (SEM) might be a better option. If the particles are not spherical but are irregularly shaped, it is difficult to define exactly which dimension should be used to define the particle size. Several semi-empirical measures have been proposed, e.g. Feret’s diameter and Martin’s diameter (see Chapter 9, Fig. 9.3 and associated text).

Powder flow

Powders must have good flow properties in order to fill tablet presses or capsule filling machines and to ensure blend uniformity when mixed with excipients. This is discussed in Chapter 12.

While poor powder flow will not hinder development of a dosage form it may prove a major challenge for commercial manufacture and so early assessment of powder flow allows time to ameliorate any problems. Assessing powder flow is easy when large volumes of material are available, but during preformulation methods must be used that require only small volumes of powder. The two most relevant methods of assessment at the preformulation stage involve the measurement of the angle of repose and measurement of bulk density. These measurements and their use in powder flow prediction are discussed in Chapter 12. The parameters of angle of repose (Table 12.1 and Table 12.2), Carr’s index (Eqn 12.14, Table 12.3) and Hausner ratio (Eqn 12.13, Table 12.3) (the latter two are both calculated from measurements of bulk density) have proved to be the most useful in predicting bulk properties when only a small amount of test material is available (illustrated in Fig. 23.17).

1. Accurately weigh three 500 mg aliquots of drug and 5 mg (~ 1% w/w) magnesium stearate as a lubricant.

2. Blend two samples (A and B) with lubricant for 5 minutes and the third (C) for 30 minutes by tumble mixing.

3. Load sample A into a 13 mm infrared compact punch and die set and compress quickly to 1 tonne, hold for 1 second and release. Eject the compact and store in a sealed container at room temperature overnight (to allow equilibration).

4. Repeat with sample B, but hold the load at 1 tonne for 30 seconds before releasing the pressure.

5. Compress sample C in precisely the same way as A.

6. After storing each compact, crush diametrically in a tablet crushing apparatus recording the crushing force.